PP

R
e
v
i
s
i
t
i
n
g

t
h
e

C
a
m
b
r
i
d
g
e

C
a
p
i
t
a
l

T
e
o
r
y

C
o
n
t
r
o
v
e
r
s
i
e
s
:

A

H
i
s
t
o
r
i
c
a
l

a
n
d

A
n
a
l
y
t
i
c
a
l

S
t
u
d
y
A
N
D
R
É
S

L
A
Z
Z
A
R
I
N
I
Edizioni dell’Università degli Studi di Pavia
€ 22,20
Texts published by Pavia University Press in the series “Editoria scientifca” have been peer-reviewed
prior to acceptance by the Editorial Board. www.paviauniversitypress.it/scientifca
Tis book deals with the Cambridge capital theory controversies both from a historical and
from an analytical standpoint.
In the frst part the author addresses the basic analytical and methodological issues underlying
the neoclassical theory, with special emphasis on how the problem of a measure of capital
arises within the marginalist theory of value and distribution. Te second part surveys the
most relevant contributions to the Cambridge debates in the light of the salient results of
‘reswitching’ and ‘reverse capital deepening’, and shows how the implications of these results,
which touch the principle of factor substitution, brought about diferent strategies pursued
by neoclassical scholars to overcome these theoretical problems.
Te book concludes that since the results of this debate touch the foundations of the theory,
and hence are of a general character, it is hard to accept that the contemporary versions of
the theory are free of capital problems, while at the same time the Cambridge controversies
are absent in the current literature used for training economics students. Te book should
therefore be useful to the undergraduate and graduate students, as well as to scholars devoted
to the History of Economic Analysis.

Andrés Lazzarini, Ph.D. in Economics from Roma Tre University, is a Visiting Fellow
at the Institute of Advanced Study – Jawaharlal Nehru University, New Delhi. He has
been Visiting Assistant Professor of Microeconomics at the University of Alicante,
Spain, in 2010, and Research Fellow and Lecturer of History of Economic Thought at
the Department of Political Economy and Quantitative Methods, University of Pavia,
in 2008 and 2009.
His main research areas are: History of Economic Analysis, Methodology of Economics
and Economic Development in Historical Perspective.
E-mail: andres.lazzarini@eco.unipv.it / alazzarini@gmail.com
Andrés Lazzarini
Revisiting the Cambridge Capital Theory
Controversies: A Historical and Analytical Study
Editoria scientifica
Pa v i a Un i v e r s i t y P r e s s
ISBN 978-88-96764-17-6























Editoria scientifica

























































ANDRÉS LAZZARINI














Revisiting the Cambridge Capital Theory
Controversies: A Historical and Analytical Study














Lazzarini, Andrés

Revisiting the Cambridge capital theory controversies: a
historical and analytical study / Andrés Lazzarini. – Pavia :
Pavia University Press, 2011. – VI, 172 p. : ill. ; 24 cm.
ISBN: 978-88-96764-17-6

1. Economia – Teorie – Gran Bretagna – Sec. 19. – 20.
330.1 CDD-22 – Economia, sistemi, scuole, teorie

© Andrés Lazzarini, 2011 – Pavia
ISBN: 978-88-96764-17-6


Texts published by Pavia University Press in the series “Editoria scientifica” have been peer-reviewed
prior to acceptance by the Editorial Board. www.paviauniversitypress.it/scientifica.


No part of this book may be reproduced or transmitted in any form or by any means, electronic
or mechanical, including photocopying, recording or by any information storage and retrieval
system, without written permission from the author.




















The cover shows: The Mathematical Bridge, Cambridge (UK), Margarita Olivera, 2007





Publisher: Pavia University Press – Edizioni dell’Università degli Studi di Pavia
Biblioteca Delle Scienze - Via Bassi, 6 – 27100 Pavia
<http://www.paviauniversitypress.it>

Printed by: Print Service – Strada Nuova, 67 – 27100 Pavia
Contents
Preface and Acknowledgements ................................................................................ V
Chapter 1. Why a study on the capital theory controversies?
1.1. Capital as a factor of production and the capital theory debates ............................. 1
1.2. The relevance of this conflict to contemporary economic analysis ......................... 3
1.3. A first phase in the capital controversies ................................................................ 5
1.4. A second phase in the capital controversies............................................................ 6
1.5. Radically different views on the meaning of the critique for the contemporary
versions of the theory ............................................................................................. 9
1.6. Sketch of this study .............................................................................................. 10
PART I. Analytical and Methodological Issues in Neoclassical Economics: Setting
the Ground to Understand the Capital Controversies
Chapter 2. Understanding the foundations of neoclassical economics: The
principle of factor substitution and the problem of capital
2.1. The data of the Marginal Theory. The direct mechanism of factor substitution.... 17
2.2. The indirect mechanism of factor substitution...................................................... 18
2.3. From ‘employment’ functions to ‘demand’ functions (I)...................................... 19
2.4. From ‘employment’ functions to ‘demand’ functions (II)..................................... 21
2.5. The capital supply function .................................................................................. 23
2.6. Key role of the downward-sloping shape of factors’ demand curves .................... 24
2.7. Problems connected with the heterogeneity of capital goods................................ 25
2.8. Supply-side role of capital .................................................................................... 27
2.9. Persistence character of the determining circumstances: The traditional method
of ‘normal positions’ ............................................................................................ 31
2.10. Demand-side role of capital: Conclusion............................................................ 34
Chapter 3. ‘Reswitching’ and ‘Reverse Capital Deepening’: Introduction to the
heart of the controversies and their implications
3.1. Some premises...................................................................................................... 39
3.2. Economies with heterogeneous capital goods. The wage curve............................ 42
3.3. The wage-frontier. Economies with more than one system of production............. 46
3.4. Implications for the basic premises of the theory: Conclusion.............................. 49
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

II
PART II. The Capital Theory Debates in a Historical Perspective and the
Transformation of Dominant Theory
Chapter 4. A first phase in the controversies: Samuelson’s 1962 intervention, the
1966 Symposium, neoclassical reactions and the highest point of the debates
4.1. Introduction.......................................................................................................... 55
4.2. Samuelson’s (1962) model with heterogeneous capital goods .............................. 57
4.3. Samuelson’s ‘special sub-class of realistic cases’................................................. 59
4.4. The neoclassical ‘parable’: ‘Surrogate’ capital and the ‘Surrogate’ production
function ................................................................................................................ 60
4.5. What did Samuelson actually demonstrate?.......................................................... 63
4.6. There is no heterogeneous capital goods model in Samuelson’s (1962) ............... 68
4.7. Levhari (1965) and the 1966 Quarterly Journal of Economics Symposium.......... 70
4.8. Pasinetti (1966a)................................................................................................... 73
4.9. Garegnani (1966).................................................................................................. 74
4.10. The meaning of the 1966 Symposium: A turning point in the controversies....... 75
4.11. Neoclassical reactions: Samuelson (1966a) ........................................................ 78
4.12. Further neoclassical reactions after the Symposium: Ferguson (1969) ............... 80
4.13. Garegnani’s 1970 critique of the marginal theory............................................... 85
4.14. Bliss’s 1970 intervention. Prelude to the shift to neo-walrasian equilibria and
the beginning of a second phase in the debates..................................................... 87
4.15. The zenith of the debates .................................................................................... 89
Chapter 5. A step back in the debates: Joan Robinson’s role in the controversies
and first neoclassical reactions
5.1. Introduction.......................................................................................................... 91
5.2. Robinson’s critique: Aggregate production functions as her main target .............. 91
5.3. Equilibrium and expectations ............................................................................... 93
5.4. Understanding Robinson’s way: The ‘Pseudo-production function’ and the
‘curiosum’ ............................................................................................................ 95
5.5. Conclusion ......................................................................................................... 101
5.6. Champernowne’s (1953) contribution ................................................................ 103
5.7. Further replies to Robinson: Solow (1955; 1956) and Swan (1956) ................... 105
5.8. Robinson rejoinders............................................................................................ 109
Chapter 6. A second phase in the capital controversies and radical changes in the
neoclassical theory
6.1. Introduction........................................................................................................ 111
6.2. Rehabilitation of Walras’s treatment of capital as physically heterogeneous
goods and Walras’s inconsistency ...................................................................... 112
6.3. Walras’s way-out to the inconsistency................................................................ 116
6.4. Implications of the Walrasian specification of capital ........................................ 118
6.5. Hahn’s (1975a; 1982) and Bliss’s (1975) views on the question of capital: A
first critical excursus .......................................................................................... 119
Contents

III
6.6. Uniform profit rates and intertemporal prices. Hahn’s ‘special case’ ................. 124
6.6.1. Introduction .............................................................................................. 124
6.6.2. Commodity own rates of interest .............................................................. 124
6.6.3. Uniform rates of return on supply prices in intertemporal equilibria ........ 125
6.6.4. Further discussion of Hahn (1975a) .......................................................... 127
6.6.5. Conclusion: some obscurity in the debates ............................................... 131
6.7. Do ‘reswitching’ and ‘reverse capital deepening’ bear theoretic problems on
intertemporal general equilibrium? Two views: Bliss and Garegnani ................. 132
6.7.1. Bliss.......................................................................................................... 132
6.7.2. Garegnani ................................................................................................. 134
6.7.3. Confrontations .......................................................................................... 136
Chapter 7. Conclusions ........................................................................................... 139
Appendixes
A. The curve of the net marginal product of labour ................................................... 143
B. Discussion of the ‘supply-side’ of capital in Wicksell’s general equilibrium
system................................................................................................................ 147
C. Sraffa’s landmark in the critique of the marginal theory....................................... 153
D. ‘Reverse capital deepening’ without ‘reswitching’ ............................................... 155
Bibliography ............................................................................................................ 157
Index of Figures....................................................................................................... 171
Preface and Acknowledgements
This book is the result of my research on the post-war Cambridge capital theory
controversies conducted at the Department of Economics, Roma Tre University, to obtain
my Ph.D. in Economics (Dottorato di Ricerca in Economia Politica), which I eventually
achieved in April 2008. Since then I have presented and discussed several works drawn
from my original dissertation which immensely helped to improve the present version for
publication at the Pavia University Press.
The so-called Cambridge controversy in the theory of capital took place between the
beginning of the 1950s and the mid-1970s, though arguably it got its heyday after the
publication of Sraffa’s 1960 book. That there existed a controversy between Cambridge
(UK) and Cambridge, Massachusetts (US), could hardly be ignored by any practitioner of
the discipline. Yet the recognition of its most relevant results is still lacking even though
they touch upon the core of the basic economic principles underlying the core of
mainstream neoclassical economics.
Certainly, after more than fifty years since the first exchanges in these debates many
articles and books have been written about this theoretical conflict in the economic
theory, and in fact the present book has dealt with most if not all of them in our
reconstruction. But, unlike many of the reviews available, our reconstruction attempted to
present an account which, first, does not follow a chronological survey of the
contributions to the Cambridge controversy but rather focuses on the analytical
arguments necessary to understand the implications of the phenomena of ‘reswitching’
and ‘reverse capital deepening’. In order to understand why these phenomena affect the
very core of neoclassical economics, we have thoroughly examined the ultimate reason of
their theoretical existence: the specification of capital in value terms among the data of
the theory in order to determine equilibrium variables through supply and demand forces
which ultimately rely on the factor substitution principle. These arguments, from a
chronological viewpoint, were visibly recognised in the last contributions of a first phase
of the debates. Secondly, we present and discuss some contributions of a later phase of
the controversy in which the treatment of capital among the data of the theory had already
been shifted to its Walrasian specification, but which have not been thoroughly dealt with
by previous surveys and which deserve due attention to better understand the present state
of the controversy between the neoclassical authors and the critical side.
Another issue examined in the present study, and which also deserves due attention
for understanding the evolution of the controversy, is the different – and sometimes we
may say radically different – positions not only between the two camps of the debate but
also within the critical side of the controversy, in particular during its later phase. Despite
the fact that the critically oriented (Cambridge, UK) side agreed that the main results of
the controversy had put the foundations of neoclassical economics at serious risks – and
hence the latter had to be replaced with an alternative economic theory – differences in
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

VI
the method among the critical side (and also the silence on the part of the marginalist
authors in the second phase as to accepting the controversy’s results) helped to prevent
that alternative from surfacing at that time.

I am indebted to many people who in one way or another helped me in the endeavour
to study capital theory and the ensuing controversies. First of all, I should like to
acknowledge Prof. Roberto Ciccone from Roma Tre University for his supervision, his
encouragement and his invaluable critiques and observations on my research; and Prof.
Pierangelo Garegnani for his commitment to science, his guidance at the initial stages of
my research and his enlightening criticisms.
Special gratitude goes to my partner, Margarita Olivera, who not only helped me to
complete and edit this book and shared her constructive comments on my work but also
supported me with her endless love during many years in our European life experience.
But I am also indebted to a number of colleagues and friends (from different corners
of the world) with whom I have been discussing over the last two years or so, and who no
doubt made many technical discussions here examined far more interesting. In particular
I would like to thank Professors Geoff Harcourt, Enrico Bellino, Gianni Vaggi, Amit
Bhaduri and Irina Eliseeva; and Denis Melnik, Ariel Dvoskin, Eduardo Crespo, Scott
Carter, Fabian Amico and Alejandro Fiorito.
I have also to thank all my colleagues from Roma Tre University; Stefano Finamore
for our long talks and travels in Italy; Juan José Pradelli for long discussions (in Rome
and elsewhere) on economic theory; and Frederic Lee who provided me with very
valuable material in the last phase of completion of this study. I have to acknowledge the
Alan Office (scholarship no: E04D040397AR) for providing me with the financial
support to undertake my Ph.D. studies in Roma Tre University from 2004 up to 2008. I
would also like to thank my family and friends from Argentina for their emotional
support over these years in my efforts to pursue the Dottorato di Ricerca in Rome.
This book would have never existed if I had not got the warm, friendly and endless
support and reception that my friends Luca Mantovan, Valeria Abbonato, Ariel
Wirkierman and Nadia Garbellini provided me in my many recent visits to Pavia.
Last but not least, I would like to thank Dr. Annalisa Doneda, Prof. Piero Mella
and an anonymous referee for the possibility to publish my research at the Pavia
University Press.
Of course, I am the only responsible for all remaining errors contained in this work.

Andrés Lazzarini
Pavia, January 2011
Chapter 1
Why a study on the capital theory controversies?
mainstream / 'meInstri:m / n [the+S] the main or most widely
accepted way of thinking or acting in relation to a subject
Longman Dictionary of Contemporary English (1991)

1.1. Capital as a factor of production and the capital theory debates
Beginning in the mid-1950s and for the following twenty years or so, a debate
concerning the neoclassical treatment of capital turned apparent in the discipline. This
gave rise to a series of exchanges between scholars associated with Cambridge,
England, and Cambridge, Massachusetts, (US). This debate is broadly known in the
literature as the ‘Cambridge capital theory controversies’.
1

The relevance of this controversy lies in that the criticisms of neoclassical theory
raised by Cambridge (UK) concern both the theoretical illegitimacy of measuring
‘capital’ as a single magnitude in value terms to determine prices and distribution, and
the foundational premise underlying the dominant supply and demand approach: the
factor substitution principle. In the controversies it has been shown that this principle
cannot in general be posited to explain the distribution of the social product in terms of
supply and demand. This result, discovered by means of the analysis of the relation
between prices and distribution in economies with heterogeneous capital goods, has
been revealed as theoretically irrefutable, and, as this study will argue, concerns the
hard core of the neoclassical or marginal theory
2
both in its traditional (capital
measured in value terms) and in its contemporary formulations (capital as a set of
heterogeneous goods).
Paradoxically, current economics has put up this historical debate in a hermetically
sealed compartment of its teaching programme. As can be seen by leafing through any
undergraduate or graduate economic theory textbook, neoclassical theory is the only

1
This label was christened in print by the first survey written by Harcourt (1969). See also n. 9,
this chapter, below.
2
In this book we shall refer to the neoclassical or marginal approach interchangeably – though the
author should like to emphasise that the label ‘neoclassical’, used in the literature to refer to a
theory that is essentially anti-classical, is a misfortune in the history of economic ideas.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

2
one theory being taught. At best, we are sometimes informed by the mainstream of
economics that the Cambridge debates would only be relevant to an ‘old’ or incomplete
version of the theory that treats capital in value terms but not to the contemporary
versions, which specify the factor capital as a set of heterogeneous goods. However, the
results drawn from the controversies are also relevant for a better understanding of the
present state of neoclassical capital theory, because, as we will argue, it has been due to
the implications of the debates that dominant theory shifted from the traditional to the
contemporary treatment of the factor capital.
Indeed controversy in capital theory was not unusual in the neoclassical school.
Neoclassical capital theory has been one of the most controversial areas of economic
science since its inception in the discipline in the last quarter of nineteenth century.
While former controversies arose out from different viewpoints on the supply side of
capital within the marginal approach, the Cambridge controversies, on the other hand,
entailed a criticism of the whole marginalist analysis.
3
At the time of the Cambridge
controversies, though, capital as factor of production was still conceived as a single
magnitude in value terms among the data of the theory. Although serious misgivings
regarding this notion of capital started to arise in the late 1930s and over the 1940s, still
the centre of economic theorising treated capital as a single magnitude in value terms.
4

Capital as a single magnitude has been the treatment on which, under conditions of
free competition, traditional theory had to resort to explain the distribution of the social
product between wages and profits (and rents) in terms of supply and demand – in other
words, by applying the factor substitution principle. As we shall see in the following
chapter, capital in value terms not only is needed in order to allow for factor
substitutability, but also to determine a uniform rate of profits on the supply prices of
the several capital goods. The condition of a uniform rate of profits on the capitals’
supply prices was as universally accepted a condition by the founders of marginal
analysis as it was by the classical political economists since Adam Smith. Satisfaction

3
See Garegnani (1970a), pp. 424-427; and Chapter 4 below. For the different controversies on
capital theory within the neoclassical approach in the twentieth century see Kurz (1987), Huth
(1998), and Kaldor (1937). According to Kurz (1987) the two former controversies within the
neoclassical theory involved J.B. Clark and E. Böhm-Bawerk at the turn of the twentieth century,
while the second one, in the 30s, mainly involved Knight, Hayek, and Kaldor.
4
Although this study has not systematically dealt with the general situation of neoclassical capital
theory before the Cambridge controversies, we shall assume, by relying on Garegnani (1976a;
1990) and Petri (2004, section 5.4), that the notion of capital shared by the centre of economic
theorising was, in the main, the traditional quantity of capital, in spite of the increasing
reservations arisen in the period referred to as to the legitimacy of this notion. (See e.g. Malinvaud
1953). It is interesting to recall that Knight’s entry “Capital and Interest” in the Encyclopaedia
Britannica, which considers capital as a single magnitude expressed in value terms, was reprinted
in 1950 by the American Economic Association (see Knight 1950), thus revealing that until those
years capital as a single magnitude appeared to be the central notion. Cf. Chapter 2, below; Dewey
(1965) Chapter 8, in particular p. 114, and Robertson (1957). Problems associated with the
treatment of capital in value terms started to be recognised by neoclassical economists, such as
Hicks (1946), who argues to dispense with the traditional notion of capital altogether (see
Garegnani 1976a).
Chapt er 1 – Why a st udy on t he capi t al t heory cont roversi es?
3
of this condition ensures that the equilibrium determined by neoclassical supply and
demand forces could conceivably be taken as a long period equilibrium in that the
persistent nature of the conditions of the economy can guarantee that the theoretical
variables become gravitational centres of the actual ones (trends or average) over time.
Thus, that treatment of capital
5
allows determining a normal position of the economy’s
quantities and prices characterised by a uniform rate of profits on the capitals’ supply
price under free competition both in consumption-goods’ and factors’ markets. For that
determination it is crucial that the theory accounts for downward-sloping demand
curves for factors. As we shall see in this study, the Cambridge controversies have
shown that the belief on an inverse relation between demand for factors and their
rewards cannot in general be postulated.
1.2. The relevance of this conflict to contemporary economic analysis
The importance of this controversy for the discipline is indubitable. To the vast amount
of contributions which were published in that period by many scholarly journals on
economic theory
6
– not to mention the oral debates
7
– we have to add the ensuing
reviews and surveys from both sides assessing the significance of this remarkable
episode.
8
Hardly could any practitioner of the discipline deny the fact that over more
than twenty-five years the controversy in the theory of capital occupied a prominent
magnitude in contemporary history of economic analysis.
9
However, after more than

5
This treatment of the capital as a single magnitude in value terms is the way in which capital was
traditionally conceived as a factor of production, and perhaps it might have been seen by some
scholars as an aggregate factor. But it will be seen below that this conception has nothing to do
with the use of ‘aggregate’ production functions for the determination of distribution. Capital in
value terms as traditionally conceived is analysed in Chapter 2. Endless misunderstandings in the
controversies were created as to whether the critique only bears on versions of the theory using
aggregates production functions. On this issue cf. Chapter 4, below.
6
Mainly in The Quarterly Journal of Economics, The Review of Economic Studies, The Journal of
Economic Literature and The Economic Journal. See bibliography below.
7
For instance the oral debates between Paul A. Samuelson and Pierangelo Garegnani in 1961-62
in Cambridge, Massachusetts (US) during a research visiting period of the latter. This oral
exchange later gave rise to Garegnani (1970a). See Chapter 4 below and also Samuelson (1962, p.
193, n. 1) and Robinson (1970a, p. 311).
8
See Harcourt (1969; 1972; 1976), Bhaduri (1969), Robinson (1970a), Garegnani (1976a) for
assessments from the critically oriented sides of the debates and Blaug (1974), Hicks (1974),
Stiglitz (1974), Ng (1974) for neoclassical surveys. More recent assessments from the critical side
also include Cohen and Harcourt (2003; 2005), Garegnani (1990; 2003 “Appendix II”), Petri
(2004, Chapter 6; 2007), while for the neoclassical side we should refer to Hahn (1982),
Burmeister (1980, Chapter 4; 2000), Bliss (2005). See also Bliss (1975, Chapter 15), Hahn (1972;
1975a), Birner (2002) and the entries provided by Kurz (1987), and Ahmad (1998).
9
Though the Cambridge-Cambridge divide is not strictly accurate to represent the division of the
two different approaches in the history of economic analysis behind these controversies, still it is
acknowledged in the literature that the American group may be regarded as the neoclassical side of
this debate, while the British one as the non-neoclassical or the critical side. This is so because,
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

4
fifty years since the first exchanges between Cambridge (UK) and Cambridge (US),
10

we might perhaps be asked by a marginalist scholar something like “Why to rekindle
this debate to-day?” Well-trained in mainstream ambiences he would remind us that the
completely developed theory of capital is the one which treats ‘capital’ as a set of
physically heterogeneous goods.
According to this view, the Cambridge debates would have petered out once
neoclassical theory started to rely on a physical specification of capital (‘Walrasian’
treatment)
11
thus rendering, apparently, the supply and demand approach immune from
capital theoretic problems. Thus, neo-Walrasian authors argue that the debates would
have no relevance to the mainstream of economic thought; in fact neoclassical scholars
deem the theory of supply and demand relying on a Walrasian capital as the one and
only theory, without even showing the slightest effort to inform students of this
episode
12
that is the capital theory controversies, which, as we shall see in this study,
have serious implications on the contemporary versions of the theory. But, on the other
hand, we can see today, for instance in some recent neoclassical growth literature like
Real Business Cycle and Endogenous Growth theories, that ‘aggregate production
functions’ are still used by the bulk of mainstream economists working in growth and
aggregate investment issues. This reveals that a part of the discipline
13
accepts the
supply and demand approach treating capital as a single magnitude as if nothing had


broadly speaking, the English school in the 1950s, 1960s and even 1970s hosted some critics of
orthodoxy, chiefly A. Bhaduri, Krishna Bharadwaj, P. Garegnani, G.C. Harcourt, Nicholas Kaldor,
Luigi L. Pasinetti, Joan Robinson, Piero Sraffa. On the other hand, key neoclassical participants,
like Paul A. Samuelson and Robert M. Solow (this one still is) were affiliated to the MIT (US).
Both Christopher J. Bliss and Frank H. Hahn, who became the most active neoclassical
participants during the 1970s, were affiliated to Cambridge, UK (in Bliss’s case at least part of the
time). There have of course been many other contributors that worked elsewhere at that time but
were associated, in terms of theoretical membership, to either the neoclassical or the critical side.
We keep to the geographical divide suggested by the literature in this study. Also, it is worth
noticing that the label ‘neo-neoclassical’ was sometimes used by some participants (especially J.
Robinson 1962b; 1971a) to refer to the fourth generation of neoclassical economists that not only
includes the ones referred to above but also K. Arrow, C. Bliss, C. Ferguson, F. Hahn, D. Levhari,
J. Stiglitz, among others. In this work we shall refer to critics as the ‘critically oriented side’ or,
more simply, the ‘critical side’. Note also that some literature refer to the critically oriented side as
the “neo-Keynesians”, Harcourt (1972); “neo-Marxist-Keynesians”, Ng (1974); “Post-
Keynesians”, King (2002); or “Anglo-Italian critics”, Bliss (2005).
10
There is almost complete agreement that Joan Robinson’s (1953) article kicked off the debates.
See Chapter 5 below.
11
It has been preferred to use the term ‘Walrasian’ treatment of capital because Walras, unlike his
contemporary peers, treated ‘capital’ as a collection of physically heterogeneous capital goods, and
not capital as a single magnitude; cf. Chapter 2 and n. 100, Chapter 4 below.
12
Cf. Hahn’s (1975a, p. 364) global assessment of the debates: “Great charade”. See Chapter 6,
below, where we deal with this author’s view of the controversy. A view similar to Hahn can be
found in Sen (1974).
13
See Felipe and Fisher (2003) who have drawn attention to why economists still use aggregate
functions in theoretic and empirical research despite the capital controversies. Cf. Barro and Sala-
i-Martin (1995). See Chapter 5 below.
Chapt er 1 – Why a st udy on t he capi t al t heory cont roversi es?
5
ever occurred in the recent history of economic analysis. So a very important question
arises: What happened to the Cambridge debate that today, it seems, nobody talks
about it any longer?
1.3. A first phase in the capital controversies
The answer that will be attempted here suggests considering a breakdown of the period
of the debates into two phases. The first phase (1953-1970/1) regards the period when
salient discoveries like ‘reswitching’ and ‘reverse capital deepening’ started to gain
notoriety in the discipline,
14
after the first exchanges between the two Cambridge much
centred on the use of aggregate production functions in economic growth models. The
second phase (1970/1-1976)
15
regards a situation which might be characterised, for
reasons we shall analyse below, as an obscure, unclear chapter in the history of the
debates where many misunderstandings damaged communication between the two
parties. In this second phase neoclassical capital theory had already shifted from the
traditional treatment of capital in value terms to the Walrasian specification of it and
this helps explain why some of the exchanges in this phase were at cross-purposes.
From a historical standpoint, our identification of two phases in the controversies goes
along with a breakdown in the development of the debates as has been documented by
the literature of the critical side of this theoretical conflict.
Regarding the first phase it has been natural for the present author to take into
account one of the most frequent references to the Cambridge controversy in the
literature, namely, Harcourt’s 1972 survey.
16
This work has spread the controversies in
the discipline and is the most frequently cited work by the bulk of practitioners when
referring to the first phase of the debates. In Harcourt (1972) we find an all-inclusive
review of the main contributions to the debates from both sides, which, starting from
Joan Robinson’s 1953 kick-off, follows a chronological account of the different
contributions to the controversies.
17
Harcourt (1972), following Joan Robinson’s target

14
See Chapters 3 and 4 below.
15
As will be seen later (Paragraph 4.14, Chapter 4 and Chapter 6), we have considered 1970/1 as
the end of the first phase because in 1970 Bliss (1970) replied to Garegnani (1970a) on the
grounds of a different notion of equilibrium and thus misunderstanding Garegnani’s criticisms of
the theory at stake which was, and is, the traditional long period supply and demand equilibrium
theory. As to 1971, it is due to Robinson’s (1971b) suggesting title “The End of the controversy”.
On the other hand 1976 has been chosen as the ending date of the second period for reasons that
are described in the text.
16
Harcourt’s (1972) book is an extended version of his original survey article Harcourt (1969).
17
Beginning with Joan Robinson’s kick-off 1953 article criticising the use of the aggregate
production functions, Harcourt’s account continues by reviewing later replies that Robinson’s
article stirred up. His survey focuses on issues regarding aggregates production functions, vintage
models, technical progress and economic growth models, the debate on the rate of return involving
Solow and Pasinetti; then it continues by presenting the ‘reswitching’ debate, and concludes by
summarising the Cambridge alternative distribution theory as represented by the works of Kaldor
(1955), Robinson (1962a) and Pasinetti (1962). Though we shall deal with Robinson and her
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

6
of criticisms, focuses on the so-called ‘aggregate production functions’ rather than on
the problems for the theory owing to the treatment of capital in value terms it needs rely
on.
18
Moreover, from the criticisms centred on the aggregate production function, along
with Robinson, Harcourt draws methodological implications in that both these authors
argue that equilibrium comparisons are not good tools for an analysis of changes.
19
But,
as we shall see later, phenomena like ‘reswitching’ and ‘reverse capital deepening’ cast
serious doubt on the robustness of theory itself independently of whether it uses
aggregate functions or not, thus bringing about important reactions from the
neoclassical side that tried to save the basic principles underlying that theory.
Thus, even if Harcourt points out that ‘double switching’ and ‘capital reversal’
damage the foundations of the marginal approach, it is yet not clear, in Harcourt (1972),
the consequences, if any, of these theoretic results on the versions of the theory relying
on a Walrasian treatment of capital. Then, when neoclassical writers started to argue
against the critics by contending that a “general equilibrium model of capital does not
require aggregation”,
20
counteractions and replies from the critical side were not
unified. It is the opinion of the present author that when these arguments started to be
raised by neoclassical participants in the debates, a second phase in the controversies
started being recognised both by direct participants and commentators. Indeed our
contention can be verified by some different views in this regard in some relevant
literature of the critical side.
1.4. A second phase in the capital controversies
If we now turn to Harcourt’s updated account of the Cambridge controversy in 1976 we
shall notice a very different opinion regarding whether the results of ‘reswitching’ and
‘reverse capital deepening’ drawn from the first phase of the controversies bear on the


immediate contestants in Chapter 5 on issues related to the debates on capital theory, the so-called
Cambridge theory of distribution is not considered by us here.
18
Harcourt’s concern about the aggregate production function is particularly clear when Harcourt
(1972, p. 3) states that the aggregate production function “is indeed by far the most common
version” of the neoclassical theory. But, as we shall see later, the problem of capital does not
uniquely entail using aggregate production functions. For instance, Wicksell’s system does not use
aggregate production function and yet has to resort to a quantity of capital in value terms. See our
discussion of Wicksell’s general equilibrium system in Appendix B, below.
19
See Chapter 5, below where we deal with Robinson’s critique.
20
Bliss (2005, p. xii). This is the position that Bliss also maintains in his (1975). We shall argue
later (Chapter 6) whether Bliss’s contention is due to a supposedly ‘more complete’ analysis which
takes as data the many capital goods there are in the economy (i.e. a Walrasian treatment of
capital) or to another reason. Note also that neoclassical authors have almost always taken
Robinson’s target of criticisms at aggregate production function in order to contend the critics. On
this, see also Chapter 6.
Chapt er 1 – Why a st udy on t he capi t al t heory cont roversi es?
7
versions which rely on a Walrasian treatment of capital. Thus, we can find in Harcourt
(1976) a quite different assessment of the relevance of the debates:

We now conclude: if the arguments of this article are accepted, while general
equilibrium may emerge logically intact as ‘the theory of inter-temporarily
efficient paths and their price duals’ [Hahn 1972, p. 3] it is not the theory which is
relevant for the issues raised in the Cambridge controversies. The attempts to use
the other versions of neoclassical theory flounder both on the results of the
reswitching and capital-reversing debates, with which is allied the problem of
measurement of ‘capital’, and the (more fundamental) criticisms that stem from
the distinction between comparisons and changes.
21


This different appreciation of the results of the debates on the ‘intertemporal
general equilibrium’ theory as viewed by Harcourt (1976)
22
has induced the present
author to delve into this second part of the controversies and place at the centre another
view on the relevance of the critique from the critically oriented side.
In fact, in Garegnani (1976a) we can find that his appraisal of the controversies is
quite different from Harcourt’s comments. Thus, while Harcourt (1976) suggests that
the neo-Walrasian theory “may emerge logically intact” from capital theoretic
problems, Garegnani argues that after Hicks’ (1946 [1939]) Value and Capital an
increase proportion of practitioners persuaded in the neoclassical approach started to
shift to a “short period general equilibrium.”
23
What is important for our purposes to
notice here is that the consequent abandonment of the long period equilibrium method

has not been due to weaknesses of the method as such, but rather to weaknesses of
the dominant theory of distribution, and in particular, of the conception of capital
it relies on.
24


21
Harcourt (1976, p. 58, emphasis in the original, underlines added). Harcourt splits the theory in
two versions: the aggregate versions (“the other versions of neoclassical theory” in the passage)
and the ‘general equilibrium’ one. Harcourt means the latter to be the neo-Walrasian approach of
simultaneous equations. This breakdown in different versions between an ‘aggregate’ and ‘general
equilibrium’ one is misleading since it does not take care of Wicksell’s system which is a general
equilibrium, is disaggregated, and, however, needs a quantity of capital in value terms for the
determination of equilibrium (not an ‘aggregate production function’ which is what Harcourt
seems to suggest in the passage). On the meaning of an intertemporal general equilibrium, see
Chapter 6.
22
Similar positions can be found in Cohen and Harcourt (2003, p. 207; 2005, p. xxxiv). See also
the protestations raised against Cohen and Harcourt (2003) by Pasinetti et. al. (2003); and see
Bellino (2005), Reati (2006) and Petri (2007) for comments on Cohen and Harcourt (2005). Cf. n.
66, Chapter 4, below.
23
Garegnani (1976a, p. 26).
24
Garegnani (1976a, p. 26).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

8

According to this view, due to the Walrasian specification of capital the marginalist
theory was forced to adopt a short period general equilibrium method as a last refuge
from the results of the controversies regarding the treatment of capital which the long
period equilibrium neoclassical theory need rely on. Although in Garegnani (1976a)
“the question of whether the two kinds of analysis [intertemporal and temporary
equilibria] have in fact overcome the deficiencies of demand-and-supply theory with
respect to capital [is left] aside”,
25
once we take into account that a radical shift in
method undergone by the theory has taken place, then it is hard to accept that the new
versions of the theory may emerge logically intact, as Harcourt (1976) seems to
conclude.
Thus, whatever the different reasons, there does not seem to be agreement in the
critical literature on whether the results of the Cambridge debates would have forced
the theory – in its attempts to preserve the supply and demand approach and to
overcome the deficiencies of capital – to adopt a method which is radically different
from the traditional one and hence whether the debates bear on the new versions of the
theory.
26
But, it is interesting to note that if the controversies compelled the theory to
adopt a method of short period general equilibrium,
27
which lacks the necessary
persistency of the equilibrium
28
and was in the main confined to a minority group of
mathematical economists, then that shift would be representing the very strength of the
critique. Thus, I submit, it is not the case that Walrasian theory is not the version of the
theory which is relevant for the issues raised in the Cambridge debates; the shift to
Walrasian theory, with all its methodological implications, is the result of the capital
theory debates.
These different views also point to a gap within the critical side with respect to
whether ‘reswitching’ and ‘reverse capital deepening’ are the “more fundamental”
criticisms in the debates.
29
Indeed Harcourt (along with Cohen) have recently argued
that the shift from ‘aggregate’ versions to a theory relying on a Walrasian capital would

25
Garegnani (1976a, p. 38). Garegnani (1990; 2000; 2003) has taken up this issue and has shown
that the same kind of difficulties regarding the conception of capital in the traditional versions of
the theory relying on a notion of capital in value terms also reappears in the new versions in the
investment and savings markets. Garegnani’s arguments are briefly dealt with in Chapter 6 below.
26
This apparent halfway situation of the controversies is also recorded by Backhouse (1985, p.
328) who says: “the implications of reswitching for disaggregated neoclassical models [of the
Arrow-Debreu type] were still not clear.” Note that ‘disaggregated neoclassical models’ are meant
as the neo-Walrasian equilibrium theory.
27
Cf. Garegnani (1976a; 1990) and Milgate (1979).
28
Among other problems of the theory which relies on a Walrasian treatment of capital. See
Chapter 6 below.
29
Harcourt (1976, p. 58). It is apparent that in this work Harcourt has given more importance to
the “criticisms that stem from the distinction between comparisons and changes.” Echoing
Robinson’s concerns, it is then no surprise that Harcourt has regarded the ‘reswitching’ and
‘reverse capital deepening’ debates as a “Dead End” (id.); cf. in particular Robinson (1975a).
Chapt er 1 – Why a st udy on t he capi t al t heory cont roversi es?
9
have represented a shift to ‘more disaggregated analysis.’
30
They thus neglect other
critics’ views such as Garegnani’s (1976a), who argues that the key to understand the
shift to a Walrasian specification of capital is to realise that the theory had to adopt that
way – which had already been provided by Hicks’ (1946)
31
– due to the deficiency of
the traditional notion of capital as revealed by the implications of the results of the
debates.
32

One of the aims of this book is to clearly show how the Cambridge capital theory
controversies endanger the specification of capital as a single magnitude on which the
traditional marginal theory relies, in order to contribute for a better understanding of the
results of the capital debates and their implications for the neoclassical theory
nowadays, which relies on a Walrasian treatment of the factor capital.
1.5. Radically different views on the meaning of the critique for the contemporary
versions of the theory
At the turn of the 1970s, neoclassical authors started to oppose the critique by
contending that capital theory need not have recourse to any aggregation,
33
thus
suggesting that the contemporary versions of the supply and demand approach would
not be subject to theoretic capital problems as in the traditional versions (capital in
value); at the same time, however, the critical side did not immediately provide a
unified rejoinder to the above neoclassical contention.
In the present author’s opinion, neoclassical economists took advantage of the non-
unified position with respect to the implications of ‘reswitching’ and ‘reverse capital
deepening’ on the new versions of the theory and that these neoclassical authors, by
disseminating misleading ways to treat the issues,
34
have put the debates in a quite
obscure chapter of contemporary history of economic analysis, which suggests

30
This idea is implicitly suggested by Cohen and Harcourt (2005, p. xxii). But this view seems not
to have considered that e.g. Wicksell’s analysis is as disaggregated as were analyses conducted by
Böhm-Bawerk (1891), or by Hicks (1932), thus broadcasting a misleading idea, as Petri (2007, p.
604) has recently remarked, that the sole ‘disaggregated’ analysis would appear to be the new
versions of neo-Walrasian theory. In addition, Cohen and Harcourt (2005 p. xxiv) suggest that the
“characterisation of simultaneous equations [in disaggregated analysis of neo-Walrasian theory] is
correct”, without however discussing the economic meaning underlying those formal systems. In
Chapter 6 below, we shall discuss the economic meaning of Walras’s system of capitalisation.
31
In this book we have considered Hicks (1946), the 2
nd
ed. of Hicks’ (1939) Value and Capital.
32
See Chapter 6 below. Garegnani’s (1976a) insights into the issue at stake, that is, how far
neoclassical theory relying on a Walrasian capital is immune from capital theoretic problems, gave
rise to a series of contributions attempting to clarify some issues of the controversies in capital
theory in its second phase. See: Petri (1978; 2004); Milgate (1979); Eatwell (1990).
33
This is the position shared by Hahn (1972; 1974; 1975a; 1975b; 1982); Bliss (1970; 1975;
2005); Dixit (1977).
34
For instance, the misleading way in which neoclassical economists received the critique was as
if it concerned the so-called aggregate production functions. Cf. the works cited in n. 33, this
chapter, above, and see also Chapter 6, below.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

10
regarding this period as the second phase of the Cambridge controversies. As will be
seen later,
35
this obscurity in the debates, I submit, were brought about by marginalist
authors in their attempt to reverse the sign of the critique by introducing misleading
interpretations of previous results (first phase) and reinforcing further
misunderstandings that “have marred the controversies on capital.”
36
Thus, misleading
ways used by some participants in the second phase of the debates have fogged
communication and might have therefore detained a full appreciation of the strength of
the implications of the controversies that touch the general principles of dominant
theory.
In order to understand the main issues debated over the capital theory debates,
misunderstandings included, the method adopted in this study embraces, on the one
hand, a historical standpoint, in that we deal with historical aspects of this conflict in
the discipline. However, on the other hand, we have not chosen to proceed in a
chronological order: we have rather preferred to place at the centre the analytical
arguments necessary to understand the implications of the phenomena of ‘reswitching’
and ‘reverse capital deepening’ which, from a chronological viewpoint, were visibly
recognised in the last contributions of the first phase of the debates (particularly after
Sraffa’s 1960 book). And this methodological decision has been made by the present
author for a very important purpose: to rescue the true nature of the critique produced in
the controversies in order to contribute to a better understanding of the current situation
of neoclassical capital theory.
1.6. Sketch of this study
This study is divided in two parts: while the first part is devoted to introduce the
analytical and methodological issues in neoclassical economics, the second part
analyses and discusses the historical aspects of the Cambridge controversies in the light
of the analytical elements provided in the first part.
Since we will argue that the capital controversies touch the basic premises of the
neoclassical theory, essential notions of the marginal approach are introduced in the
first part of this study. Thus, in the next chapter (Chapter 2), one of the first key notions
in neoclassical theory examined is the factor substitution principle, which is explained
in simple terms for the sake of exposition (corn-consumption, corn-capital and labour
economy). Following the description method developed in Garegnani (1990), the
chapter aims at providing the basic analytical elements to understand the importance of

35
See Chapter 6, below.
36
Garegnani (2005), p. 411. Of the many misunderstandings created along this controversy, the
one that confuses the neoclassical treatment of capital as a single magnitude in value terms with
the illegitimacy of using aggregate production functions is arguably the most dangerous when
assessing the importance of this theoretical clash. Of course, many other misunderstanding (e.g.
the confusion between steady state equilibrium and long period equilibrium) were due to historical
aspects which are not completely covered by this study.
Chapt er 1 – Why a st udy on t he capi t al t heory cont roversi es?
11
the factor substitution principle for the neoclassical approach so as to grasp the notion
of a downward-sloping demand curves for ‘factors of production’ which is key to
derive a plausible explanation of distribution in terms of supply and demand (Chapter
2, Paragraphs 2.1-2.6). Then, by surveying some key marginalist authors’ views on the
matter, we turn to economies with heterogeneous capital goods and the treatment of
capital measured in value terms is examined. In fact in economies with heterogeneous
capital goods a problem of measurement of capital arises. In this connection we shall
face that problem from two angles: we distinguish a supply-side or supply-aspect of the
problem – which is related to the inescapable need to treat ‘capital’ as a factor in value
terms – from a demand-side or demand-aspect of the problem, which is related to the
possibility to conceive of that factor’s demand such that owing to changes both in the
methods of production and in output the quantity of that factor demanded could be
conceived to be inversely related to its rate of remuneration. The chapter also
introduces the meaning of the persistent circumstances underlying the method of the
long period equilibrium, that is, the equilibrium notion that was at the heart of the
controversies. (Chapter 2, Paragraphs 2.7-2.10).
Once the basic notions of the theory are laid down, Chapter 3 introduces the
meaning of the most relevant results of the Cambridge controversies: ‘reswitching’ and
‘reverse capital deepening’. The purpose of this chapter is to introduce the reader to the
concepts and, more importantly, the implications of these phenomena which were at the
centre of the most salient exchanges between the two sides.
In the second part of this study – which consists of three Chapters (4, 5, and 6) –
we turn our attention to the salient historical aspects of the capital controversy. While
Chapters 4 and 5 deal with the themes, issues and episodes developed in the first phase
of the controversies, Chapter 6 considers the questions and problems of the second
phase of the debates. The present author should like to emphasise that Chapters 4 and 5
do not follow a chronological order. Rather, we have preferred to firstly analyse and
discuss the group of contributions that provide the essential nature of the criticisms, and
the neoclassical replies that the latter gave rise to. We thus begin with (Chapter 4) some
of the most important contributions in these debates by surveying Samuelson’s (1962)
intervention and the limits of his particular defence of the neoclassical theory. The
chapter then goes on to analyse the attempts made by Samuelson’s Ph.D. student
Levhari (1965), who tried to find out the conditions to rule out ‘reswitching’. These
interventions were very important from a historical perspective of the controversies,
because they subsequently gave rise to a famous Symposium on Paradoxes in Capital
Theory held by The Quarterly Journal of Economics in 1966. Thus, the importance in
the development of this conflict is analysed by bringing to the fore the interventions
raised in that Symposium by Pasinetti (1966a) and Garegnani (1966). Yet, as we shall
argue, neoclassical scholars, e.g. Ferguson (1969), though recognising theoretic
difficulties with respect to capital for the whole approach, attempted to minimise or
downplay the relevance of the results produced in the capital debates by invoking the
‘empirical’ way-out. Moreover, it is worth recalling an important question in the
aftermath of the Symposium: while the heart of the controversies have referred to the
traditional, long period equilibrium versions of the neoclassical theory, we shall see that
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

12
some reactions on the part of some neoclassical scholars actually relied on a short
period general equilibrium; that is why the present author has deemed appropriate to
briefly deal with Bliss’s first intervention in 1970 against Garegnani (1970a), in order
to point out the way adopted at that moment by neoclassical authors to overcome the
problems concerning the treatment of capital.
The following chapter (Chapter 5) turns back in our historical reconstruction of the
controversies: we survey Robinson’s (1953; 1956) first interventions in the Cambridge
controversies and her immediate contestants Champernowne (1953), Solow (1955;
1956; 1957), Swan (1956). It will be argued there that over the first exchanges between
1953 and 1960 the debate rather centred on problems regarding measurements of
‘aggregate’ capital in ‘aggregate’ production functions, without however pointing to the
general problems entailed by a treatment of capital in value terms for the derivation of
supply and demand functions which are essential to the marginalist approach –
although, as we shall see, some scholars had tangentially faced the issues of
‘reswitching’ and ‘reverse capital deepening’ at that time.
So far the study devotes to what the present author has above referred to as the first
phase of the debates. Thus, in Chapter 6, we present and discuss the particular reaction
from the neoclassical side that has tried to avoid the criticisms by returning to a
treatment of capital in physical terms (Walrasian treatment). Because the relevant
literature regards Bliss and Hahn as the most notably exponents of this type of reaction
we shall deal with how both these authors faced the results of the controversies. Since
one of the most frequent misunderstandings is the confusion between uniformity of the
rate of profit on the capital goods’ supply prices (the traditional rate) and uniformity of
commodities-own rates of interest, we shall examine in this chapter the problems
related to the issue of such different rates by discussing Walras’s system of
capitalisation
37
and the critique provided by Garegnani (1958; 1990; 2008). Also, it will
be argued that the problem of the non-uniformity of the rates of return poses the
problem of which way-outs could be pursued by dominant theory if the theoretical
framework of supply and demand were still to be maintained: either capital is expressed
in value terms, or the traditional method of long period equilibrium has to be
abandoned. Thus, following Garegnani (1976a), we shall see how these neoclassical
authors took refuge from the criticisms by confining the analysis to short-run general
equilibrium and pinpointing that the Cambridge critique only bears on the ‘aggregate’
versions (read it as the ‘aggregate’ production function)

but not on the modern versions
of the theory, thus starting to create misunderstandings and confusions which did not
help communication and which in fact characterises this second part of the debates.
However, as it will be argued, short period general equilibrium versions of the theory
represent a radical transformation of the dominant approach in the wake of the capital
controversies. Finally, the last word will be devoted to a provisional examination
regarding whether or not ‘reswitching’ and ‘reverse capital deepening’ bear on the new

37
Walras (1954).

Chapt er 1 – Why a st udy on t he capi t al t heory cont roversi es?
13
versions of intertemporal and temporary general equilibrium models. Accordingly, two
positions will be confronted: Bliss (who has attempted to overcome some of the
implications of ‘reswitching’ on the intertemporal theory) and Garegnani (who has
extended the critique to these new versions).
38

The study finally ends in Chapter 7, where some conclusions are advanced.




































38
The confrontation is very provisional and relies mainly on Bliss (1975) and Garegnani (2000;
2003).




PART I








Analytical and Methodological Issues in Neoclassical Economics:
Setting the Ground to Understand the Capital Controversies


Chapter 2
Understanding the foundations of neoclassical
economics: The principle of factor substitution and the
problem of capital

2.1. The data of the Marginal Theory. The direct mechanism of factor substitution
To understand the basic features of the critique of capital theory, it is worth considering
firstly the foundations on which the principle of factor substitution relies. To begin
with, it can be useful to list the three groups of the data taken as independent variables
– as ‘given’– by the marginalist theory in order to determine prices, output and
distribution. Since the data are of an exogenous nature, their determination falls outside
the purview of analysis of the theory (e.g. tastes and preferences are rather studied by
psychological and sociological theories). The endogenous variables to be determined,
however, depend crucially on those data; therefore the data must not change while the
endogenous variables are being determined.
1

The three groups of data taken as independent variables by marginal theory are:

i. Consumers’ tastes and preferences.
ii. Alternative techniques of production (methods of production available) at
a certain attained state of technical knowledge.
iii. The endowment of each ‘factor of production’ available in the economy.

By using these data, the marginalist theory can determine the relative prices, output,
and distribution simultaneously. It must now be seen how this determination comes
about, and which roles the above data play in it. In what follows, by introducing a very
simple economy, it will be explained how the two mechanisms of factor substitution,
which underlie the whole marginal approach, operate.
Let us assume an economy where the only consumption good is corn and the only
capital good in it is corn too.
2
We further assume that corn is produced by means of

1
Cf. Paragraph 2.9, this chapter, below.
2
This simplifying assumption could be replaced by the one establishing that the composition of
the output – actually composed of many goods – is given.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

18
labour and corn-capital in continuously variable proportions, which are given by the
group (ii) of the data. There is free competition, land is free (so we disregard the issue
of rent), and entrepreneurs are the capitalists. All transactions are assumed to occur in
terms of corn. To simplify matters, both productive factors – labour and corn-capital –
are assumed to be in fixed supply.
Under this very simplified economy, each capitalist is faced with a curve of the net
‘marginal product’ of labour, which results when the variable quantities of labour she
may hire is applied to the given quantity of corn-capital she possesses. We know that,
given the quantity of corn-capital, the net ‘marginal product’ curve of labour will be,
first, constant – because the entrepreneur applies the exact amount of corn-capital per
unit of labour so as to optimise the combination of both inputs in order to obtain the
maximum profit, not implying therefore to use ‘all’ the given quantity of corn-capital –
and then decreasing (otherwise, the increment of output due to hiring one more unit of
labour would always be higher; therefore diminishing returns – by keeping fixed one of
the factors – must prevail within each firm of the industry).
3
(See Appendix A)
The net marginal product of the labour curve indicates the entrepreneur how many
labourers it would be profitable to hire with her capital at any given real wage. Thus
this curve becomes a ‘labour employment’ curve. Then, by adding up the individual
‘labour employment’ curves of each capitalist it is can easily be obtained a ‘labour
employment’ schedule for the whole economy at any given real wage. If distribution
changes (e.g. the wage rate falls), cost-minimising firms tend to adopt production
methods that utilise a higher proportion of labour relative to corn-capital. Therefore, by
allowing group (ii) of the data to work, we have illustrated the first, direct, mechanism
of factor substitution.
2.2. The indirect mechanism of factor substitution
Now let us introduce a second consumption good, ‘cloth’, which is also produced with
labour and corn-capital. We now however assume that each of the two goods is
produced with only one method of production (fixed coefficients of production), with
‘cloth’ requiring the higher proportion of corn-capital to labour, that is:
/ /
cloth cloth corn corn
K L K L > . Now it is consumer’s choice through group (i) of the data that
plays a similar role to producer’s choice and marginal products in providing the basis
for a decreasing ‘labour-employment’ schedule. Let us see how it works.
In fact, since any fall in the wage rate will make the price of ‘corn’ fall relative to
‘cloth’, then, after introducing consumer’s choice, it will follow that more corn will be

3
The importance for the marginal theory of the diminishing returns principle is expressed e.g. by
Marshall (1959, p. 337): “The notion of the marginal employment of any agent of production
implies a possible tendency to diminishing return from its increased employment. Excessive
applications of any means to the attainment of any end are indeed sure to yield diminishing returns
in every branch of business.”
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
19
consumed. Therefore, given the quantity of corn-capital employed in the economy – a
crucial assumption to derive demand functions, as we shall see below
4
– a higher
proportion of it – of that constant corn-capital – will be employed in the production of
corn, with a consequent increase in the quantity of labour employed in the economy as
a whole, since, by assuming fixed production coefficients, corn needs more labour for
any quantity of corn used as capital than cloth does.
The important point to notice here is that even when there is no alternative methods
of production available (i.e., group (ii) of the data only provides one fixed-coefficients
production method), substitutability among consumption goods can give rise to
decreasing employment curves for factors, analogous to the employment factor curves
derivable from technical substitution, because the substitution among consumption
goods implies an indirect substitution among the factors employed in their production.
If we reintroduce the assumption of variable coefficients of production, it can be
envisaged that the concepts of marginal utility and marginal productivity will reinforce
each other in ensuring the downward sloping labour-employment schedule. This helps
us understand why both marginal productivity and marginal utility are both key
concepts in the foundations of neoclassical theory, as will be examined in more detail in
the following paragraphs.
2.3. From ‘employment’ functions to ‘demand’ functions (I)
To use the ‘factor-employment’ curve as a basis for demand functions we should find a
justification for assuming that the amount of corn-capital employed tends to remain
constant. Or, more precisely, as Garegnani has pointed out:

we need establish that the amount of ‘corn-capital’ employed tends to vary in
accordance with its ‘supply function’ as the wage rate, and therefore the interest
rate, vary.
5


That said, let us meanwhile go on assuming that supply of corn-capital is fixed. So,
a further condition to ensure that the quantity of labour ‘demanded’ and ‘supplied’ are
equal at any level of the wage rate and the interest rate (whatever those levels may be)
is to assume some kind of flexibility on both variables. What we need, in other words, is
that the shape of those ‘employment functions’ will not be in contradiction with the
conception of free competition. Accordingly, it is assumed that workers will tend to bid
down their wage so long as some of them are unemployed, whereas capitalists unable to

4
Cf. Paragraph 2.4, this chapter, below.
5
Garegnani (1990), p. 6.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

20
hire the quantity of labourers they want will bid the wage up.
6
This implies that the
‘employment-curves’ are downward-sloping.
7

By referring to those functions – and particularly to their properties referring to
their shapes – not only we can plausibly support how free competition works or tends
to work (as shown in Figures 2.1-2.2 below), but even more importantly, from the
economic theory’s viewpoint, we have arrived at the resulting explanation of
distribution of the social product in terms of the forces of supply and demand. The
economy will, therefore, tend to gravitate around the equilibrium wage rate w* and
interest rate i*. It is actually this gravitational process which variables undergo that
allows us to establish that demand and supply schedules determine both w* and i*.




6
As Wicksell wrote (1934, pp. 113-114, emphasis in the original): “So long as wages are
materially lower than the marginal product of the sixteenth labourer [the last], it will be to the
advantage of every landowner [being the entrepreneur] to employ more than sixteen labourers [the
maximum]. But all the landowners cannot simultaneously succeed in this object, and consequently
their endeavour must result in a rise of wages. Again, if wages are higher than the marginal
product, each of the landowners will content himself with less than sixteen workers, which will
result in unemployment and a fall in wages through the competition of the unemployed (…) The
final wage must therefore lie somewhere between the marginal product of the sixteenth and that of
an imaginary seventeenth labourer.”
7
Consider Figure 2.1. Suppose that the market wage rate is w’, the number of labourers employed
is L’, which is lower than L* (the quantity of labourers seeking employment). Now, the assumed
fall in the wage rate will lead to such a situation that the quantity L* will be employed at w*,
because the existence of unemployment u (that is u=L*–L’) will push workers to bid w down, as
assumed by free competition. By the same token, at an initial wage rate at w’’, the tendency of
capitalists to bid the wages up, when they are unable to hire the workers they want, will lead to the
point (w*, L*). See Appendix A below.

Figure 2.1. Labour employment curve Figure 2.2. Capital employment curve
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
21
2.4. From ‘employment’ functions to ‘demand’ functions (II)
Yet, the marginalist theoretical framework is not complete. It has been said above that
when w varies (e.g. w drops) the quantity of corn-capital employed should not change
with the wage, or change in accordance with the labour-employment schedule, in order
to derive the resulting increase in the labour employed and, therefore, to arrive at an
explanation of equilibrium by the forces of supply and demand through the direct and
indirect substitution mechanisms.
In fact if we assume that the quantity of corn capital employed in the economy is
kept constant as the wage rate falls, corn (in its role as consumption commodity) will be
demanded in a higher proportion relatively to cloth; hence, since more corn produce is
needed a higher proportion of corn capital will be employed in the corn sector (we keep
on assuming fixed coefficients of production). But, since corn production is more
‘labour intensive’ than cloth production, a consequent increase in the quantity of labour
employed in the economy will be reached. This is the (indirect) substitution mechanism
derived from consumers’ tastes and preferences.
As the wage rate falls the production of cloth becomes more expensive relatively to
corn production, because the production of a unit of cloth requires more corn-capital
than labour relatively to a unit of corn output. Through consumer choice corn is
demanded in a higher proportion to cloth. Production of corn must be increased to
satisfy the new consumer demand. But this assertion makes sense so long as the corn-
capital employed in the economy remains constant. The following diagram illustrates
our argument.
8

corn
D Y D D D
corn cloth cloth corn
w P P Corn Corn K K L L + ¬+ ¬| ¬| ¬| ¬+ ¬| ¬| ¬|
Diagram 2.1. Labour market when w falls

Not only the assumption of constant capital employment, or, more generally, that
the quantity of capital changes in the way expressed by its supply function (compatible
with the labour employment schedule) is plausible because the entrepreneurs are the
capitalists. Even if workers (or a third party) are the entrepreneurs that assumption is
still plausible because a similar mechanism to the one corresponding to labour market
will also be at work in the corn-capital market.
9


8
In diagram 2.1, P
corn
and P
cloth
are the relative prices of corn and of cloth respectively, Corn
D
is
the demand for corn as a consumption good, Corn
Y
is the output of corn, K
D
cloth
and K
D
corn
are the
demand for corn-capital from the cloth industry and from the corn industry respectively, L
D
corn
is
the demand for labour from the corn industry, w is the wage rate in terms of corn, and L is the
quantity of labour employed in the economy. Upwards arrows indicate ‘increase in’, while
downward arrows ‘decrease in’.
9
In other words, suppose i(r) to be the interest rate that capitalists as entrepreneurs earn for each
unit of corn-capital employed, and i to be the interest rate of the market – that is the interest rate
which the capitalists would obtain by supplying corn-capital in case the workers (or a third party)
were the entrepreneurs. Due to competition, therefore, i(r) = i in equilibrium. Otherwise, e.g., if
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

22
In fact, as Wicksell had put it, it is “the law of constant returns”
10
that ensures that
both the wage and the interest rates will not be affected by which person acts as
entrepreneur. Both labourers and capitalists are free to employ their labour or corn-
capital on their own account or to hire it out to others. If the share, say, of labour in the
product is different according as which party acts as entrepreneurs (that is w(r) = w),
the difference will soon be cancelled out by competition. And this reasoning is similar
for the case of corn-capital.
11



i(r) > i, then every capitalist would find it more profitable to act as an entrepreneur than to supply
corn-capital to firms managed by workers or a third party. But this would contradict our
hypothesis of the existence of a corn-capital market, since corn-capital, in this case, would not be
supplied. Competition prevents this contradiction from taking place. Consider the following
diagram.
( )
D Y D
cloth corn cloth
i P P Cloth Cloth givenL K K | ¬| ¬+ ¬+ ¬+ ¬ ¬+ ¬+
Diagram 2.2. Capital market
In diagram 2.2, Cloth
D
is the demand for cloth, Cloth
Y
is the output of cloth, K
D
cloth
is the demand
for corn-capital from the cloth industry, i is the interest rate in terms of corn, and K is the quantity
of corn-capital employed in the economy.
10
Wicksell (1934), p. 129.
11
The profits of the entrepreneur must therefore be zero. In fact, if the entrepreneur devotes his
human capacities to the management of production, then he must receive a wage like any other
worker. If this person also employs corn-capital in production, then he will receive interest. If, on
the contrary, he might obtain a share of product not based on either corn-capital or labour, then
everybody would like to enter this ‘easy’ business. Therefore, the profits of entrepreneur must be
nil. For this, we need the condition of constant returns, or, as Wicksell (1934, p. 126), puts it that
“large-scale and small-scale operations are equally productive, so that, when all factors are
increased in the same proportion, the total product also increases exactly proportionally”. Constant
returns can therefore be seen as the necessary condition for an explanation of distribution
according to marginal productivity principles; were it otherwise (i.e., constant returns do not
prevail at the level of the industry) the levelling influence of competition could no longer be
invoked by the theory. Consider the following example. Suppose P to be corn-output, K corn-
capital, and L labour. (Symbol © stands for ‘together with’, and symbol ÷ stands for ‘produce’).
Suppose, further, that one unit of P can be obtained with hundred units of each factor:
100 100 L K P © ÷ . Then suppose a 1% increment on each factor: 101 100
L
L K P P © ÷ + and
101 101
L k
L K P P P © ÷ + + . As is well known, the marginal productivity of the last unit of each
factor, in equilibrium, equals the respective rates of remuneration (both expressed in terms of P).
With constant returns, hence, we have
100
L K
P
P P + = which immediately implies that
100 100
L K
P P P + = so that the wages of 100L and interests of 100K together exactly correspond to
the whole product; the latter being distributed between both factors according to their marginal
contribution to production. Now, on the other hand, consider that an increment of 1% on each
factor will yield an increment of P higher than 1%, (that is (1 + o)%, with o >0):
101 101
100 100
P P
L K P
o
© ÷ + + . It follows, accordingly, that the rates of remuneration to pay for
cannot be paid from the total output, for the marginal contribution of the last unit of each factor –
(cont.)
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
23
2.5. The capital supply function
To use the labour employment curve as a basis to derive the demand function for
labour, we need the amount of corn-capital employed tend to vary in accordance with
its ‘supply function’ as the wage rate, and with it the interest rate, vary. The point here
is to ensure that the supply function of corn-capital be not in contradiction with the
demand schedule for labour. However, in this regard, a very important question arises:
is the assumption on constancy of capital employed coherent? Or, more generally, can
we a priori establish that the supply function of capital will always be compatible with
the downward-sloped employment schedule of labour?
When w falls (and i rises),
12
what we need is a supply function that allows us to
relocate a part of that constant corn-capital from the cloth industry to the corn industry,
and so to ensure that the labour employed in the economy will be increased. For this to
be so, however, the supply schedule of corn-capital must not be in contradiction with
that equilibrium process. The compatibility of the capital supply function with the
downward-sloped employment schedule of labour will be ensured, if the supply
function of corn-capital is, e.g., a fixed one.
13
Moreover, the downward-sloped shape of
the demand curves would guarantee the equilibrium stability. The capital supply
function must then be pre-determinable.
14



which governs the remuneration’s equilibrium levels – is higher than the product:
100 100 (1 )
L K
P P P o + = + . Consequently, to pay for P
L
and P
K
, the entrepreneur would lose (-o).
(Similarly, with decreasing returns, the entrepreneur would earn extra-profits).
12
As we have argued in the text, constant returns to scale ensure that the wage rate and the interest
rate will not be affected by which person acts as entrepreneur. Furthermore, due to competition,
factor payments will exhaust the product in equilibrium, so if the wage rate in terms of corn
decreases, the interest rate increases. In the case with fixed coefficients that inverse relationship
can be easily seen. With alternative methods of production, let Y=F(L,K) be the production
function, with Y corn-product, L labour, K corn-capital; let F
L
and F
K
indicate the marginal
products of labour and corn-capital respectively equal to the wage rate and the interest rate. Then
L K
dF F dL F dK = + . Moreover, constant returns to scale imply ( , )
L K
F L K F L F K = + and by
differentiation of both sides:
L L K K
dF dF dL F dL dF dK F dK = + + + . Hence, in equilibrium,
0
L K
dF dL dF dK + = so dF
L
and dF
K
are of opposite sign and therefore, also with variable
coefficients does the interest rate increase as the wage rate falls.
13
More generally, the supply function of capital should not be less negatively inclined than is the
demand curve at any level of the interest rate i.
14
Apart from both an upward-sloped supply and a fixed supply function, it would be possible for
the theory to encounter some difficulties in that, as distribution changes (e.g. the wage rate falls),
the capital supplied varies in such a way that the quantity of labour employed is not increased –
even though a huge number of labourers seeking for a job can be assumed to exist. If that were the
case, one may assert a negative answer to the question posed above. Nevertheless, we are not
concern with these further, though plausible, difficulties which the marginalist theory might come
across. The same reasoning can then be applied to the compatibility condition between the labour
supply function (L*L*’ in figure 2.1 above) and the downward-sloped demand schedule of corn-
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

24
To sum up, we could say that, in deriving the ‘employment curve’ for a factor of
production (first labour as in the example) the reasoning does not require that the
second factor (corn-capital) be fully employed; it only requires that the second factor’s
employment be given. However, since at whichever level the second factor’s
employment (corn-capital) is given, a tendency toward the full employment of the first
factor (labour) will be at work, and since this holds true for either factor, it can be
argued that the economy will be able to reach the full employment of both factors.
Then, actually, the assumption that the second factor (corn-capital) is fully
employed comes out to be the correct assumption, when one wants to determine the
price (wage) of the first factor (labour) which competition will tend to establish. Thus,
under free competition, the marginal product curve of a factor (or, what is the same, the
‘employment’ factor curve) derived under the assumption of full employment of the
other factor can then be seen as a demand curve, since the assumption of constancy of
one of the factors when distribution changes is already entailed by the full employment
of such a factor.
(The analysis we have so far undertaken can be extended to any number of factors
of production).
2.6. Key role of the downward-sloping shape of factors’ demand curves
So far it has been shown the role played by the factor substitution principle to derive a
plausible explanation of the downward-sloping factors’ demand functions. The
downward-sloping shape of such curves do in fact play a key role both in the definition
of free competition as assumed by the marginalist approach (i.e., if there is some
unemployed labourers, they will bid down their wages) and in the determination of a
supply-and-demand explanation of distribution of the social product between wages
and interests. Moreover, the tendency towards full employment of productive factors in
an economy depends crucially on the negative-sloping shape of the demand functions.
15

This is why if the shape of those demand functions were not inversely related to its
factor price (or, what is the same, if the shapes of the demand functions were upwards-
sloping), the explanation of the distribution of income between wages and interests in
terms of supply and demand would be at serious risk. And this risk would affect the
conception of a unique and stable equilibrium that is so important for an analysis of
changes in the permanent conditions of the economies.



capital. So, in general, the supply functions of factors must reflect their relative scarcities in
accordance with their demand functions.
15
We shall see below in Chapter 3 how, for particular ranges of values of the rate of interest, the
shape of the demand function for capital may not be downward sloped and therefore, it would no
longer be possible to make the assumption e.g. that unemployed labourers would bid wages down
indefinitely so long as they remain unemployed – a basic definition of free competition.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
25

Figures 2.3.-2.4. Employment function of labour and capital with upwards-sloping shapes

A very important implication can now be drawn from this concise analysis of the
basic mechanisms at work underlying the marginal theory, which are decisive to be
borne in mind to assess the Cambridge controversies in the theory of capital. The
implication is that two fundamental pillars of the marginal approach to explain the
determination of distribution in real economies through the forces of supply and
demand are:

(i) the downward-sloped shape of the demand for labour vis-à-vis the real wage rate.
(ii) the downward-sloped shape of the demand for capital vis-à-vis the interest rate.
2.7. Problems connected with the heterogeneity of capital goods
In the simple economy described in Paragraphs 2.1-2.6 above, capital consisted of a
single good, and, therefore, a problem of measurement cannot arise since the units in
which capital is expressed are the same in both its roles: a role describing the
alternative methods of production [group ii) of the data] and the role defining the
‘factor endowment’ [group iii) of the data]. Turning the analysis to more realistic
economies it must no longer be assumed that capital homogeneous both in the
alternative processes of production and in the production of heterogeneous consumption
goods. Accordingly it is now worth providing the three basic characteristics of those
commodities normally treated in the literature as capital goods:

(a) They are reproducible goods – unlike land (or to labour) since either of these
original factors cannot be reproduced.
(b) They are utilised for further production both of consumption and of capital
goods.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

26
(c) They wear out within a sufficiently short period of time.
16


When capital consists of heterogeneous goods bearing the properties just listed, a
contradiction between the roles pointed out above arises. This contradiction lies in the
fact that, on the one hand, the capital goods which enter in the argument of the
individual production functions defining the methods of production are (and must) be
expressed in technical units to define the alternative methods of production,
17
while on
the other, the endowment of ‘capital’ in the economy, as it can be tracked down in the
history of marginalist economic analysis, has been expressed not in technical units but
in value terms. If this is so, there is a serious problem for the theory in that the ‘factors
of production’ should be measured – and hence known in advance – before knowing
distribution, which is the aim of the theory.
18

Traditionally, as we shall see in the remainder of the present chapter, capital as a
‘factor of production’ has been conceived in value terms. Moreover, in the explanation
of supply and demand equilibrium functions for productive factors capital has been
specified among the data in traditional general equilibrium analysis as a single
magnitude in value terms. The definition and specification of the endowment of capital

16
One can better regard the long-life capital goods as rent goods, as Wicksell (1934, p. 118,
emphasis added), puts it: “capital in many cases also advances rent. A farmer who breeds cattle for
meat, for milk, (…) must pay rent for his pasturage for many years before he can employ the
animals in question”. Likewise, some pages later we find in Wicksell’s Lectures the following
distinction: “nearly all so-called fixed capital (houses, durable machinery, etc.) are, economically
speaking, on the border line between capital in the strict sense and land […] [T]he operation of the
laws of capital depends upon the assumption of a constant adjustment of concrete capital goods in
an endless repetition of the same process of investment and production. But this is only of
practical importance in capital investments of relatively short duration.” Wicksell (1934, p. 186,
emphasis added). Nowadays one might compare the depreciation that, say, a lathe would undergo
through its utilisation life with the depreciation of a dam or a windmill in order for the former to
be replaced through a “constant adjustment in an endless repetition”.
17
Both these roles of capital can be seen in Wicksell’s general equilibrium system developed in
his Lectures (See Appendix B to this chapter.) In that system, capital is, on the one hand,
expressed into the production function in technical units (saved-up labour and saved-up land used
in previous periods for the production of capital goods necessary for the production of the
consumption good, cf. equation [1.B] in the appendix already referred to). On the other, together
with the equilibrium of supply and demand functions for original factors (labour and land
expressed in technical units) are the supply and demand equilibrium functions of capital.
18
Note that the identification of factor’s endowments can be easily satisfied by the factors
traditionally grouped as labour and land. What we call ‘labour’ is actually composed of different
kinds. However, it can be assumed that, for long periods of time, the hierarchy of wages relating to
the different kinds of labour does not steeply change, so one can regard the different kinds of
labour as different powers of a standard labour. But even if wages of different kinds of labour
cannot be taken as constant, those kinds of labour can always be dealt with without resorting to
value magnitudes, by treating each as a factor to which there correspond separate supply and
demand functions. Similarly, different kinds of ‘land’ can also be dealt with in the same way as we
have seen for the case of different kinds of labour when the wages of those kinds cannot be taken
as constant.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
27
is not independent of distribution, however; thus a contradiction between those roles
arises.
19

A very important question arises: What are the consequences of treating capital in
value terms when it is used as part of the argument to derive supply and demand
equilibrium functions for the productive factors, and why capital was so treated in the
specification of the endowment?
20

Since the Cambridge critique is intrinsically connected with the role of capital as a
‘factor of production’ by the marginalist theory, it might be useful to approach the
difficulties associated with the question raised above from two angles. On the one hand,
a ‘supply-side’ role of capital, which is associated with the identification of the capital
endowment, and, on the other, a ‘demand-side’ role of capital, which is associated with
the conditions required to derive demand functions of capital.
2.8. Supply-side role of capital
21

Some difficulties associated with the measurement of capital may immediately be
envisaged from a ‘supply-aspect’, or a supply-role, of capital. The question is, how to
specify, and hence to measure, the given ‘capital’ as an endowment independently of
distribution in case it does not consist of a single good? Should we regard, in a world of
heterogeneous capital goods, each single capital good as a component of the
endowments, as we did before when capital consisted of a single good? If one goes
back in time to see how traditional economists had considered capital among the data of
equilibrium, we shall see, however, that economists such as S. Jevons, E. Böhm-
Bawerk, K. Wicksell, J.B. Clark, T.N. Carver, or A. Marshall (among others) conceived
of capital as a single magnitude in value terms susceptible to changing form without
however changing its value.
22


19
As Garegnani (1990, p. 10, emphasis added) has pointed out: “if we cannot define the quantity
of a factor before knowing distribution and relative prices, we shall clearly be unable to use the
theory, which allows us to determine distribution and relative prices only when we know the
methods of production and the factor endowment of the economy in question.”
20
The problems, which arise out by treating capital in value terms as arguments of the production
function, are analysed in Chapter 5 below, where Joan Robinson’s participation in the Cambridge
controversies is analysed.
21
See Appendix B where the supply-role of capital is treated in some more detail by considering
the general equilibrium system which Wicksell put forth to analyse production in economies with
capital goods- that is “capitalistic production” as he called it (Wicksell 1934, p. 144).
22
These examples constitute the core of the traditional neoclassical approach which can be
reassumed by taking as datum of the equilibrium the ‘factor’ capital in value terms. Cf. equation
(4.B) of Wicksell’s system, Appendix B, below. For the treatment of capital by Pigou see
Hennings (1990, p. 166). Along this work we shall return to this topic, that is why these
marginalist authors conceived of capital in value terms to define the equilibrium of supply and
demand – and not as a vector of the heterogeneous capital goods which we may come across in an
economy as is the case of Walras’s treatment of capital.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

28
For instance we have the case of Marshall who in his time-honoured Principles of
Economics stated:

the chief trouble [regarding ‘the general rate of interest’] comes from the fact that
the income derived from capital […] invested in particular things […] is properly
a quasi-rent and can be regarded as interest only on the assumption that the capital
value of the investment has remained unaltered.
23


On the same page, the following conclusion is raised by the renowned English
economist:

thus then interest […] tends towards an equilibrium level such that the aggregate
demand for capital in that market, at that rate of interest, is equal to the aggregate
stock forthcoming there at that rate.

Marshall’s ideas, however, were not isolated. Wicksell, too, regarded ‘capital’ (the
endowment) as a single factor in value terms susceptible to adopting any physical form.
In fact, when he deals with “the concept of Capital”, he says:

Capital also includes the raw materials […] and other commodities which must be
saved-up. […] This, of course, is the commonly accepted sense of the term. [A]ll
these [different capital goods] […] have only one quality in common, namely that
they represent certain quantities of exchangeable value so that […] they may be
regarded as a single sum of value, a certain amount of the medium of exchange,
money.
24


Both authors – along with other representative traditional economists mentioned
above – conceived of capital the endowment as a substance in value terms. Thus, this
“single sum” is taken as given in the group (iii) of the data; capital, a quantity measured
in a single unit (a consumption good for instance, or also in terms of a capital good)
25
is
taken as datum in the determination of output, prices and distribution in terms of supply

23
Marshall (1959, p. 443, emphasis added). The adjective which I chose to use in the text refers to
the several reprints of Marshall’s book.
24
Wicksell (1934, pp. 144-145, emphasis added). Cf. also Paragraph B.3 (Appendix B), below.
25
If the value of a particular capital good were made constant by using the capital good in question
as its own unit value, still the problem would evidently be there for all other capital goods.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
29
and demand functions.
26
What was the justification of such a measure in marginal
theory?
Although this measurement does not satisfy the requirement of independence from
changes in prices and distribution, it was traditionally believed that owing to group i)
and ii) of the data, the theory would somehow determine distribution according to the
factors substitution principle, establishing an inverse relation between the relative
quantities employed of a productive factor and its rate of remuneration. In a world of
heterogeneous capital goods, capital was specified as a single factor in value terms
which could adopt the physical form according to the equilibrium conditions of the
economy, which in turn were thought to be sufficiently persistent so as to regard the
theoretical variables determined by the theory as gravitational centres of actual
variables. The determination of the physical form of the capital stock was envisaged as
a process through which a uniform rate of profit (interest) on the supply prices of the
several capital goods could be ascertained.
The idea underlying this process is the conception that the several capital goods
can embody different amounts of the substance ‘capital’; in other words, ‘capital’ is a
kind of ‘stuff’ which can change form without changing its total quantity in value. For
example, imagine for a moment a non-equilibrium situation; we can envisage an ‘old’
and a ‘new’ situation, before and after some change in the conditions (e.g. a change in a
pattern of consumption or a change in the quantity of people seeking for jobs due to a
process of immigration): in that equilibrium process the physical composition of that
substance would be endogenously determined, if, given a sufficient long period of time
for adjustment, the specific physical capital goods to be produced corresponding to the
new conditions of the economy could be produced by the most profitable use of the
available resources (techniques), which must go in accordance with the factors’
substitution principle. In fact, a change in the ‘form’ of the single factor ‘capital’ was
thought to be brought about by the employment of factors, namely – labour and
‘capital’, which may had been previously used to produce the previous capital goods no
longer profitable to the new conditions of production (and consumption), in the
production of new capital goods. Thus the new capital goods are produced by
employing the same economic resources of the economy which produced the hitherto
capital goods and, therefore, would have the same total cost of production and hence
the same total value as the no longer produced capital goods.
27
In this way, the uniform

26
The outstanding exception among traditional authors is Walras, who conceived of ‘capital’ as a
physically heterogeneous capital goods vector existent in the economy. Cf. Walras (1954); see also
Garegnani (1960, part II; 2008). We shall deal with this author later (Chapter 6) and see the
problem arising out from such a conception of capital (Paragraph 6.2).
27
The clue to such an explanation provided by marginalist authors (that is the justification of the
principle of factor substitution) can be traced in the attempt at a generalisation of the principle of
intensive rent applied by Ricardo however to ‘land’ and ‘labour and capital’ to explain profits: the
generalisation was then applied to any number of factors, in particular to labour and ‘capital’
(assuming free land) such that both labour and ‘capital’ can be seen as receiving their marginal
product. Therefore the analysis is thought to apply symmetrically to factors of production, land,
labour, and ‘capital’, and it is re-interpreted as evidence of the two substitution mechanisms we
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

30
rate of profit (interest) on the supply prices of the several capital goods is ascertained.
This process was thought to work somehow through the sufficiently persistency of the
data of equilibrium.
As noted above, capital in value terms not only ensures the uniformity of the rates
of profits (interest) on the supply prices of the capital goods, but also it allows for the
substitutability of the factors of production.
28
In the wake of any change on the
permanent conditions of the economy (e.g., new quantities available of some resource),
the economy would reach a new position of equilibrium by the supply and demand
forces which work through the data of the theory. The new position to be reached
implies a process where production, demands, prices, quantities are actually in
disequilibrium; however, after certain period of time the supply and demand forces
would put the economy again in balance. For such a result, it is key to assume that the
capital ‘substance’ be expressed in a single value magnitude susceptible to change form
but not changing in quantity in order to let the supply and demand functions work
throughout. Yet that ‘quantity’ in value, like any other value magnitude, depends on the
variable which theory attempts to determine – the rate of interest, prices, etc.; thus a
problem, which we may identify as the ‘supply-aspect’ problem, regarding the
measurement of capital as the data consistent with the logic of the theory arises.


have considered in the text. Accordingly, were it not for conceiving that ‘capital’ in constant value
terms susceptible to adopting any physical form as the economy conditions change, an explanation
of distribution relying on ‘factor’ substitution – from which in turn we can derive the equilibrium
supply-and-demand functions – would not be possible. Were it otherwise, this would entail the
impossibility to derive demand functions for factors from employment factor’s curves, as we have
already argued in our simple economy’s example (cf. Paragraph 2.2-2.4 above.) This, as we argue
in the text, also explains why the physical composition of the stock of capital is left to be
determined endogenously, while the value of that stock does not change. For an analysis of
Ricardo and the classical political economy, see in particular Garegnani (1960, part I; 1984a).
28
See Appendix B where the issue of the uniformity of the rate of interest of capitals’ supply
prices is discussed within the framework of Wicksell’s general equilibrium system.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
31
2.9. Persistence character of the determining circumstances: The traditional
method of ‘normal positions’
29

We have seen above the meaning of the data taken as given by the theory to determine
the dependent variables output, distribution and prices simultaneously; how the forces
of supply and demand work in the factors’ markets have been explained in the first part:
those forces are supposed to work in real economies, since they are thought to be
founded on the basic principle of factor substitution. We have finally seen that the
factor substitution can be brought into analysis both through the alternative techniques
of production (direct mechanism of substitution) and through consumers’ tastes and
preferences (indirect mechanism of substitution). The resulting equilibrium (let us
assume it to be unique for the sake of exposition) of supply-and-demand (i.e., the level
of the dependent variables, at which it is arrived through the working of those forces) is
ensured to be stable – provided the ‘right’ shape of the demand curves for factors. But
to apply this approach to real-world economies the data through which one arrives at
equilibrium must account for some basic features: the data must be thought of as having
a persistence nature through time, in that they are supposed to be those variables on
which the working of the mechanisms of factor substitution rest.
In actual markets, however, prices, quantities, the level of the wage rate are all
magnitudes that depend on circumstances that are at work due to many causes that can
change on a daily (or hourly) basis. A specialised journalist, say, in agriculture
products’ prices, does every morning try to provide his audience (in general perhaps
investors) with the level of those prices due to the variables he considers the most
relevant at that particular moment in that particular place. They run from the drought of
agriculture fields in American prairies to the level of the oil price, on which the
production costs of corn depend and which in turn is conditioned by the political
situation in some oil exporter’s countries.
30
But these particular considerations this

29
The term ‘normal positions’ has been adopted by Garegnani (1990) in order to refer both to the
marginal long period equilibrium of supply and demand, which defines a ‘long period position’ of
supply and demand, and to the ‘normal’ or ‘natural’ positions of the classical economists (Smith;
Ricardo), which are not derived from supply and demand forces, as in marginal theory. The latter
school of economic analysis, which has been revived by Sraffa (1960) and the successive works
undertaken by the surplus approach, has grounded the explanation of normal or natural position of
the economy by relying on the key notions of a subsistence wage (determined by social and
political circumstances) and the existence of a surplus, which arises after the necessaries of the
workers and the replacement of means of production are deduced from the social product. (See
Garegnani 1984a) Though in this study we shall deal with the marginal approach, it is worth
reminding that, nevertheless both theories have relied on entirely different forces to explain
distribution of income, they have shared the method of normal positions – although the marginal
theory has since recent decades shifted to a different notion of equilibrium. We shall argue that
this shift has been one of the consequences of the debates on capital theory, which is analysed in
Chapter 6.
30
Cf. the following argument raised by Marshall (1920, p. 291): “The actual value at any time, the
market value as it is often called, is often more influenced by passing events, and by causes whose
action is fitful and short lived, than by those which work persistently. But in long periods these
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

32
journalist is concerned with are not, at a theoretical level, thought to be persistent
circumstances on which those prices ultimately depend. So, in general, supply-and-
demand equilibrium prices (and quantities, and distribution), as envisaged by the
theory, are not daily prices, but rather average, ‘normal’, ‘natural’ ones. The actual
variables, or better still, the average of actual variables, if the theory were correct,
would follow the trend described by those theoretical equilibrium values. But to arrive
at such averages does take time: it is thanks to the persistence of the conditions of
economic phenomena that ensures the gravitation process, where divergences between
the values of the theoretical and actual variables are cancelled out over time, so long as
repetition of transactions in markets is guaranteed;
31
(following the analogy, temporary
droughts in American prairies are not a persistent causes in the determination of corn
prices). Thus an explanation of distribution of the social product in real economies in
terms of equilibrium of supply and demand is further grounded on the persistent nature
of the economic forces meant to be at work.
From the foregoing discussion on the persistent nature of economic conditions, a
very important question can be posed: Does persistency imply that the data of the
theory must not change?
Traditional marginalist authors did obviously not deny that the data might change.
However, thanks to the persistence of the causes of economic phenomena, they
envisaged that their changes were generally either pretty slow enough so as to be
negligible relative to the presumable speed of adaptation of demand and supply for
factors and products, or, if they were rapid, changes were assumed to be once-and-for-
all changes, whose effects could be dealt with through comparative statics, that is by
assuming a tendency to the corresponding new equilibrium in the new situation. In other
words, the same persistent nature of the causes allows the theory to deal with changes
by relying on the method of comparison of long period equilibrium positions.
32
Let the
present author explain it a little further, if only briefly.
Imagine an economy where a process of immigration is taking place. This process
will certainly bring about a change in the quantity of labour supplied. Before the
newcomers got into the labour market, there was an equilibrium in such market, say
(w*, L*), to which there also corresponded a determined set of equilibrium prices and


fitful and irregular causes in large measure efface one another’s influence so that in the long run
persistent causes dominate value completely.”
31
The variables that the theory determines cannot be the actual magnitudes observable moment by
moment in the market, which are influenced by many accidental factors, since those factors could
not be known in advance. See, Garegnani (1990, p. 46).
32
The usefulness of the method of long period equilibrium lies in the fact that the theory provides
a logic framework to analyse changes through an analysis of comparison of different ‘normal
positions’ by means of the mechanisms pointed out in the text. It is however very important to bear
in mind that for this method to be used the capital must be conceived in value terms, thus an
illegitimacy is entailed. Yet we will return along this work to these uses of the theory, in particular,
when we come to approach the ‘new’ versions of the general equilibrium theory in its
intertemporal or temporary versions.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
33
outputs. As the immigration sets off, imagine that workers’ wages remain at w*.
However, population at the age of getting a job is changing. As time goes by,
population is (re)adapting itself to the new conditions in the labour market: some new
arrived people go to some firms to seek jobs, other ones go elsewhere, both new and old
labourers are modifying the labour supply: however, it is necessary time for that supply
to arrive at a determined level compatible with demand and free competition. In that
process, there of course is a ‘quantity of capital’ in value terms which is slowly
changing, as those forces are pushing the system towards equilibrium, that is – as the
physical composition of that ‘quantity’ is being adapted to the new conditions. In that
very process the theory assumed that quantity to remain constant. That period of time
must be a long one, in order to let the physical composition of capital be adjusted, even
though, individually perhaps, the period required by a ‘new’ arrived person (or family
or group) to get a job was short due to specific circumstances (e.g. s/he or they
was/were given a job in advance before arriving in situ) and so on. But, at a general
level, the period of time must be long. How then is the equilibrium reached?
As the population at the age of getting a job changes, the supply of labour will
change (some of the data of the theory change).
33
Accordingly, the wage rate falls, and
the quantity employed rises, as expected by the theory. But for this we have to assume
that the quantity of ‘capital’ employed remains constant (in fact in full employment).
However, the composition of that capital will also be changing, as the more ‘labour
intensive’ methods of production come to be adopted in that very process of adjustment
(a time consuming process which runs from a disequilibrium situation towards an
equilibrium one). In between, that is, since the immigration sets off but before reaching
the new equilibrium, transactions, contracts, etc., take place at non-equilibrium values
(with prices changing very gradually, which turn out to be almost imperceptible) but,
however, they take place by thinking at the same time of a tendency to a future
equilibrium position.
34
As the immigration process takes place, more ‘labour-intensive’
processes of production are adopted – given the technical knowledge of our imaginary
economy – since they are the most profitable. When the composition of ‘capital’ is then
adjusted to the new conditions of the economy, we would have reached a new long
period equilibrium. Once the uniform rate of return on the supply prices of the capital
goods is reached, the composition is endogenously adjusted, thus ensuring the
persistence of the equilibrium position itself.

33
The theory can give an appropriate picture of the process if the data are stable in comparison to
the speed of adjustment. For a detailed methodological analysis, see Schlicht (1985, §1.2.7, p. 6).
34
A logical necessity of the traditional method – which attempts to explain actual magnitudes as
oscillating around centres of gravitation – is that these gravitational centres must not be affected
by the ‘day-by-day’ disequilibrium production and consumption activities undertaken by the
economic agents. Traditionally, it was thought that in the definition of long period equilibrium of
supply and demand, prices’ changes could be ignored. The argument for such a thesis is the
assumption that, when adjustments are taking place, the allocation of savings among capital goods
could not appreciably influence the equilibrium position. See Garegnani (1990, p. 51); cf.
Appendix B below, in particular equation (3.B) regarding Wicksell’s treatment of savings.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

34
Under conditions of free competition in factors’ markets, and due to the properties
of the capital goods, which in general are commodities with a short utilisation period,
investors will, in the wake of some change in demand conditions, seek the most
profitable capital goods wherein to invest their capital. The resulting equilibrium capital
goods’ prices will tend towards the level of their costs of production (supply prices);
thus the rates of return over capital goods’ costs will tend to be uniform: the differences
in profitability among the several capital goods will cause some investors to sell their
capital, whilst others to buy other capital goods. This process, in turn allowed by free
competition among investors throughout the different industries and capital goods, can
therefore be seen as an equilibrating process in that the several goods will pay off a
uniform return over costs, independently of the particular physical capital good
concerned. Thus provided, we may understand this basic reason why in traditional
writers the endowments of the several capital goods cannot be taken as part of the data
of the equilibrium,
35
if a process involving disequilibrium adjustments is taken into
account: in general those adjustments need long periods of time.
2.10. Demand-side role of capital: Conclusion
In the example about immigration above, we have said that once the composition of the
stock of capital is endogenously determined a new long period equilibrium will be
reached, by satisfying the condition of uniformity of the rates of profits on the capitals’
supply prices. The substitution of factors when capital consists of a family of
heterogeneous capital goods entails, in general, that it is not a change in the proportions
of ‘factors’ that allows theory to arrive at a new equilibrium, but a change in the kinds of
the capital goods due to the adoption of new methods of production.
In fact, whether an economy would produce particular kinds of capital goods will
depend on a) the methods of production available in the economy, and on b) outputs. In
other words, production of the specific capital goods (investment) depends upon the
demand conditions of the economy. In the wake of some change in the demand
conditions, marginal theory still needs a treatment of ‘capital’ as analogous to that of any
‘factor of production’ in the direct and indirect factor substitution mechanisms, to allow
the demand for ‘capital’ to be a decreasing function relative to its ‘price’ – the interest
rate. This is the demand-role or demand-aspect of capital. This demand side concerns two
related aspects, which are very important to bear in mind when we shall come down to
the results of the controversies in the theory of capital (ch.3 and part II below).
First, capital could be treated as any other factor of production to ensure that the
substitution mechanism work in the ‘right’ direction, i.e. the demand function for capital
is inversely related vis-à-vis its remuneration rate – the rate of interest. The conception of
capital of variable form in turn makes it possible to assume the existence of a sufficiently
substitutability between capital and labour (or land). This sufficient substitutability – and

35
With the outstanding exception of Walras as noted earlier.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
35
here we come to the second related aspect referred to above – is compatible with a
conception of a ‘quantity of capital’ as a single magnitude in value terms but
incompatible with a specification of the physically heterogeneous capital goods among
the data of the theory. The importance of this sufficient substitutability between the
factors of production can be traced e.g. in Hicks’ 1932 The Theory of Wages, where the
author argues that such a substitutability would be impossible to conceive, in the long
period, for single capital goods:

In the short period it is reasonable to expect that the demand for labour will be
very inelastic, since the possibility of adjusting the organisation of industry to a
changed level of wages is relatively small. (...) Since the whole conception of
marginal productivity depends upon the variability of industrial methods, little
advantage seems to be gained from the attempt which is sometimes made to define
a ‘short period marginal product’ – the additional production due to a small
increase in the quantity of labour, when not only the quantity, but also the form, of
the co-operating capital is supposed unchanged. It is very doubtful if this
conception can be given any precise meaning which is capable of useful
application.
36


In fact that variability in the form of capital cannot be conceived as an instantaneous
process; actually, at each moment of time, capital is ‘crystallised’ in specific capital
goods, and can take the best form adapted to demand conditions: changes in demand due
to changes in consumer patterns, technical progress, etc. But that change in ‘form’ is
gradual: it occurs once the existing capital goods are depreciated and the reinvestment of
these allowances into new capital goods takes place, which are physically different from
the depreciating ones. Thus ‘capital’ – the value substance – embodied in the old capital
goods becomes free to be re-invested in the new, different, ones. This is the idea of a
‘capital’ as a single factor susceptible to changing form. Thus, the ‘marginal product of
capital’ (see the basic scheme in Paragraphs 2.1-2.6, above) can manifest itself in
investment in the new plants: the flow (investment) of free capital would be met with the
labour ‘freed’ by the closure of the old capital goods’ plants. This linking is what allows
us to consider the investment demand as a manifestation of the demand for capital.
37
Then

36
Hicks (1932, pp. 20-21, emphasis added).
37
That is to say: the supply and demand functions for ‘capital’ as a fund were envisaged by
traditional authors to analyse the forces operating in the investment and savings market – where
capital appears as a flow. To understand this it is sufficient to think about the traditional supply and
demand functions for capital the single factor. As it was shown in the text the physical
composition of the capital demand function will be determined by the production methods and
outputs and hence are unknowns of the problem. On the other hand the physical composition of
the capital supply function cannot be but what is commonly known as the stock of the economy. In
general, both compositions will differ; this incompatibility between both compositions is resolved
itself precisely in the investment savings market: what is underlying the capital supply curve is
that the physical composition of the stock appropriate to the equilibrium will be assumed by the
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

36
if this connection is admitted, and, provided the ‘right’ shape of the demand curve for
capital, the interest rate could then be seen as the price bringing into equality the demand
for and supply of ‘capital’ the flow.
To conclude. The introduction of a ‘quantity of capital’ (a single magnitude) into the
general equilibrium supply and demand functions makes it possible to arrive at long
period equilibrium positions, while leaving the endowments of the several capital goods
to be endogenously determined by the tendency toward a uniform rate of return on the
supply prices of the capital goods through factor substitution. This allows the theory to
apply the method of comparison of normal positions for an analysis of changes.
Notwithstanding the supply-side problem regarding the illegitimacy of conceiving of
capital in value terms to determine distribution,
38
it was once believed by the founders of
the neoclassical approach that, in order to ascertain an equilibrium position as a result of
applying supply and demand general equilibrium functions, the quantity of capital per
worker in the economy had to be inversely related to the rate of interest, by means of
changes in demand associated with groups i) and ii) of the data of the theory.
This ‘belief’ regards the demand-aspect of the problem: that aspect resolves itself in
the necessity for the theory to have downward-sloped demand curves for the productive
factors. If downward-sloped demand curves for factors of production can be rigorously
built on, then the way to the explanation of distribution in terms of supply and demand
through the mechanisms of substitution of relative ‘scarce’ factors would be achieved.
In the next chapter, we shall introduce some main results drawn on the capital
controversies in order to see whether the marginal theory can determine distribution on
the grounds of an inverse demand curve for factors in economies with heterogeneous
capital goods. What is key to keep in mind is the role of the downward-sloping demand
curve for capital as it was traditionally conceived – a single magnitude in value terms –


existing stock over a period of time when part of the capital goods has to be replaced by
reinvesting the depreciating allowances into new capital goods. As Garegnani (2000, pp. 395-396)
has argued, it would seem that the traditional authors’ focus was on analysing the forces of the
fund concept of capital rather than the flow, in order to analyse the “substitution of factors in a
purer form”. We shall see the link between the savings-investment market with the fund capital
market in more detail later, Chapter 6.
38
This is the most evident contradiction which arises from the measurement of capital in value
terms for the marginal theory, since it is that the datum from which we should depart in order to
arrive at the equilibrium values of output, prices, and hence the very distribution. Therefore, the
circularity in the reasoning is not avoided; however, it is capital conceived in value terms that
allows the theory to leave the physical composition of the stock be endogenously determined, thus
being coherent with the traditional method of a uniform rate of profits on the capitals’ supply
prices (see the case of Wicksell in Appendix B).
Later in this work we shall also present the other treatment of capital namely that of Walras who,
even sharing with his contemporary peers the traditional method, considered the capital
endowment in the data of equilibrium as a vector of heterogeneous capital goods of the economy:
this led Walras to contradict his own general equilibrium system, since in general the arbitrary
composition of capital in his treatment will prevent the rates of return of the several capital goods
from being uniform on the supply prices of that goods. See Chapter 6.
Chapt er 2 – Underst andi ng t he f oundat i ons of neocl assi cal economi cs
37
and with it the supply and demand aspects of the problem of capital. The role of the
demand curve’s slope is decisive for uniqueness and stability of the equilibrium and
hence for the plausibility of the traditional supply-and-demand theory of distribution.
(Cf. Figures 2.3 and 2.4 above).


Chapter 3
‘Reswitching’ and ‘Reverse Capital Deepening’:
Introduction to the heart of the controversies
and their implications
3.1. Some premises
In the previous chapter it has been introduced the key role of the downward-sloping
demand curves for productive factors that can be derived according to the factors’
substitution principle. This principle can univocally establish, e.g. in a labour and corn
(both as consumption and as capital-good) economy, an inverse relationship between
factors’ intensities and factors’ rates of remuneration. However, can we derive the same
kind of relations when one of these factors is ‘capital’? As sketched above, the bulk of
well-known marginalist authors have conceived of capital as a single magnitude in
value terms susceptible to adopting the physical form of the several capital goods –
hence their quantities as well – determined by the equilibrium condition of a uniform
rate of profit on the supply prices of the capital goods. Nevertheless the illegitimacy of
the notion of capital in value (the supply-side aspect), the founders of the neoclassical
approach somehow believed that, by considering changes in demand due to changes in
the production methods and outputs – associated with groups i) and ii) of the data –, an
inverse relation between the rate of interest and the demand for ‘capital’ could in
general be postulated. It was believed that owing to changes in demand the factor
substitution principle could be also taken for granted even when factors were measured
in value terms so as to derive a monotonic relation between changes in distribution and
factors’ intensities, i.e. to derive a negative relation between ‘capital’ and the rate of
interest. This is the demand side of the problem. As we shall see in this chapter, the
discovery of the phenomena of ‘reswitching’ and ‘reverse capital deepening’ in the
Cambridge controversies has put that neoclassical belief at stake.
‘Reswitching’ and ‘reverse capital deepening’ started to reach prominence in the
Cambridge controversies in the 1960s, particularly after Sraffa’s (1960) famous book,
which was followed by a series of contributions that started to spread these discoveries
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

40
over the whole decade.
1
Rather than following a chronological description of the
episodes around the Cambridge debates, it has been preferred in this part of the study to
bring to the foreground the meaning and implications of ‘reswitching’ and ‘reverse
capital deepening’, because these results touch the hard core premises of the marginal
theory, despite that these phenomena started being acknowledged only in a later stage
of the first phase of the controversies. The aim of this chapter is therefore to introduce
the reader to the meaning of such phenomena and, more importantly, their implications
for the marginal theory. Thus, this chapter provides the basic ground through which
further exchanges between the two sides evolved.
2

Is it possible, in a heterogeneous capital goods economy, to derive a univocal
relation between capital intensity and changes in distribution? Neoclassical traditional
theory certainly took on board such question and traditionally provided its solution by
putting forth the belief that, at a low rate of interest (r) and from a spectrum of
alternative techniques (say A and B both using heterogeneous capital goods), producers
– e.g. of a single final consumption good
3
– would be led to use a ‘more capital
intensive’ technique (say A) so that this technique would, in competitive long-run
equilibrium, minimise the cost of production of the final good. Further, it was also
believed that a rise in r might lead producers to change (switch) to a ‘less capital’
intensive technique (B) – so long as this technique minimised costs – leading the
economy to account for a lower quantity of ‘capital’ per worker. But it was also
believed that before a further rise in r the old technique could not be adopted again
because that technique would have been categorised by the theory as a ‘more capital’
intensive one, which could have never again been chosen at a higher rate of interest. In
the Cambridge controversies it was shown that that possibility cannot be excluded a
priori. Since capital goods are reproducible and also require further capital goods in
their production, it could be the case that at a higher r the technique that had been used
at lower values of r could be chosen again – why? Because this technique might also
minimise costs at the new situation of prices and distribution. This theoretical
possibility has been known in the debates as ‘reswitching’ of techniques. Thus, it would

1
See Appendix C. Among the several contributions to the ‘reswitching’ debate we should cite:
Sraffa (1960) and the articles published in the 1966 Symposium: Pasinetti (1966a), Bruno et. al.
(1966), Morishima (1966), Garegnani (1966), Levhari and Samuelson (1966), Samuelson (1966a).
That Symposium was the result of the intervention from the neoclassical side of Levhari (1965).
We deal with the historical aspects of the Symposium in the next chapter. The content of the
present chapter is mainly drawn from Samuelson (1962) and Garegnani (1970a).
Note, however, that the phenomena of ‘reswitching’ and ‘reverse capital deepening’ had been
noticed in the literature in the controversies before 1960, in particular in Robinson (1953) and
Champernowne (1953); however these authors gave these phenomena little prominence at the time
(Cf: Robinson (1953) p. 106 where ‘reswitching’ is treated as a “curiosum”; and also Robinson,
(1956) p. 109 where the Cambridge author regards the ‘curiosum’ as an “intricate piece of analysis
which is not of great importance.”) Nevertheless we shall deal with these authors in Chapter 5.
Finally note that, after Sraffa (1960), the first work where the terminology of ‘switching’
commenced to be used was – to the present author’s knowledge – McManus (1963).
2
See Chapter 4, below.
3
Or a given bundle of final commodities.
Chapt er 3 – ‘ Reswi t chi ng’ and ‘Reverse Capi t al Deepeni ng’
41
seem, a technique a priori considered as ‘more capital’ intensive is chosen for a range
of values of r and also is chosen for other range of values of r, while another technique
is chosen in between.
4
If this is so, then there will not be any monotonic relation
between changes in distribution and ‘capital’ intensities. The idea of an ordering of
techniques according to ‘more’ or ‘less’ capital intensity disappears with changes in
distribution, for a change in distribution entails a change in the way in which – if such
way existed – those techniques were ordered according to their relative factor
intensities. But, if that were the case, then there would be no theoretic support to think
of an inverse relation between factors’ demands and their rates of remuneration, and
hence of the factor substitution principle in general, which is key to the marginal
theory. Thus, with ‘reswitching’, hard core premises of the approach are at stake.
‘Reverse capital deepening’ indicates, in economies with production involving
heterogeneous capital goods, the possibility that the demand for capital be in a direct
rather than in an inverse relationship to the rate of interest. Why? Thanks to
‘reswitching’, as we shall see below in more detail,
5
the value of capital per worker in
the economy may rise with a rise in r as the economy ‘re-switches’ from one technique
to the one that had been previously used at lower values of r. But also, even without
‘reswitching’, ‘reverse capital deepening’ may arise. In fact, due to changes in
distribution it might be the case that the value of capital per worker changes in the
opposite direction as traditionally conceived.
6

The novelty of these results, revealed by the Cambridge, UK, side of the debates,
lies in the fact that even if the illegitimacy of the supply side is put aside, and by taking
into account the very premises of the theory, serious difficulties would arise in order to
derive downward-sloping demand curves by relying in demand changes, because this
demand aspect is undermined by ‘reswitching’. Hence it turns out to be impossible to
derive any measure of capital in value terms in a univocal direction with changes in
distribution, but, for reasons already explained, this notion is central for the whole
approach (cf. Chapter 2).
In what follows the demonstration of the results of ‘reswitching’ and ‘reverse
capital deepening’ is introduced by using very simple graphical analysis as used by
some of the participants in the debates.
7




4
Or that technique may be ‘more capital’ intensive for some ranges of values of r and ‘more
labour (or other factor)’ intensive for other ranges of r.
5
See Paragraph 3.4, this chapter, below.
6
See Appendixes C and D.
7
For the present exposition we shall follow Samuelson (1962) and Garegnani, (1970a). System of
production involves two methods of production: the one for the consumption good and the other
for the capital good. It is assumed that for each system the capital good is produced by itself and
labour which is a very simple case chosen by Samuelson (1962) in the controversies, then also
used by Hicks (1965) and fully studied and criticised by Garegnani (1970a). For more general
models, see Sraffa (1960), Pasinetti (2003, ch.6), Garegnani (1970a, pp. 417-421).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

42
3.2. Economies with heterogeneous capital goods. The wage curve
In order to show the results of ‘reswitching’ and ‘reverse capital deepening’ it is
assumed an economy with the following characteristics.

• Yearly cycles of production of the consumption good and of the capital
good.
• Capital is circulating capital.
• Constant returns to scale.
• No scarcity of natural resources.
• The wages are paid at the end of each productive cycle.
• There exists a surplus obtainable over the pure replacement of machines:
that is to say, the economy is viable, hence the proportion of the capital
good entering the production of the capital good itself must be lower than
unity (i.e. c
c
<1).
8

• The capital good enters the production of both commodities.
• Some positive amount of labour is required directly or indirectly to
produce both commodities.

Let us first consider production with only one system of production. Each system
consists of two ‘methods of production’, that is, one method to produce the
consumption good, A, by means of fixed quantities of labour (l
a
) and of the capital
good, C, (c
a
), and the other method to produce the capital good by means of fixed
quantities of labour (l
c
) and of the capital good itself (c
c
).
9

Positing the equilibrium conditions of a uniform rate of interest and a uniform
wage we shall be able to write the following two price equations:

p
a
=1= l
a
w+c
a
p
c
(1+r)
p
c
=l
c
w+c
c
p
c
(1+r) (3.1)

where p
a
and p
c
are the consumption good’s and the capital good’s prices respectively.
As usual, w is the wage rate in terms of the integrated industry of the consumption good
and r is the interest rate in the same terms. The consumption good has been considered
as our numéraire, hence p
a
=1.
There are two equations and three unknowns (p
c
, w, r). The degree of freedom will
allow us to take one of the distributive variable as independent (r from now on), and
then both w and p
c
will be univocally determined for any given value of r.

8
Lower than one because we assume capital goods to be circulating. (See Garegnani 1970a, p.
408, n. 1.) We shall not deal with fixed capital since our aim is basically to show these results of
the debates as clear as possible. The reader may be referred to Sraffa (1960). For an analysis of the
historical aspects of fixed capital and its distinction from circulating capital see Varri (1998).
9
Other systems of production involve, apart from homogenous labour that is used by the different
systems, different capital goods for the production of the consumption good A.
Chapt er 3 – ‘ Reswi t chi ng’ and ‘Reverse Capi t al Deepeni ng’
43
From this system it can be obtained the following relation between the wage and
the interest rate corresponding to the current system of production:

1 (1 )
[ ( )(1 )]
c
a c a a c
c r
w
l l c l c r
÷ +
=
+ ÷ +
(3.2)
10


From equation (3.2), and for any value of r, there corresponds a unique value of w,
hence (3.2) is a function. Moreover, one of the distributive variables is negatively
related to the other.
11
A rise in r (the rate of interest), while the other elements remain
constant, will make the numerator of (3.2) fall, whereas the denominator will rise:
consequently w (the wage rate) will fall.
12

To compute the value of capital per worker, it may be useful to introduce the
integrated industry of the consumption good. This imaginary industry is such that the
proportion, K, of the capital industry relative to the consumption good industry is the
one which ensures the replacement of the capital goods consumed in the integrated
industry. Accordingly, this industry must re-integrate the quantity of the capital good
necessary for producing the consumption-good directly (c
a
), and the quantity which is

10
From the second equation in (3.1) we obtain
1 (1 )
c
c
c
l w
p
c r
=
÷ +
(3.3). Then by substituting p
c

on the Right Hand Side (RHS) of the first equation of (3.1) for (3.3), we obtain (3.2), which is
equivalent to
1 (1 )
(1 )
c
a
c r
w
l D r
÷ +
=
+ +
(3.2’), where D is the determinant of the productive coefficient
matrix, that is
c a
c a
l l
D
c c
= . This determinant is central to the entire analysis, in particular, for the
graphical representation of the relation between w and r which allow us to easily grasp many of
the results of ‘reverse capital deepening’ and ‘reswitching’.
11
Cf. equation (3.9) n. 16, this chapter, below.
12
Likewise, it can be verified that the price of the capital good (p
c
) is positive for every 0 r R s s
(where R is the maximum rate of profits obtainable when w=0 then
1
1
c
R
c
= ÷ ). In fact from
(3.2’) and (3.3) we can obtain
(1 )
c
c
a
l
p
l D r
=
+ +
(3.4) . Then if 0 D > hence p
c
> 0 for 0 r > .
And if 0 D < , then p
c
> 0 too. In fact, proving by contradiction we may consider p
c
be negative;
in this case denominator of (3.4) should be negative (since l
c
is positive throughout). Then
1
a
l
r
D
> ÷ ÷ but we know that r R s so 1
a
l
R
D
> ÷ ÷ , then
1
1 1
a
c
l
c D
÷ > ÷ ÷ hence
a c
D l c < ÷ and hence finally 0
c a
l c < , but this is impossible since l
c
and
c
a
are both positive. So p
c
>0 for every 0 r R s s .
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

44
necessary for producing itself (c
c
) – that is the proportion of the capital good consumed
in the production of the proportion of the capital good necessary for producing the
consumption good. In formal terms:

K=c
a
+c
c
K
Then,
1
a
c
c
K
c
=
÷
(3.5)

Equation (3.5) provides the quantity of capital of the integrated industry of the
consumption good. Recourse to the integrated industry is necessary if the net product
(that is the product to be divided between wages and interest) is to consist of a physical
quantity of the consumption good. Therefore, the net physical product (of the integrated
industry) per worker,
13
q, can be measured as follows:

1
a c
q
l l K
=
+
(3.6)

Then, by replacing (3.5) in (3.6), we obtain:

1 1
(1 )
1
c
a c a c a
a c
c
c
q
l c c l c
l l
c
÷
= =
÷ + | |
+
|
÷
\ .
(3.7)

But, (3.7) will be equal to (3.2) or to (3.2’), whenever r=0.
Hence:

1
( 0)
(1 )
c
a c a c
c
w r W q
l c c l
÷
= = = =
÷ +
(3.8)
14


If r is zero then the segment 0W measuring the maximum wage rate (see Figure
3.1 below) will also measure the net physical product of the integrated industry per
labourer. If the wage rate is w
1
and the rate of interest r
1
, the value of capital per worker
can be measured in the following way:

13
Consumption-good production is normalised to one so q is one unit of net A divided by the
quantity of labour employed both directly and indirectly in the production of such a unit of net
physical output.
14
Since
1 (1 )
[ ( )(1 )]
c
a c a a c
c r
w
l l c l c r
÷ +
=
+ ÷ +
with a zero r becomes equation (3.8), where W is the
maximum wage, and equation (3.8) is equal to equation (3.7).
Chapt er 3 – ‘ Reswi t chi ng’ and ‘Reverse Capi t al Deepeni ng’
45
Since w
1
W indicates the quantity of consumption good received as profits for each
worker employed in the integrated industry, and 0r
1
on the other hand indicates the
level of the interest rate, then the value of capital k can be obtained as the ratio
between:
15


k = Interests per worker / Rate of interest

The system of production gives the relationship between r and w.
16
We shall label
this relationship as the wage-curve,
17
whose graphic can be obtained from equation
(3.2), and is drawn in Figure 3.1.


Figure 3.1. The wage-curve: 0W is the net physical product per worker; tan(u) is the
value of capital per worker (k) when the wage rate (w) is 0w
1



15
Mathematically, the way to obtain it is by measuring the trigonometric tangent of the angle u –of
the triangle formed by points P, W and w
1
. (See Figure 3.1). We have in the abscissa the interest
rate value 0r
1
, while in the vertical axis, correspondingly, the value of the interest per worker
(w
1
W). Then, the trigonometric tangent of angle u will give us:

1 1
1
1 1
tan( ) tan( )
0
wW wW
w PW
w P r
u = = = (P is point w
1
r
1
), and then, tan(u)=tan (w
1
PW) = interest per
worker / rate of interest = value of capital per worker, and this is the result at which we wanted to arrive.
16
The inverse relationship between w and r in (3.2) is shown by its graphic representation. Some
formal (and economical) considerations are in order: if, as we have already seen, r=0 then w=W.
On the same assumptions, as r rises from zero, w falls reaching zero when r reaches R – its
maximum value. And this is so because of the assumptions here assumed for the economy, which
ensure that w be a continuous differentiable function of r. In fact, the inverse relationship between
w and r can be shown by differentiating (3.2’):
2
[ (1 )] [1 (1 )] ( )
[ (1 )]
c a c
a
c l D r c r D dw r
dr l D r
÷ + + ÷ ÷ +
=
+ +
=
2
[ (1 )]
c a
a
c l D
l D r
÷ ÷
+ +
2
[ (1 )]
a a
a
l c
l D r
÷
=
+ +
(3.9)
which is negative provided that the production coefficients (l
a
, c
a
) are both positive.
17
Garegnani 1970a. Other authors have called this relationship in different ways, e.g. Samuelson
(1962) named it “factor-price frontier”.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

46
In Figure 3.1, it has been assumed D<0. If D=0, then the function (3.2), or (3.2’),
would define a straight line. If D>0, the price of the capital good will fall as the rate of
interest rises, and the form of the wage curve will be convex.
18

3.3. The wage-frontier. Economies with more than one system of production
Thus equipped with the fundamental relationships between the wage rate, the interest
rate, the value of capital and the net output per worker, it may be considered two
systems of production, for instance o and |. Each system involves two methods of
production – one for the consumption good, one for the capital good.
19


18
In other words, the form of the function will crucially depend on the sign of determinant D (i.e.
on the relative factor coefficients in the production of each of the goods). In fact, by differentiating
equation (3.3) with respect to the rate of interest we obtain:
2
[ (1 )]
c c
a
dp l D
dr l D r
÷
=
+ +
…..(3.10).
The case D>0 (equation 3.10 is negative and will indicate that the price of the capital good per
worker increases as the rate of profits falls) was called in the debates a negative ‘price Wicksell
effect’. Instead, if D<0, the price of the capital good will rise as the rate of interest rises, and the
form of the w curve will be concave. This was called a positive ‘price Wicksell effect’. Lastly, if
D=0, the price of the capital good does not change as the rate of interest changes, and the form of
the wage curve will be a straight-line (a neutral ‘price Wicksell effect’). The terminology here
adopted is that used by Burmeister (1987). On the other hand Harcourt (1972, pp. 39-45) has
instead preferred to call a negative price ‘Wicksell effect’ the case when D<0 and vice versa. The
Wicksell effect was a term coined during the controversies by Robinson (1953), who actually
picked it up from Uhr (1951). See also Swan (1956), Robinson (1958), Bhaduri (1966), Pasinetti
(1978), Garegnani (1984b). The connection between changes in the value of capital when
distribution changes with the name of Wicksell can be seen in terms of the RHS of equation 4.B in
Appendix B; for instance if w rises – assume that the rate of rent does not vary for the sake of
exposition – then the interest rate must fall. The point is that the absolute increment in wages may
be higher, or lower (or equal) than (to) the absolute decrement in interest and hence the value of
capital (on the RHS of the equation) may increase or decrease (or remain unchanged) as the rate of
interest fall, even when the physical quantities there involved vary in a direction in accordance
with the theory (A
1
rises, etc.) These problems regarding changes in the value of capital due to
changes in distribution are discussed by Wicksell in his Lectures (1934, pp. 149-passim and pp.
180-181) where the Swedish economist argues against von Thünen’s theorem that holds that the
‘marginal productivity of capital’ equals the interest rate. Wicksell shows in those pages that due
to “unproductive absorption” of the increased output von Thünen’s theorem is false.
19
In Alpha system the coefficients regarding the methods of production of the consumption good
are (l
ao
, c
ao
) and the ones regarding production of the capital good are (l
co
, c
co
). For the case of
Beta system we have (l
a|
, c
a|
) and (l
c|
, c
c|
). The curvature of each wage-curve corresponding to
each system crucially depends on the value of the determinant D, which actually depends on the
value of the production coefficients of each system. For example if c
oc
/l
oc
>c
oa
/l
oa
, that is a rise in
interest cost affects the production of the capital good more than that of the consumption good –
then D<0. Note that we can confront both of the relations referred to, since the capital good used
in both sectors is the same, that is c
oc
and c
oa
are proportions of the same capital good C
o
(the
capital good associated with system o).
Chapt er 3 – ‘ Reswi t chi ng’ and ‘Reverse Capi t al Deepeni ng’
47
Under long period conditions, producers’ plan as to which technique
20
s/he might
choose depends on the costs associated with the production of the consumption good.
The cost of the consumption good will in turn depend on the price system which
happens to be in use.
If system o is in use, for a given r we have w
o
and p
co
, then p
c|
(o)
and p
a|
(o)
can be
calculated at that r and at w
o
(prices of capital good and consumption good using | -
technique at prices and w of system o). If, on the other hand, | is in use instead of o at
that level of r, we have w
|
and p
c|
, then p
co
(|)
and p
ao
(|)
can be calculated at that r but
at w
|
. In other words, the costs of the consumption good will be different according to
which system happens to be in use. However it can be shown that producers’ tendency
to switch from one system to another more convenient (cheaper) one in the existing
price situation will bring them to the one giving the highest wage rate at the given rate
of interest.
21


20
The term technique is used as synonymous of system involving two methods of production.
21
Assume system o to be in use:
System o
p
ao
=1
p
ao
=l
ao
w
o
+c
ao
p
co
(1+r)
p
co
=l
co
w
o
+c
co
p
co
(1+r) (3.11)
p
a|
(o)
=l
a|
w
o
+c
a|
p
c|
(o)
(1+r)
p
c|
(o)
=l
c|
w
o
+c
c|
p
c|
(o)
(1+r)
The last two equations of system o give us the estimated costs both of the consumption and of the
capital goods with system | but under the prices and wage of system o.
In case | is in use we have:
System |
p
a|
=1
p
a|
=l
a|
w
|
+c
a|
p
c|
(1+r)
p
c|
=l
c|
w
|
+c
c|
p
c|
(1+r) (3.12)
p
ao
(|)
=l
ao
w
|
+c
ao
p
co
(|)
(1+r)
p
co
(|)
=l
co
w
|
+c
co
p
co
(|)
(1+r)

The last two equations of system | give us the estimated cost of both the consumption and the
capital goods with system o but under the prices and wage of system |. The producer knows that
if system o is in use, and p
a|
(o)
<p
ao

is verified, then there would exist a convenience to change to
the method of producing A given by system |, since the cost of producing the consumption good is
lower than the current price p
ao

. On the other hand, if technique | is in use, and p
ao
(|)
<p
a|

is
verified, then he would change technique | for technique o. So it would seem to appear that, a
priori, given two methods of production, the choice will depend upon where the producer happens
to be initially.
However, if we consider each system separately, for instance system o, we shall have four
independent equations to solve for five unknowns. The problem here is that we cannot determine
the value of those unknowns, since p
a|
(o)
and p
c|
(o)
, as such, are qualitatively different from the
remainder unknowns (p
co
, w
o
) because the system used to produce both A and the capital good,
that is the system underlying unknowns p
a|
(o)
and p
c|
(o)
, is |, not o. The same applies to both p
ao
(|)

and p
co
(|)
in case we consider system |.
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

48
Since producers will change systems according to this rule, it can be generated an
outer envelope (Northeast envelope) by intersecting wage-curves of the two systems of
production considered in the present exposition, and will represent the relation between
r and w. This curve may be labelled as the wage-frontier, which has a ‘corner’ where a
switch occurs (‘switching point’), and in the corner itself the two systems can co-exist
(“adjacent”
22
systems). (See Figure 3.2).
23




Figure 3.2. The wage-frontier (thicker line) and ‘reswitching’ of techniques. At r=r
2

producers switch back to technique o




Even though we cannot determine the value of the unknowns, we can determine the ratios between
them, since both systems are structurally identical. By considering the relations between the
unknowns:
( ) ( ) ( ) ( )
: : : : : : : :
a c a c a c a c
p p p p w p p p p w
o o o | | |
o o | | | | o o
= (3.13)
and by discarding each of the first equation of both systems, we have:
w
o
/w
|
=p
ao
/p
ao
(|)
(3.14)
w
o
/w
|
=p
a|
/p
a|
(o)
(3.15)
Whenever the ratios (3.14) and (3.15) are equal, the cheaper system will be the same at both wage
rates and price systems. Thus if w
o
>w
|
, then p
a|
>p
a|
(o)
i.e. system o is cheaper for producing the
consumption good than |, and allows at the same time to have a higher wage. This is the objective
method. Note that in the process of switching, entrepreneurs who first happen to switch to the most
profitable technique might get some extra profits. But when the economy has already adopted the
cheapest method, no more extra profits can be obtained. See: Garegnani (1970a), Pasinetti (2003),
Kurz and Salvadori (1997).
22
Samuelson (1962, p. 198); Garegnani (1970a, p. 412).
23
This envelope could also be generated by intersecting a family of systems of production. See
Chapter 4 below.
Chapt er 3 – ‘ Reswi t chi ng’ and ‘Reverse Capi t al Deepeni ng’
49
3.4. Implications for the basic premises of the theory: Conclusion
With the aid of the wage-curves and the wage-frontier the intuitive explanation of
‘reswitching’ introduced in this chapter may be reinforced by analysing the relations
between the net physical product, the wage rate, and the value of capital per worker as
the rate of interest changes.
24

For our purposes here it will be sufficient to consider what happens to the value of
capital per worker as the rate of interest changes. Assume that, as required by the
theory, an a priori criterion to establish which technique is more (less) capital intensive
might be chosen. Assume that a method is ‘more’ capital intensive because it is the
method yielding the highest net product per worker of the integrated industry. Assume
that o gives the highest net product and is a straight-line curve (i.e. D=0),
25
while the
second technique, |, which is concave in its entire trajectory (i.e. D<0), yields a lower
net product and intersects system o at two points.
26
Then, according to the method
explained above (cf. n. 21, this chapter, above), when 0<r<r
1
technique o (‘more
capital’ intensive, as assumed) is chosen; then as r rises further such that r
1
<r<r
2

technique | (‘less capital’ intensive) is chosen; so far neoclassical long period technical
choices go in direction as hypothesised by the theory, namely: the higher the rate of
interest, the lower the ‘capital’ intensity. However, if r increases further such that r>r
2
,
it will be found that the ‘more’ – and not the ‘less’ – capital intensive technique is
chosen according to the rule governing the choice of techniques.
27
(See Figure 3.2)
Thus, even if a criterion could be found so as to order ‘intensities’ according to
different levels of distribution, the ordering itself will change with changes in
distribution.
28


24
It is also possible to consider w as the independent variable, and the results will be as similar as
those reached when we regard r as the independent variable.
25
The purpose pursued by Samuelson (1962), who introduced this case, are analysed in Chapter 4
below.
26
On the issue of the number of switch-points cf. Bharadwaj (1970).
27
In the text it has been illustrated ‘reswitching’ undermining the ‘direct’ factor substitution
mechanism, which is derived from the group (ii) of the data. As to the ‘indirect’ mechanism,
consider two goods (corn and cloth) and decide, whichever the criterion adopted, which
consumption-good is more ‘capital’ intensive. Then, whichever the more ‘capital’ intensive
consumption-good, there will be one or more ranges of values of the interest rate, where an
increment of r makes the more ‘capital’ intensive good relatively cheaper, implying a shift of the
composition of consumption in the direction of an increase, rather than a decrease, of the average
‘capital’-intensity of the demand for consumption goods. Cf. Appendix D where it is also
illustrated ‘perverse’ behaviour when there is no ‘reswitching’.
28
In the following chapter we shall survey the views of critics and defenders of the theory at stake
on these results. In the controversies many examples have been built (cf. Symposium 1966:
Pasinetti 1966a; Garegnani 1966 and see also Levhari 1965) in order to generate concave or
convex, or concave and convex curves if each system involves one or more capital goods in such a
way as to give rise to more than one intersect between different wage curves. We shall also see in
that chapter, that Samuelson’s (1962) attempt at rescuing neoclassical theory implies the fact that
each wage curve intersects each other only once.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

50
In the foregoing analysis capital-inputs have been measured in technical units
independent of distribution; yet there is no guarantee that changes in the rate of interest
will cause changes in the opposite direction in the quantity of capital per worker
29

employed in the economy. However, some might argue that the substitution
mechanisms work ‘in the right direction’ when capital is measured straightforwardly in
value terms, thus raising the claim that the rate of interest might bring investment
(capital demand) into equality with full-employment savings (capital supply).
30

However, in the light of ‘reverse capital deepening’ and ‘reswitching’, there seems to
be no guarantee to derive a downward-sloping demand curve (or employment curve for
capital). To show that the demand for capital might be directly related to r, suffice it to
use the graphical device below (Figure 3.3) which corresponds to values of Figure 3.2.



Figure 3.3. ‘Reverse capital deepening’: k (capital in value) is not inversely related to r

As can be seen from Figure 3.3, the ‘employment curve’ for capital (a basis to derive
the demand function for capital) is no longer inversely related to the rate of interest.
31-32


29
Measured in terms of the consumption good of the integrated industry.
30
Cf. Paragraph 2.10, Chapter 2, above.
31
For
1
0 r r s s there is a vertical curve indicating a constant value of capital due to the straight-
line wage curve (o is in use, cf. Figure 3.2) which corresponds to these levels of the rate of
interest. Then, when r=r
1
, along the wage frontier there is a switch point; hence both systems can
co-exist (adjacent systems) and the value of capital per worker assumes different levels at that rate.
Then, as r increases from r
1
up to r
2
, system | is in use and the value of capital per worker
gradually increases because the value of each successive tangent along the | wage curve increases
as r increases until r
2
. At r=r
2
, there is ‘reswitching’ of techniques. At that point both techniques
can co-exist, and that is why the value of capital assumes different levels at that rate. Then, as r
rises from r
2
up to R
o
the value of the capital is constant, since the o wage curve associated to that
range of values of r is a straight-line – hence o is being used again, hence ‘reswitching’.
32
Some (e.g. Solow 1967, p. 30) have argued that the form of the demand curve with ‘reverse
capital deepening’ and the discontinuities in the values of k is due to discrete technologies; so if
(cont.)
Chapt er 3 – ‘ Reswi t chi ng’ and ‘Reverse Capi t al Deepeni ng’
51
So far it has been seen that even if the illegitimacy of the supply aspect of the
problem of capital (the necessity to conceive of capital in value terms to explain long
period supply-and-demand equilibrium, expressed, in the capital market, through the
equilibrium condition of the uniformity of the rates of return on the capital goods’
supply prices) is put aside, ‘reswitching’ and ‘reverse capital deepening’ undermine the
demand aspect so that it turns out at least doubtful to believe in an inverse relationship
between the rate of remuneration and the quantity of the factor employed. But, if we are
reminded of the key role played by the downward-sloping demand curves so as to
provide a plausible basis for the theory to postulate a stable and unique long period
equilibrium, then it would not be so difficult to envisage the scope of the implications
of the Cambridge (UK) critique based on ‘reswitching’ and ‘reverse capital
deepening’.
33

The results described in this chapter cast serious doubt on the issue concerning how
the distribution of the social product between wage-earners and profit-earners can be
explained. But these results do not arise out from outside the neoclassical assumptions.
On the contrary, the critical side in the arguments between the two Cambridge has
shown, by using the analysis of production theory (minimisation of costs), that
distribution might be explained by forces other than the supply and demand ones – the
explanation of which we have seen in Chapter 2 above. On the grounds of supply-and-
demand theory, in economies with heterogeneous capital goods the demand curves
could be of any form, not necessarily negatively sloping. As we have also seen in this
chapter, supply and demand forces were conceived by the founders of the marginal
theory to operate in markets under free competition by relying on the belief of the
factors’ substitution principle. We thus conclude this first part of this study with the
following question: Could we then expect from this theory predictions and forecasts on
real economies, if e.g. demand curves are not downwards-sloping? Can we rely on the
factor substitution principle in order to determine the distribution of the social product
between wages and interests?
As we shall see in the second part of this study, the theoretic possibilities of
‘reswitching’ and ‘reverse capital deepening’ have serious implications for the


the number of techniques becomes larger and larger, switch from a technique to another would
imply an infinitesimal change in the production coefficients (l
ao
, c
ao
, l
a|
, etc.) so as to ensure such
a continuous relation between k and r (and q) along the technical frontier (the wage frontier) that
the neoclassical relations could be sustained. However, it has been shown by Bellino (1993) that
the same wage frontier – which entail ‘enveloping’ wage curves and would have ensured changes
in the coefficients so as to keep neoclassical relations among variables – could also be generated
by other wage curves, without however implying changes in those coefficients, and which do not
necessarily entail a neoclassical relation between k, r and q. Cf. also Pasinetti (1969, p. 523) and
Garegnani (1970a, pp. 412-414.)
33
The author would like to highlight that the negative implications of these results for the theory
have nothing to do with the utilisation of any ‘aggregate’ production function, as Joan Robinson
has continuously insisted. See Chapter 5 below where Joan Robinson’s role in the Cambridge
debates is surveyed. Cf. also Paragraph 4.6, below.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

52
dominant theory, which one may appreciate as soon as it is realised why notable
neoclassical scholars, such as Samuelson, got in the capital debates. As the following
chapter will discuss, this author’s participation together with the replies and
counteractions it gave rise constitute, from a historical standpoint, the heart of the
Cambridge controversies in its first phase of its development.



PART II








The Capital Theory Debates in a Historical Perspective
and the Transformation of Dominant Theory



Chapter 4
A first phase in the controversies: Samuelson’s 1962
intervention, the 1966 Symposium, neoclassical reactions
and the highest point of the debates
“No alchemist can turn one capital good into the other”
P. Samuelson (1962) p. 196.


4.1. Introduction
In the previous chapter, one of the most relevant results of the controversies has been
introduced, namely: ‘reswitching’ of techniques and ‘reverse capital deepening’. We
have seen that these phenomena seriously concern the core foundations of neoclassical
theory: it cannot in general be postulated a negative relation between demand for
capital and the rate of interest. The aim of the previous chapter, it is fair to say it, has
been to introduce the reader to the analytical grounds on which the controversies
evolved, so as to understand the arguments raised by the participants in the successive
debates in the light of those results.
Although by the beginning of the 1960s the most salient arguments raised in the
debate were not clear with respect to the traditional role ascribed to ‘capital’ as a factor
of production in deriving the equilibrium supply-and-demand functions for that ‘factor’,
serious doubts concerning its theoretical tenability had started to arise.
1
As we shall see,
the way in which Joan Robinson sparked the ensuing controversies in her famous, 1953
article much characterised the debates that immediately followed it.
2
In particular, the
first critiques in the controversies were mainly concerned about the use of aggregate
production function as an instrument to determine distribution by the partial derivatives

1
Eventually the period previous to 1960 (i.e. Sraffa’s publication date) may be characterised as a
‘pre-reswitching’ period of the debate – although, of course, the phenomenon had already been
noticed by Robinson (1953; 1956) and Champernowne (1953), but little importance was given to it
by these and other authors at that time. See Chapter 5 where we treat Robinson’s role in the
controversies as well as Champernowne’s contribution.
2
Robinson (1953).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

56
of that construct.
3
Yet, it had not been discussed in a clearer ground the consequence
for the marginal theory of distribution the fact that, as Sraffa demonstrates in his 1960
book, it is impossible to talk about ‘more’ or ‘less’ capital-intensive production
processes independently of distribution. Indeed, the marginal theory relies on the factor
substitution principle, a theoretical principle that should indicate the cheaper or dearer
production processes of the net output according to ‘relative factor’s intensities’ as
distribution changes. But, due to the phenomena examined above (Chapter 3), it turns
out to be impossible to specify, independently of distribution, which production process
is more or less ‘capital’ intensive among the alternative productive processes, because
the very classification of techniques according to ‘factor intensities’ changes as the
distribution of income between wages and profits changes too. This problem, we
submit, is totally independent of the aggregate production function and shows that a
negative relation between ‘capital’ and the rate of interest cannot in general be
postulated, thus casting serious concerns on the determination of a ‘quantity of capital’,
which is however essential to the traditional theory. That is why the present author has
deemed more appropriate to deal with the first phase in the capital debates by
discussing the contributions which appeared in the 1960s decade, rather than having
adopted a chronological standpoint onto the discussions. (However, we shall turn back
to Robinson’s criticisms and her immediate contestants of the 1950s in the following
chapter).
The aim of this chapter is therefore to put at the centre the salient exchanges that
surrounded the phenomena of ‘reswitching’ and ‘reverse capital deepening’ since
Samuelson’s intervention in the controversies in 1962.
4
In the wake of Samuelson’s
article the debates followed an interesting path both from a historical as well as from an
analytical viewpoint. As to the first aspect, Samuelson’s article appeared soon after
Sraffa’s book; while, as to the latter aspect, Samuelson laid the ground for the
discussion by presenting systems of production with heterogeneous capital goods –
though from a particular viewpoint as we shall see below. Also interestingly is to note
that Samuelson’s article triggered off a critique by Garegnani, who, having visited MIT
at that time, had criticised the former in an oral intervention around 1961-62 in MIT,
where a manuscript of Samuelson’s article circulated. It was there that Garegnani
developed his criticisms, which finally reached readers somewhat eight or nine years
later in (1970a)
5
(Paragraphs 4.2-4.6).
Also, this chapter deals with the reactions of neoclassical economists to the
criticisms raised in the 1966 Quarterly Journal of Economics Symposium on paradoxes
in capital theory (Pasinetti 1966a; Garegnani 1966), which came into being thanks to
the intervention of one of Samuelson’s go-between at that time: Levhari (1965), who
pretended to have shown that ‘reswitching’, though possible at the level of single

3
Champernowne (1953) tried to save the production functions from being a multi-valued relation
due to value measurement of capital by the ‘chain index’ measurement of capital, a measurement
which, however, is also dependent on distribution. See Chapter 5, in particular Paragraph 5.6 below.
4
Samuelson (1962).
5
See n. 26, this chapter, below.
Chapt er 4 – A f i rst phase i n t he cont roversi es
57
industries, cannot arise for the economy as a whole (Paragraphs 4.7-4.10) This was
contested in such a detail and accuracy by the critical side that ‘reswitching’ turned out
to be issue for the discipline. Some successive reactions (Samuelson; Ferguson; Bliss)
after the Symposium and Garegnani’s 1970 critique are analysed (Paragraphs 4.11-
4.15).
4.2. Samuelson’s (1962) model with heterogeneous capital goods
Although Samuelson’s contribution is a direct response to Robinson’s concerns about
the theoretical tenability of the production function – in fact Samuelson dedicates his
paper to Joan Robinson’s “memorable 1961 visit to MIT”
6
– it is important to
emphasise the fact that in Samuelson’s contribution one can find the discussion of the
relation between distribution and demand for factors in heterogeneous capital goods
economies on a ground which is clearer than the ground on which Robinson (and her
immediate contestants) had put it.
7
Indeed that ground was the one on which Sraffa had
already laid the argument of the choice of techniques in his 1960 book, where the
important result of ‘reverse capital deepening’ emerged out of the analysis of the
relation between prices (choice of techniques) and distribution. (See Appendix C
below.)
That we have decided to deal with Samuelson (1962) in this chapter (which is the
first of the second part of the study) is justified by the fact that in that article it is
possible to find one of the most serious attempts from the neoclassical camp to deal
with heterogeneous capital goods without any recourse to a production function, and
that in Samuelson’s surrogate note there is the subtle message that heterogeneity of
capital goods could not bring big headaches to the traditional theory. A similar view
can be found in King (2002) who writes:

Ironically the first to expose the depth of the Sraffian critique were defenders of
the neoclassical theory. […] [It was] Samuelson’s attempts to solve [Sraffa’s
results on reswitching and reverse capital deepening which were pointed out by
former PhD student in Cambridge, P. Garegnani] that initiated the Cambridge
Controversies.
8



6
Samuelson (1962, p. 192, n. 1).
7
For the way in which discussions evolved over the first stages of the debates see Chapter 5
below. See Chapter 3 above, where we have introduced the ground for the forthcoming discussions
in these debates.
8
King (2002, p. 95).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

58
Samuelson starts his note by pointing out that capital theory can be rigorously
developed without any recourse to a ‘Clark-like’ concept of aggregate capital.
9
The
attempt pursued by Samuelson is to show that, in order to determine distribution,
economists can use a “new concept”, the “surrogate production function”, which is
supposed to consider the heterogeneous capital goods involved in the different
techniques of production. Moreover, he claims that by using the surrogate production
function and the associated “surrogate capital” one can predict exactly how
“complicated heterogeneous capital models will behave”.
10

Thus, Samuelson begins with the introduction of a heterogeneous-capital-good
model:

I begin with a concrete model in which there are a great variety of capital goods:
call them: alpha, beta,… 999 or anything else.
11


and in the same page some rows below he explains (emphasis in the original):

One need never speak of the production function, but rather should speak of a
great number of separate production functions.

In Samuelson’s model there are concrete different capital goods that can produce
an output composed of a consumption-good and of a capital-good in fixed proportions
with homogeneous labour. Samuelson’s model is, apparently, quite similar to that
present in Sraffa (1960) though there are quite important differences which are worth
mentioning. In Samuelson’s model there is a fixed coefficient production process for
each component of output. Both production processes encompass the system of
production of the economy where constant returns to scale and perfect competition
prevail. Samuelson generalises the fixed-coefficient model for any system of
production, so that the consumption good can be produced with homogeneous labour
and a particular capital good associated with each system of production, while each
particular capital good is produced by itself and labour. Then he assumes that each

9
‘Clark-like’ concept of ‘capital’ refers to the idea that the production of net output of an economy
can be represented by a ‘production function’ whose argument contains homogeneous labour and
‘capital’ (in terms of output) such that relative rates of remuneration can be determined by the
partial derivatives of such a function with respect to each factor. Concepts like ‘Clark-Ramsey
parable’, ‘Clark parable’ and the like refer to the same idea.
10
Samuelson (1962, p. 194).
11
Samuelson (1962, p. 194).
Chapt er 4 – A f i rst phase i n t he cont roversi es
59
capital good does not enter the production of any other capital good except for the
production of itself.
12

As is well known, different production systems will in general imply different
wage-curves or “Factor-price frontiers” as Samuelson has put it. Different wage-curves
may or may not intersect, depending crucially upon the relative production coefficients.
Now we must discuss a crucial assumption in Samuelson’s model.
4.3. Samuelson’s ‘special sub-class of realistic cases’
To be coherent with the hypothesis of production with strictly heterogeneous physical
capital goods, Samuelson puts forth a “special sub-class of realistic case”
13
in which
every capital good is different according to the different systems of production, and
enters the production process in fixed proportions with labour. In this regard Samuelson
makes a highly restrictive assumption: that the proportion of labour to the means of
production for the system of production concerned must be the same for all goods
produced, i.e. for the final consumption good and for the capital good (l
a
/c
a
=l
c
/c
c
).
14

No doubt, then, that the “factor-price frontier” built for a single capital-good
technique where the capital-labour ratio is constant as distribution changes must result
in a straight line. The straight line connects two points, namely, the maximum wage
rate for the process at zero profit rate, and the ratio of technical surplus over
replacement to means of production when the wage rate is zero. As the labour-means of
production ratio is assumed to be equal in both sectors of the economy, there will only
be one possible capital-labour ratio in the economy as a whole. Then the value of
capital is constant and independently of changes in distribution. Samuelson finally
extended his restrictive assumption to all systems of production in the economy (“book
of blueprints”).
15
Then, as already explained, the points on the ‘north-east frontier’

12
Samuelson’s model is similar to that discussed in Chapter 3 with the difference that we have
considered only circulating capital whereas Samuelson considers fixed capital, though his treatment
of fixed capital evades the problem of joint production. See Garegnani (1970a, p. 408, n. 1).
13
Samuelson (1962), p. 196.
14
Samuelson (1962), pp. 196-197. Cf. Chapter 3, above. The case of the straight-line wage curve
is D=0. It must be noticed that not only is Samuelson’s model quite arbitrary because of the
equality of the capital-labour ratio in each industry, but also because he assumes that every activity
needs just only one particular capital good – a very unrealistic case. The fact that in Samuelson we
have switch points in his very special cases is due to the existence of only one good common
among techniques, also a very unrealistic case; on this issue see Baldone (1984).
15
Thus one can plot in the wage-profit rate layout the different straight-line “factor price frontiers”
according to each system of production so that, if a range of production systems (each involving
only one capital-good) is considered, each with either a higher proportion of ‘capital’ to labour, the
“envelope” (Samuelson 1962, p. 198.) of the outer portions of the straight-lines will depict the
most profitable techniques at any profit rate. (See Figure 4.1.)

(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

60
envelope, i.e. the wage-frontier, are of equilibrium. Because of the particular
assumption made by Samuelson this envelope, for a finite number of techniques, has
straight segments (when one technique is the one most profitable) as well as corner
points (‘Switch points’, ‘adjacent techniques’) where two techniques can coexist. (Cf.
points A, B, or C of the right-side graph in Figure 4.1) The switch points on
Samuelson’s envelope have the basic neoclassical property of being all ‘forward
switches’, i.e. as the profit rate rises the capital-labour ratio falls.
The successive step that Samuelson took was to associate the Clark-Ramsey
parable with his ‘more realistic’ heterogeneous capital-good model. Samuelson, after
having drawn the envelope corresponding to the book of blue prints (see Figure 4.1.b
above), maintains that the envelope generated by straight-line wage curves is
“generally similar”

to the function relating the wage rate and the profit rate derived
from the “Clark-Ramsey parable”.
16
Samuelson then claims that

if we invent the right fairy tale, we can come as close as we like to duplicating the
true blue-print reality in all its complexity.
17


4.4. The neoclassical ‘parable’: ‘Surrogate’ capital and the ‘Surrogate’ production
function
Did Paul Samuelson succeed in inventing the ‘right fairy tale’? Samuelson believed in
having found a rationalisation of the ‘Clark parable’ because the envelope generated
by the latter (see Figure 4.2.c) is generally similar to the envelope derived from his
supposedly ‘real’ heterogeneous capital goods model. (Cf. Figure 4.1) Thus, he



Figure 4.1. A straight wage-curve and a family of straight wage-curves

16
Samuelson (1962), p. 201.
17
Samuelson (1962), p. 201, emphasis added.
Chapt er 4 – A f i rst phase i n t he cont roversi es
61
pretended to have supplied a demonstration that the formal properties of the ‘parable
system’ are the same as those derived from his “quasi-realistic complete system of
heterogeneous capital goods”. That is why Samuelson believes that his “new concept”
18

of Surrogate production function, with its associated Surrogate capital,
19
which
replicates the parable, could be a useful tool for analysis of heterogeneous capital
models.
It is correspondence between the formal properties of both constructs – that is an
‘imaginary’ surrogate function (or parable)
20
and the ‘real’ economy of straight wage-
curves – that quickened Samuelson to conclude:

18
Samuelson (1962), p. 201.
19
Samuelson (1962, p. 200) assumed that labour and homogeneous physical capital (jelly) produce
a flow of homogeneous national product, which consists of consumption goods or of net capital
formation “the two being infinitely substitutable on a one-for-one basis”
20
Samuelson (1962, p. 200). The ‘Clark parable’ entails using a production function, which
represents the whole techniques of production of the net output that is produced with labour and
‘capital’ homogeneous with the output. Samuelson (1962, p. 201) thinks that ‘K’ is surrogate
capital “that is to go into the Surrogate Production Function that will predict all behaviour.”


Figure 4.2. The ‘parable system’

Samuelson pretended to analyse the relations between the wage rate, the profit rate and the net
output per worker that prevail in the ‘real’ economy by analysing the properties of the surrogate
production function. In formal terms we have:

( , ) Y f K L = (4.1)
because of constant returns we may re-write the function as follows:

( ) y f k = (4.1’)
where y is net output per worker and k is capital per worker.
Then, as Samuelson assumed, from this information the family of the wage-curves can be derived,
once wages and profits are determined by the marginal productivities of each factor respectively,
so that:
( ) '( ) w f k kf k = ÷ (4.2)
'( ) r f k = (4.3)
If we derive both of these equations with respect to k we obtain:
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

62
[T]he Surrogate Homogeneous Capital gives the same result as does the shifting
collection of diverse capital goods in our more realistic model.
21


(Again, it is worth looking at Figure 4.2 above.)
So, apparently, Samuelson would have dealt with heterogeneous capital goods
economies and then would have shown that, by studying the formal properties of the
surrogate production function, one could analyse the relations between w, r, and q (i.e.
the net physical consumption output of the ‘real’ economy).
22
Even more, these
apparently similar formal properties of the ‘parable system’ and of the ‘real’ model
seem to have made Samuelson maintain that

[the surrogate function] is not an approximation but a rigorous equivalence. This
justifies the Surrogate Production Function as a useful summarising device.
23




''( )
dw
kf k
dk
= ÷ (4.4)
and
''( )
dr
f k
dk
= (4.5)

If we divide both sides of (4.4) and (4.5) we finally get:

dw
dw
dk
k
dr
dr
dk
= = ÷ (4.6)

which is the slope of the envelope (family of wage-curves) and represents the capital-labour ratio
of the economy. (Cf. Samuelson, 1962, p. 202 where Samuelson develops the surrogate function.)
21
Samuelson (1962), p. 201.
22
In fact Samuelson derived from equations (4.4) and (4.5) a functional relation between w and r,
which would be the wage-frontier (envelope) representing the whole competitive techniques.
Then, as output and surrogate capital are homogeneous, Samuelson concluded that the envelope is
a convex curve and therefore could be compared with the wage-curves derived from Samuelson’s
‘quasi-realistic’ cases where, in actual fact, every single wage-curve is a straight line. As we well
know, if we consider an infinite number of straight wage-curves, then the ‘envelope’ will be a
smooth convex envelope (see Figure 4.1 and cf. Figure 4.2.c).
23
Samuelson (1962), p. 203.
Chapt er 4 – A f i rst phase i n t he cont roversi es
63
4.5. What did Samuelson actually demonstrate?
Before answering that question, it is interesting to note to what extent Samuelson
parable had been accepted by the discipline at that time. In this regard it is worth citing
the following views from a well-known protagonist of the controversies, Joan
Robinson:

At first the neo-neoclassicals were happy to accept this parable [the surrogate
production function] […] For some years they remained cooped up in this
position.
24


Five years later Robinson provides further information on the historical aspects
surrounding Samuelson’s participation in the controversies:

For several years, everyone (except Piero Garegnani) was somewhat baffled by the
surrogate production function.
25


Joan Robinson’s appraisals suggest that among the critical contributions to the
Cambridge controversies one may find in Garegnani’s (1970a)
26
article a critique of
Samuelson’s construct. We shall explore in the following why this is so.
As we saw (Paragraphs 4.3-4.4), Samuelson’s analysis assumes the existence of a
‘real’ economy, one feature of which being an envelope function relating the wage rate

24
Robinson (1970a), p. 311. Robinson is thinking of Ferguson (1969) when she asserts that the
neo-neoclassicals “were happy” with the surrogate function. See Paragraph 4.12 below, where we
deal with Ferguson (1969).
25
Robinson (1975a), p. 37.
26
Garegnani (1970a). The article finally published in 1970 is an extension of a formerly accepted
paper submitted to the editors of The Review of Economic Studies in April 1963, which had been
accepted subject to revision. Garegnani was a visiting Rockefeller Fellow at MIT in 1961-62,
where he started writing the manuscript finally published in 1970, that is to say, in the period
immediately previous to the publication of Samuelson’s ‘Surrogate’ article (1962). In the
discussions with Samuelson, Garegnani already pointed out the extreme assumptions on which the
surrogate function is based. Though Samuelson acknowledges Garegnani for having saved him
“from asserting the false conjecture that my extreme assumption of equi-proportional inputs could
be lightened and still leave one with many of the Surrogate propositions” (Samuelson 1962, p.
202, n. 1), yet that did not prevent Samuelson from asserting that the Surrogate function is a
“rigorous equivalence” (Samuelson 1962, p. 203) of the ‘real’ heterogeneous capital goods model.
The issue under dispute has been precisely how far that ‘real’ model actually represented
production with heterogeneous capital goods. This issue was finally settled in Garegnani’s 1970
article. Thus Garegnani’s (1970a) became a very important reference, from the critically oriented
side of the controversies, in the history of the contemporary capital debates, as is recognised e.g.
by Joan Robinson. On the influence of Garegnani’s criticisms of Samuelson cf. Bhaduri (1969) p.
536, and n. 62, this chapter, below.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

64
and the profit rate, derived from the alternative systems of production. Then he
introduced the “surrogate production function”,
27
and showed that the results derived
from this ‘new’ device are the same as those obtained from the ‘realistic’ model. Is that
true?
Following Garegnani, to come to grips with the particular defence of traditional
theory attempted by Samuelson with his Surrogate function we need set out a more
familiar scheme of traditional theory: it will be necessary to assume that each system of
production can change continuously with r. By this it is meant i) that any change in r,
however small, causes a change of system, and ii) that one system only is in use at each
possible level of r. Accordingly, the family of systems
28
does not show any corner-
point, and the wage-curve of each system contributes only points to the wage-frontier
and not stretches. Thus, it is guaranteed that this wage-frontier becomes a smooth
envelope
29
enclosing the whole family of the wage-curves from above, and determining
a continuous function between the wage rate and the rate of interest, that is

w = e (r) (4.7)

As a matter of fact, we perfectly know that the higher the rate of interest the lower
the wage rate (
dr
r de ) (
< 0), but its curvature can be of any shape.
In general this function envelops several wage-curves of different curvatures, and
may therefore alternate convex with concave segments. For example, if all the wage-
curves are straight lines or convex ones the envelope will be convex, whereas the latter
will be concave or straight when the wage-curves are concave.
30

As it was assumed that at any possible value of r a single system is in use, a given
physical net output and physical capital per worker corresponding to each system can

27
The ‘as if’ aggregate production function, Samuelson (1962, p. 194, n. 1).
28
The assumptions underlying the present discussion are those present in Paragraph 3.2 above.
29
This smooth envelope entails avoiding switch points.
30



Figure 4.3. Concave, convex and straight ‘envelopes’
Chapt er 4 – A f i rst phase i n t he cont roversi es
65
therefore be found for any given r. This net product per worker, q, is therefore function
of r:

q=q (r) (4.8)

It is important to notice that equations (4.7) and (4.8) reported above belong to the
heterogeneous capital goods economy (the actual ‘real’ economy), and must have, as
Samuelson hypothesises, a parallel with equations (4.2) and (4.3) above, which belong
to the imaginary economy (the ‘Clark parable’). It is a confrontation between the
meanings associated with both groups of equations that will actually allow us to see
how ‘rigorously equivalent’, as Samuelson so decidedly claimed, are the Clark system
and the heterogeneous capital-goods model. Since according to Samuelson both the
‘Clark system’ and his heterogeneous capital-goods model are ‘rigorously equivalent’,
then heterogeneous capital-goods might be expressed as quantities of homogeneous
capital. The core of Samuelson’s attempt was to apparently determine the relations (in
the ‘real’ economy) between r, w, and q through a production function, the imaginary
parable.
In analytical terms, we are asked to find a production function homogeneous of the
first degree:
S=S(J,L) (4.9)

with S as the quantity of net product, J that of capital, and L that of labour. This is the
‘surrogate function’, and must satisfy the following two conditions:

i)
L
S
c
c
= e(
J
S
c
c
) (4.10)

The marginal product of labour (
L
S
c
c
) must be equal to w, which must be equal to
the ‘envelope’ function e(r =
J
S
c
c
) of the ‘real’ economy at the given r corresponding to
that equilibrium.
ii) S/L = q(r =
J
S
c
c
) (4.11)

The net product per worker, defined for the ‘parable’, must be equal to the function
q for the real economy defined above.
It is useful to re-write equation (4.9), by using Euler’s theorem:
( )
S S S J
L L J L
o o
o o
= + (4.12)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

66
Since in the imaginary economy the equilibrium profit (interest) rate must equal the
‘marginal productivity of capital’, then, by using condition i) noted above, equation
(4.12) can be written as follows:
( )
S J
e r r
L L
= + (4.13)

So far we have obtained functions expressing the net output per worker in the
imaginary economy. Now let us turn to the capital-labour ratio of that imaginary
economy.
If we differentiate function (4.12) with respect to
S
J
o
o
we obtain:
31

( )
( )
S
d
J
L
S
L
d
J
o
o
o
o
= ÷ (4.14)

But expression (4.14), by using again condition i), is:

J/L = – e’(r) (4.15)

Function (4.15) establishes that the slope of the ‘envelope’ curve measures the
capital-labour ratio of the economy.
32
But now let us compare this result with the
information we can trace from our real economy thanks to knowledge of the curvature
of the envelope.
In effect we know that in any real economy the envelope may be concave to the
origin in parts or throughout if the wage-curves are concave. Therefore, J/L would rise
with r =
J
S
c
c
as long as the envelope is concave, but this implies that the marginal

31
If we derive (4.12) with respect to
S
J
o
o
we immediately obtain
( ) ( ) ( )
( ) ( ) ( )
S S J
d d d
J S
L L L
S S S
L J
d d d
J J J
o
o
o
o o o
o
o o o
= + + ; by multiplying and dividing the LHS of this equation by ( )
J
L
d
then
( ) ( ) ( )
( ) ( ) ( )
S J J
d d d
S J S
L L L
S S S
J L J
d d d
J J J
o
o o
o
o o o
o o
o o o
= + + ; from which
( )
0
( )
S
d
J
L
S
L
d
J
o
o
o
o
= + that is equation (4.14).
32
Cf. equation (4.6) above.
Chapt er 4 – A f i rst phase i n t he cont roversi es
67
product of capital rises as J/L (the ratio of ‘capital’ to labour) rises
33
and, therefore, “the
function S could not be a production function”
34
because equilibrium in the ‘imaginary’
economy requires an inverse relation between ‘marginal product of capital’ and the
capital-labour ratio.
Furthermore, should we consider some straight-line part of the ‘envelope’ – which
is theoretically admissible when we have concave wage-curves – the contrast with the
‘Clark-Ramsey parable’ would be even more striking, for the ‘marginal products’
would change when the ratio of ‘capital’ to labour does not. This would also imply that
condition i) above cannot be satisfied with any differentiable homogeneous function as
it is impossible to talk about marginal changes when factor’s proportions do not vary
(they cannot vary by definition of a straight line envelope). The foregoing
considerations are sufficient to dismiss the existence of such a surrogate production
function, but it is still of some interest – as Garegnani points out – to consider the case
of convex envelopes.
When the ‘envelope’ is a convex curve we have two alternatives, either the wage-
curves are convex or they are straight lines. But S/L will equal q=q(r) of the real
economy only when the wage curve is a straight line but not when the wage curve is a
convex one.
35
So the only one case in which function (4.16) correctly estimates q, while

33
In concave parts of the envelope we have e’’(r) < 0, and accordingly we would have that J/L
increases when r rises. In fact we must rule out concave wage-curves because they imply that,
even when there is only one system of production from which a concave wage-curve is derived,
the value of capital increases as the profit rate increases, a behaviour that goes in contradiction
with the marginalist premises establishing an inverse relationship between demand for ‘capital’
(investment) and the interest rate.
34
Garegnani (1970a), p. 415.
35
To see that a convexity relation between the wage rate and the profit rate does not ensure the
existence of the surrogate function, it is necessary to re-write equation (4.12), which, combined
with result (4.15), becomes:
( ) [ '( )]
S
e r r e r
L
= + ÷ (4.16)
In fact, function (4.16) underestimates q as can be seen from the following graph. Function (4.16)
will not equal q when the wage curve is convex either. In this case that function will over-estimate
the real value of q.

(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

68
function (4.9) satisfies conditions i) and ii), is when the wage-curves are all straight
lines, and only in this particular case does a surrogate function exist.
36

4.6. There is no heterogeneous capital goods model in Samuelson’s (1962)
Drawing on our knowledge of the systems of production involving heterogeneous
capital goods, a straight wage-curve actually means that the proportion of capital goods
to labour is the same both in the consumption industry and in the capital good industry.
It is in this particular case that the ‘marginal product of capital’ (the partial derivative
of the production function with respect to ‘capital’) would measure the exact level of
the rate of profit. No doubt, now, that Samuelson has pretended to generalise the
‘parable’ to apparently ‘real’ heterogeneous capital goods models by precisely turning
some particular heterogeneous models into the parable system.
In this regard Garegnani clears the ground:

[B]ut for the arbitrary choice of the capital-good unit, the input coefficients of the
two industries are identical.
37


Since heterogeneity of commodities can be defined only as differences in the
production conditions (that is different systems of productions involving different capital
goods), a straight wage-curve does actually mean that the consumption good is produced



Figure 4.4. q is not equal to S/L when the wage-curve is convex

36
It is important to note that Samuelson’s way of measuring the relative shares of labour and
capital through the ‘elasticity’ of the envelope is incorrect for general cases (i.e. for convex and
concave curves). For example, consider the convex wage-curve of Figure 4.4: the elasticity on
point P is 0w
1
/w
1
A, whereas the relative share actually is 0w
1
/w
1
W.
37
Garegnani (1970a, p. 415).
Chapt er 4 – A f i rst phase i n t he cont roversi es
69
by itself and labour. So Samuelson’s system (i.e. a straight-line wage-curve representing
the production methods of the consumption good and of the capital good) becomes
indistinguishable from one where the consumption good is produced by itself and labour.
Samuelson’s quasi-real model, whose input coefficients are equal in both sectors, is not
more real than the jelly world he believes in. This jelly world is the only one world in
which an aggregate production function, or surrogate production function, can exist.
Garegnani also noted
38
that the aggregate production function is arrant nonsense,
because if there is only one consumption commodity in the economy, there is no need to
‘aggregate’, since there is no more than a single commodity in that economy and hence
there is no aggregate production function. If, on the other hand, there are many
consumption goods as well as capital goods, the aggregate production function does not
exist either, because it can only be built if and only if the national net output is composed
of only one commodity produced by itself and homogeneous labour. In a word,
Samuelson’s claim – that establishes that to deal with heterogeneous capital goods
economies there is no need to recourse to a ‘quantity of capital’
39
– is absolutely
dependent upon the highly restrictive assumptions of equi-proportionality of inputs, as
Garegnani has shown it.
40

In Samuelson’s jelly world the demand aspect problem of capital (see Chapter 2
above) in the context of the traditional method of long period equilibrium or normal
positions is totally saved, because at any given r, the ratio J/L and the ‘slope’ of the
envelope is at the same time the ‘slope’ of the wage-curve tangent to the envelope at that
r, so that the capital per labourer demanded is independent of changes in the profit rate.
Moreover, in Samuelson’s model the switch points would all be ‘forward’ switches, i.e.
neoclassical switches establishing a lower value of capital per labourer as the profit rate
increases, since each technique is single-valued in the capital-labour that it contains (the
value of capital is always independent of changes in distribution).
Therefore, where is the ‘rigorous equivalence’ between the ‘Clark system’ and the
quasi-real heterogeneous capital-good model that Samuelson would have demonstrated?
The point is that, in our opinion, a ‘rigorous equivalence’ must take into account the
general cases (convex, linear, concave wage-curves) not only a very particular one:
indeed both ‘worlds’ turn out to be equivalent because Samuelson disguised the ‘Clark
parable’ into his ‘quasi-real’ model – though, actually, linear and hence homogeneous.
Samuelson would have demonstrated, therefore, that the ‘parable’ provides the same

38
Garegnani (1970a, p. 421).
39
Samuelson (1962, p. 200) wrote: “An interesting point […] is that even in our discrete-activity
fixed-coefficient model of heterogeneous physical capital goods, the factor prices can still be given
various long-run marginalism (i.e., partial derivative) interpretations. And all this without our ever
having to pretend there is any quantitative aggregate homogenous ‘capital’ that itself truly
produces anything” (emphasis added).
40
The following conclusion raised by Garegnani seems to go in accordance with what we have
suggested above, with the help of Robinson’s views, regarding whether and how far the parable
system had been accepted by most of the practitioners of the discipline at that time: “Samuelson’s
‘surrogate production function’ is thus nothing more than the production function, whose existence
in such an economy no critic has ever doubted.” (Garegnani, 1972a, p. 416, emphasis added.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

70
properties of a heterogeneous capital-good model if and only if there had existed only one
good produced by itself and labour but then there is no heterogeneous capital good
economy at all. The ‘Samuelson parable’ is not less unrealistic than Clark’s. Thus in
Samuelson’s model the crucial problems of ‘reswitching’ and ‘reverse capital deepening’
cannot occur by definition: the consumption good is produced by itself and labour thus
the systems of production under use (each using heterogeneous capital goods) are
indistinguishable.
To conclude, it is true that in Samuelson’s world “No alchemist can turn one capital
good into the other” (1962, p. 196), not because there are physically heterogeneous
capital goods, but, on the other hand, because in that world the alchemist, if existed,
would not have any work to do at all, since there exists only one capital good
homogeneous with the consumption good. The alchemist would therefore find that his
métier in that world is no longer profitable, and will resign himself to pass his mysterious
task on neoclassical economists. Samuelson’s 1962 model actually avoided dealing with
heterogeneous capital-goods economies.
It is logical to follow our visit to the controversies by looking at some neoclassical
reactions and critical counteractions over the period 1965-1970.
4.7. Levhari (1965) and the 1966 Quarterly Journal of Economics Symposium
In 1966 the MIT-based Quarterly Journal of Economics gathered together a series of
contributions which conclusively showed that ‘reswitching’ and ‘reverse capital
deepening’ are phenomena which can in general exist either in economies with
indecomposable or decomposable technology matrices.
41
In terms of Sraffa,
‘reswitching’ is possible in economies where all commodities are basic (an
indecomposable technology matrix) as well as in economies where not all the
commodities are basic but for at least one. This story regarding the possibility of
perverse behaviour with indecomposable or decomposable technologies is due to the
claim carried forward by Levhari, who, in 1965,
42
published an article in which he
pretended to have proved that ‘reswitching’ is not possible at the economy level for
indecomposable techniques. Levhari’s analysis is briefly analysed here below, while the
important criticisms raised in the Symposium by Pasinetti (1966a) and Garegnani
(1966) in the wake of Levhari’s attempt will be analysed later (Paragraphs 4.8-4.10).
Levhari’s purpose has been to demonstrate that


41
Pasinetti (1966a), Bruno et. al (1966), Garegnani (1966), Morishima (1966), Samuelson
(1966a), Levhari and Samuelson (1966).
42
Levhari (1965).
Chapt er 4 – A f i rst phase i n t he cont roversi es
71
though it is quite possible that we could use some technique of producing good i at
a high rate of interest, then switch to another technique and, as we reduce it
further, return to the old technique, this cannot happen with the whole matrix.
43


His ‘whole matrix’ refers to a technology wherein every commodity enters directly
or indirectly in the production of all the commodities of the economy. In terms of
Sraffa that means that all commodities are basic. The input-output matrix representing
this type of economies is indecomposable. So while Levhari recognises the possibility
of ‘reswitching’ for single industries, it is not possible in an economy as a whole where
all commodities are basic. Levhari himself acknowledges this conjecture to
Samuelson.
44
Because Levhari’s article gave rise to the Symposium, and because the
issue of ‘reswitching’ became publicly acknowledged in this Symposium, in what
follows the main argument carried out by Levhari will be very briefly reproduced.
Levhari assumes an economy where all capital is circulating and is in stationary
equilibrium; all commodities are basic in the sense of Sraffa (so technological matrices
are indecomposable), and prices are expressed in terms of labour commanded, that is
w=1. Wage-curves represent different systems of production in terms of real-wages
(w/p=1/p) and the rate of discount ì (ì=1/1+r). So, for a given rate of profit, the
technique will be chosen according to which one provides the lowest price (analogous
to the highest wage). To simplify matters assume that there are only two techniques, a
and b which produce n basic commodities. Each technique differs from the other in at
least one activity (method of production). Row-vectors a
0
and b
0
refer to the labour
input associated with each technique respectively. As is well known, we can write the
price equations as follow:

1
0
1
0
( )
( )
a
b
p a I a
p b I b
ì
ì
÷
÷
= ÷
= ÷


The condition for a to be chosen is that
a b
p p s , so

1
0
( ) a I a ì
÷
÷ s
1
0
( ) b I b ì
÷
÷
then
1
0
( ) a I a ì
÷
÷ ( ) I b ì ÷ s
0
b

and, with some manipulation we can finally get:


1
0
( ) a I a ì
÷
÷ ( ) a b ÷ s
0 0
b a ÷ (4.L)

43
Levhari (1965, p. 103).
44
Levhari (1965, p. 103 n. 5).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

72

If techniques switch from a to b, then
b a
p p < or what is the same
1
0 0 0
( ) ( ) a I a a b b a ì
÷
÷ ÷ > ÷ . If a is not to return, as speculated by Levhari, then this
inequality must continue to hold. Levhari attempted to prove this by finding sufficient
conditions such that
1
0
( ) ( ) a I a a b ì
÷
÷ ÷ be a monotone non-decreasing function of ì.
This crucially depends on the sign of the difference between matrices a and b, i.e., the
sign of ( ) a b ÷ .
As “all the elements of
1
0
( ) a I a ì
÷
÷ are monotonic decreasing functions of ì”
45

the elements composing the difference ( ) a b ÷ must be negative or zero to ensure
1
0
( ) ( ) a I a a b ì
÷
÷ ÷ to be a monotonic non-decreasing function.
Levhari then introduces a key condition for his whole argument:

Now, for two positive indecomposable matrices, there exists a semi-positive
vector x such that either (a-b)x >0, (a-b)x s0, or (a-b)x=0.
46


The economic meaning of the above condition is expressed by Levhari in the
following way:

there exists some activity level x such that we need more circular capital of all
goods either with a or with b, or we are indifferent.
47


The above mathematical condition would ensure, Levhari believes, that either
technique a is in use and b is never to be chosen ((a-b)x > 0) or that the economy
switched from technique a to technique b but a is never to return ((a-b)x s0). Before
discussing the validity of this conjecture, which has been precisely the issue of the
Symposium, let us introduce the last part of Levhari’s argument.
Levhari’s next step was to multiply both sides of (4.L) – the inequality showing
that system a is at least as profitable as system b – by the semi-positive vector x*:

1
0
( ) a I a ì
÷
÷ ( ) a b ÷ x*s(
0 0
b a ÷ )x* (4.L’)

45
Levhari (1965, p. 104). This Levhari has already demonstrated in his proof of the non-
substitution theorem, an issue we are not going to deal with in this study. See Levhari (1965 pp.
100-102.).
46
Levhari (1965, pp. 104-105). In Levhari’s article these inequalities are written as strong
inequalities; but we have here introduced them as reported by Pasinetti (1966a, p. 511), who had
been told by Levhari that that had been a misprint.
47
Levhari (1965, p. 105).
Chapt er 4 – A f i rst phase i n t he cont roversi es
73

The left side of the above inequality is called by Levhari function ¢(ì). If we
differentiate this function with respect to ì we will get:

2 *
0
'( ) ( ) ( ) a I a a b x ¢ ì ì
÷
= ÷ ÷ ÷

Hence if
*
( ) 0 a b x ÷ > , ¢(ì) is monotonic decreasing; whereas if
*
( ) 0 a b x ÷ < ,
¢(ì) is monotonic increasing.
48
Finally if
*
( ) 0 a b x ÷ = , ¢(ì) is a constant, actually
zero.
Levhari then argues that in such case where system a has been used at a given ì
and then at a higher ì (lower r) the economy switches to system b, we need that ¢(ì)
be monotonic decreasing in ì so as a not to return, and hence ¢’(ì) must be negative at
the switch point. So if Levhari’s argument were true, ‘reswitching’ would, therefore,
take place only for single goods production but not for an economy as a whole. Thus
Levhari attempted to minimise the critical strength of ‘reswitching’ and, once more, to
rescue received theory. Now we turn to see whether Levhari’s argument is valid by
visiting some key works gathered in the Symposium.
4.8. Pasinetti (1966a)
Levhari’s – and Samuelson’s – conjecture was firmly contested by some participants of
the controversies in the already introduced Symposium.
The first participant in the capital theory debates that replied Levhari was Pasinetti,
who, in 1965, had already presented a refutation of Levhari’s lemma at the Rome
Econometric Society meeting by providing a counter-example which rebutted the ‘non-
reswitching’ theorem. Pasinetti’s contribution to the Symposium is an improved version
of that paper, as Pasinetti himself recalls in his article.
49

The first part of the article is again devoted to his counter-example which
invalidates Levhari’s lemma.
50
Pasinetti concludes in the first part that, since the
example which “satisfies all the requirements of Samuelson-Levhari analysis can yield
results that contradict their conclusions”, his illustration “is sufficient to disprove their
theorem”.
51
Albeit this would have sufficed to show that Levhari’s argument is totally
unfounded, Pasinetti also provides a “direct logical criticism” of that construct.

48
Every element of
2
0
( ) a I a ì
÷
÷ ÷ is negative. In the article Levhari (1965, p. 105) makes a
mistake when he establishes that ¢’(ì) is either monotonic increasing or decreasing according to
whether (a-b)x*>0 or (a-b)x*<0. What is relevant, instead, is function ¢(ì).
49
Pasinetti (1966a, p. 503, n. *). Note, however, that in Pasinetti’s example the underlying
technological matrices are not indecomposable. See Pasinetti (1966a, p. 505).
50
Pasinetti (1966a, pp. 540-510).
51
Pasinetti (1966a, p. 510).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

74
In the discussion about Levhari’s mathematical construct, we have seen that, for
two given techniques that differ in only one activity, the elements of the matrix
corresponding to the difference between both of them are all zeroes except for those
located in one column, which will contain both positive and negative elements. “Thus”,
Pasinetti explains,

the matrix elements which are responsible for making all prices change, when the
switch takes place, are all contained in that nonzero column.
52


However in order to get Levhari’s theorem right, it is necessary that the elements
of the semi-positive vector x*, which are multiplied by the non-zero column of ( ) a b ÷ ,
be all zero. This is the one trivial case for which Levhari’s argument would stand. But
the implications of such a trivial case are quite strong, as Pasinetti argued.
If the column responsible for the switch – which we have referred to above – is
multiplied by the zero elements of x*, then

this will simply eliminate the column. And if precisely that column is eliminated
which contains all the differences between the two matrixes, it will become
impossible to tell which of the two is more profitable. […] Levhari’s device would
thus lead us to the straight equality
*
( ) 0 a b x ÷ = , and to the false conclusion
that matrix a and b are indifferent for all levels of the rate of profit.
53


Therefore the exercise of comparison between two series of prices after eliminating
the components which actually make them differ is totally absurd. Pasinetti shows that
Levhari’s theorem is false.
4.9. Garegnani (1966)
Garegnani (1966) demonstrates that if technique a is more profitable than b then the
strong inequality in (4.L) must hold at least for one sector. But, because of having
multiplied inequality (4.L) by a semi-positive vector x*, we may, therefore, have
multiplied by zero one or more relations in (4.L’) to which the strong inequality
previously applied, hence getting 0=0 in their place. Then “the sign = would apply to
all the relations”
54
in (4.L’). The same reasoning can be applied when it is technique b
at a higher ì that is the most profitable, and again the sign = would apply to all the

52
Pasinetti (1966a, p. 511).
53
Pasinetti (1966a, p. 512).
54
Garegnani (1966, p. 558).
Chapt er 4 – A f i rst phase i n t he cont roversi es
75
relations in (4.L’). Then, assuming that r’<r’’ and that at r=r’ a is the most profitable
technique, but at r=r’’ b is so, Garegnani concludes,

[since] ( '') ( ') r r ¢ ¢ = and the fact that ¢(r) is a monotonic function (in this
case, a zero constant) does not exclude that producers may switch back to A as r
rises beyond '' r .
55


So the validity of Levhari’s theorem requires “stricter conditions”,
56
namely, that
nonzero elements of vector x* be “applied to at least one of those industries for which,
at r’ or r’’, the methods of systems A and B are not indifferent”.
57

Despite that these results would have sufficed to invalidate Levhari’s conclusion,
Garegnani in addition points out which Levhari’s “main error” is. In fact Garegnani
shows that Levhari’s proposition, which establishes that there exists a semi-positive
vector x such that either (a-b)x >0, (a-b)x s0, or (a-b)x=0, cannot “be proved since it is
false”.
58
What Garegnani shows is that, for two given techniques (a and b) which
produce, say, two commodities, and both techniques differing in two methods
(activities), then, after multiplying the matrix containing the elements of the differences
of both techniques (matrices a and b) by a semi-positive vector x*, the resulting vector
( ) a b ÷ x*=y* can exhibit elements such that one of them is negative and the other
positive, thus contradicting Levhari’s lemma i.e. (a-b)x >0, (a-b)x s0, or (a-b)x=0.
59

Levhari’s argument is, therefore, false: ‘reswitching’ (with all its implications) can exist
in economies where techniques are either indecomposable or decomposable. Samuelson’s
conjecture – as Levhari recalls in his article – is, once more, proved wrong.

4.10. The meaning of the 1966 Symposium: A turning point in the controversies
Both Pasinetti and Garegnani end their respective contributions to the Symposium by
drawing attention on the implications of ‘reswitching’ and ‘reverse capital deepening’
for the marginalist theory. Thus, they conclude, once it is shown that those phenomena
can exist either in decomposable or indecomposable technologies, as well as in any
single industry, it has to be admitted that there is no logical ground on which supporting

55
Garegnani (1966, p. 558). Garegnani uses function ¢(r) instead of ¢(ì), as did Levhari.
Remember that Levhari uses ì=1/1+r.
56
Garegnani (1966, p. 559).
57
Garegnani (1966, pp. 558-589, emphasis in the original).
58
Garegnani (1966, p. 561, emphasis added).
59
See Garegnani (1966, p. 558 n. 9); and Bruno et. al. (1966, p. 543), where these authors give
Solow credit for having correctly “pointed out that one cannot in general find a semipositive
vector x such that either (a-b)x >0, (a-b)x s0, or (a-b)x=0.”
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

76
the traditional premise of marginal theory, which establishes that there is a negative
relationship between the rate of profit and the quantity of capital per labour. Thus,
Pasinetti (1966a, p. 514) concluded:

The general conclusion is that, at any given state of technical knowledge, switches
of techniques due to changes in the rate of profit do not allow us to make any
general statement on changes in the ‘quantity of capital’ per unit of labour. The
new technology may require a lower ‘quantity of capital’ per unit of labour, or it
may require a higher ‘quantity of capital’ per unit of labour, whether ‘capital’ is
measured in terms of value or in terms of any chosen physical unit, whether we
consider any single industry or the economic system as a whole.
In the same direction Garegnani (1966, p. 564, emphasis added) thus concluded:

In reality, once we correctly recognize how the value of capital goods depends on
distribution, the very possibility of ordering the techniques according to the
proportions of capital to labour slips from our hands; that order may change as
prices and distribution change. And no way-out of this difficulty can be found by
relating capital per man to some magnitude independent of distribution, which
would then be taken as the measure of capital intensity: the ‘return’ of a technique
shows that any such measure, even if it could be found, would lead to
contradicting the principle of an inverse relationship between rate of interest and
capital intensity.

Both authors raised very strong concerns about the admission of these results by the
hitherto literature.
60
On the one hand we have Pasinetti, who argues that “the whole
theory of capital seems to have been caught in the trap of an old mode of thinking”.
61

Garegnani, on the other, concludes that those results were admitted only “sporadically”
and that, as he also pointed out in his long gestated reply to Samuelson’s Surrogate
note, the mainstream approach to economics has ever had the tendency to regard those
possibilities as “exceptions”.
62


60
Both authors refer to Hicks (1965), p. 154.
61
Pasinetti (1966a, p. 516).
62
Garegnani (1966, p. 565). We have written ‘long gestated’ reply to Samuelson partly because
the reasons we have already seen (see n. 26, this chapter, above). By the way, the present author
would like to call attention on the issue about when ‘reswitching’ became publicly important in the
debate. For the reasons we explain in this chapter, that importance arose in the wake of the 1966
Symposium. Nevertheless, this does not mean that previous to that date ‘reswitching’ has not been
used by critics as a powerful instrument of critique of the marginal approach nor does it mean that
neoclassical economists have not realised their implications. An example is precisely the case of
the oral controversy between Garegnani and Samuelson then crystallised in the 1970 paper.
However, in the literature on the controversies, there does not seem to be total agreement on this.
Recently, Birner (2002, p. 107, n. 2) writes that when he asked Paul Samuelson (in an interview in
(cont.)
Chapt er 4 – A f i rst phase i n t he cont roversi es
77
Pasinetti affirms that the traditional principle (inverse relationship between demand for
capital per worker and the profit rate) has been taken for granted

without any justification, except that this is the way economists have always been
accustomed to think.
63


On the other hand Garegnani points to the problematic consequence that traditional
theory would undergo had those negative results been admitted. This may allow us to
envisage what could happen to the discipline if the problematic issues like
‘reswitching’ and ‘reverse capital deepening’ had been frankly admitted:
64


[T]he consequences of admitting this, however, are far-reaching, because on that
principle has been erected the dominant theory of distribution.
65




April 1989) whether or not the latter had read Sraffa (1960) in the time of Garegnani’s visit to MIT
in 1961-62 and whether or not Garegnani had orally used ‘reswitching’ in his critique of
Samuelson, “Samuelson replied that he did not want to say that (at that time) he understood all the
things that were in this cryptically written book. But he was rather emphatic in his recollection that
‘reswitching’ had not come up in Garegnani’s oral criticism”. Actually what Birner wants to argue
in that passage, as well as in the chapter referred to the controversy between Garegnani and
Samuelson (Birner 2002, pp. 66-83) is that, by relying on the support of Samuelson’s answers to
Birner’s questionnaire, ‘reswitching’ “did not play a major part in Garegnani’s criticisms” (p. 77)
of Samuelson. It is hard to go along with Birner’s position once it is realised that, despite the fact
that “reswitching is only mentioned in a footnote of [Garegnani’s] manuscript” (Birner 2002, p.
77) – which is, by the way, the only one reason other than Samuelson’s answers pointed out by
this author for justification of his thesis – the ‘reswitching’ case is, just to argue in the same terms
as does Birner, actually illustrated by Garegnani at the outset (p. 411) of section II of (1970a),
which had already been sketched in the 1961-62 MIT manuscript. That ‘reswitching’ became
publicly important some years after 1961-62 (i.e., precisely in the Symposium of 1966 owing to
Levhari’s 1965) is not under dispute (see this chapter). But this is radically different from saying
that ‘reswitching’ “did not play a major” role, as Birner (2002, p. 77) conjectures, in the criticisms
raised by the Italian economist against one of the most important representative of the marginal
school in the world, as is Samuelson. (See Paragraph 4.5 above for the consequences that
Samuelson’s surrogate would have created among the discipline and how that situation was
viewed by Joan Robinson).
63
Pasinetti (1966a pp. 516-517). As Sraffa (1960, preface) clearly points out the classical political
economy had already been ‘submerged’ by the marginal revolution at the end of the nineteenth
century. The “way economists have always been accustomed to think” to which Pasinetti refers is
the marginal theory. Cf. also Keynes (1936) p. viii.
64
The word frankly was italicised for reasons that will appear clearer immediately below (See
Samuelson’s reactions).
65
Garegnani (1966), p. 565.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

78
What Pasinetti and Garegnani unveil in their respective interventions in the
Symposium is that the columns upon which the edifice of the dominant theory has for
so long been built are not so strong, as it was once believed. The problem of capital
undermines even the idea of thinking of that building (and re-building) of those
columns because there is no appropriate – even less definable – material for such a
construct. It is therefore hard to conceive of an explanation of profits determined by the
scarcity of capital considered as a factor of production.
66

4.11. Neoclassical reactions: Samuelson (1966a)
The 1966 Symposium conclusively showed that the traditional principle establishing a
negative relation between the rate of interest and the ‘quantity of capital’ could not be
postulated in economies with heterogeneous capital goods. Defeat was then admitted by
defenders of marginalism. Let us see now how this admission has appeared soon after
the Symposium.
An authoritative neoclassical view, as represented by Samuelson, admitted this by
pointing out that in the light of the results of the Symposium, there exist serious
problems for the theory. Thus Samuelson (1966a), in the well-known “Summing up”,
admits that ‘reswitching’ is a theoretically possible phenomenon:

Pathology illuminates healthy physiology. Pasinetti, Morishima, Bruno-
Burmeister-Sheshinski, Garegnani merit our gratitude for demonstrating that
reswitching is a logical possibility in any technology, indecomposable or
decomposable. Reswitching, whatever its empirical likelihood, does alert us to
several vital possibilities.
67


One of these possibilities for Samuelson is that there would no longer be any
guarantee to ensure an inverse relationship between consumption per capita and the
interest rate. However, Samuelson totally minimises the relevance of ‘reswitching’ and
‘reverse capital deepening’ by pointing to a problem of a ‘dynamic stability’ in steady-

66
For the reasons we have here examined, the importance of the Symposium in the Cambridge
controversies is indubitable. However, recently, Cohen and Harcourt (2003) seem to have
forgotten the turning point of this episode by completely ignoring it in their accounts of the
controversies. In fact the lack of this issue in their review immediately provoked the reactions of
Pasinetti, who argues that the Symposium “was the central point of the controversy and which is
inexplicably ignored by the authors” (Pasinetti in Pasinetti et. al. 2003, p. 227). This recent
exchange on issues which took place around forty years ago reveals the current importance of
setting in perspective both the Cambridge debates in general and, specifically, the 1966
Symposium.
67
Samuelson (1966a), p. 582, emphasis added.
Chapt er 4 – A f i rst phase i n t he cont roversi es
79
state equilibria.
68
Quite incomprehensibly, when referring to the consequences of
‘reverse capital deepening’, Samuelson writes:

going to a lower interest rate may have to involve a disaccumulation of capital,
and a surplus (rather than sacrifice) of current consumption, which is balanced by
a subsequent perpetual reduction (rather than increase) of consumption as a result
of the drop in interest rate.
69


Samuelson of course is taking for granted that accumulation is created by “sacrifice
of current consumption”, that is by full employment savings. He does not seem to be
worried, none the less, about how the accumulation of capital, in the light of ‘reverse
capital deepening’, would be brought about: in the presence of ‘reverse capital
deepening’, a situation showing savings exceeding investment would need a rise in the
interest rate so as to increase investment to the level of the former, whereas the
tendency in the savings-investment market, according to supply-and-demand analysis,
should be to lower that rate. This argument is what we have illustrated above by
introducing the meaning – and their importance in economic terms – of ‘reverse capital
deepening’ and ‘reswitching’. (See Chapter 3.). Despite the fact that Samuelson admits
the problems for the theory as a result of the definitive critiques raised in the
Symposium, his admission does not mean abandoning the theory, as it is the theory
relying on factor’s substitution what is shown to be unfounded. Then it is no wonder
the way in which the famous MIT economist receives the criticisms:

If all this causes headaches for those nostalgic for the old time parables of
neoclassical writing, we must remind ourselves that scholars are not born to live in
an easy existence. We must respect, and appraise, the facts of life.
70


Did successive theorists appraise and respect “the facts of life”, as Samuelson
stated to the practitioners of the discipline?
As we shall see below, some Samuelson’s disciples, and even Samuelson himself,
actually tried to judge the validity of the criticisms, increasingly relying on the
empirical ‘importance’, so as to minimise their relevance, rather than to respect them –
let alone to accept them. Ferguson (1969) is a good example (see below).

68
Samuelson (1966a, p. 578).
69
Samuelson (1966a, p. 581, emphasis in the original). A similar position can also be found on p.
579, when Samuelson writes: “[Due to the possibility of reverse capital deepening] after
sacrificing present consumption and accumulating capital goods, the new steady-state equilibrium
can represent a rise in interest rate!” Interestingly, Samuelson’s position is not isolated: a similar
attitude can be found in e.g. Burmeister (2000), p. 310.
70
Samuelson (1966a, p. 583).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

80
Needless to repeat that Samuelson, and others (e.g. Ferguson 1969; Blaug 1974),
when admitting the possibilities of irrefutable discoveries like ‘reswitching’, have had
the common tendency to associate those results, which undermine the ‘aggregation’ of
capital, with a problem of ‘aggregate’ production function: in fact they characterise
these results as elements raising ‘suspicion’ of the neoclassical ‘parables’,
71
a term that
in the controversies was associated almost exclusively with the aggregate production
function ‘version’. Despite the fact that this version of the theory has clearly been
shown to be theoretically secondary in Garegnani’s critique of Samuelson as far as the
building of the roots of received theory is concerned, it is a matter of persistent
misunderstanding that the damaging implications of ‘reswitching’ on the conception of
capital as a single magnitude (which in a sense might be seen as a problem of
‘aggregation of capital’) – which is what actually is relevant for the long period supply
and demand theory – is misread as damaging the ‘aggregate’ production function, as if
the results of the controversies would have only affected a ‘less rigorous’ version of the
theory. These misunderstandings were mostly created by neoclassical economists –
though sometimes owing to responsibility to some critical economists like Joan
Robinson.
72

4.12. Further neoclassical reactions after the Symposium: Ferguson (1969)
We deal with Ferguson (1969), because this author has been one of the most notable
representatives of the “neo-neoclassicals cooped up by the surrogate function” to whom
Robinson refers in some above citations.
At the outset of his book, Ferguson recognises that the Cambridge criticisms are
valid. This does not mean, nevertheless, that he has not attempted to minimise the
significance of the criticisms of the theory he tried to shield:

[The validity of the] Cambridge Criticism is unquestionable, but its importance is
an empirical matter that depends upon the amount of substitutability there is in the
system. Until the econometricians have the answer for us, placing reliance upon
neo-classical economic theory is a matter of faith.
73


The terms in which Ferguson disputes the relevance of the Cambridge criticism are
hard to accept. Firstly, because he reduces the problem of the basic principle on which
the theory relies to an ‘empirical’ matter. But the point is, on the contrary, that marginal

71
Samuelson (1966a) p. 574.
72
Cf. Chapter 5, below.
73
Ferguson (1969) p. xvii. Later on the book Ferguson also claimed: “The crucial point to
emphasize is that the validity of neoclassical theory is an empirical, not a theoretical, question”
(Ferguson 1969, p. 258, emphasis in the original).
Chapt er 4 – A f i rst phase i n t he cont roversi es
81
economic theory is not grounded on the probabilities of having downward-sloping
demand curves for factors. When the theory meets insurmountable theoretical flaws
some of its defenders appeal to the empirical way-out; still the downward-sloping
demand functions were basically derived from logical deductions, in turn derived from
self-evident facts.
74
Secondly, Ferguson dodges the problem at all, because, since what
matters is the empirical probability to derive ‘well behaved’ demand curves, we must
look forward to the “econometricians’ answer” in order for us to assess whether or not
one can rely on the traditional supply and demand functions. In the meantime the author
proposes to “have faith” in the neoclassical theory, hence in the downward-sloping
demand curves, etc.
75
Thirdly, it is worth noting that the problem is the
“substitutability” itself, not just a problem of “amounts of” as Ferguson believes. The
substitutability principle is a basic premise of the theory, a basic principle that, being
necessary for the theory, had long been taken for granted since the inception of
marginal economics at the turn of the twentieth century (see Chapter 2 above). Yet this
extreme opinion is quite different from some other Ferguson’s views on the debates:

Shortly after this [Samuelson’s ‘surrogate’ article] was completed, the
‘Symposium on Capital Theory’ appeared in the Quarterly Journal of Economics;
and it then became quite apparent that the Cambridge Criticism, as I call it, must
be accorded more careful consideration.
76



74
These facts (alternative methods of production and tastes and preferences) allow deriving logical
deductions resulting in inverse relations between quantities of ‘factors of production’ used and
their rates of remuneration. What is behind the consideration of the two facts mentioned above
regarding the data of the theory, is that these data might directly and/or indirectly guarantee the
substitution of factors as distribution changes. This general principle is what is so necessary for
the theory rather than the ‘empirical probability’. To try to ascertain the probabilities of ‘perverse’
behaviour would mean to try to ascertain the probabilities of ‘non-perverse behaviour’ that is of
downward-sloping demand functions for factors. However, this attempt would imply a “drastic
change in the nature of the theory”, as Garegnani (1990, p. 72) pointed out, because “that shift of
basis would occur in a field like economics, where experiment is impossible and the material is so
highly complex as generally to limit considerably the weight one can attribute to any observed,
purely empirical regularity”. Although we do not deal in detail with this drastic consequence that
the theory would undergo if the ‘empirical’ foundation were adopted, it is worth noticing that all
these attempts appeared soon after the controversies in capital theory.
75
The ‘empirical’ route in the wake of the controversies was not only taken by the neoclassiclal
side. We also have to note that some critically oriented authors in the controversies somehow
attempted to ascertain the ‘probabilities’ of ‘reverse capital deepening’ in some heterogeneous
capital-goods models. See Mainwaring and Steedman (2000), D’Ippolito (1987). The ‘empirical
issue’ as viewed by these authors is not examined in this study, since we have preferred to focus
on the neoclassical reactions against the critical Cambridge. For a discussion of these works, see
Ciccone (1996) and Petri (2000; 2004, pp. 252-255).
76
Ferguson (1969, p. xvii).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

82
Anyone who happens to come across Ferguson’s reaction grounded in terms of
‘faith’ would probably have expected for another kind of ‘more careful consideration’
of the ‘Cambridge Criticism’. We are not arguing that basing economic principles on
‘faith’ is unfair, but Ferguson does not provide any convincing argument in order for
economists to have faith, and his only attempt to that purpose is “to invoke the weight
of Samuelson’s authority as represented by [the following words]:”
77


Until factors cease to have their rewards determined by bidding in quasi-
competitive markets, I shall adhere to generalized neoclassical approximations
(…) somewhat limited factor substitution can fail to have some of the simple
properties of the idealized J.B. Clark models.
78


These words reveal an embedded belief of neoclassical economists, namely: that
the sole way to envisage a process of “bidding in quasi-competitive markets” is by
invoking marginal, or neoclassical, supply and demand functions and not any other
alternative. To logically derive such functions it is indispensable the factor substitution
principle, which would act through the data of the theory. But the critiques of marginal
theory precisely showed how groundless is to postulate a general principle by relying
on ‘factor substitution’ when one of those ‘factors’ is measured in value terms – capital.
That principle is the ultimate foundation on which the determination of “rewards in
quasi competitive markets” relies – namely ‘well behaved’ supply and demand
functions, which are what Samuelson means when this author speaks of “bidding in
markets”.
Ferguson later introduces “Samuelson’s authority” again in his book
79
when he
attempted to give the criticisms “more careful consideration”. Ferguson’s purpose is to
show that “a small deviation from Samuelson’s [surrogate] model” will result in a
“slope of the frontier [which] does not give the capital-labour ratio [of the economy].”
80

The small deviation alleged is that the relative factor production coefficients are the
same in both sectors (Samuelson parable), but the absolute magnitudes of the
coefficients differ. This is called by Ferguson the “Hicks case” because, according to
Ferguson, Hicks had emphasised this case “so strongly”.
81


77
Ferguson (1969, p. xviii, emphasis added).
78
Samuelson (1966b), pp. 444-445. This passage is reproduced in Ferguson (1969, p. vi) as a
‘flyleaf’ and in p. 250.
79
Ferguson (1969), pp. 251-270.
80
Ferguson (1969), pp. 259-260.
81
Cf. Ferguson (1969), p. 259. Ferguson refers to Hicks (1965) pp. 148-159. However, it is hard to
go along with Ferguson here as well, since Hicks’ chapter Ferguson refers to actually treats those
cases in which the relative factor proportions in the ‘corn’ and in the ‘tractor’ industries may
differ. On the point Ferguson introduces, Hicks (1965, p. 153) pointed out that, in cases where
both factor proportions are equal and by comparing two straight-line wage curves, “a fall in the
rate of profit would lead to a shift in technique such as […] the technique [chosen] must become
more capital intensive.” Hicks justified that by arguing that as the profit rate falls the technique
(cont.)
Chapt er 4 – A f i rst phase i n t he cont roversi es
83
In Samuelson’s parable the wage-curve is
(1 ) 1
c
a a
c r
w
l l
+
= ÷ hence
c
a
c dw
dr l
÷ = ,
which is equal, if the absolute magnitudes are the same in both sectors, to
a
c
c K
l L
= ,
which is of course equal both to
c
c
c
l
and to
a
a
c
l
. (In the example it is being considered
only circulating capital). Now Ferguson by introducing his “Hicks case” argues that
c
a
c K
l L
= . However this is a more general condition. In fact since in Samuelson’s model
the relative factor production coefficients are equal in both sectors, then the capital-
labour ratio of the economy will be equal to the relative coefficient and, moreover, will
be constant as distribution changes because the price of the capital per labourer is
constant throughout. The capital-labour ratio of the economy, when absolute
magnitudes differ, will be equal to
c a
a c
c l
l l
| |
|
\ .
– that is a physical magnitude multiplied by
a value magnitude, which is, in Samuelson’s model, equal to the relative ratios of
quantities of labour in both industries.
82

Ferguson’s conclusion is that “the factor price frontier may not yield the
neoclassical relation between factor proportions and relative factor prices even in the
absence of reswitching.”
83
Of course ‘reswitching’ cannot arise in Ferguson’s model.
For linear wage curves there is no heterogeneous model at all. A linear wage curve
actually implies that the consumption good is produced by itself and labour; now if a
second linear wage curve is introduced, though apparently involving a heterogeneous
capital good, it is actually implied that the same consumption good is being produced
and therefore, since this technique is depicted by a linear curve, again, the consumption


chosen must account for a higher “capital-capital coefficient” and a lower “labour-consumption
good coefficient” – in other words, as r falls, cc/la must increase. Of course, Hicks here does not
realise that he is actually dealing with homogenous systems since, as we have seen above with
regards to Samuelson’s parable and the criticisms raised by Garegnani, a straight-line wage curve
actually means production of ‘corn’ by means of ‘corn’ and labour. And this is valid for any
number of techniques whose wage equations are linear. In fact, then, Hicks (p. 154) considers
cases in which relative factor proportions are not equal in both productive sectors but minimised
the relevance of the possibility of ‘reswitching’ by pointing out that “such cases [could be
regarded] as exceptional.”
82
For the sake of the argument, assume an economy where 4
c a
c a
c c
l l
K
L
= = = . Then
c a
a c
a a
c c
l l
l l
=
and from this equation we can obtain:
c a c
a a a
c c l
l l l
= so that 4
c a
a c
c l
l l
K
L
= = .
83
Ferguson (1969, p. 260, emphasis in the original).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

84
good is produced by itself and labour. Independently of Ferguson’s explicit
minimisation of the ‘reswitching’ results for the theory, his analysis fails from its very
beginning, as he thinks of having dealt with heterogeneous capital goods economies
with systems of production described by linear wage-curves. Yet, as we have
previously seen, the problems for the theory not only involve ‘reswitching’ but also
‘reverse capital deepening’, a problem, that is, which does not seem to disturb so much
Ferguson.
Ferguson like many, if not most, neoclassical economists, downplayed the
relevance of ‘reswitching’ by incessantly pointing out that the Cambridge criticisms
only “invalidate the simple Clark fairy tale” (Ferguson, 1969, p. 269), show that “the
Clark parables may not hold” (Ferguson, 1969, p. 269), “point up a definite potential
weakness of simple neoclassical theory” (p. 270), and that “simple neoclassical results
concerning the relation between production and input and output markets may not hold”
(p. 265). Is there, perhaps, a more robust or more sophisticated version than the ‘fairy
tale’, on which Ferguson is relying his ‘faith’? The point is that the criticisms destroy or
invalidate the ‘simple relation’ from which “all neoclassical results follow”, as
Ferguson himself put it: “the lower the rate of interest, the greater the capital intensity
of production.”
84
Ferguson recognises this, but manifests his concerns by pinpointing
“that economists may be unable to make any statement concerning the relation of
production to competitive input and output markets.”
85
However, Ferguson believes that
economists can still use the theory under attack, and, once again, he justifies his
claiming on his own faith. He argues that all that is needed to establish the neoclassical
parables is the existence of a sufficient substitutability establishing capital intensity
uniqueness, but the latter is “an econometric question susceptible of resolution in a
probabilistic sense.”
86
Hence, “until the econometricians have the answer for us,
placing reliance upon neoclassical economic theory is a matter of faith”, as we noted
above. This was the way chosen by Ferguson to rebut the criticisms of capital.
87
As
noted, Ferguson’s reactions are very extreme. And yet it would seem that many
practitioners of the discipline have taken up this route, as it is pretty evident the great
deal of work done in econometrics relying on the traditional theory – and all this as if
nothing had ever happened in the theory of capital after the post-war controversies.
88





84
Ferguson (1969, p. 252).
85
Ferguson (1969, p. 269).
86
Ferguson (1969, p. 270).
87
It is worth noticing that the ‘empirical’ way out was emphasised by some commentators from
the neoclassical side such as Blaug (1974), pp. 39-40.
88
See n. 74, this chapter, above.
Chapt er 4 – A f i rst phase i n t he cont roversi es
85
4.13. Garegnani’s 1970 critique of the marginal theory
Garegnani’s (1970a) article not only provided the analytical elements necessary to
definitely dismiss Samuelson’s surrogate function that “somewhat baffled everyone” –
as Robinson pointed out
89
– but also shows that expressing the production conditions in
terms of a production function (either aggregate or individual)
90
with ‘capital’ as one of
its arguments, while the other one being labour, has been a “feature of only some
versions of the traditional theory.”
91
In fact what is being argued by Garegnani is that
marginal theory reduces its core explanation of distribution to the functions of supply
and demand for productive factors. The basic premise on which the traditional theory
ultimately relies is the ‘belief’ that a fall of the profit rate cheapens the more ‘capital-
intensive’ method either by means of the alternative methods of production available
(direct factor substitution mechanism) or by changes in consumer substitution in favour
of the ‘more capital intensive’ goods. And this problem is quite independently of
considering the systems of production as represented by an aggregate production
function.
92

In addition it is interesting to note how Garegnani uses the implications of
‘reswitching’ and ‘reverse capital deepening’, when he argues that

[Theoretical] examples do not seem to indicate that the conditions in which a fall
of r results in a relative cheapening of the less capital-intensive productive
processes are any less plausible than those in which the opposite would be true.
This appears to undermine the ground on which rests the explanation of
distribution in terms of demand and supply for capital and labour.
93


The fact that the factor substitution principle is undermined by the phenomena of
‘reswitching’ and ‘reverse capital deepening’ implies that, in systems of production
with heterogeneous capital goods, a fall in the rate of profit may well bring about a fall
rather than a rise in the capital-labour ratio of the economy. Thus, with ‘reverse capital
deepening’ there would not be any support to the claim that it is only by postponing

89
See Paragraph 4.5 above.
90
It must be said that nor can individual production functions exist such that they could represent
the production of a single commodity by means of labour and ‘capital’, because of the same
reasons of why it cannot exist an aggregate function for the economy as a whole, that is to say,
they can only exist if and only if the net consumption output is produced by itself and labour, so
this single commodity would be produced by itself and labour, hence the question which arises is:
What kind of representation of different systems of production – involving heterogeneous capital
goods – can these individual production functions be? Yet these striking conclusions seem not to
disturb the authors who are devoted to use aggregate production functions in e.g. modern growth
theory, see Barro and Sala-i-Martin (1995).
91
Garegnani (1970a) p. 422, emphasis added.
92
The particular target of attack at production functions pursued by Robinson in the controversies
is analysed in Chapter 5, below.
93
Garegnani (1970a), p. 425.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

86
consumption (that is by saving) that investment can be brought about, because as the
profit rate falls it is a less ‘capital intensive’ method which is chosen (because it is the
most profitable at the new distributive situation) with a lower value and not a higher
‘quantity of capital’ per worker.
94

We have seen above (Chapter 1), the consequences for the traditional marginal
theory of having upward-sloping demand curves for productive factors. In this regard,
Garegnani’s article also provides a clear-cut example
95
which conclusively
demonstrates that “a change, however small, in the ‘supply’ or ‘demand’ conditions of
labour or capital (saving) may result in drastic changes of r and w.”
96
These drastic
changes in distribution actually mean that the profit rate might fall to zero while wages
to infinite or vice versa, as supply and demand for each respective ‘factor’ will not be
brought into equality. The point is that, had one ever detected in any real economy, for
example, a zero wage rate with unemployment, the variables determined by the theory
would have served as theoretical guides of actual ones, and explanations in terms of
supply and demand would have resulted in a plausible theory of distribution. But, as
Garegnani concludes

no such instability of an economy’s wage- and interest-rates has ever been
observed. The natural conclusion is that, in order to explain distribution, we must
rely on forces other than ‘supply’ and ‘demand’ […] [T]he theory becomes
implausible once it is admitted that this principle [that a fall in r would always
raise the proportion of ‘capital’ to labour in the economy] is not always valid.
97


What, Garegnani asks, then to an economic theory which cannot describe any
actual economy? In a word, with ‘reswitching’ and ‘reverse capital deepening’ we
would have to admit, as theoretical possibilities, extreme values of distributive
variables; an infinite value of, say, the interest rate – owing to an upward-sloping
demand function for ‘capital’ – would imply zero wages: it is hard to conceive of any
explanation establishing an equilibrium in terms of supply of, and demand for, labour
resulting in a zero wage rate. In the light of these results it is clear that the traditional
belief does not work, and therefore the forces “in order to explain distribution”, as

94
Cf. Samuelson’s reactions in the Symposium, Paragraph 4.11 above.
95
Garegnani (1970a), p. 424.
96
Garegnani (1970a), p. 426. emphasis added.
97
Garegnani (1970a, p. 426). Garegnani in this passage talks about the existence of a neo-classical
‘belief’ in the traditional principle. ‘Belief’ is here taken to mean that, independently of the supply
aspect of the problem of capital (that is the illegitimacy of conceiving of capital in value terms),
the idea of an inverse relation between demand for ‘capital’ and the profit rate would somehow
have been taken for granted by the theory; the ‘belief’ being ‘supported’ by two groups of the data:
the alternative methods of production, and tastes and preferences of consumers. This ‘belief’ is
what is behind the demand aspect of the problem of capital, and it is this belief that is totally
destroyed by ‘reswitching’.
Chapt er 4 – A f i rst phase i n t he cont roversi es
87
Garegnani strongly stressed,
98
must be sought in a theoretical approach alternative to
marginal theory.
4.14. Bliss’s 1970 intervention. Prelude to the shift to neo-walrasian equilibria and
the beginning of a second phase in the debates
In the same issue of The Review of Economic Studies where Garegnani (1970a)
published his critique of Samuelson, there appeared the sole reply that Garegnani’s
article ever received for the record from the neoclassical camp. It was the case of
neoclassical Christopher Bliss, who in a brief comment
99
completely misunderstood the
theoretical background in which Garegnani’s argument relied – the conception of
capital in value terms. In fact Bliss takes Garegnani’s argument as if the latter would
have referred to a short-period, neo-Walrasian equilibrium.
100

Bliss argues that one of the most striking Garegnani’s conclusions – the possible
non-existence of equilibrium in the savings-investment market due to ‘reverse capital
deepening’ – is contradictory with Debreu’s results, who, according to Bliss, would
have considered the same required conditions as did Garegnani for the existence of
equilibrium.
101
But, in a Debreu’s economy, capital is conceived as the heterogeneous
capital goods vector of the economy. In Debreu’s words:


98
Garegnani (1970a), pp. 424-427.
99
Bliss (1970).
100
Bliss (1970) refers to a “momentary equilibrium”. We call ‘neo-Walrasian’ equilibrium the
notion of equilibrium professed by the modern versions of the theory (temporary/momentary
equilibria or intertemporal equilibria) which, as we shall explain in Chapter 6 below, take as datum
of the equilibrium the physically heterogeneous capital goods of the economy instead of the single
magnitude ‘capital’ as it was traditionally conceived (with the exception, as is well known, of
Walras). The label ‘neo-Walrasian’ is not misleading, because it indicates that latest versions of
the theory (i.e., the temporary and intertemporal equilibrium versions associated with the names of
Arrow, Debreu, Hahn, Bliss) conceived, as Walras did, the data regarding capital as the
heterogeneous vector of the several capital goods (the ‘Walrasian’, so to speak, ‘arm’ of that
label); while ‘neo’ refers, instead, to a completely different notion of the equilibrium owing to the
not satisfaction of the condition of a uniform rate of returns on the supply prices of the capital
goods, thus renouncing to the traditional long period (or normal position) method, which had
characterised received theory until recent decades (Cf. Garegnani 1976a): this Walras was very
concerned on, because despite he had conceived of capital as a heterogeneous vector, he shared, on
the other hand, the traditional method of the normal positions, but the latter was contradictory with
the conception of capital as a collection of heterogeneous goods. It is important to notice that
Debreu and other economists of mathematical orientation pursued the research programme
commenced by Hicks in Value and Capital (1946[1939]), where the author after having criticised
the traditional method of long period equilibrium proposes a ‘temporary’ equilibrium by relying in
a Walrasian conception of capital. Cf. particularly Paragraph 6.1, Chapter 6, below.
101
Bliss is referring to the well-known work of Debreu (1959). These conditions are continuity of
demand functions and convexity of technology.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

88
The total resources of an economy are the given quantities of commodities that are
available. They include the capital of the economy at the present instant, i.e., all
the land, buildings, mineral deposits, equipment, now existing. All these are a
legacy of the past, they are a priori given.
102


Then, Bliss goes on arguing that “[w]ith more than one capital good current
demand for investment goods will depend not only upon present prices but also upon
future expected prices.”
103
Evidently Bliss is taking Garegnani’s argument regarding
investment as if it were a short-run problem of expectations.
104
Bliss’s
misunderstanding is finally completed by his conclusion:

the major fault in [Garegnani’s] argument is the illicit importation of long-period
equilibrium theory into the analysis of a short run situation.
105


But, as it has been described in the present chapter, the background of the
controversies was completely another one. In fact Garegnani immediately replied
106

Bliss, pointing out that he had never treated in his argument the discussion of such
short-period equilibrium and argues that:

the demand and supply explanation of distribution seems faced, at the present
moment, by two alternatives. On the one hand, there is a long-run analysis
undermined by the inconsistencies of the notion of capital it must use. On the
other hand, there are short-period analyses where the attempt to avoid these
inconsistencies (by referring to heterogeneous capital goods) leaves room for the
doubts about the fruitfulness of the whole approach. In this situation, the
temptation might be strong to use the short-period approach to provide some
rationalisation of results which derived their plausibility only from a long-run
analysis.
107


Unfortunately Garegnani’s arguments were not replied by Bliss or by any other
marginalist scholar at that moment. Although it cannot be assumed that communication
between the different sides of the debates has been an easy matter for the participants,

102
Debreu (1959), pp. 74-75.
103
Bliss (1970), p. 438.
104
Bliss points out that “Garegnani’s error” would have been due to an overlook of “expected changes
[of prices] over time” (Bliss 1970, p. 438). However, price changes “over time” is a matter that the
traditional approach of long period equilibrium took into account. Cf. Chapter 2, Paragraph 2.9 above.
105
Bliss (1970, p. 438).
106
Garegnani (1970b).
107
Garegnani (1970b, p. 439, emphasis added).
Chapt er 4 – A f i rst phase i n t he cont roversi es
89
because communication is per se a question with many ramifications, yet this kind of
neoclassical reaction, which no doubt obscures the argument for an otherwise better
communication, is raised for a purpose. While it is worth noticing the misunderstanding
surrounding Bliss’s appreciation (though, it may seem, not respect, in the light of
Samuelson’s view after the Symposium)
108
of the results of the controversies, on the
other hand Bliss is pointing to a way-out to those negative results, the reason of which
Garegnani has already envisaged clearly in his brief rejoinder.
These kinds of exchanges reflect what then turned out to be an increasing attitude
by defenders of the neoclassical theory soon after “more careful consideration” – to use
Ferguson’s wording – the phenomena of ‘reverse capital deepening’ and ‘reswitching’
raised in the debates was “accorded”. As we shall see in Chapter 6 in much more detail,
“the temptation to use the short-period approach”, as Garegnani points out, was in fact
the way out adopted by the neoclassical approach to the inconsistencies of the
traditional theory unveiled in the controversies by the critics.
4.15. The zenith of the debates
It has been seen at the end of the previous chapter the theoretic problems surrounding
the notion of capital for the theory. In this chapter, in addition, we have seen how
defenders of neoclassical theory reacted against the threats of ‘reswitching’ and
‘reverse capital deepening’. In this regard, it is worth mentioning Samuelson’s
influence in stimulating the debates: on the one hand we have seen how Samuelson’s
Surrogate article resulted in what we can here call one of the cornerstones of this
controversy – namely Garegnani’s criticisms of the neoclassical theory, and his
struggling for an alternative. On the other hand, we have again Samuelson – though
indirectly – trying to defend dominant theory but at the same time bringing about one of
the most notorious results of these controversies for the discipline as a whole – the 1966
Symposium. It is interesting to look at a retrospective view on Samuelson’s role in the
controversies from another authoritative voice from MIT, Robert Solow:

Samuelson had the notion that universal diminishing returns, together with the
assumption that every line of production involves every commodity directly or
indirectly, would rule out the paradoxes of reswitching even in models with many
capital goods. […] [T]he notion is false. But sometime in 1964 Samuelson planted
it in the mind of David Levhari. With such a gardener it would be hard not to grow
something; in 1965 Levhari published a proof of the notion. Since the notion is
false, the proof contains a slip. Counterexamples were soon produced and
reswitching became an issue instead of a curiosum, like the Giffen good.
109


108
See Paragraph 4.11 above.
109
Solow (1983) p. 184. In Chapter 5 below, Solow’s participation in the controversies is analysed.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

90

The Symposium in paradoxes in capital theory definitely revealed that the seeds
planted by Samuelson – to follow Solow’s analogy – were totally sterile to fence in the
traditional theory from the critical Cambridge. The lesson to be learned from the
Symposium is that ‘reswitching’ and more generally paradoxes in neoclassical capital
theory cannot be ruled out in any one of the imaginable worlds proposed by
neoclassical authors. The debates were gaining momentum and defeat from the
neoclassical side had to be admitted – though, as noted, hitherto sporadically. The
critically oriented side of the controversies was winning the debate in the theory of
capital by showing how groundless the idea of conceiving of capital as a factor of
production is.
In the next chapter we make a step back in the history of the debates. We shall
discuss whether and how the question of capital appeared in the literature prior to 1960.
Then, Chapter 6 returns to the consequences of the tensions created by the implications
of ‘reswitching’ and ‘reverse capital deepening’ in a second phase of the controversy.


Chapter 5
A step back in the debates: Joan Robinson’s role
in the controversies and first neoclassical reactions
5.1. Introduction
1

Joan Robinson was a renowned participant in the Cambridge controversies for various
reasons. One of them is her famous and provocative article published in 1953 by the
Review of Economics Studies, which has been considered as the kick-off for the so-
called Cambridge debates in capital theory.
It has been said above that we have preferred to place at the centre the results of
‘reswitching’ and ‘reverse capital deepening’ and their implications because these
phenomena allow us to clearly grasp the relevance of the debates. Now, in the light of
the implications of the critique, we turn to the analysis of the first arguments raised in
the Cambridge controversies by Robinson and her immediate contenders. The object of
this chapter is not to provide an exhaustive account of all the contributions of the
Cambridge economist but rather to critically focus on two points that we have deemed
the most relevant for the problems of capital. While the first point is focused on why
Joan Robinson might have chosen the aggregate production function as her main target
of attack, the second one is to explain as simple as possible Robinson’s analysis in
order to understand the kind of criticisms she raised and how far those criticisms
actually touch the heart of the problem of capital as we have discussed above (Chapters
3 and 4). Lastly some neoclassical reactions raised against Robinson critique in the
mid-1950s (Champernowne 1953; Solow 1955, 1956; Swan 1956) will be reviewed.
5.2. Robinson’s critique: Aggregate production functions as her main target
Joan Robinson starts her critique of the neoclassical capital theory in her vividly written
article of 1953. In that article Robinson accuses the neoclassical economist of not

1
We have considered for the part regarding Robinson the following works: Robinson (1953),
(1956) chapters 8-10 and “Diagrams”, and (1962a). See also Naqvi and Robinson (1967); Harcourt
(1972) chapter 1.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

92
making clear, when he teaches students the production function O=f(L,C), “in what
units C is measured.” In fact, Joan Robinson’s critique of capital has almost always
been associated with the problem of considering this ‘factor’ as an argument, together
with labour, of an aggregate production function, in which, according to this economist,
“the relative prices of the factors of production [i.e. the wage and interest rates] are
exhibited as a function of the ratio in which they are employed.”
2

The aggregate production function Robinson criticises is a very particular, and
indeed theoretically secondary, representation of the neoclassical long period
equilibrium theory, where both output and capital per worker are assumed to be
homogeneous.
3
Within this construct it is assumed that the physical composition of
capital is instantaneously adjusted by the supply and demand mechanisms. So the use of
aggregate functions already entails accepting the supply and demand approach to value
and distribution. But Robinson does not focus her critique on the supply and demand
mechanisms as such, but rather on the long run equilibrium conditions which must
prevail – always according to Robinson – in order to use that production function.
Robinson is criticising the fact that in real life the composition of capital cannot be
instantaneously adjusted due to problems of uncertainty and expectations underlying
problems of effective demand. Relying her arguments on Keynesian premises,
Robinson believes that the production function is a very poor instrument to analyse
process of accumulation occurring in real life because the “pseudo-production
function”
4
(and hence the neoclassical theory, according to Robinson)
5
could be only
used in conditions of long run equilibrium, which already implies that expectations
about the uncertain future must be fulfilled.
In fact Robinson sees the problem of capital only limited to this representation of
the long period equilibrium, when she says that in equilibrium the value of capital could
be measured either in terms of costs or current purchasing power or future earnings

but to do so, we have to begin by taking the rate of interest as given, whereas the
main purpose of the production function is to show how wages and the rate of
interest are determined by technical conditions and the factor ratio.
6


The citation makes it clear that for Robinson it is only in the aggregate production
function representation that the problem of valuation might arise; but this view is
devoid of a critical examination of the problematic issue entailed by a value measure of

2
Robinson (1953), p. 81.
3
Cf. Samuelson’s construct in Chapter 4, above.
4
Originally Robinson labelled the production function as a “pseudo-productivity curve”
(Robinson 1953, p. 106). See Paragraph 5.4, Chapter 5, below.
5
Robinson (1970a) considers that neither Wicksell nor Walras have a theory of distribution at all,
and that the sole neoclassical theory (‘neo-neoclassical’ in her terms) is that represented by the
aggregate production functions.
6
Robinson (1953, p. 81).
Chapt er 5 – A st ep back i n t he debat es
93
the endowment of capital for the theory independently of the aggregate production
function. In fact, as we have seen above, it is neither necessary nor sufficient the
existence of any such construct to realise the necessity of a value measure for capital in
marginal, long-period equilibrium theory.
7
On the other hand, it would seem that the
strength of her arguments crucially lies in her notion of long-run equilibrium.
5.3. Equilibrium and expectations
Robinson conceives of equilibrium as a situation where the composition of the physical
capital stock is instantaneously fully adapted. The core of her arguments on equilibrium
lies precisely in those Keynesian premises encompassing issues such as investment,
expectations and uncertainty. In this regard, it is worth quoting her definition of
equilibrium, which characterised all her further works on the Cambridge controversy:

Equilibrium requires that the rate of profit ruling today was expected to be ruling
today when investment in any plant now extant was made, and the expectation of
future profits obtaining today was expected to obtain today. Thus the value of
capital in existence today is equal to its supply price calculated in this manner.
The heavy weight which this method of valuing capital puts upon the assumption
of equilibrium emphasizes the impossibility of valuing capital in an uncertain
world. In a world where unexpected events occur which alter values, the point of
view of the man of deeds, making investment decisions about the future, and of
the man of words making observations about the past, are irreconcilable, and all
we can do is botch up some conventional method of measuring capital that will
satisfy neither of them.
8


According to Robinson, equilibrium of realised expectations in a world where
uncertainty prevails implies that the capital stock composition was adjusted to the
conditions expected by investors in the past, when investment actually was made. Using
Robinson’s words, the “equilibrium rate of profit” today (which will equal the interest
rate) can be used to value capital in terms of current purchasing power, but, also, in
terms of the costs of production because that rate “was expected to be ruling today”
when investment (in the past) was made; then it will be also equal to the value of
capital in terms of future earnings, because that rate of profits prevailing today, is
expected to prevail in the future.
But the corollary of Robinson’s message would be that in a state of realised
expectations the theory would not be faced with any problem regarding the issue of

7
Our visit of Wicksell’s general dis-aggregated, long-period equilibrium system has illustrated
this issue. See Appendix B, below and Chapter 2, above.
8
Robinson (1953), p. 96.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

94
conceiving of capital in value terms for the determination of a neoclassical equilibrium.
Moreover, according to Robinson:

In a position of equilibrium […] there is a quantity [of capital] which can be
translated from one number to another by changing the unit. This is the definition
of equilibrium. It entails […] that the human beings in the situation are expecting
the future to be just like the past.
9


These citations suggest that the theory would meet the question of capital (the
illegitimacy of conceiving of capital in value terms to explain prices and distribution in
terms of supply and demand, as it has been analysed in Chapters 2-3-4 above) only in a
state of not realised expectations. The problem of having one of the data in value terms
for the determination of equilibrium – outside the one-commodity aggregate production
function – is totally got rid of by Robinson’s analysis. Rather, as Robinson also informs
us in her 1956 book, in equilibrium capital can be given a meaning:

Under the protection of our assumptions [of equilibrium], there is no difficulty in
measuring the stock of capital within any one economy.
10


Under realised expectations (and free competition) Robinson believes that the
traditional adjustment mechanisms of supply and demand work properly:

an accidental increase in the stock of capital above the equilibrium quantity would
depress the rate of profit, and cause the additional capital to be consumed.
11


Nevertheless Robinson criticises the neoclassical theory on the grounds that it does
not provide the appropriate tools for an analysis of accumulation owing to the
Keynesian problems such as expectations and uncertainty, on the other hand the
neoclassical relation between accumulation of capital and the rate of interest seems not
to be put under dispute by the English economist.


9
Robinson (1953), p. 82.
10
Robinson (1956), p. 117.
11
Robinson (1953), p. 87.
Chapt er 5 – A st ep back i n t he debat es
95
5.4. Understanding Robinson’s way: The ‘Pseudo-production function’ and the
‘curiosum’
The production function is considered by the Cambridge economist as her main target
of attack because she believes that the analysis proposed by a production function is
nothing more than an exercise of static comparison of long period equilibrium positions
of the economy, where the profit rate is what it is because it has for a long time been so,
and it is expected to be at that level in the (infinite) future as well. As early as in her
1953 article, we are informed by Robinson that this kind of exercises are not
appropriate for an analysis of changes (of factor’s supplies, of inventions, of technical
knowledge) running in historical time.
12
As she persistently insisted, to compare two
equilibrium positions on the curve due to a shift in some parameter does not tell us so
much about how those positions were actually reached, because as we pass from one
point to another one, all the history (the form of the function, the position of the
function) is being changed.
13

To show that the production function is ridiculous for an analysis of changes,
Robinson analyses the relation between output, labour, capital, distribution and
technical knowledge in an economy under equilibrium conditions.
14
In fact it is only in
equilibrium of full realised expectations and knowing distribution in advance that that
construct can be built. That is why Robinson called it the ‘pseudo-production function’.
Outside equilibrium of full employment that construct is impossible to build, Robinson
thinks. It is in connection with the building of that construct – which is only useful to
make comparisons but not for analysing changes – where Robinson’s analysis
tangentially touches upon the question of capital. To such constructs we now turn to.
To begin with it is necessary to introduce the ‘productivity curve’. This curve,
which has been used by Robinson since 1953, is similar to a neoclassical one-good
production function, with the difference that it is drawn for a given value of r.


12
Cf. Robinson (1953) p. 100: “In short, the comparison between equilibrium positions with
different factor ratios cannot be used to analyse changes in the factor ratio taking place through
time, and it is impossible to discuss changes (as opposed to differences) in neo-classical terms.
The production function, it seems, has a very limited relevance to actual problems.”
13
Cf. e.g. the following critique extracted from Robinson (1974, p. 58): “The problem of the
‘measurement of capital’ is a minor element in the criticism of the neo-classical doctrines […]
What they pretend to offer is nothing but an error in methodology – a confusion between
comparisons of imagined equilibrium positions and a process of accumulation going on through
history.”
14
See Robinson (1953), pp. 85-87.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

96

Figure 5.1. The productivity curve

This curve represents, for a given r, the relation between the (net) output per man
(vertical axis) and the value of capital goods per man (horizontal axis) in terms of
wage-units (labour commanded), or as Robinson preferred to call it, “real capital”.
15
For
the case illustrated, each corner on the curve represents the use of a determined
technique (o, |, ¸, o, etc.); while stretches on it represent combination of techniques.
For the given r (1/0N, represented by the slope of the tangent drawn) the equilibrium is
reached only at the stretch where ¸ and | are used, the other techniques being not
eligible at that level of r.
The ‘pseudo-production function’
16

Robinson draws, for each feasible value of r, a determined productivity curve. Then by
generalising the productivity curves for all values of r we may draw what Robinson
calls the ‘pseudo-production function’.
17
This generalisation is done by choosing the
maximum rate of (net) output at the minimum costs (‘real capital’) for all feasible
values of r. The pseudo-production function is the result of joining all equilibrium
segments for every r.


15
Notice that the two aggregates are expressed in different numéraire. Yet it is not a problem,
since Robinson is just measuring those magnitudes in a long-run equilibrium in order to compare
the different positions on that curve at a given rate of interest (profits) – and hence a given wage
rate. On this see Salvadori (1996) and Pasinetti (1996).
16
Robinson (1953; 1956) also calls it factor-ratio curve.
17
Once we have got these productivity curves corresponding to the whole possible values of r we
must draw tangents from every level of the wage rate onto the former curves. Any point lying both
on each tangent and on the corresponding productivity curve represents equilibrium positions.
Chapt er 5 – A st ep back i n t he debat es
97

Figure 5.2. Robinson’s ‘pseudo-production function’

It is the present author’s opinion that in the above analysis clarity is lacking. In fact
the stretches of the productivity curves that do not belong to the pseudo-production
function
18
do not bear any economic interpretation and so it is hard to understand them
with the analytics of Robinson. The construct of the pseudo-production function
crucially depends on the relation between one productivity curve and the next one as r
is modified – a relation which depends crucially on the reaction of the cost of
equipment to differences r. Again, this depends, we are told by Robinson, on the
“gestation period” and “length of life of equipments”. But, since in Robinson’s analysis
the production conditions of capital goods associated with each technique are not
specified, her productivity curves are entirely arbitrary. In fact Robinson in her analysis
prefers to take the traditional premises for granted in that the value of capital (in terms
of output) in the economy will be higher as the rate of profit falls due to the use of a
‘more mechanised’ technique, which becomes profitable for lower rates of profits.
19

It is possible to overcome the deficiency around Robinson’s construction of the
pseudo-production function. To such a purpose, we can precisely use the analysis we
have described in Chapter 3 regarding wage curves and the wage-frontier (or envelope).
My reinterpretation of Robinson’s construct in the light of the analysis of wage curves
will allow us to draw the pseudo-production function, reflecting the relation between
output and capital per man, without showing points out of equilibrium in some
stretches, which might result misleading for readers otherwise. In fact by drawing on
the analysis provided by the wage curves, which supplies us with the relationships

18
It is the measurement of capital in wage terms associated with non profitable techniques at a
given r.
19
Regarding the relation between one productivity curve and the next, Robinson (1953, p. 94)
comments that “there is little to be said a priori” and she adds, on the other hand, that “it is
reasonable to suppose that the most mechanised techniques are the most sensitive to the rate of
interest, so that the family of curves fans out laterally [to the right] as it rises.” Cf. also Pasinetti’s
(1978, p. 186 n. 2) critical view on Robinson.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

98
between value of (net) output per man, capital per man, and distribution, a pseudo-
production function can be rebuilt. We shall analyse two cases: i) a typical neoclassical
case, and ii) ‘reswitching’ (Robinson’s ‘curiosum’.)
i) Neoclassical case. Four techniques are considered. Every production techniques
have values of productive coefficients such that D>0. (cf. Paragraph 3.2 above)



Figure 5.3. Wage curves and the ‘pseudo-production function’ (neoclassical case)

On the right hand side of Figure 5.3 we draw four wage curves, depicting
techniques o, |, ¸, o. All these techniques show the neoclassical property that the value
of capital in terms of output corresponding to each technique increases as r falls. This is
the example with which Robinson exemplifies her analysis.
Assume that equilibrium conditions entail r=r
¸o
then we can univocally pass from
our measure of capital in terms of output to the value of capital in wage units (real
capital). We know that at that r both ¸ and o are equi-profitable. At that point the value
of capital in terms of output varies from the value associated with o to the value
associated with ¸, which is higher than the former. As we pass to measure capital in
wage units, real capital increases from the value associated with o to the one associated
with ¸ at w=w
¸o
: the value in wage units is changing due to a change in the composition
of the stock of capital from o to ¸ at that given r and w. The pseudo-production
function is built by joining each value of real capital to the corresponding level of
output: for example the minimum value of real capital is associated with level D, and
the maximum real capital at that rate of profit corresponds to C. So when we have a
switch point we have inclined stretches in Robinson’s pseudo-production function.
As we move from right to left onto the ‘envelope’ (from higher to lower values of
r) we find that it is only ¸ the only profitable technique. Then, at some r just a little
Chapt er 5 – A st ep back i n t he debat es
99
higher than r
¸o
the value of capital in wage units decreases as w at which ¸ is now
profitable is higher. For higher w, ¸ technique continues to be the only profitable one
and therefore the value of capital in wage units decreases, thus showing a horizontal bar
joining different real capital levels to the corresponding level of output associated with
¸, that is C. So when we have dominance of a technique over the rest we have a
horizontal bar in Robinson’s pseudo-production function.
At a lower r such that ¸ and | are both profitable, the value of capital in wage units
varies from the value associated with ¸ to the value associated with |. Again we can see
that ‘real capital’ at that r increases due to the change in the composition in the stock
from ¸ capital-goods to | capital-goods.
This analysis may be extended to lower values of r where | is the only profitable
technique; then we can see that at an even lower r, o is also profitable and so on.



Figure 5.4. The ‘pseudo-production function’ (from Figure 5.3)

Thus, thanks to this re-interpretation a ‘pseudo-production function’ can be built on
the grounds provided by the analysis of choice of techniques in terms of the wage
curves. Because Joan Robinson tried to show the equilibrium positions, the relevant
curve is the following ‘pseudo-production function’ (and not Figure 5.2):
20


20
The important thing to note is that we need not draw the productivity curves which are not
competitive. For example the stretches ¸
3
-|
3
-o
3
or |
2
-o
2
in figure 5.2 are not drawn on an
objective basis. Robinson might have drawn them with a determined slope to show that those
positions are not of equilibrium (and that the techniques associated with those points are not
profitable) but no objective method can support that drawing, because they are arbitrarily drawn
on the basis of having already assumed that a determined technique is cost minimising at a given r
while others are not. The value of capital goods (like the value of any other produced commodity)
might not have a univocal relation with changes distribution. In fact this is the general case. What
is interesting to notice is that Robinson, even until 1969 (Accumulation of Capital, 3rd ed.) when
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

100


Figure 5.5. Robinson’s ‘pseudo-production function’ amended

ii) ‘Reswitching’ of techniques (implying ‘reverse capital deepening’) or Joan
Robinson’s “Curiosum”. There are two techniques such that D
o
=0, D
|
<0 (W
o
>W
|
and
R
o
>R
|
for every r and w)




Figures 5.6-5.7. Wage-curves (‘reswitching’) and the ‘pseudo-production function’ (from
Figure 5.6)

When r<r
1
o is more profitable than | and this is reflected by a horizontal bar
measuring different values of capital in wage units corresponding to net output level A.


the ‘reswitching’ debate had got already started, continued to use the productivity curves as her
main instruments. Cf. Robinson (1969) “Diagrams” p. 427.
Chapt er 5 – A st ep back i n t he debat es
101
Since the value of capital in terms of output when o is used is a constant value for any
level of r so that as r falls, real capital falls. When r=r
1
, we have co-existence of
techniques. At that rate of profit real capital falls, provided that entrepreneurs are
changing the composition from o capital-goods to | -ones. When r increases further,
but does not reach r
2
yet, | is the only profitable technique. We have a horizontal
stretch here representing a higher real capital as a result of dividing an increasing value
of capital by a lower and lower w. When r=r
2
we have ‘reswitching’, since o is also
profitable at this r and will continue to be so as that r increases further and further.
When r>r
2
only o is the most profitable technique. ‘Real capital’ increases as a result
of dividing a constant value of capital in terms of output by a lower and lower w.
The form of the factor-ratio curve resembles a U-like line shape.
21
This is what
Robinson labelled a “Ruth Cohen’s curiosum”.
22

5.5. Conclusion
The main question which emerges from our brief analysis of some main criticisms
raised by Robinson is: Does her particular way of criticisms point to the heart of the
Cambridge controversy, as we have seen in above chapters?
What emerges from our review of some Robinson’s works is that she might have
chosen the aggregate production function as her main target of attack because she might
probably have thought that it was a didactical way to spread the critique against this

21
We could draw the pseudo-production function in terms of capital measured in terms of output
by plotting in the output-capital (in output terms) diagram the values of capital and net output
obtained from the wage curves. Basically, the function will have a triangle-like form. As the rate
of profit rises from its minimum (r=0) to its maximum (R
a
), points (y, k) move anticlockwise.

Figure 5.8. The ‘pseudo-production function’ in terms of capital per worker in terms of output

22
Robinson (1953), p. 106, n. 1. This phenomenon, later known as ‘reswitching’, was ascribed to
the agricultural economist Ruth Cohen by Robinson; however, Robinson (1970a, p. 309) later
revealed that it was a “personal joke” and recognised that the critical arguments raised in her
article were a “preview” of Sraffa’s (1960). Also, Robinson (1970a, p. 309) recognises the
influence of the famous Italian economist, since the clues for her arguments had been “picked up
from Sraffa’s Preface to Ricardo’s Principles.”
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

102
particular representation of the long period neoclassical equilibrium. In addition, her
remarks reveal the intuitiveness of her arguments used to broaden the view of
traditional theory by Cambridge (UK).
23
Lastly, Robinson’s arguments also reveal her
revival of Keynesian economics at a time when some neoclassical scholars, such as
Samuelson, Modigliani, and Hicks,
24
had started to bring Keynes in the neoclassical
theoretic framework.
25

Notwithstanding Robinson’s harsh criticisms of the neoclassical production
function for analysis of changes, it is hard to find in her writings an actual critical
assessment of the supply and demand mechanisms which underlie the core of the theory
she attempted to deny. In particular, on the relation between ‘capital’ and the rate of
interest Robinson says:

The production function has a very limited relevance to actual problems and after
all these labours we can add little to the platitudes with which we began: in
country Gamma, where the road builders use wooden shovels, if more capital had
been accumulated in the past, relatively to labour available for employment, the
level of real wages would probably have been higher and the technique of
production more mechanised, and, given the amount of capital accumulated, the
more mechanised the technique of production, the smaller the amount of
employment would have been.
26


where the relation between ‘accumulation’ and the interest rate is taken by the critical
economist on traditional grounds. Moreover, in Robinson’s analysis it is hard to find
the analytical roles of capital conceived as a single magnitude for a determination of
equilibrium of supply and demand. The necessity of capital in value terms to determine
a uniform rate of profits and the key role of a downward-sloping demand curve for that
factor are aspects that are not discussed by Robinson, even when Robinson herself
faced what later became known as ‘reverse capital deepening’ and ‘reswitching’, which
destroy the belief on an inverse relation between capital and the rate of interest. In fact
the famous Cambridge economist seemed to downplay those phenomena when she
speaks of a ‘curiosum’, though these phenomena rather appear to be general. The
‘curiosum’ case is regarded by Joan Robinson as unimportant and hence dismissed:


23
Cf. “The student of economic theory […] is instructed to assume all workers alike, and to measure
[labour] in man-hours; he is told something about the index-number problem involved in choosing a
unit of output; and then he is hurried on the next question, in the hope that he will forget to ask in
what units C [capital] is measured. Before ever he does ask, he has become a professor, and so
sloppy habits of thought are handed on from one generation to the next.” Robinson (1953) p. 81.
24
E.g. Hicks (1937).
25
And this could also be appreciated by recalling the post-Keynesian literature which arose in the
wake of the controversies. See Pasinetti (1962), Kaldor (1955), Robinson (1956; 1962a).
26
Robinson (1953), p. 100.
Chapt er 5 – A st ep back i n t he debat es
103
A good deal of exploration is needed before we can say whether the above
[curiosum] is a mere theoretical rigmarole, or whether there is likely to be
anything in reality corresponding to it.
27


Nevertheless this economist did not bestow much importance on this line of
criticism, her preferred way of attacking neoclassical theory reveals weaker in
comparison with the arguments of ‘reswitching’ and ‘reverse capital deepening’, it was
due to Robinson’s vivid intervention that later on stirred up a series of replies from the
neoclassical camp. To some of those first reactions we turn in what follows.
5.6. Champernowne’s (1953) contribution
David Champernowne’s 1953 article was the first echo of Joan Robinson’s kick-off of
the controversy. Champernowne’s endeavour was to provide a kind of measure of
capital alternative to the one adopted by Robinson (‘real capital’), because he
considered her measure as being “not convenient”, for two “anomalies” might arise.
What Champernowne rejects, if capital is measured in “Joan Robinson units” (in wage
units), are two possible phenomena: i) that the same physical capital equipment may
have different values at two different ‘stationary’ positions due to the two different
values both of w and of r associated with each position, and ii) that the production
function may not preserve a 1-1 relationship between output per capita and capital per
capita. The last possibility may lead to the “paradoxical result” that the relation
between the output per capita and the capital per worker might be negatively related.
28

Champernowne wants to overcome Robinson’s puzzle in order for the production
function to determine distribution.
For such a purpose Champernowne built a ‘chain-index’ measure of capital. The
‘chain-index’ is a device to measure the capital as an argument in the production
function in such a way as to make it possible to determine the distribution of product
between wages and profits through the partial derivatives of that function.
29
To form

27
Robinson (1953), p. 106. In Robinson (1956) she totally dismisses the pursue of this line of
critique: “[The curiosum is] a somewhat intricate piece of analysis which is not of great
importance” (1956, p. 109, n. 1).
28
Champernowne (1953), p. 112. In terms of figure 5.2 above, the first “anomaly” pointed out by
Champernowne can be seen in the horizontal bars of Joan Robinson’s ‘pseudo-production
function’, where only one technique is being used but bear different values owing to different
values of r and w. The second anomaly, on the other hand, occurs when to identical values of ‘real
capital’ there correspond different levels of net output per worker – hence the ‘pseudo-production
function’ is actually not a function for some intervals of real capital.
29
Cf. “[T]he use of [the chain index] removes the more glaring difficulties in the way of regarding
aggregate output as a function of the amounts of labour and capital employed. In particular, we
shall show that the rewards per unit of the factors are once again given by the partial derivatives of
the aggregate production function if stationary state conditions with perfect competition are
assumed.” Champernowne (1953), p. 115.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

104
such a construct in a historical sequence,
30
the index of capital is increased at each step
of the sequence by the proportion in which the cost of capital at current w and r at the
end of the step exceeded the cost of capital at the beginning of that step, calculated at
those w and r. This construct has the apparent advantage, Champernowne believes, of
removing the effects of the interest rates on the costs, and thus “paradoxical results” or
“anomalies” would be prevented.
31

But how sure was Champernowne that his device would have removed the “glaring
difficulties”? As we may now speculate, in the light of ‘reswitching’ and ‘reverse
capital deepening’, there is no univocal relation between the costs of capital and
changes in distribution. However, what is interesting to notice is that, in order to
construct the ‘chain-index’ for Champernowne’s purposes, he had to assume that:
32

a) if any one technique, having been the most (or equi- most) profitable technique
at a given r (or range of values of r), could never reappear again at another value (or
range of values); and that
b) of two techniques, which are equiprofitable at a given r, it is the one with higher
output per capita and higher capital per capita that is more profitable (and hence
chosen) at lower values of r.

In other words the only way to keep the value of capital per worker to move in
accordance with the traditional principles, so as to achieve a one-to-one relation
between output and capital per worker, is by a priori assuming that phenomena like
‘reswitching’ (point a) and ‘reverse capital deepening’ (point b) not to arise. (Cf.
Appendix D, below)
So, in general, Champernowne’s construct cannot avoid “anomalies”. In fact, he
produced examples of “exceptional cases” which may lead to “paradox results” (a triple
valued production function) and points out that

formidable difficulties arise, unless we rule out cases in which a lowering of
interest rates can cause the introduction of techniques with a lower productivity
than those used up until then.
33


Nevertheless Champernowne recognises that these cases cannot be ruled out “on
logical grounds”, he says that “intuition suggests that the excluded case is unrealistic”
(p. 119). It is interesting to note that it is “intuition” what might have led the author to

30
Champernowne regards the long period equilibrium positions as ‘sequence of equilibria’ or as
historical sequences of equilibria.
31
Intuitively, Champernowne’s chain-index attempts to eliminate, so to speak, the ‘zags’ (the first
anomaly in Robinson’s pseudo-production function) and to diminish the slope of the successive
‘zigs’ (the second anomaly) in Robinson’s production function. The present author owes the ‘zig-
zag’ analogy to Harcourt (1972, chapter 1).
32
Champernowne (1953), pp. 115-116.
33
Champernowne (1953), p. 130, emphasis in the original.
Chapt er 5 – A st ep back i n t he debat es
105
believe in the basic, foundational, premise of the theory i.e. that “a gradual fall in the
rate of interest would entail increases in the quantity of capital per head”
34
namely – a
belief that it has for long been taken for granted in the discipline at the time of the first
blows of this controversy.
In conclusion, the ‘chain-index’ measure of capital appears to be designed only to
analyse a process of capital accumulation by constructing the path of sequence
equilibria
35
by already assuming that the neoclassical theory works in the ‘correct way’;
it does not, however, seem to address the basic question on which the issues of the
measure of capital turn: whether or not the marginalist supply-and-demand functions
provide a logically valid explanation of prices and distribution.
36
On the contrary, this
construction as well as the so-called aggregate production functions, assumes the
explanation in terms of supply and demand to be valid.
37
Finally, it is worth noticing
that Champernowne, like Joan Robinson, did not give his “unrealistic” case (that is
Robinson’s Ruth Cohen’s curiosum) much importance.
5.7. Further replies to Robinson: Solow (1955; 1956) and Swan (1956)
Besides Champernowne’s response to Robinson’s first complaints are two further
authors from the neoclassical side who contributed to the debates in the first exchanges.
Anyway the attempts of these responses have not been to solve for the problems that

34
Champernowne (1953), p. 118.
35
As Champernowne calls the different long period equilibrium positions of the economy.
36
It might be also pointed out, as Petri (2000) has suggested, that Champernowne’s chain index
assumes that changes of technique involve the rate of profit stopping at a value in which two
techniques are equiprofitable. But changes of techniques, in disequilibrium adjustment processes,
do not necessarily involve that the change in the rate of profit has to stop at the level in which two
techniques are competitive. And, what is more, when economies change techniques (production
systems), the ‘form’ or kind of capital goods will in general change, and so will the value of
capital. Hence the chain index is not a “reliable measure of the quantity of capital”, since the
theory needs this ‘factor’ susceptible to changing form but keeping the quantity unaltered. (Petri
2000, p. 31).
37
On the issue of capital as an argument of the production function, Garegnani (1960) has pointed
out: “[If the] aggregate of capital goods is measured so as to appear as a variable ‘quantity of
capital’ when distribution changes, it will no longer be possible to establish a univocal relation
between the quantity of capital and the physical quantity of the product. And it turns out to be so,
after measuring capital in value terms: given the quantity of the other factors of production, both
the same technique and the same physical quantity of output would be compatible with more than
one quantity of capital.” (p. 85, emphasis in the original, my translation.) Once a production
function is so built, its partial derivatives, which should have resulted in the ‘marginal products’,
are however an entirely different construction from the traditional meaning given to the ‘marginal
products’ of the theory, which needs a physical measurement of all factors.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

106
Robinson had raised, but merely to dismiss them altogether.
38
We consider here Solow
(1955 and 1956) and Swan (1956).
Solow (1955) considers that the process of production can be visualised as a
process divided into two stages: first, an ‘index’ quantity of capital, K, can be a
function of the physically heterogeneous capital goods, say, C
1
and C
2
. Secondly, the
consumption good, Q, is produced by means of labour and the services of the several
capital goods through function F. However, if the first function can be logically built,
then Q will be a function both of labour and of an ‘index’ of capital. Solow justifies this
procedure by resorting to Leontief’s weak separability condition (the substitution rates
between the factors that one wishes to aggregate – here C
1
and C
2
– must be
independent of the quantities employed of the other factors, L), which is necessary and
sufficient for the representation of the production function of the form Q=F(L,C
1
,C
2
)
into the form Q=H(L,| (C
1
,C
2
))=H(L,K), if K=| (C
1
,C
2
). However, there are two points
worth noticing: first, a priori, there is no justification why C
1
, C
2
should be services of
heterogeneous capital goods. Rather, they may stand for amounts of original factors
such as (different types of) labour or land both measured in physical units. So this is
sufficient to say that Solow’s attempt does not answer to Robinson’s problems.
Secondly, in terms of real economies, it is very unlikely to find situations such that the
marginal substitution rate of C
1
and C
2
are independent of the quantity of labour. In fact
Solow himself commented on this: “[the condition of independence of the marginal
substitution rate from the quantities employed of the other factors] is not satisfied [e.g.]
for one-ton trucks and two-ton trucks; technical substitution possibilities will depend on
the number of drivers available.”
39
However, Solow believes that “there is a whole class
of situations in which the condition may be expected to hold […] For example, we can
imagine that C
1
and C
2
to be two kinds of equipment for generating electricity which is
then used in further production. [E]lectric power would be an index of the capital
inputs” (Solow 1955, p. 103). To conclude. Solow’s attempt is nothing but a clear-cut
defence of the validity of the marginalist propositions. Nevertheless, for such a defence,
he cannot overcome the difficulty entailed by the fact that any index of capital depends
on changes in distribution.
40
Moreover, and more important, this relation between the
index and changes in distribution is not univocal.

38
To dismiss the problem that distribution cannot be determined by the marginal productivities of
the factors. Solow dismisses it by considering the problem raised by Robinson as “antiquated”, see
Solow (1955), p. 101 and Robinson (1955), p. 247.
39
Solow (1955), p. 103.
40
Actually, in the article, Solow (1955, p. 106) recognises that his “quantity-of-capital-index” is
also the result of the transformation of a “price index”. Certainly, in the literature, we might come
across many instances in which “prices and quantities indices come in pairs” (Solow 1955, p.
106); cases like this are known as the so-called ‘duality problems’. Solow (1955, p. 106)
comments: “Apart from the comforting thought that at this late date no one should be surprised to
find price-quantity dualities, I have no explanation to offer”. But the problem is well another one:
if capital is expressed in value terms (and it has to be so, if we keep in the long period tradition),
there is of course no guarantee to “justify [Solow’s] statement: that there is a perfectly definite and
(cont.)
Chapt er 5 – A st ep back i n t he debat es
107
One of the most known works from the neoclassical side that indicates that the first
exchanges much centred around aggregate production functions has been Solow (1956).
In that work Solow represents an economy that produces a single output by means of
itself and homogenous labour. The commodity produced can be used either as
consumption good or as an input of itself. It is assumed continuous variability in the
proportions of the factors necessary for producing the output. Full employment of
labour is guaranteed, provided a sufficient flexibility of the wage in perfect
competition. In fact, since it is produced only one commodity there is no indirect
mechanism of factor substitution. The only mechanism available therefore is the direct
one: through technical conditions of production. Since technical conditions are assumed
to be variable, ‘capital’ can always adapt itself as to employ the whole quantity of
labour of the economy. Savings determine investment: the consumption forgone is
saved and hence to be added to the stock of capital. Savings, investment and capital are
all measured in the same technical units. (There is, under these restrictive assumptions,
no problem regarding the irreversibility of investment – another of Robinson’s concerns
in the 1953 article).
Of course, there is no problem regarding the measure of the marginal productivity
and of the measure of capital. Robinson’s concerns about capital are plainly dismissed.
We are in an economy just as the same as the one we have described at the beginning of
this work in Chapter 2: the only way to save the neoclassical propositions, in economies
with heterogeneous capital goods as between techniques of production for the
consumption good, is to assume that the output, actually composed of heterogeneous
goods, is produced as if it were a single commodity which is produced by itself and
labour. His belief in aggregate production functions ultimately relies on the acceptance
of the basic premises of the theory.
41



Let us now focus on Swan’s (Swan, 1956) reactions to Robinson’s 1953 kicking-off
paper. Swan (1956) is concerned with Robinson’s notion of capital as a factor of
production which is accumulated over a period of time by savings. In his long appendix
to that article, he regards capital as “the number of meccano sets”. This ‘meccano’ can


consistent sense in which it can be said that the relative factor price ratio P
K
/P
L
is a decreasing
function of the ratio of capital to labor K/L.” (Solow 1955, p. 105).
41
In Solow (1957), where he attempts to apply the (1956) model to US data for the period 1909-
1949, his belief in aggregate production function (and hence in neoclassical theory) is expressed in
the following way: “Were the data available, it would be better to apply the analysis to some
precisely defined production function with many precisely defined inputs. One can at least hope
that an aggregate analysis gives some notion of the way a detailed analysis would lead.” Solow
(1957) p. 312, n. 1. We have, however, seen that even when the heterogeneous capital goods are
measured in technical units according to the different techniques for producing the consumption
good, there is no guarantee that capital employed in the economy (gross investment of full-
employment savings) be an inverse function of the rate of interest. (Cf. the discussion of
Samuelson’s ‘surrogate’ function in Chapter 4 above.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

108
adapt itself instantaneously and without costs to any form, according to economic
profitability and technical progress. This attempt too avoids the irreversibility of
investment.
42

Swan assumes the existence of more than one commodity, but he assumes that
prices are constant when distribution changes. This is not very different from Solow’s
(1956) model: the only way to assume that prices do not change when distribution does,
is by assuming an economy where the commodities, which compose the output, are
produced as if it were a single commodity produced by itself and labour – that is
Solow’s model.
Swan affirms that his device does not solve for Robinson’s problems; instead, they
can directly eliminate them. That is to say, Swan takes the assumption of a
“malleable”
43
capital (‘meccano’ sets) as a “scarecrow” assumption in order “to keep
off the index-number birds and Joan Robinson herself”. Nevertheless, this author
considered that, due to the assumptions on constant relative price (including that of the
capital goods ‘meccano’), accumulation – the level of which is, in equilibrium, equal to
savings and investment
44
– will exactly be the current output of ‘meccano’ sets, and
accordingly can always be measured in terms of any kind of consumption good.
On the other hand, Swan agrees with Robinson insofar as she was complaining that
the neoclassical tradition contains no indication of how a technical unit may be devised
in order to measure capital, but, however, he concludes: the neoclassical theory, when
aiming at explaining the marginal variations about the stationary position, does not
depend on the existence of such a unit in which measure ‘capital’, because the “natural
unit of Capital is simply an equilibrium dollar worth regardless the physical variety of
capital goods.”
45
This last statement can hardly be satisfactory, because of the problems
that we have already seen when capital in value terms is considered as argument of
production functions. (Cf. Chapter 2 above.) And also, because the original meaning of
marginal productivity is related to the description of the methods of production of the
consumption good – measured in physical terms – according to the different
combinations of the factors of production which must be measured in physical terms
too. It is by drawing on this, that we can derive the direct factor substitution mechanism
as a base for an explanation of the equilibrating forces of supply and demand.

42
Swan (1956), p. 344.
43
Swan (1956), p. 343. Similar other names were used during this period in the literature
concerned with economic growth models to refer to ‘capital’ in aggregate production functions
such as steel, leets, butter, ectoplasm.
44
Cf. Swan (1956), p. 349.
45
Swan (1956), p. 350. Were a technical unit measure of capital existed, the marginal product of
capital, measured in that unit, would be equal to the rate of profit multiplied by the price (in output
terms) of a unit of capital (in technical units). But, since Swan assumes that its price does not
change, capital could be also measured in value terms, in which case its marginal product equals
the rate of profit. However, this does not overcome one of the criticisms raised by Robinson,
which concerns the nonsense of talking about marginal productivity in equilibrium, whereas it is
only in equilibrium that, Robinson assumes, capital can be aggregated.
Chapt er 5 – A st ep back i n t he debat es
109
To conclude on these authors we may say that the only way that neoclassical
economic growth models can be rigorously formulated is in economies which produce
either a single commodity by means of itself and labour (Solow) or a composite good
redistributed without costs as between the different uses (Swan). Outside these
instances, the shortcomings entailed by the dependence of the value of capital upon
distribution cannot be shunned – nevertheless Robinson trifled with the role of a
quantity of capital as a determining one for the definition of the supply-and-demand
functions for productive factors.
5.8. Robinson rejoinders
On Solow’s reply: Robinson does of course not go along with Solow’s view. For her,
the economist from Massachusetts does not deal with the problem of capital. She
correctly asserts that his endeavour to collapse “C
1
and C
2
” into K is rather a way to
represent heterogeneous natural resources into a single index and not a way to represent
capital goods because “nothing is said about the time which it takes to produce them.”
46

In her reply to Solow (1955), as well as in her further interventions in the debates,
47

there is no reference to the problematic attempt to derive inverse functions between
factors’ remuneration and factors’ quantities (employed hence demanded), when capital
is measured in value terms – inverse relationships that Solow of course takes for
granted.
48

On Swan’s reply: Robinson replied to Swan in 1957.
49
As to the question of not
having considered a single-commodity output as Solow did, but an output composed of
many goods, Robinson does not accept the assumption that prices can be kept constant
in Swan’s model, since, if distribution varies, prices will vary too: “Swan must come
off the fence, and say which kind of good is taking as numeraire.”
50

On the other hand, Robinson correctly accuses Swan of thinking of “the wage rate
as being determined by the physical marginal product of labour, which in turn is
determined by the amount of meccano in existence” without, however, stating the units
which it is a quantity of.
51
Therefore she considers Swan’s way as unsatisfactory to

46
Robinson indicates how the issue may be considered: “[C]ompare two machines exactly alike in
all respects except that one is drawn from an economy with a higher product-wage rate. The value
of the two machines is different, and the investment required to create them is different. A
difference in value remains if we deflate them by the wage rate (obtaining capital in labour time),
for in two economies with different wage rates the rate of profit and the rate of interest are
different.” (1955) p. 247. Solow (1955) does not address his answer to that issue.
47
In particular cf. Robinson (1962b; 1970a; 1971a; 1971b; 1974; 1975a).
48
Solow (1955), p. 105.
49
See Robinson (1957).
50
Robinson (1957, p. 104, emphasis added). Cf. Samuelson’s very special case in Chapter 4.
51
Robinson (1957, p. 105). Then, some pages later, Robinson (1957, p. 108) scathingly concludes
her reply to Swan in the following way: “My scheme of analysis [in The Accumulation of Capital,
1956] is intended to make it possible to discuss accumulation in spite of the fact that capital cannot
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

110
answer the questions which she had raised in her 1953 paper. According to Robinson,
equilibrium is obtained, for a given marginal productivity of labour, if the conditions of
the amount and form of capital, the rate of investment, and the propensity to save are
such as to make the wage rate equal to that marginal productivity. This is the kernel of
her reply; a reply relying on Keynesian principles about investment and accumulation.

In this chapter we have seen the particular characteristic that the debates took
during the first exchanges between the two Cambridge. We have seen that discussions
seemed to be related to find a measure of aggregate capital in order to be considered
into aggregate production functions, which would in turn determine distribution.
Swan’s attempt was to consider the malleability assumption (his ‘meccano’ sets).
Solow, instead, plainly regards social output as if it were a single commodity produced
by itself and labour. Champernowne’s attempt does not overcome the anomalies for
which the ‘chain-index’ was built. In sum, in the first reactions to the issues raised by
Robinson the analytical problems connected with the notion of capital for the marginal
theories are still unclear, and that is why we have introduced them in advance (Chapters
3 and 4) without following the chronological path of the many issues debated in the
Cambridge controversies. From a historical standpoint, our brief review over these first
exchanges indicates us that participants in the controversies had to await Sraffa (1960)
from the critical side and Samuelson (1962) from the neoclassical one, in order to find
more clarity in the issues at stake, as shown in Chapter 4.
The following chapter restarts by discussing a new phase in the capital debates,
which was initiated, as I shall argue, by some neoclassical participants soon after the
most salient results analysed in Chapter 4 were acknowledged. In particular we shall
focus on the way in which these advocates of dominant theory received the critique; a
preview of that way was indicated in Chapter 4 too, when Bliss (1970), in his reply to
Garegnani (1970a), drastically changed the background on which the critique had
hitherto been conducted (cf. Paragraph 4.14 above).







be measured in physical units. […] It is not really my fault that capital does not consist of pieces of
meccano.”


Chapter 6
A second phase in the capital controversies and radical
changes in the neoclassical theory
6.1. Introduction
In the previous two chapters we have seen the main discussions of the first phase of the
Cambridge controversies. Especially, the attention has been focused on the results of
‘reswitching’ and ‘reverse capital deepening’, the implications of which touch upon the
basic premises of the neoclassical theory of production, value and distribution
(Chapters 3 and 4). The neoclassical reactions reviewed have shown reluctance to admit
the consequences of the debates; in fact, advocates of the marginalist theory have tried
either to find out the ‘conditions’ to rule out ‘reswitching’ or to minimise their
relevance by pointing to the empirical issue. However, the first phase of the
controversies clearly showed that phenomena of ‘reswitching’ and ‘reverse capital
deepening’ are theoretically irrefutable as some neoclassical contenders were
compelled to admit (e.g. Samuelson 1966a).
In the aftermath of the Symposium and related contributions to the capital debates,
dominant theory increasingly started to rely on a different notion of capital and hence
of equilibrium.
1
This indicates that a different phase in the controversies, in comparison
with the first one, had begun. Actually, the main characteristic of this second phase was
the fact that almost the majority of the core of neoclassical theorists had already
embraced the theoretical path provided by Hicks’ (1946 [1939]) Value and Capital,
who incorporated the Walrasian conception of capital into the supply and demand
approach. The history underlying the incorporation of Walrasian capital into dominant
theory cannot be developed in full detail in this book, as it exceeds the scope of the
present research. However, by relying on related works,
2
we have to bear in mind that
both before and during the Cambridge controversies pure theoretical work related to the
rehabilitation of a Walrasian conception of capital started to be further developed by
neoclassical economists – most of them of mathematical orientation
3
– and this is why a

1
Cf. Paragraph 4.14 above.
2
Garegnani (1976a), Petri (2004, pp. 136-205).
3
e.g. Debreu (1959), Morishima (1960; 1969).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

112
different version of the marginal theory was, so to speak, available to those willing to
use it and renounce to the traditional notion of capital, though still keeping the supply
and demand approach to value and distribution as the only one economic theory. In the
wake of the controversies’ results raised in the first phase, which no doubt put the
dominant theory at stake, neoclassical participants started to argue against the critics of
marginal theory in terms of this version of the theory (which we call Walrasian capital).
In this way, in order to avoid the problematic issue of capital, neoclassical theorists
increasingly started to rely precisely on a physical specification of capital. But this
shift, as we shall see in what follows, embodies a radical change in the nature of the
theory.
The aim of this chapter is to discuss the way in which the neo-Walrasian theorists
received the critique centred around the notion of capital, attempting at the same time to
clear up some misunderstandings created in this phase which, as we shall argue, might
have damaged both the relevance of the capital debates and the communication between
the two sides of this theoretical conflict. For this purpose, in the following three
paragraphs, we shall deal with Walras’s treatment of capital as a set of physically
heterogeneous goods and the inconsistency with the traditional conception of
equilibrium which is entailed by this treatment of capital (6.2), the suggested way-out
to that inconsistency (6.3) and the implications for the approach of adopting the
Walrasian treatment of capital and the transformation of the theory (6.4). Then we shall
return to the historical aspects of the controversies in this second phase (6.5, 6.6 and
6.7).
6.2. Rehabilitation of Walras’s treatment of capital as physically heterogeneous
goods and Walras’s inconsistency
Because the modern versions of the supply and demand theory rely on a Walrasian
treatment of capital we shall here briefly deal with Walras’s theory of capitalisation.
4

As is well known, Walras treats capital as a physically heterogeneous capital goods’
vector. Then every single capital good (from ploughs to screws, from lathes to
tweezers, etc.) is taken as a ‘factor of production’ in the same footing as ‘labour’ and
‘land’.
As has already been analysed in the literature,
5
Walras general equilibrium system
would admit solutions whenever capital services are considered as original factor’s
services. However these solutions do not take into account the fact that capital goods
are reproducible and, therefore, the former would only refer to exchanges and
production without capital.

4
Walras (1954). We shall henceforth refer to the 5
th
edition of 1926 (in French), which was
translated by Jaffé and published in 1954. That edition was strictly alike the 4
th
and definitive
edition of Walras’s Éléments published in 1900. In this paragraph we follow Garegnani (1958;
1960; 1990; 2008).
5
Garegnani (1958; 1960; 1990; 2008). Cf. also Eatwell (1990).
Chapt er 6 – A second phase i n t he capi t al cont roversi es
113
To be coherent with the fact that capital goods are reproducible, Walras introduces
their production into the analysis and hence the equilibrium condition of uniformity of
rates of returns on capitals’ supply prices – a condition, that is, which is implicit in the
equality signs of the price equations of those goods. However, as soon as production of
capital goods is introduced into the general equilibrium system, problems as to
economically meaningful solutions of the system arise.
In order to deal with the production of capital goods Walras introduced the price
equations of each capital good (k, k’,… ) reflecting the costs of production (supply
prices), which, in equilibrium, must equal the demand prices of these goods (which are
the capitalisation of the service prices p
k
). Note that satisfaction of this condition
implies determining a uniform rate of return on the capitals’ supply prices, i – which, as
will be underlined, is crucial to the whole analysis. Further, there will annually emerge
out of the annual output a flow of savings (F
e
(.)) which must be equal to the investment
flow namely – valued quantities of the new capital goods produced
(
' '
...
k k k k
D P D P + + ). All these conditions can be summed up in the following two
blocks of equations, that is, (Y) and (Z).

' ' ' ' '
( ...) ( ... ... ...) ( ' ... ' ... ' ' ..)
e b k t t p p k k k k k t t p p k k k k
F p D k p k p k p k p D k p k p k p k p = + + + + + + + + + + + + + (Y)

(Note that the total amount of ‘capital’ i.e. the savings (investment) is endogenously
determined; see equation Y)

' '
'
' '
... ... ...
' ... ' ... ' ' ..
'
.
k
t t p p k k k k
k
t t p p k k k k
p
k p k p k p k p
i c
p
k p k p k p k p
i c
= + + + + + +
+
= + + + + + +
+
(Z)
6

… … … …


6
Equations in (Z) are the result of equalising supply prices
' '
... ... ...
k t t p p k k k k
P k p k p k p k p = + + + + + + ;
' ' '
' ... ' ... ' ' ...
k t t p p k k k k
P k p k p k p k p = + + + + + + ;
etc. and demand prices
k
k
p
P
i c
=
+
;
'
'
'
k
k
p
P
i c
=
+
; etc.; where p
k
indicates the gross price of the
capital good k (the price from which a net price is obtained after subtracting the amount
corresponding to depreciation and insurance, both expressed by Walras as fraction c>0 of P
k
– the
price of the new capital good). We are not going to discuss the incorrect way in which Walras
treats depreciation of fixed capital since his formula does not take into account the role of interest
upon the very depreciation through time. For a clarification in this regard see Garegnani (1960,
part II; 1990, p. 14) and Sraffa (1960). For the purposes of our present discussion we will use
Walras’s original formula. Up until the 3
rd
edition of Éléments Walras regarded the raw materials
as being consumed as soon as they were produced and not stocked in advance. In the 4
th
edition
circulating capital is treated in a strictly similar way as fixed capital thus assuming for these capital
goods a value of c=1.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

114
where k
t
, k
p
, k
k
, k
k’
… k’
t
, k’
p
, k’
k
, k’
k’
…are the quantities of productive services
necessary to produce k, k’,… .
(Note that the RHS of equations in Z reflect the costs of production of the several
capital goods.)
Although this system has solution, if only formally, it is interesting to see whether
it admits economically meaningful results.
The issue which concerns us here is the economic meaning attached to equations
(Z), because these equations establish the equilibrium condition of equality between
demand and supply prices of the capital goods produced – hence uniformity of rates of
return on the supply prices of those goods. Then, to understand the meaning, we may
consider two cases. First, let us assume that consumers’ tastes led up the yearly gross
savings to be at zero. This being so, according to (Y), we shall see that gross investment
in k, k’’,… , represented in (Y) by D
k
, D
k’
,… , must be all zero, since they cannot, of
course, be negative magnitudes. In this case, the system is over-determined: in general
Walras’s capital formation system is inconsistent and hence has no solution.
7

This case, which considers gross savings zero, would however be sufficient to
argue that there is no general solution to that system because, if a theory boasts itself of
being a general economic theory, it must account for general cases, including also the
case in which savings happen to be zero. However it is interesting to consider the case
when savings are positive.
The only possible way that the rates of return could be uniform throughout the
system lies in the influences of the quantities currently produced of capital goods on the
prices of the capitals’ productive services. To show as clear as possible that in this more
general case (positive savings) there does not exist economically meaningful solutions
either, let us begin by imagining that, owing to the arbitrariness of the endowment
supply of, say, two distinct capital goods, k and k’, i’ > i in (Z) so that we have one
more unknown.
8
To re-establish equality between number of unknowns and
independent equations let us further assume the following relation:

' k
k
D
R
D
=

Since we assume i’ > i free competition of capitals’ bearers would make R increase
so as to bring about effects on both the costs of production of the capital goods and the
productive services’ prices. Of course those effects will crucially depend on the relative

7
Cf. Garegnani (1958, part II; 2008). If a uniform rate of interest were determined under savings
being zero it would be a very particular case, where the configuration of the relative scarcities of
the several capital goods taken as data should have to be appropriate so as to be in accordance with
the consumers’ tastes and preferences which had determined the prices of the productive services
of the capital goods. But in general the arbitrariness of the initial configuration of the capital goods
deprives their quantities from being the appropriate ones determined by consumers’ conditions.
8
These transformed equations we call (Z’), see below.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
115
proportions of the capitals’ productive services in the production of themselves.
9
Since
those proportions in general differ, a change in R will make relative scarcities of capital
goods change, and with it the productive services’ prices and hence the rate of return
over the costs of production.
10

A rise in R will cause a rise in the relative costs of production of k’. Hence to re-
establish equilibrium, and by keeping constant the gross prices of both capital goods, i’
will tend to fall while i will tend to rise, so that the initial disparity between both
interest rates will tend to narrow.
On the other hand the effect of a rise in R on the prices of the productive services
of the capital goods is not univocal. In particular the sign of the effects on those prices
will crucially depend on the technical conditions of production which characterise
capital goods. If, in our hypothesised economy, the productive requirements of k’
(relatively to k) are higher in the production of k’ than in the production of k, then an
increase in R – an increase in production of k’ relatively to k – will make the relative
scarcity of k’ rise (because k’ needs ‘more’ of itself than of the other capital good – this
is the auto-intensive case k’
k’
/k’
k
> k
k’
/k
k
) hence the price of the service associated with
k’ will tend to rise as well. If this is so, then the original disparity between both rates of
interest will tend to widen.
For the sake of completeness we can see that in the opposite case to the one
referred to above (the hetero-intensive case when k’
k’
/k’
k
<k
k’
/k
k
), the initial disparity in
the rates of interest will tend to narrow as we make R rise. Still, this compensation is
not guaranteed, because the initial disparity between the different rates of interest might
be of any width, since it depends on the arbitrary given endowments of capital goods
which are at disposal of producers in the economy. The ultimate result is therefore
totally uncertain.
11
Thus, we have a further negative element in Walras’s system which
undermines the equilibrium conditions he set out there – equilibrium conditions in
accordance to the long period equilibrium method or normal position. In our example,
we have assumed an increase in R – one may even imagine that rise further and further
to the point where all the gross savings are invested in k’ alone while k approaches zero.
At that moment, as a matter of fact, the difference between both interest rates cannot be

9
The particular case in which those proportions might happen to be equal is devoid of meaning
here, for the effects referred to above would not exist at all (the relative scarcity of capital goods
would not be modified as R changes). Cf. Garegnani (2008).
10
Let us see this point analytically:
' '
'
' '
... ...
' ... ' ... ' '
' '
k
t t p p k k k k
k
t t p p k k k k
p
k p k p k p k p
i c
p
k p k p k p k p
i c
= + + + + +
+
= + + + + +
+
(Z’)
11
Cf. Walras (1954, p. 292) where he analyses the effects of the services prices of the capital
goods on the costs of capital goods to be produced and their services’ supply. Cf. also Garegnani
(1962, p. 19).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

116
reduced
12
since the economy is run out of savings to be invested in k. That the capital
good k would not be (re-)produced later has important implications that deserve our
attention. (See Paragraph 6.3 below.)
To conclude, the Walrasian system is prevented from having an economically
meaningful solution and hence fails in its purposes.
13

6.3. Walras’s way-out to the inconsistency
But Walras did not limit himself only to count the number of unknowns and equations
in order to seek economically meaningful solutions of his system. In fact in the part
regarding capital formation in the 4
th
edition of the Éléments Walras somehow went by
intuition that equations (Z) might undergo some modifications, namely – to substitute
the inequality “ s” sign for the equality sign “=”.
14
Before discussing the meaning of
such formal changes, it is important to notice that in Walras’s system the data of the
equilibrium were assumed not to change as the time-consuming process of adjustment
towards equilibrium was taking place. This can be noticed when he describes the
tâtonnements process (to grope towards equilibrium by trial and error):

Our problem is to reach equilibrium in capital formation ab ovo […] we propose
to start by assuming the arbitrary data of our problem to be constant over a certain
period of time, and subsequently we shall suppose them to change in order to
study the effects of such changes.
15



12
Formally one can get a solution if, say, D
k
is negative; but since D
k
is gross investment it would
be totally meaningless, in economic terms, to account for some ‘negative’ investment.
13
In the words of Collard (1973, p. 472): “If there is production of new capital goods [in Walras
system] the approach is open to the charge of sterility [because] almost anything might happen.”
14
See Walras (1954), pp. 286-295.
15
Walras (1954), p. 282. Interestingly enough, in lieu of “over a certain period of time [pendant
un certain temps]” (as written in the 4
th
edition) we read, in the 2
nd
edition, “pendant tout le temps
que dureront nos tâtonnements” [for the entire period of time that the tâtonnements will last; (my
translation)]. If we regard the tâtonnements as a purely analytical device for an analysis of
solutions which were to be stable, it is clear that the data of equilibrium could not change
drastically in the very process of time-consuming adjustment towards equilibrium. If doubts,
reasonably, still remain, it is worth reading some further words from Walras, which are full of
significance in this regard: “Thus equilibrium in capital formation will first be established in
principle [through the tickets]. Then it will be established effectively by the reciprocal exchange
between savings to be accumulated and new capital goods to be supplied within a given period of
time, during which no change in the data is allowed. Although the economy is becoming
progressive, it remains for the time being static because of the fact that the new capital goods play
no part in the economy until later in a period subsequent to the one under consideration” (Walras,
1954, pp. 282-283; emphasis in the original).
Chapt er 6 – A second phase i n t he capi t al cont roversi es
117
For Walras it is crucial that the data of equilibrium do not change over the course
of the trial and error actions carried out by agents; that the tâtonnements process must
be seen as an analytical device for an analysis of an equilibrium is clear when Walras
introduced the so-called “tickets” (bons) in the 4
th
edition so that they could allow for
transactions to be provisional until equilibrium is reached. But if this is so, namely, if
the purpose of Walras was precisely to determine a long period equilibrium (i.e.
attempting at determining the economy’s normal position) then it is at least not clear
the suggested way out put forth by Walras himself in the 4
th
edition too, as to
(meaningful economic) solutions of equations (Z).
In fact alteration in (Z) would imply that when the inequality sign applies then the
production of the capital good(s) concerned will be zero because the rental gained from
investment in such capital good(s) will fall short of rentals obtainable from investing
elsewhere. Then as Walras indicates:

all the remaining equations of system [Z] will be satisfied after the exclusion of
those new capital goods which it was not worth while produce.
16


So Walras seems to suggest that a solution to his system – namely to determine a
long period equilibrium of supply and demand in free competition – can be obtained
only with regards to those capital goods which can be (re-)produced, thus determining
the rate of interest at its highest level.
17

The fact that a capital good’s demand price falls below its supply price (costs of
production) actually means that such capital good is not to be produced thereafter.
The consequences of this change cannot be overlooked: it will of course entail a
change in the composition of the (initial) capital stock of the economy, thus implying
changes in prices and quantities in the economy throughout; hence in that process of
groping towards equilibrium, changes in the quantities of capital goods, as a result of
an equilibrating process in the wake of any initial disequilibrium position, can well be
in contradiction with the assumed endowments of capital goods which the theory
takes as given. Thus the equilibrium reached would not therefore bear the necessary

16
Walras (1954, p. 294).
17
Walras is coherent when he excludes from production the capital goods whose demand prices
would fall below their supply prices. In fact that is the way of the traditional method which, by
conceiving of capital as a single magnitude, could endogenously determine the composition of
such a stock. Walras keeps the system within the method of the long period equilibrium
characterised by a uniform rate of return on costs – and this is what is done when system (Z) is
modified by letting only those capitals whose supply and demand prices equal at the highest rate.
However, the problem is that the notion of capital as the physically heterogeneous vector as a
given of the problem is in contradiction with that method, which Walras of course shared with all
his contemporaries peers such as Böhm-Bawerk, Wicksell, Marshall, Clark. Cf. Garegnani (1990)
pp. 20-22.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

118
persistence that it is however required so as to regard the theoretical variables as a
gravitational centre of actual ones both prices and quantities.

6.4. Implications of the Walrasian specification of capital
On acknowledging the results of the first phase of the Cambridge debates, it is not
worthless to remember that the theory started to increasingly rely on a Walrasian
conception of capital.
18
The latest versions of the neoclassical theory, which have
become dominant since comparatively recent decades,
19
have tried to overcome
Walras’s inconsistency by dropping the equilibrium condition of uniformity of rates
of returns on capitals’ supply prices.
But, now, relying on a Walrasian treatment of capital and dropping the condition
of the uniform rate of profits bear drastic implications on the method of the theory
itself. In fact this new notion of equilibrium could no longer refer to the normal
positions of the economy, because the equilibrium variables are prone to rapid
changes owing to any initial disequilibrium situation such that, when groping towards
equilibrium, some of the data will also change, and therefore it would be hard to
conceive of that equilibrium as a gravitational centre taken by agents in their groping
towards equilibrium.
This leads us to the problems of lack of persistence of the data of equilibrium. In
fact, that lack might prevent the repetition of the markets – transactions – which are
however enormously necessary in order to ascertain gravitational centres such that
deviations between theoretical and observable – trends or averages in time – variables
could be compensated each other. Moreover, the datum of capital in the definition of
equilibrium by the theory relying on a Walrasian capital is the physically
heterogeneous vector, which is arbitrarily given and hence will not allow the
necessary substitutability among the capital goods – particularly among fixed capital
goods – when time-consuming adjustments take place so as to determine their amount
required by equilibrium.
Thus, in the wake of the results drawn from the Cambridge controversies,
marginal theory abandoned the notion of capital in value terms and adopted the
Walrasian specification of capital; however, to keep analysis within the supply and
demand framework, the theory was compelled to abandon the long period equilibrium

18
It is a misfortune, as Bellino (2005, p. 24) and Petri (2007, p. 597) have both remarked in their
respective reviews, that the recent 3-volume collection of works on Capital theory edited by Bliss,
Cohen and Harcourt (2005) does not include anything about Walras’s theory of capital, important
as it is for understating the contemporary versions of the neoclassical capital theory.
19
As Garegnani (1976a) has suggested, the restoration of Walras’s conception of capital was not
new since the path adopted after the controversies had already been laid by Hicks (1946[1939]),
who proposes to dispense with the notion of capital in value terms because that notion would have
only considered steady states. Cf. also the discussion of Paragraph 1.4, Chapter 1 above.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
119
method, rendering the very theory of doubtful usefulness, as Garegnani (1976a) has
shown in his assessment of the results of the controversies. (Cf. Harcourt’s views,
Chapter 1 above, on the shift to Walrasian theory: a ‘more disaggregated’ analysis,
which may emerge ‘logically intact’ from the capital debates).
The point which must be clear is that this new notion – short-period – of
equilibrium has essentially to do with capital, and not with the long-period method
itself. (cf. Paragraph 1.4 in Chapter 1 above.)
Hence the Cambridge debates are relevant to the current theory because it forced
the latter to a theoretic apparatus of very doubtful results for explaining actual
economies. The controversies put in doubt the whole approach and compelled the
theory to adopt a short period general equilibrium method, while at the same time
modern marginalist authors tried to reverse the sign of the critique by pinpointing that
“general equilibrium does not require aggregation” (Bliss 2005, p. xii).
Whether or not neo-Walrasian theory is free of capital theoretic problems is an
issue we shall deal with in Paragraph 6.7, although provisionally; but now we attempt
to clear up some misunderstandings raised by neoclassical authors in the second phase
of the controversies, who argue that neoclassical capital theory “does not require
aggregation” (Paragraph 6.5) and that the uniformity condition of the rates of profit on
capitals’ supply prices is not important (Paragraph 6.6).
6.5. Hahn’s (1975a; 1982) and Bliss’s (1975) views on the question of capital:
A first critical excursus
To side-step the critique concerning capital and the problematic consequences for the
whole approach some neo-Walrasian theorists have entertained that rigorous theory
does not need of ‘aggregates’. Hahn and Bliss, as introduced in our introduction,
20
are
the most notably representatives of this reaction.
21

In the wake of results raised in the Cambridge debates Hahn and Bliss reacted in
such a way so as to minimise their relevance for the theory under attack. One of these
authors’ most noticeable reaction
22
has been the claim that the results of ‘reswitching’
and ‘reverse capital deepening’ would only bear negative consequences only on the
so-called ‘aggregate’ versions of the theory, by which these authors have meant the
particular and theoretically secondary version of a representation of a long period
theory of distribution of supply and demand which regards production of commodities
by an ‘aggregate’ production function. This reaction however reveals a
misapprehension of the traditional meaning attached to the notion of capital. As we
shall see below, these authors have had, up to-day,
23
the tendency to regard the issue

20
See Chapter 1, above.
21
Hahn (1975a; 1982); Bliss (1970; 1975; 1976; 2005).
22
And others, e.g. Dixit (1977), von Weizsäcker (1971).
23
See e.g. Bliss (2005).

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

120
of capital as a problem of ‘aggregation’ which would not be different from the issue
of aggregation of original factors like land or labour. By thus arguing, these authors
have created a dense fog which did not allow scholars to come to grips with the actual
critique, which we have visited in previous parts of this study. To attempt to clear up
some confusions surrounding the ‘aggregation’ issue we may begin with F. Hahn
(1982), who have endeavour to counterattack the critics of neoclassical theory in the
following way:

In general, there does not exist a function from the vector of endowments to the
scalars such that knowledge of the scalar (and of preferences and of technology) is
sufficient to allow one to determine a neoclassical equilibrium. If you put it the
other way round, it is even more obvious. In general, the neoclassical equilibrium
can be found given the vector of endowment which may have, say, 10
8

components. It would be surprising if there were a single number which gives the
same information as the 10
8
dimensional vector. In fact, sometimes and in very
special cases, this surprising property holds. But neoclassical economists have
shown these special cases to be without interest.
24


In this passage Hahn refers to a “neoclassical equilibrium” where the endowments
of the several capital goods are taken as given by the theory. The author affirms that
there is no “single number” that can occupy the place of the endowments of the several
capital goods in the determination of that equilibrium. That “single number” is the
quantity of capital the single magnitude in value terms which, as we have seen, was at
the centre of attack in the debates. Hahn then seems to suggest that ‘aggregation’ would
be the substitution of the vector of endowments for that single number. This would then
suggest that the version that relies in the quantity of capital the single magnitude would
not be interesting for the determination of the “neoclassical equilibrium” because it
would be hard to believe that a “single number” may supply “the same information” as
does, according to Hahn, the vector of the endowments of the capital stock. However,
the traditional versions of the theory – long-period equilibrium or normal positions
method – could, by relying on a notion of capital in value terms, endogenously
determine the composition of the capital stock, i.e. the kinds and quantities of the
several capital goods endowments subject to the condition of the uniformity of rates of
returns on capitals’ supply prices. Capital as a single magnitude in value terms among
the data is essential and indispensable, so long as the variables to be determined by the
theory must refer to normal values, thus defining normal positions. Therefore, in the
traditional version of the theory the “single number” does not imply to do away with
the endowments; the latter are determined according to the former and the satisfaction
of the uniformity of rates of returns on their supply prices. The “10
8
components” are
not erased in the traditional equilibrium framework, but endogenously determined. Yet,

24
Hahn (1982), p. 369.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
121
to achieve that result, the theory has to rely on an unacceptable conception of capital in
value terms, which is what has been uncovered over the controversies. This point seems
to be missed in Hahn’s argument.
In fact, Hahn, who had actually published the core of his 1982 message against the
“Neo-Ricardians” sevens years earlier in a Symposium organised by the Economic
Record,
25
wrote:

Why do people balk only at aggregation of machines and not of people?
26


where it is clear that the author refers to the question of capital by pinpointing that
it would not be a distinct problem from the technical aggregation of different kinds of
labour. But the issue of ‘aggregation’ of items of heterogeneous capital goods must not
be seen as a problem analogous to that of aggregation of original ‘factors’. While
“aggregation of people” may conceivably be assumed since the endowment of each type
of labour can be taken as given since they change pretty slowly with respect to the
speed of adjustment in their respective markets, “aggregation of machines” on the other
hand poses very different theoretic problems since those endowments of “machines”
will change with changes in outputs and demand – hence the endogenous determination
of the endowments of the stock, hence the essential need for a quantity of capital in
value terms.
Hahn’s position has not been isolated at the time of this phase of the debates which
started somewhat at the turn of 1970s. In effect, his rather misleading view on the
centrality of the notion of capital in value terms for the traditional approach is shared
by the second spokesman of the supply and demand approach we are considering here:
C. Bliss.
Bliss says that the problem of ‘aggregation’ he tries to work out refers to

what is involved in replacing a heterogeneous collection of inputs by an
aggregate.
27


But, again, it is not a question of a simple replacement of the endowment by a
single number – the quantity of capital. The point is, on the other hand, a matter that in
the traditional versions was by far clearer: the quantity of capital is crucial for the
definition of the equilibrium in that this notion allows the stock to adapt the form
required by the equilibrium conditions entailed, under free competition, by the
condition of uniformity of the rates of returns on the capitals’ supply prices. It may be

25
Hahn (1975a). See also Harcourt (1975a). See Paragraph 6.6, this chapter, below.
26
Hahn (1975a), p. 364.
27
Bliss (1975), p. 147.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

122
of interest to remind what we have seen in Chapter 2 regarding the supply aspect and
the demand aspect of the problem of capital. In the traditional version the ‘supply’
aspect is expressed by the quantity of capital in value terms which has to be taken as
datum to determine the very value of the rate of interest, among the other ‘unknowns’.
The value of the total stock does not change, although its composition of course does
according to the equilibrium conditions. This is the only way in which the physical
form of the several capital goods can be adapted such that their total demand prices can
be made equal to the quantity of capital – a given amount of value; yet there will still be
an endowment composed of heterogeneous capital goods in that equilibrium position.
Bliss’s “heterogeneous collection of inputs” is not erased from the traditional
neoclassical blackboard by an ‘aggregate’ but rather both, so to speak, live together so
long as that ‘aggregate’ is taken to mean the quantity of capital the single magnitude in
value terms.
Of course, it would have been clearer for us to grasp the centrality of conceiving of
capital in value terms as datum had Bliss provided the crucial reasons why “capital
cries out to be aggregated”, as he himself wrote in the outset of his work dealing with
the theory of capital.
28
However, not only did this author provide no hint at such
statement but also, to repeat, takes the problem of capital as a problem of ‘aggregation’
analogous to aggregation of non-produced factors of production.
In fact, Bliss has devoted a good deal of his celebrated 1975 book to broadcasting
the misleading idea that ‘capital aggregation’ is of little importance for the theory of
capital (read it – the marginal theory needn’t rely on the quantity of capital). His
contention seems to be based in the acceptance of the framework of ‘aggregation’
introduced by Solow in the debates by relying on the so-called Leontief separability
conditions
29
and then he argues that:

[T]he investigation has provided no support whatsoever for the idea that the
aggregation of capital is relatively difficult. The conditions for general capital
aggregation are identical to the conditions for the aggregation of labour, or of
output.
30


28
Bliss (1975), p. 8.
29
These conditions establish that the rates of substitution of the factors one wishes to aggregate must
be independent of the quantity employed of the other factors, and hence allow technically to
aggregate a number of factors into a ‘single’ factor in a production function. As discussed above,
there is no reason to assume that the factors to be aggregated be services of capital goods. Cf. Chapter
5 above where we have seen the reactions of Solow (1955) to some Robinson’s (1953) complaints.
30
Bliss (1975, p. 147, emphasis in the original). Note, however, that what has been at issue in the
controversies is not a question of more or less difficulty, as Bliss seems to argue; it is on the other
hand a crucial question to the supply and demand theory when explaining distribution by factors’
substitution forces. See n. 18, in Chapter 2 above, where I have discussed the entirely different
meanings of aggregation of different kinds of labour or land with respect to that of capital.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
123
So what Bliss is saying is that, since ‘capital aggregation’ involves the same nature
of technical difficulties – which underlie the strong conditions – as does the aggregation
of labour or land or output, then the ‘aggregation of capital’ is actually not a problem.
31

And Bliss concludes:

We may thus conclude that the widespread belief that there is a notable, particular
and distinct problem posed by capital aggregation is at best an ill-formulated idea,
and at worst is based simply on ignorance.
32


But, was actually that the real problem for traditional theory? The “notable
problem” Bliss refers to is precisely the problem to define and to determine and hence
to measure the quantity of capital in value terms so long as we keep in the long period
equilibrium version of neoclassical theory. However it is worth mentioning that the way
in which Bliss refers to the problem of a ‘capital aggregation’ is the way in which
Robinson has in her huge collection of critical works raised her concerns about capital
– attacking the aggregate production function. But, we have also seen that the question
of capital as uncovered in these debates, which touches the core of the traditional
approach to distribution, has not to do with an aggregate production function (see
Chapter 4). The actual arena where we can appreciate the relevance of the struggling
between the two sides of the controversies is not that associated with aggregate

31
It might perhaps be interesting to recall that the process of selection of the data by a theory is a
matter of the highest importance, which might perhaps have not been entirely grasped by some
authors who have argued that a notion of ‘capital’ in value terms is not ‘relatively difficult’ when
it is taken as a datum of the equilibrium. Cf. the following extract from Schlicht (1985, p. 19,
emphasis added) regarding methodological issues in economics: “In freezing some factors by
means of the ceteris paribus clause, we transform them into the data of our analysis. The choice of
these data is not, however, simply a matter of caprice if the isolation is intended to be of
substantive relevance. A substantive isolation requires those factors transformed into the data by
means of the ceteris paribus clause have to be sufficiently stable with regard to the process we
want to explain such that the movements of the data do not destroy or supersede the relations we
are studying.” If one confronts this view vis-à-vis Bliss’s, one may understand that, in the
traditional versions of the marginal theory, to regard ‘capital’ in value terms as datum is an
exercise of “substantive relevance” when studying the forces explaining equilibrium. On the other
hand, the author of this study asks himself what the “substantive isolation” undertaken by neo-
Walrasians is, when taking as datum the physically heterogeneous capital goods’ vector present in
the economy.
32
Bliss (1975, p. 147). It is interesting to notice that these authors have left a lamentable imprint in
successive work carried forward by neo-Walrasian theorists when dealing with the controversy in
mid-1970s. In particular it is noticeable that authors who followed that path concentrated much of
their efforts on presenting the issue of the notion of capital as a single magnitude by confusing it
with the issue of ‘aggregation’ and hence contributing to increase misunderstandings on all these
matters. For example we have the opinion of Dixit (1977, p. 18), who, in his review of Bliss
(1975), affirmed: “I for one would not wish to make too much out of the fact of the relative
simplicity of capital aggregation.” However, it is not a convenient simplifying device, as Dixit
pretends to show, but essential to that theory.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

124
production functions, but the supply and, particularly, the demand functions for
productive factors, since what has been conclusively shown by critics in the
controversies is the impossibility to generalise, in heterogeneous capital goods’
economies, the factor substitution principle – the basic premise of the theory.
6.6. Uniform profit rates and intertemporal prices. Hahn’s ‘special case’
6.6.1. Introduction
In what follows it will be discussed the meaning of own-commodity rates of profits and
the (traditional) rate of profit on the capitals’ supply prices, because the uniformity of
the latter has been at times confused with uniformity of the former.
33
Once we have
cleared up those different two meanings we shall discuss why this confusion brought
“obscurity to the controversies”, as Garegnani has pointed out,
34
which, as we shall
argue, has eventually marred the relevance of the debates (and that is why the present
study aims at reviving the meaning and relevance of the controversy for the economic
theory). In order to clarify the issue we have to define, to begin with, what an own-rate
of interest means in an intertemporal general equilibrium.
35

6.6.2. Commodity own rates of interest
We shall consider a two-period intertemporal economy (the first from t=0 to t=1, and
the second from t=1 to t=2; for brevity we shall refer to these periods as t=0 and t=1
respectively) with two goods (a and b), each being both a consumption and a capital
good (we shall assume them to be circulating capital).
36
In addition, we assume that
production is viable (i.e., there exists a surplus from the pure replacement of that
capital), and wages (w) are paid ex-post (after the production cycle) in terms of one
commodity.
37


33
See Bliss (1987).
34
Garegnani (2000), pp. 392-393.
35
We are not considering the case of a temporary equilibrium, which takes as datum the
expectations functions of the agents, since the meaning of own-rates of interest does not change if
we consider this version.
36
We are following the model present in Garegnani (2000; 2003) which is the same presented by
Hahn (1982).

37
Actually we have in this economy four commodities. In an intertemporal economy there are nT
commodities; while n refers to the number of heterogeneous goods, T does to the number of
periods the whole economy contains. This has been developed by Debreu (1959), who
distinguishes the commodity as to location and time; for example, bread in London in 1998 is
different from bread in Newcastle in 1998, or bread in London in year t is different from bread in
London in, say, year t+s. However we limit the analysis only to time and do not consider location.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
125
In an intertemporal framework, all contracts happen in markets at the initial
moment (t=0) of the first period. For instance, let us consider P
b0
=1 (good b in t=0 is
taken as numéraire). Imagine that someone wishes to buy one unit of that good in t=0
(when the contract takes place) but to be delivered in t=1. Accordingly, this agent will
disburse P
b1
, i.e., the amount of commodity b disbursed in t=0 against the promise that
this agent will receive one unit of that good. In other words, this agent has to give up 1
unit of b in t=0 to consume 1/P
b1
in t=1. However, it is pretty clear that this operation
might be regarded as a loan. In fact, that operation is equivalent to that where someone
lends (in t=0) the amount P
b1
of b to some other agent, who will pay that debt back at
the following period by providing one unit of b. This operation entails a rate of interest
which we can call the own-rate of interest in terms of good b.
Following the foregoing example, let us now assume that P
b1
=10/11, namely, the
amount of commodity b to be disbursed in t=0 in order to receive in t=1 one unit of
that good. In other words, a loan in terms of b has been made in t=0, and the rate of
interest there involved is 10%:

1 0
(1 )
b b b
P r P + = hence 10%
b
r =

So, in general, own-commodity rates will be defined by:

0
1
1
a
a
a
P
r
P
= ÷
0
1
1
b
b
b
P
r
P
= ÷

6.6.3. Uniform rates of return on supply prices in intertemporal equilibria
Having defined the two own-rates of return in our economy, a very important question
arises: which is the form that the condition of a uniform rate of return on the supply
prices of the capital goods takes, when the equilibrium relative prices change over
time?
Now let us assume that one unit of commodity a in t=0 has a value of 2 (i.e., one
unit of a is exchangeable by 2 units of b), but paying for it in the moment of its delivery
(t=1), it would cost 1.8 (notice that we are assuming this price of good a to be
decreasing, and to be undiscounted). The question that arises is how to determine P
a1

(the discounted price over time).
But before going into that problem, we have to face the compelling issue regarding
the (traditional) uniformity of rates of interest. In effect, it will first be necessary for us
to consider that basic condition: the rate of interest, in equilibrium, cannot be different
for the different loans made in terms of the different commodities. We have seen that
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

126
we can obtain with 10/11 units of b in t=0 one unit of itself in t=1, the rate of interest
there involved being 10%.
38
Therefore, such a rate of interest will have to be the same
as in the case we could obtain 10/11 units of b by renouncing to the acquisition of
commodity a in t=0, and with that quantity of b ‘saved’ (since the 10/11 were not
disbursed) buying commodity a
1
in that period. In that period (t=1) we know that one
unit of a costs 1.8 of b, therefore, in order to know how much we must pay in t=0 for
obtaining one unit of a (but to be delivered in t=1 namely – P
a1
), we must discount 1.8
at the rate of 10%.
Therefore:
P
a1
=1.636

This will be the value of commodity a in terms of b, namely – the amount of b
obtained in t=0 by selling one unit of a which will be delivered in t=1. That amount of
b could be lent to someone, then obtaining an interest rate of 10% when the borrower
pays back in t=1 one unit of commodity b. On the other hand, since we know that
r
a
=P
a0
/P
a1
– 1 and that P
a0
=2, then r
a
=22.2%.
Clearly both own-rates of interest are different; moreover, the own-rate
corresponding to the good a is higher than b’s, for the relative value of the former
commodity to the latter is decreasing over time, as was assumed above. On the other
hand, if the loan were made in terms of b, while its relative price increases over time, it
will follow that to compensate those relative advantages (of the lender) and
disadvantages (of the borrower) r
b
would have to be lower (and r
a
higher) as shown in
the example. This is no more than a simple triangular arbitrage operation.
39

The above example also shows us that both own-rates of interest would have the
same value if and only if P
a0
/P
a1
=P
b0
/P
b1
, that is, when both intertemporal prices are
equal, from which it can be deduced P
a0
/P
b0
=P
a1
/P
b1
namely – when both relative
contemporary prices are equal. In general, however, relative prices will tend to be
different and so will the own-rates of interest. So, when equilibrium prices are assumed
to change over time, the condition of a uniform rate of interest on the capitals’ supply
price will in general entail different own-rates of interest. Accordingly it is convenient
to call the former uniformity of rate of interest as the effective rate as Garegnani has
christened it in his attempt at clarification on some misconceptions created in the
second phase of the controversies:

38
Of course, since we have assumed commodity b in t=0 to be the numéraire (P
b0
=1), then both
the own rate of return of that good and the (traditional) uniform rate of interest will coincide. It is
clear, therefore, that such equality will entirely vary according as the choice of the good taken as
numéraire.
39
In any intertemporal economy there are as many own-rates of interest as are goods for each pair
of ‘periods’. In our case discussed in the text we have two own-rates since we have two goods (a
and b) for only one pair of periods, namely – the only pair involving t=0 and t=1. In general, we
may account for own-rates of interest by relating to the rates for the succession of adjacent periods
linked by the so-called ‘Hicksian condition’, that is, the connection of two generic own-rates by
the following condition: (1+r
i(t, t+1
)) / (1+r
j(t, t+1
)) = (P
it
/ P
jt
) / (P
it
+1 / P
jt
+1) where i and j are
two generic commodities. Cf. Bliss (1975), pp. 53-56.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
127

The word effective by which we qualified the long-period notion of a uniform rate
of return is meant to take care of the fact that the definition of such a uniform rate
will entail a non-uniformity of the own rates of returns of the several capital
goods, when changes in their relative prices are considered.
40


Now, having clarified the different meanings between the uniformity of the
effective rate of interest on the capitals’ supply prices and the uniformity of own-rates,
it must be seen in this connection what the critical side of the debates unveiled
regarding the non-satisfaction of the uniformity of the effective rate owing to the
Walrasian specification of capital namely – to take as datum of the equilibrium the
physically heterogeneous capital goods’ endowment, instead of the quantity of capital
the single magnitude in value terms.
In fact the Walrasian specification of capital is not compatible with the traditional
method. Yet in Walras’s equations we can explicitly see that the equilibrium condition
underlying the traditional method, namely the uniformity of the effective rates of profit
on the capitals’ supply prices, is present as soon as we recognise in those equilibrium
price equations the equality sign between demand and supply prices of those capitals.
41

But, because of assuming capital in physical terms, Walras was then compelled to
admit solutions with price equations bearing inequality signs, instead of equality signs,
thus leading some capital goods not to be produced. ‘Walras’s inconsistency’, as started
to be revealed in the years of this controversy (see 6.6.4 below), has been disregarded
by the new versions of the theory, because the latter disregard the traditional meaning
attached to the normal positions or long period method entailed by the uniform rate of
profit on the supply prices.
42
The dropping of the uniformity of the effective rate of
profit is apparent when we recognise the inequality sign in the price equations of those
versions of the theory.
43

6.6.4. Further discussion of Hahn (1975a)
44

What do all these digressions have to do with the Cambridge controversies? In the
course of the debates the criticism of Walrasian theory as contained in Garegnani

40
Garegnani (2000), pp. 392-393, n. 2.
41
In fact, as soon as capital goods’ demand prices are equalised with the corresponding supply
prices it is assumed both i) production of all capital goods and ii) the introduction of the traditional
condition of uniformity of the rate of profit on those supply prices. Cf. Garegnani (2003), p. 151.

42
Of course, it is worth noticing that this path was increasingly adopted by latest versions of
marginal theory in the wake of the discoveries of ‘reverse capital deepening’ and ‘reswitching’ in
the course of the controversies.
43
See e.g. Hahn (1982).
44
Part of this paragraph is based on Garegnani (2003), pp. 149-157.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

128
(1958; 2008) started to be spread by Robinson (1970a).
45
Robinson’s views were then
echoed by Harcourt (1975a) whose arguments became the target of Hahn’s (1975a)
attack, as we shall see below. The issue, not to be forgotten, is the uniformity of the
effective rate of return. The following Harcourt’s message – though literally echoing
Robinson’s (1970a) wording – reflects Garegnani’s critique of Walrasian theory:
46


the Walrasian model of general equilibrium has the implications that it is not
possible to have a theory of the rate of profits.
47


where, no doubt, the impossibility in the ‘Walrasian model’ referred to in that passage
clearly means the non-uniformity of the effective rate of return on capitals’ supply
prices. This is what Hahn (1975a) took up in his reply to Harcourt (1975a) – anyhow
intended to reverse the sign of the critique – to argue that this issue would only affect a
“Special case”
48
of the theory under attack.
Hahn’s contention can indeed be derived from what this author says in the
following passage:

General equilibrium theory is general and so can discuss the equilibrium of an
economy whatever its ‘initial conditions’, e.g. outfit of goods inherited from the
past. For most such specifications it will not be the case that the equilibrium price
of a good for future delivery in terms of the same good for current delivery will be
the same for all goods.
49



45
Also, before 1970, we can read in Robinson (1964a, p. 60): “there is a notorious difficulty about
introducing a rate of profit into the Walrasian general equilibrium system.”
46
To assess the evolution and results of the controversies, it must be clear that Walras’s treatment
of capital was exposed and criticised by Garegnani in his Cambridge Ph.D. dissertation (1958),
and which some years later (1962) he took up again and drafted a manuscript only published very
recently (Garegnani, 2008). The argument was known by Joan Robinson, who was the internal
examiner of Garegnani’s thesis (and L. Robbins the external one), so that Robinson was well
acquainted with Garegnani’s work. Also it must be noted that some of Robinson’s closest
collaborators, in particular, John Eatwell, made the reading of Garegnani’s dissertation in the
Cambridge Library almost compulsory for all his students. (I owe this information to professors P.
Garegnani and G.C. Harcourt) Therefore, Robinson’s claim that Walras does not have any theory
of the profit rate at all can be taken as an echo of the inconsistency between the condition of the
effective uniformity of rates of return on the supply prices of capital goods and regarding capital as
the physical stocks. It might be noticed that perhaps the form chosen by Robinson to refer to the
traditional authors who conceived of capital in value terms – ‘Walrasian leets’ – have misled some
participants in the controversies, when it is well known that Walras does not use capital as a single
magnitude.
47
Harcourt (1975a), p. 350.
48
Hahn (1975a), p. 360.

Chapt er 6 – A second phase i n t he capi t al cont roversi es
129
It is thus apparent that Hahn’s “special case” is not but the result of a confusion
involving the two kinds of uniformities of the rate of profit that we have previously
discussed.
According to Hahn, then, the issue of the impossibility to determine the effective
rate of profit owing to a Walrasian specification of capital is misleadingly confined to a
‘case’ in which relative intertemporal prices are equal over time. In other words, by
recalling the example we have previously used, Walras’s inconsistency would only
affect the ‘case’ such that P
b1
/P
b0
=P
a1
/P
a0
(“the equilibrium price of a good for future
delivery in terms of the same good for current delivery will be the same for all goods”)
namely – when the own-rates of profit happen to be equal owing to the constancy of
equilibrium prices over time. However, we have seen that the effective uniformity takes
the form of non-uniform own-rates of profit when we admit changes in the equilibrium
prices.
In addition there is a further issue. While it is clear that, according to Hahn,
‘general equilibrium theory’ is the neo-Walrasian theory which takes as datum of
equilibrium the physically heterogeneous capital vector, it is totally hidden in Hahn’s
intervention that Walras’s inconsistency affects that ‘general theory’ independently of
whether we consider changes in equilibrium prices over time. But Hahn’s hideaway can
be easily discovered as soon as we recognise that the effective uniformity of the rate of
profit of “Wicksell (and others)”
50
is completely misunderstood by the uniformity of
own-rates of profit due to constancy of prices. Yet the condition of uniformity will be
present in the new formulations of the theory as soon as we reckon in the price
equations of the capital goods the equality signs.
To rebut criticisms in this second phase of the controversies, Hahn attempted to
reverse the sign of the critique by arguing that the effective rate of profit in a ‘Walrasian
model’ is not an issue:

The crudest empirical observations will convince one that there is no unique rate
of profit to be observed in the economy. Do we conclude from that that
competition is functioning badly? Answer: No. Consult any general equilibrium


49
Hahn (1975a), p. 360, emphasis in the original.
50
Hahn (1975a), p. 360. This reference to “Wicksell (and others)” in Hahn’s words is related to
the way in which Hahn somehow acknowledges the results of critics in the controversies as raised
by Robinson –echoed by Harcourt (1975a) – concerning the Walras’s inconsistency. In fact we can
read the entire Hahn’s passage: “to get a uniform rate of profit Wicksell (and others) had to work
with Robinsonian leets” (Hahn, 1975a, pp. 360-361) where it is clear that ‘Robinsonian leets’
refers to the homogeneous capital in value terms, which was actually what Wicksell, Jevons,
Böhm-Bawerk, Marshall, Clark had to work with in order to get the effective uniformity of rate of
profit. Nevertheless Hahn shows in those very passages awareness of the issue at stake, yet he
confuses both kinds of uniformities as the main passage quoted in the text shows it.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

130
text. Why is ‘the rate of profit’ an interesting unknown? Ask the neo-Ricardians.
If we knew all relative prices from now to doomsday, could we ask more?
51

Note that Hahn has apparently tried to show the existence of divergences in the
own-commodity rates in reality; but those observations would be actually indicating
different rates of profits in different businesses compared in terms of money or other
numéraire; but, then, this divergence will have nothing to do with divergences of own-
commodity rates because, when the price of the numéraire is changing, own
commodity rates will be entailing the same effective rate of profits expressed in
different numéraires. So Hahn’s contention shown in the passage misinterprets the
divergence between own-commodity rates: own-rates may diverge each other when
prices change over time, while the effective rate of profit is being ascertained itself
when the physical composition of capital is adjusted. On the other hand, we may have a
situation where the effective rate of profit is not ascertained, that is the Walras
inconsistency due to the arbitrariness of the capital endowments, as the critics revealed
in the controversy, and yet account for uniformity in the own-commodity rates (this
case being when prices are assumed constant over time).
Notwithstanding Hahn’s confusions, what is clear is that the essential role of the
uniformity of the effective rate of profit is totally minimised by Hahn because he
believes that one could not “ask more” if “all relative prices” are known up to the end
of the world – that which implies to know all the intertemporal prices and accordingly
the own-commodity rates those prices would establish.
The “rate of profit” is however “an interesting unknown” in order to determine
long period equilibrium of the economy, as the founders of the marginal approach
believed. And the meaning – and the determination, principally – of that “interesting
unknown” was at the heart of the controversies. For Hahn this would be a very special
case of a general theory. In fact Hahn seems to insist that his ‘general theory’ is quite
safe and

that a very extreme specialisation of a general model somehow shows the latter to
be inapplicable requires the very summit of incomprehension.
52


where the “extreme specialisation” would be that associated with a uniformity of own-
rates of profit, which is taken by Hahn, however, to mean the effective rate of profit.
53


51
Hahn (1975a), p. 361.
52
Hahn (1975a), p. 360.
53
The fact that Hahn has in this intervention confused the uniform effective rate of profit with the
uniformity of own-rates of profit due to constancy of prices over time might perhaps be traced in
the fact that in the traditional versions of the theory, when determining the long period
equilibrium, the prices were assumed to be constant in that very process; so in that very process of
adjustment when prices are assumed constant both uniformities coincide. However this method
centred on the persistence of the equilibrium itself, which in turn is ensured by the effective
(cont.)
Chapt er 6 – A second phase i n t he capi t al cont roversi es
131
With the lenses of modern neo-Walrasian equilibrium theory, which treats capital in
physical terms, it would only appear as a “very extreme specialisation” the case with
which “Wicksell (and others)” had to deal with – namely capital the homogeneous
magnitude.
54

However Hahn’s ‘general model’ cannot determine the relevant variables as
intended not only by “Wicksell (and others)” but also by Walras – to determine long
period equilibrium or normal positions of the economy.
6.6.5. Conclusion: some obscurity in the debates
In the hope that the use of Walrasian capital might have overcome the deficiencies
associated with the notion of capital for marginal theory, the neo-Walrasian theorists, as
exemplified by Hahn, have minimised the results raised by the critics in the debates by
confining the issue of the effective uniformity of the rate of profit to steady-state
equilibrium where relative prices are constant over time. However, the arbitrarily given
physically heterogeneous capital goods vector cannot ameliorate the situation under
which the marginal theory has fallen, unless the traditional notion of equilibrium be
abandoned – that is what actually happened in marginal theory since recent decades.
In this connection – together with our excursus through Hahn’s misleading views
on the issue of uniformity of the effective rate of profit – one can appreciate the
following remark recently raised by Garegnani:

A question which contributed to the opacity of the capital controversies of recent
decades [has been] that inequality of own rates has often been confused with
inequality of effective rates on the supply prices of the capital goods, which is
instead due to the arbitrary initial composition of the capital endowment of
Walrasian theory. This has had the result of obscuring both the necessity of the
concept of a ‘quantity of capital’ for the traditional equilibrium, and the causes of
the abandonment of the latter.
55




uniformity, allows such an assumption. Not being acquainted with such a key notion of persistence
might perhaps misled Hahn into confusing both kinds of uniformities. The point to remark here is
that Hahn and other authors, by introducing such a misleading argument, in turn confused the
traditional long period equilibrium or normal positions with equilibrium of steady state (constancy
of prices). Thus a further misunderstanding was created as well. Cf. Garegnani (1976a, p. 32) and
Robbins (1930).
54
In the neo-Walrasian authors’ ‘comprehension’ this extreme case would reveal itself as the
constancy of equilibrium prices over time.
55
Garegnani (2000), pp. 392-393, n. 2.

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

132
As some reflection on communication might suggest, ‘obscurity’ per se is one of
the worst appropriate grounds for debating. But, sometimes, obscurity might be the
weapon used by some both to hide the issue at stake and to escape the critiques from
their contenders. Yet the obscurity brought in by Hahn into the debates has had the
further purpose to reverse the sign of the critique. In fact, Hahn (1975a) ended his
intervention by proposing to substitute one of Harcourt’s (1975a) appreciations of the
debates – “The Great debate” – by “The Great Charade”.
56
However, as we have seen, it
was the neo-Walrasians who actually produced an endless series of misunderstandings
on key questions for the theory. On visiting the foregoing exchanges in this second
phase of the debates, we would not be guilt of slandering if we pointed out Hahn, Bliss,
Dixit, von Weizsäcker and many other neo-Walrasian adherents (such as J. Stiglitz
1974; Burmeister 1980), as the ones who actually contributed to foster such an image
of the final results of the controversies that might have appeared as a sham.
6.7. Do ‘reswitching’ and ‘reverse capital deepening’ bear theoretic problems on
intertemporal general equilibrium? Two views: Bliss and Garegnani
6.7.1. Bliss
According to Bliss (1975)
57
the ‘paradox’ of ‘reswitching’ in an intertemporal general
equilibrium concerns a key assumption underlying a maximisation problem of the linear
programming type:
58
the possibility that two price systems could be solution to the
maximisation problem. Bliss explains that if two prices systems maximise the present
value of consumption, a “convex combination” of them should also be solution of the
problem, and therefore “it is not possible to have two disjoint sets of prices (i.e.
complete price systems) at each of which a development of the economy is a
production-price equilibrium.”
59
However, with ‘reswitching’, the author also notices,
two such price systems “may not be connected.”
60
This phenomenon would undermine
the convexity properties underlying price systems solutions of linear programming
problems and it might probably have been this feature Bliss is concerned about;
nevertheless this author contends:


56
Hahn (1975a), p. 364. A similar position can be found in Hahn (1972), p. 5.
57
Cf. Bliss (1975, chapter 3).
58
The linear programme maximisation of present consumption entails a dual minimisation of costs
problem. Cf. Dorfman, Samuelson and Solow (1958).
59
Bliss (1975, p. 238, emphasis added). The idea of a convex combination is the following: if p
and p’ are both price systems (or “weights”, or “discount factors”, in Bliss’s words) which are
solution to a maximisation linear programming problem then any convex combination of those
‘weights’ op+(1-o)p’, with 0<o<1, will be solution as well. Cf. Bliss (1975) p. 238, n. 8.
60
Bliss (1975), p. 238. This could be understood as if the same technique, in our well-known wage
curves graphs, is in use for two disjoint intervals of the rate of profits.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
133
Surely the excitement caused by the demonstration that double switching
[reswitching] was a ‘common’ phenomenon, in the sense that the examples do not
depend upon choosing very special values for the parameters, arose because it
seemed that the set of values [of the rate of interest] at which a technique might be
chosen was not connected. But this is an optical illusion.
61


The “optical illusion” Bliss refers to would have been caused because economists
have had “the habit of working with a space of price sequences corresponding to
constant rates of interest.”
62
Since, according to Bliss, phenomena like ‘reswitching’
may only arise when economic analysis is “confined to constant rate of interest price
systems”, the “illusion” would then disappear as soon as it is realised that “[firms
choose their investment plans] in the light of complete intertemporal price systems.”
63

Thus, it would seem, by widening the range of values of the interest rates (implicitly in
a “complete intertemporal price system”), Bliss would have re-established “the property
of convexity in the set of solutions” and so their connectedness; hence the ‘paradox’
would disappear in the intertemporal versions.
However apart from the fact that ‘reverse capital deepening’ is not accounted for in
Bliss’s disputations, it is yet not clear why ‘reswitching’ would have disappeared in
intertemporal general equilibrium theory when the set of values of the rate of interest
can be enlarged and connectedness re-established.
64


61
Bliss (1975, p. 239, emphasis in the original).
62
Bliss (1975), p. 238.
63
Bliss (1975), p. 239. Note the way in which this author sees the long period rate of interest that
is central in the controversies: a simple ‘constant’ rate.
64
Yet it may be interesting to see that even without violation of the so-called connectedness
property there is cause for concern when, in the case of ‘reverse capital deepening’ without
‘reswitching’ on the ‘envelope’ (where one technique is in use only for one convex interval of the
rate of profits), we realise that when producers switch from technique | to ¸ at r=r
2
a higher rate of
profits is associated with a higher net product per worker. (See graph below.) Bliss could warn us
of the fact that we are considering “the constant rate of interest price systems”. But even if in this
‘subset’ of values of rates of interest we may account for ‘perverse’ relations between distribution
and net output per worker, how could one be sure that by widening the ranges of values of that rate
– i.e. to regard “complete intertemporal price systems” – these ‘paradoxes’ may not appear?

Figure 6.1. ‘Reverse capital deepening’. Each technique is used for convex intervals of r
(cont.)
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

134
6.7.2. Garegnani
The critically oriented side of the controversies has started to reply to advocates of
intertemporal general equilibrium theory by arguing that problems of ‘reswitching’ and
‘reverse capital deepening’ are also present in those versions. Garegnani (2000; 2003)
is one of the most notably representatives of the critical side for having brought the
critique in the intertemporal marginal approach.
According to Garegnani the problems associated with capital are present in the current
versions of neoclassical theory. The crux of Garegnani’s argument lies in that these
difficulties reappear in the savings-investment markets.
65
How to see the problem?
Let us assume, in the intertemporal economy explained in n. 65, this chapter, that
all markets other than savings and investment are in equilibrium, such that savings
exceed investment (AS= S – I >0). In the intertemporal framework the excess of
savings over investment entails an excess supply of commodities in t=0 and an excess
demand of them in t=1.
66

… …t=0………… ……………t=1



65
Garegnani (2000) shows that despite savings and investment formally disappear in the so-called
individual (or collective) wealth equations in the standard formulations of intertemporal general
equilibrium, savings and investment still are present in the intertemporal relations. In a two-period
intertemporal economy (t=0, t=1) where agents consume all output in t=1, production will have
sense only if it takes place in t=0. Savings and investment are realised in t=0 too. In t=0 savings
(S
0
) are the result after deducting from the initial endowments of commodities a and b (A
0
, B
0
;
both used as consumption and circulating capital) the respective consumption demand (D
a0
, D
b0
):
S
0
=Y
0
– G
0
=(A
0
P
a0
+B
0
P
b0
) – (D
a0
P
a0
+D
b0
P
b0
)
S
0
=(A
0
– D
a0
)P
a0
+(B
0
-D
b0
)P
b0

where Y
0
: aggregate output in t=0
G
0
: aggregate consumption in t=0
In t=1 savings will be determined by the following relations:
S
1
=Y
1
– G
1
= (L
0
w+S
0
) – (D
a1
P
a1
+D
b1
P
b1
)=0
where L
0
is the quantity of labour employed in t=0 for producing output to be consumed in t=1,
and w is the wage rate (in terms of P
b1
). Note that L
0
w becomes available in t=1 since w is paid in
arrears.
To introduce the so-called individual wealth equations, it is appropriate to consider both the
income-side and the expenditure-side in each period:
y
0
=a
0
P
a0
+b
0
P
b0
=g
0
+s
0

y
1
=l
0
w+s
0
=g
1

By adding both equations, we obtain:
y
0
+ y
1
=a
0
P
a0
+b
0
P
b0
+l
0
w+s
0
= g
0
+g
1
+s
0

hence:
y
0
+ y
1
=a
0
P
a0
+b
0
P
b0
+l
0
w=g
0
+g
1

Note that the highlighted equation above is the so-called wealth equation, where savings for each
individual have been cancelled out, since each individual’s savings enters both the expenditure-
side in t as well as the income-side in t+1. Yet individual (and collective) decisions to save and to
invest are present in the intertemporal versions.
66
Garegnani (2000), pp. 436-439.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
135
0 0 0 0 1 1 1 1
( )
a a b b a a b b
S P D P D P D P D A = ÷ A + A = A + A (o)
where AD’s

indicates the differences in the respective quantities of goods demanded as
consumption between t=0 and t=1.
If we consider the RHS of the first equality sign of equation (o) and equalise it to
zero, we obtain the following equilibrium condition:


0 0 0 0 a a b b
P D P D ÷ A = A (t)

Equation (t) represents the case in which there is an excess demand of b
0
relative
to a
0
, i.e. the case of relative excess demand of contemporary commodities.
Garegnani argues that (t) represents the equilibrium of contemporary consumption
but that it is equation (o) that represents the intertemporal adjustment of savings and
investment.
67

In order to see the nature of the problem, now consider, as Garegnani suggests,
68

three periods in the intertemporal economy, i.e. t= -1, 0, 1. Let us assume that a
0
and
b
0
are produced with means of productions and labour of t = -1.
Now, in the case of the contemporary excess of supply of a
0
relative to b
0

(equation t) there would not appear to be strong difficulties to shift the labour and
means of productions of t=-1 freed by excess supply of a
0
in order to produce AD
b0
,
since the inputs needed to produce both consumption goods are similar, that is,
labour and the given proportions of a and b composing the means of production.
That this adjustment takes place does not seem to meet insurmountable obstacles
since it is in t=-1, and equilibrium may be therefore reached within the same period
due to changes of contemporary relative prices caused by the excess demand of b
0

relative to a
0
.
69


67
Garegnani (2000), p. 439.
68
Garegnani (2000), p. 437.
69
It may be due to this kind of analogy, however formal as it seems, that Arrow (1989) might
have written that “[in any study of] intertemporal allocation, the variable are today’s prices for
future goods, [so there is] a perfectly consistent story that does not look any different from the
story about choosing commodities today.” (p. 155). The root of seeing the problem of
intertemporal allocation as if it were a problem of contemporary consumption goods’ adjustment
might be realised by confronting two radically different principles guiding consumption and
investment (savings). Some basic economic reflection may help us understand that from the
consumers’ viewpoint, who wants to satisfy his needs, specific commodities satisfy specific
needs (for instance it would be as futile to overcome hungry by coal as it would be to consume
bread in order to get rid of the winter cold.) However, in the case of the savers, who demand for
future net income, capital goods in which investment is done are perfectly substitutable: it does
not matter whether the capital good is a lathe or a loom or a plough: what matters is the
expected rate of return that the investment might yield, independently of the particular type of
the capital good. Then, the savers expect to obtain the highest rate of return on the value of the
investment. If this is so, as Garegnani argues, then, the quantity of capital – and its problems –
will be also present in the latest versions of the theory.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

136
However, regarding the intertemporal adjustment (equation o), i.e. the savings-
investment adjustment, the nature of the problem is entirely different from that
regarding the contemporary adjustment outlined above. In this case, the intertemporal
equilibrium adjustment ought to be such that the ‘freed’ labour and means of
production in t = -1 (due to, say, an excess supply in t=0 of a and b) may be directly
employed in t=0 in order to produce in t=0 quantities AD
a1
and AD
b1
; (cf. equation
o). However, this will in general not be possible, because labour and the means of
productions in t=0, which directly produce D
a1
and D
b1
, are different from those
‘freed’ in t= -1 due to the excess supply of a
0
and b
0
. The difference lies in that the
inputs required to produce the consumption goods in t= 0 – and to be consumed in
t=1 – are physically heterogeneous from the inputs required to produce the means of
productions (in t=-1) which, in turn, are required to produce the consumption goods
(in t= 0).
Consequently, to re-establish equilibrium, the only way to increase production in
t=0 of AD
a1
and AD
b1
will be that of increasing the productivity of the already fully
employed labour in t=0 by increasing the quantity of the means of production
cooperating with that labour which directly produce a
1
and b
1
. Hence there should be
a production of AI
a
and AI
b
(investment) in t = -1 by decreasing production of
quantities AD
a0
and AD
b0
. Accordingly, the way to set off these increments in
investment will be by means of a rise in the intertemporal price P
b0
/P
b1
so as to make
agents give up consumption in t=0 (AD
a0
and AD
b0
) in order to obtain some extra in
the future t=1. But a rise in such a price is a fall of r
b
. Assuming that the latter rate is
also the effective rate of interest (r), because b is the numéraire, we can no longer be
sure that a fall in r will make investment rise, as the capital controversies have
revealed. Thus, Garegnani concludes, ‘reverse capital deepening’ and ‘reswitching’
undermine the intertemporal versions of the theory no less than they do in the
traditional marginalist theory.
6.7.3. Confrontations
The negative implications of ‘reswitching’ and ‘reverse capital deepening’ on the
contemporary intertemporal version of the theory have not been utterly considered by
many modern neoclassical advocates. So far, however, we have seen how Bliss might
have believed in overcoming capital theoretic problems in his framework without
solidly arguing his point. Moreover, our brief assessment of Bliss’s treatment of the
issue seems to show us strong reservations that make it hard to accept, without further
inquiring, the claim that “most observers consider [Bliss’s 1975 book as] the
definitive neoclassical treatment of capital theory that ended the Cambridge
controversies.”
70


70
Cohen and Harcourt (2003), p. 206, emphasis added.
Chapt er 6 – A second phase i n t he capi t al cont roversi es
137
On the other hand, the critically oriented side of the debates has recently started
to bring the critique to the intertemporal land by strongly arguing that capital
theoretic problems may reappear in the savings-investment markets, as Garegnani
maintains. Thus, in the light of this recent investigation, it is hard to go along with
many voiced interpretations, like the one quoted above, suggesting that Bliss’s 1975
is a “definitive treatment of capital theory” that would have eventually led the
controversies to an end – read it as a neoclassical end.
71



































71
The interpretation which maintains that Bliss (1975) would have led the debates to an end is also
supported by Solow (1983), p. 184.


Chapter 7
Conclusions
The research presented in this book reveals the importance of the Cambridge capital
theory debates for a better understanding of the present state of neoclassical capital
theory, and thus contributes to rescuing the relevance of the controversy from the
apparently inconclusive state at which it might have been led in its second phase.
Among the many issues debated over the capital controversies, the book focused on
the most salient and relevant phenomena of ‘reswitching’ and ‘reverse capital
deepening’. These results cast serious doubts on the hard core premises of the
neoclassical theory. In fact, once the interdependency between relative prices and
distribution is taken into full consideration, it will be realised that the implications of
those results drawn from the Cambridge controversy undermine the factor substitution
principle, namely: the general principle on which dominant theory (traditional and
contemporary) founded an explanation of distribution in terms of supply and demand.
That is why we set out to deeply analyse and put at the centre these phenomena: to
reveal how they undermine the marginal approach (cf. Chapter 3).
Then what I thought to do in this research was to selectively reconsider some
different positions in these debates by taking as the main line of argument the results
and implications of ‘reswitching’ and ‘reverse capital deepening’. As I showed in this
book, neoclassical advocates did not remain silent when the serious threats started to be
perceived by neoclassical theorists. They attempted different strategies in order to
preserve the neoclassical theory from the tension created by the capital theory debates.
As we saw (cf. Chapter 4), the range of strategies rehearsed by neoclassical participants
have been i) to build up the so-called parables (Samuelson) – but this resulted in a
totally unsuccessful theoretical device to save their theory; ii) then, once the negative
implications started to be acknowledged, they tried to find the conditions to rule out
‘reswitching’ (Levhari-Samuelson), but this too was proved wrong altogether; iii) to
minimise the implications of the results by pointing to the empirical way-out
(Ferguson) – an extreme position that completely dodges the question raised by capital
theoretic problems, and finally (cf. Chapter 6) we find a shift to Walrasian theory and
hence a shift in the notion of equilibrium. What this reveals, as I argued, is that the
counteractions from the neoclassical side arose out of the fact that the implications of
‘reswitching’ and ‘reverse capital deepening’ were perceived as very strong concerns,
and not simply as ‘curiosa’ as Joan Robinson remarked (cf. Chapter 5). As illustrated,
the fact that advocates of the traditional approach – Nobel awardees included – have
tried to rescue the supply and demand theory by recurring to these strategies indicates
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

140
the tension caused by the Cambridge controversy. Still this tension forced the theory to
make all the possible efforts at their disposal to keep alive a theory whose basic
premises cannot theoretically be supported on its own inner logic. Thus, the historical-
analytical reconstruction here developed shows the relevance of these controversies.
Moreover, we have a word to say about the critical side as well. In fact, despite that
the bulk of the critics placed themselves in opposition to the marginalist arguments
raised in the first phase of the debates, it must be pointed out that some critical
participants chose to emphasise both in the first and the second phase a kind of critique
that seems to be secondary in connection with the kinds of problems entailed by the
question of capital. As I argued in Chapter 5, an outcome of my analysis was that the
overemphasised critique of the aggregate production function pursued by some
participants could have contributed to foster misunderstandings that still persist (cf.
Bliss’s 2005 views in Chapter 1). Moreover, the phenomena of ‘reswitching’ and
‘reverse capital deepening’ have been confined to ‘dead end’ by some critical authors
as well (cf. Chapter 1). Perhaps some critics might have believed that the critique
centred on the aggregate production function would have supplied a ‘more evident’ line
of argument against dominant theory (cf. Robinson’s passages in Chapter 5). However
the disadvantages of this line of criticism turn evident when, in the second phase, the
neoclassical side started to try to escape the critique by claiming that the theory does
not need ‘aggregates’, thus supplying a defensive line to neoclassical advocates to
totally shun the implications of the controversies.
However, as I discussed (Chapter 6), the purpose of these neoclassical theorists in
the second phase was far from debating; in fact they diverted the central focus from the
question of capital to a question of ‘aggregates’ as if the former question were a
problem of considering an ‘aggregate’ capital in the same footing as labour or land (cf.
Paragraph 6.5). This view, as we saw, regards the issue of the quantity of capital as a
matter of ‘simplicity’ in presenting the theory; but the role of the quantity of capital for
the traditional theory is essential to the determination of a long period equilibrium (cf.
Chapter 2). Far from debating, some authors might have preferred to bewilder the
ground by ignoring the role of the ‘quantity of capital’ in the traditional versions and by
entrenching the theory to short period general equilibrium so as to evade the
problematic issue of capital (cf. Chapter 4, Paragraphs 4.10 and 4.15). But this confirms
what critic economist Garegnani pointed out in (1976a): a new notion of equilibrium
had become dominant in marginal theory due to inconsistencies in capital, rather than a
problem of the traditional, long period method itself (cf. par.1.4). However there was
not a unified reply from the critical side in this second phase.
In fact, as I showed in the discussion of the Walrasian rehabilitation of capital
(Paragraph 6.4), the critically oriented side of the controversies received neoclassical
counteractions in different ways. Thus, the obscurity brought by neoclassical advocates
arguing in misleading ways – together with a not rapid and unified reply from the
critical side in the context of the Walrasian theory – might have contributed to lead the
relevance of the Cambridge controversies, in the second phase, to an apparent
inconclusive point. In the effort to understand the reasons for the gap within the critical
side, this study argued that there seems not to be reasons for taking the neo-Walrasian
Chapt er 7 – Concl usi ons
141
theory as the theory that is not relevant for the issues raised in the controversies (cf.
Harcourt’s 1976 passage and recent positions of this author in Paragraphs 1.3 and 1.4)
On the contrary, what turns out to be relevant is that the results of the controversies
touch the core premises of dominant theory; thus, it is the critique that has to be
relevant to the theory and not one or another ‘version’ of the theory for the issues raised
in the debates.
The clarifications made (Chapter 6) suggest that the fact that today nobody seems
to talk about the capital debates is not because of the capital theory debates actually
ended (cf. Paragraph 6.7). Some would argue that the issues raised in the debates do not
concern the modern versions of the theory like the intertemporal general equilibrium
theory; however, advocates of these versions have not been able to show the opposite,
as analysed in Paragraph 6.7. Moreover, on the other hand, the provisional assessment
of the recent critique of these versions of the theory could also be taken as a
confirmation that there would not be grounds to dispense with the critique by
neoclassical theory relying on Walrasian capital. Rather the threat for the theory that
problems regarding capital may reappear in the savings-investment market would
indicate that neoclassical theory would have to face this issue – if that were the case.
For this, however, communication between the two sides must be re-established; given
the present situation (cf. Bliss’s 2005 views) we are still far from that. Although this
book has partially dealt with the situation of capital theory in the new versions of the
theory, our attempt at clarification, it is hoped, is intended to nurture future
communication between the two sides on these very important matters at issue.



Appendix A
The curve of the net marginal product of labour
1

A.1. In a labour and corn-capital economy, the competitive capitalist will minimise
costs when hiring the number of labourers for which the net marginal product (an
amount of corn) is equal to the corn-wage he finds ruling in the market.
To make the concept of a net marginal product clear, let us imagine that the gross
marginal product of labour (M
g
P
L
) is represented by the following function:

2
4
4 ( 2)
g L
g L
M P
M P L
=
= ÷ ÷
¦
¦
´
¦
¹

0 2
2 4
if L
if L
s s
< s


For well known reasons, we know that the relevant part of the M
g
P
L
curve is the
negatively inclined stretch (the second component of the above function), since, given
the quantity of corn-capital, the product may be increased with additional quantities of
labour – that is the resulting area under that curve is larger the higher the quantity of
labour employed. Of course, the limit of such a quantity will be determined both by the
supply of labour and by the wage rate ruling in the market. Given the quantity of the
other factor, the rate at which the output increases is a decreasing one, as the negatively
inclined part of the curve indicates. But let us return to our construct of the net marginal
productivity curve.
Let us assume that for producing a unit of net corn output it is necessary to replace
a quantity o >0 of corn-capital. This means that there will be a quota (oK) of the given
quantity of corn-capital which must be replaced for producing a given level of net
output given by the technical conditions of production. We assume o = 0.1 (whichever
the level of corn-output produced), the given capital is K = 2 units of corn, and that the
labour available in the economy is not higher than 4. The maximum gross average
product of labour, which is four, is the maximum value that can be reached by the gross
marginal product of labour. Lastly we assume that half a unit of labour is combined
with 0.5 of corn-capital, one unit of labour is combined with one unit of capital, 1.5 of
labour with 1.5 of corn-capital and so on up until labour reaches 2 units which is
combined with 2 units of corn-capital. All these combinations yield a gross marginal
product of 4. (See Figure A.1 below.)

1
Appendix to Chapter 2, Paragraph 2.1.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

144
Then what is relevant to our purposes here is to show that the ‘share’ of the amount
of corn-capital to be replaced (oK) out of the gross output will be diminishing as the
output increases for the negatively inclined part of the M
g
P
L
curve – that is when the
area under the M
g
P
L
curve is being increased by adding quantities of labour higher than
2. By subtracting that ‘share’, for all relevant values of labour employed and given the
quantity of the other factor, from the M
g
P
L
function, we shall obtain the net marginal
product of labour.
Let us make it sure we understand this construct by using the above function and
the assumptions of the relevant variables. Imagine, now, that the M
g
P
L
=3 hence L=3. In
this hypothetical situation the gross output is given by
3
2
2
8 4 ( 2) L dL + ÷ ÷
}
which is
equal to
35
3
. But that marginal product is gross, not net. To find the net M
g
P
L
it is
necessary to calculate the ‘share’ of the replacement of corn-capital when gross output
is
35
3
. This ‘share’,
35
3
K o
, amounts to
1
3
5
35
175
3
= . Then the net marginal increment in
output when the labour employed is approaching 3 units will be given by
3
3 2.982
175
÷ = . Now, in order to see that this quota out of the gross output is a
decreasing function of the quantity of labour employed (and hence of the output),
imagine that M
g
P
L
=
7
4
hence L=3.5. In this situation gross corn output is given by
3.5
2
2
8 4 ( 2) L dL + ÷ ÷
}
which amounts to 12.875. Accordingly
12.875
K o
=
1
5
0.0155
12.875
=
which is lower than
3
175
= 0.017, and the net marginal increment in output when the
labour employed is approaching 3.5 units will be given by
7
0.0155 1.7345
4
÷ = . It can be
noted that, so long as we move from north-west to south-east on the negatively inclined
part of the gross marginal productivity curve, the net marginal productivity curve will
be approaching the former, as the labour employed is increased up until its maximum
value – assumed to be 4 only for the sake of exposition.
On the other hand, for values of labour which correspond to the part of the gross
marginal product curve near to its constant part, the respective levels of the net
marginal product will be near the net marginal product of labour for the constant part of
the gross marginal product curve, which is a constant value too. This value, due to our
Appendi x A
145
example, is equal to
1
40
, resulting from, when L=1,
1 0.1
1
40 4 4
K o
= = , and when L=2,
1 0.1
2
40 8 8
K o
= = .
Thus equipped we can derive the following net marginal productivity function of
labour:
1
0
2
1
40
1
5
8 4 ( 2)
g L
g L l
l
NetM P
NetM P
L dL
=
=
+ ÷ ÷
¦
¦
¦
¦
´
¦
¦
¦
¹
}


The first component of the function is applied when 0 2 L s s , while the second
one is applied when 2 4 L < s , such that
0 1
l l < with
0
2 4 l s < and
1
2 4 l < s .

A.2. Therefore we have built the net marginal product curve, which provides the net
marginal increment in output as an infinitesimal unit of labour is increased. Then, the
capitalist, who is the entrepreneur, will minimise costs when hiring the number of
labourers for which the net marginal product (an amount of corn) is equal to the wage
rate he finds ruling in the market.
2


2
The notion of a net marginal product may be traced in Marshall’s Principles: “Every business
man indeed, according to his energy and ability, is constantly endeavouring to obtain a notion of
the relative efficiency of every agent of production that he employs; as well as of others that might
possibly be substituted for some of them. He estimates as best he can how much net product (i.e.
net addition to the value of his total product) will be caused by a certain extra use of any one
agent; net that is after deducing for any extra expenses that may be indirectly caused by the
change” (Marshall, 1959, pp. 336-337, emphasis in the original).
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

146


Figure A.1.
If the entrepreneur finds in the market w=w*, then the optimal quantity of L demanded
should be L*. Were it otherwise, e.g. if the labour demanded were L
L
which is lower than
L*, the entrepreneur would lose the shadowed area ACE



Appendix B
Discussion of the ‘supply-side’ of capital in Wicksell’s
general equilibrium system
1

B.1. This appendix attempts to illustrate, by using Wicksell’s general equilibrium
system of his Lectures, the arguments raised in Chapter 2 above concerning the supply-
side or supply-aspect problem of capital.
2
Moreover, it will be attempted to illustrate
the other problem which arises with groups (ii) and (iii) of the data: in group (ii) the
quantities of factors define the methods of production, whilst in group (iii) those
quantities appear in the role of a ‘factor endowment’, that is the role on which the
supply functions of each factor are based. In both of these roles the factors should be
measured – and hence known – before knowing distribution, in order to obtain the
quantities corresponding to any given method of production [group (ii)] and to any
given physical output [group (iii)]. The crucial point seems to be to measure the factors
of production before determining distribution, in order to avoid circular reasoning.
3

Let us not forget that the discussion of these problems meets those theorists who
regarded capital in value terms as data (and not other theorists who considered capital,
as part of the data, as the vector of the heterogeneous capital goods, like Walras and the
‘new’ versions of the theory).

B.2. To begin with, we shall simplify the exposition by assuming the production of a
single consumption good in yearly cycles.
4

The consumption good is produced by means of current labour, current land, and
the saved-up (services of) land and labour: that is ‘capital’ in the form of past labour
and land applied at different periods of time to the consumption good’s production

1
Appendix to Chapter 2, Paragraph 2.8.
2
Cf. Wicksell (1934), pp. 144-206. These problems have already been dealt with by Garegnani
(1958; 1960; 1990), and on his works shall I rely for presenting the difficulties of capital
associated with its supply-role.
3
Also, it will be useful to bear in mind how the problem of capital appears in Wicksell’s theory,
which we illustrate in the present appendix and which may be useful to discuss with Joan
Robinson’s way to deal with capital, which has been associated to a discussion of ‘aggregate’
production functions. (See Chapter 5).
4
Wicksell (1934, p. 181) does so initially, and later in the book the author (Wicksell, 1934, pp.
196-206) proceeds to the case of more than one consumption good.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

148
before the latter emerges.
5
It is here assumed, as Wicksell did, an economy where
labour and land are saved up for no more than two years.
6

Let us consider the general system set out by Wicksell. Therefore let:
Q: be the yearly output of the consumption good.
K: be the value in terms of Q of the capital stock available in the economy.
A: be the quantity of labourers available (they may be regarded as the absolute
number).
B: be the acres of land available.
A
0
: be the quantities of labour (labours-year) applied ‘0’ years before Q emerges:
that is current labour applied to Q.
A
1
: be the quantities of labour (labours-year) applied 1 year before Q emerges.
A
2
: be the quantities of labour (labours-year) applied 2 years before Q emerges.
B
0
: be the acres of land applied ‘0’ years before Q emerges: that is current land
applied.
B
1
: be the acres of land applied 1 year before Q emerges.
B
2
: be the acres of land applied 2 years before Q emerges.
w: be the yearly wage rate in terms of Q paid in arrears.
r: be the yearly rent rate in terms of Q.
i: be the yearly rate of interest in terms of Q.

Wicksell takes as given the quantities A, B, and K. He further supposes the
quantities supplied for productive use to be independent of the rates of remuneration.
The unknowns of the system are A
0
, A
1
, A
2
, (dated quantities of labour), B
0
, B
1
, B
2
,
(dated quantities of land), Q, w, r, and i. For that determination we have the following
relations:

0 1 2 0 1 2
( , , ; , , ) Q f A A A B B B = (1.B)

Equation (1B) defines the production function of the year’s consumption good, Q,
– which is produced by means of current labour and land, and (saved and capitalised)
resources during the two preceding years – and gives the combination which minimises
the cost of production that the entrepreneur has to incur: wages w of the current labour
(A
0
), rents r of the current land (B
0
), and interest i of the saved up labour (A
1
, A
2
, that is

5
E.g. If in year t, corn, the consumption good Q, is produced by means of labour, land, and
‘ploughs’, then the latter are regarded as the (past) labour and (past) land used in year t-1 to
produce those ploughs (suppose to be of one-year duration), which in turn are used to produce
corn in t. In t, therefore, we reckon capital as the saved-up labour and saved-up land necessary to
produce those ploughs.
6
Cf. Wicksell (1934), p. 158. Or, n=3 if the reader is familiar with Garegnani’s (1990, p. 29)
reconstruction of the system. Anyway, there is no loss of generality if the period of saved up
labour and land is either one, or two, or n years. As Wicksell put it: “What we have to say in this
connection can easily be extended to processes of production and capital investment over any
period whatever.” (Wicksell 1934, p. 158).
Appendi x B
149
the labours-years of the two preceding years) and of saved up land (B
1
, B
2
, that is the
utilised acres over the two preceding years).
A caveat here is in order. Firstly, to give, say A
1
, economic sense, we must assume
that w
1
>w, so that the rate of interest will be
1
w w
w
÷
= i (w
1
is the wage rate which
corresponds to the quantity of labour applied one year ago necessary to produce the
capital goods to be reckoned to-day). For investment of ‘labour’-capital for two years,
we must have
2 1
1
w w
w
÷
= i. Furthermore, Wicksell (1934, p. 156) assumed that, under
competition, “all capital will receive approximately the same return”, so:
1
r r
i
r
÷
= and
2 1
1
r r
i
r
÷
= , (r
1
and r
2
are the rent rates which corresponds to the quantity of land
necessary to produce the capital goods one and two years ago respectively, and to be
reckoned to-day). Secondly, these equilibrium conditions (however not being used to
determine equilibrium in the system)
7
indicate us that, in equilibrium, uniformity of
returns on the different capital goods is endogenously reached through equations (2.B)
and (3.B)-(4.B)
8
below, that is through the relations expressed by, on the one hand, the
equilibrium conditions supplied by the role that the factors play in the definition of the
methods of production, and on the other, by the supply and demand equilibrium
functions for factors. In addition, we would like to stress that, equation (1.B) together
with the following groups of equations do express ‘half’ of the problem which arises
between the role of productive factors in both its roles: group (ii) & (iii) of the data. In
other words, the factors’ quantities defining the methods of production in the arguments
of equations (1.B) and (2.B) are expressed in technical units.
Then we have the equations expressing the equilibrium relations between marginal
productivity of each factor of production (measured in technical units) and their
respective rates of remuneration. Then:

0
1
2
2
(1 )
(1 )
Q
w
A
Q
w i
A
Q
w i
A
c
=
c
c
= +
c
c
= +
c

0
1
2
2
(1 )
(1 )
Q
r
B
Q
r i
B
Q
r i
B
c
=
c
c
= +
c
c
= +
c
(2.B)

7
In fact, once i and r are determined, one can determine r
1
by
1
(1 ) r r i = + , and r
2
by
2
2
(1 ) r r i = + . Similarly, once w is determined, both w
1
and w
2
will be determined too. It is worth
pointing out that those equilibrium equations entail a compound, and not simple interest.
8
Discussion of equation (4.B) is postponed later in the present Appendix.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

150
This group of equations provide the equilibrium condition of uniformity of rates of
returns over costs of the capital goods.
Then there are two equations, which Wicksell derives by assuming “stationary
conditions”:
9
year by year the total sum of the quantity of labour and of land annually
consumed – saved up or current – must be equal to the supply of labour and of land
annually available in the economy. Therefore:

0 1 2
0 1 2
A A A A
B B B B
+ + =
+ + =
(3.B)
10


So far we have nine independent equations for determining our ten unknowns. The
missing equation, which Wicksell however introduced verbally, is the one that it makes
the total sum of the values of ‘labour’-capital and ‘land’-capital equal to “a certain
given quantity – the total exchange value of the capital employed.”
11

Before putting forth the formal expression of Wicksell’s assertion, let us however
remember that, under conditions of zero net accumulation, the expression K is obtained
by bearing in mind that, say, to each A
j
, j= 1, 2 there will correspond j elements of
capital, each embodying A
j
labour years. For instance, if A
2
were the quantity of labour
for producing coal for producing ploughs (of one-year duration), used in turn to
produce corn (the consumption good Q), then we would have, besides the quantity A
2

of labour embodied in ploughs, and applied one year before the moment of our
reckoning of capital, a ‘second’ quantity A
2
of labour paid for at the time of reckoning
and embodied in the coal which is going to be worked upon by labourers producing
ploughs. Hence:

1 2 1 2
[1 (1 )] [1 (1 )] K wA wA i rB rB i = + + + + + + + (4.B)
12


9
Wicksell (1934), p. 204.
10
As Wicksell put it: “[I]f conditions are to remain stationary, two quantities of labour and land
must be withdrawn from the production of the consumption goods during the current year and
devoted (1) to production of goods which will only be used in the following year, (2) to goods
which will only be used in the year after that.” (Wicksell, 1934, p. 158)
11
Wicksell (1934), pp. 204-205. The complete citation runs: “[…] the total exchange value of the
capital employed in the two industries together, expressed in terms of the first commodity […]”
(Wicksell, 1934, pp. 204-205, emphasis in the original). The importance of the quotation in the
text, together with this complement, is the realisation that Wicksell assumes a given capital in
value terms as the endowment of the economy, in order to fulfil the condition of a uniform rate of
return on the supply prices of the capital goods; capital goods, that is, which are required by the
productive conditions of each of the “two industries”. This passage also provides insight into the
issue of why the composition of the capital endowment must be endogenously determined, as it
has been argued in Chapter 2, above.
12
The term wA
2
on the RHS of (4.B) is the ‘first’ quantity of A
2
reckoned as capital (from our
example: it is the labour employed in producing ploughs), while the term wA
2
(1+i) is the ‘second’
quantity of A
2
, that is, in our example, the labour embodied in coal produced two years before
producing ploughs. Cf. also n. 18, Chapter 3, above.
Appendi x B
151
Through equations (3.B) and (4.B) we can see the other ‘half’ of the problem we
raised above, that is the roles these factors occupy in defining the endowments of the
economy [group (iii) of the data]. As to capital, it is evident that its measurement is not
independent of distribution – contrary to what it is instead necessary in order to ground
a theory of distribution like the one we are here considering.

B.3. Wicksell himself was aware of the difficulties associated to a value measurement
of capital in order to provide the foundations of marginalist theory:
13

Whereas labour and land are measured each in terms of its own technical unit,
capital, on the other hand, is reckoned, in common parlance, as a sum of exchange
value – whether in money or as an average of products. […] However good the
practical reasons for this may be, it is a theoretical anomaly which disturbs the
correspondence which would otherwise exist between all the factors of
production.
14

The “correspondence” between all factors of production is expressed in equations
(1.B)-(2.B), where ‘capital’ is measured in technical units. However, its existence is at
stake, as Wicksell recognises, as ‘capital’ is then introduced in value terms – equation
(4.B).
Despite the value measurement of K in equation (4.B), Wicksell states that “capital
is saved-up labour and saved-up land.”
15
However, he does not take as given the capital
endowment of the economy in terms of physical quantities A
1
, A
2
; B
1
, B
2
. Had he been
consistent with such a definition, A
0
and B
0
would have had resulted, in accordance
with equations (3.B), by subtracting from A and B the quantities needed to replace the
saved up labour and land as they are being used up. Therefore, the number of unknowns
would nevertheless have fallen below the number of independent equations, with the
unknowns decreasing by 4 (i.e., A
1
, A
2
, B
1
, B
2
) but only 2 equations (the first two in
equation (2.B): that is
0
Q
w
A
c
=
c
and
0
Q
r
B
c
=
c
) disappearing. This over-determinacy
would of course have engendered a conflict between the conception of capital in value
terms, as Wicksell expressed it on some pages of his Lectures (and quoted above), and
equations (2.B). Therefore, both the expression for capital (equation 4.B), and
equations (2.B) establish the uniformity of the rate of return on wages (w
1
, w
2
) and rents
(r
1
, r
2
) advanced for the several periods of time and, therefore, on the supply price of
the capital goods (the rate of interest i).
In fact, it is not that Wicksell would have been more satisfied with his results had
he used a physical measurement of capital to express the endowment. In this regard it is
worth bringing in the following lengthy passage:

13
Cf. also Garegnani (1960, p. 136 n.; p. 143 n.).
14
Wicksell (1934, p. 149, emphasis in the original).
15
Wicksell (1934), p. 154.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

152
If capital also were to be measured in technical units, the defect [of measuring
capital in value terms] would be remedied and the correspondence [between factor
of production demanded and remuneration] would be complete. But, in that case,
productive capital would have to be distributed into as many categories as there
are kinds of tools, machinery, etc., and a unified treatment of the rôle of capital in
production would be impossible. Even then we should only know the yield of the
various objects, but nothing at all about the value of the goods themselves, which
it is necessary to know in order to calculate the rate of interest, which in
equilibrium is the same on all capital.
16

One can perceive, through the perusal of this long quotation, that Wicksell was
aware that the only way to reach the uniformity of return on the supply prices of capital
goods is by treating capital as a single magnitude in value terms, leaving the particular
endowments of the stock to be endogenously determined. Further, he gives the “unified
treatment of the rôle of capital in production” the meaning that most of traditional
economists had already supplied to economic theory: capital is conceived in value
terms to determine the uniform equilibrium rate of interest over the costs of the capital
goods. And this is the supply aspect of the problem of capital.
17

















16
Wicksell (1934), p. 149, emphasis added.
17
Wicksell, as Garegnani (1960) has noted, “falls back” to a value measurement of capital. He
“falls back” because he had already provided in Value, Capital and Rent (1954 [1893]) a value
measurement of ‘capital’ based, as had Böhm-Bawerk (1891), on the notion of an average period
of production. Wicksell, indeed, was very unsatisfied with the assumptions necessary for such a
measurement, in particular: i) the rate of interest involved is a simple one, ii) the absence of fixed
capital, iii) the existence of only one original factor to derive the average period of production. In
the Lectures Wicksell drops the first and the third assumptions, but, however, for the reasons we
have provided in the text, he falls back, again, on a value measurement of capital. Some of the
quotations provided in the text of this appendix somehow reflect how unsatisfied the Swedish
economist felt with the definition of the capital endowment in value terms.
153
Appendix C
Sraffa’s landmark in the critique of the marginal theory
18

C.1. Underlying the results of ‘reswitching’ and ‘reverse capital deepening’ is Sraffa’s
outstanding contribution Production of Commodities by means of Commodities,
Prelude to a Critique of Economic Theory (1960). In that work the famous Italian
economist, who had established residence in Cambridge since the mid-20s, reached the
results then known in the Cambridge debates as ‘reswitching’ and ‘reverse capital
deepening’. In that work (though we are told in the preface that the bulk of the ideas
present in the book had been developed in the 1920s and 1930s) Sraffa shows that as
the rate of profits
19
(or the real wage) varies, both the relative prices and the value of
capital per labourer have to change.
20
This is of course not new. In fact, the value of
capital will change with changes in distribution when neither the techniques of
production nor the output change. But Sraffa goes further. He demonstrates that, given
the technical conditions of production and outputs, when the rate of profits is increased,
the movement of the price of a commodity relative to another need not be monotonic.
Sraffa shows that a commodity’s price can be obtained by “reduction to dated
quantities of labour”, that is the sum of an infinite series of terms, representing the
wages paid, to direct labour, to the direct labour necessary to produce the means of
production, to the direct labour necessary to produce the means of production of the
means of production, and so on. Each term of this sum must be multiplied by 1, (1+r),
(1+r)
2
, etc. (Wages of current labour are paid in arrears) For a generic commodity j its
price can be represented as follows:
2 3
( ) ( 1) ( 2) ( 3)
(1 ) (1 ) (1 ) ...
j j t j t j t j t
p wL wL r wL r wL r
  
        (1.C)


18
Appendix to Chapter 3, Paragraph 3.1.
19
It has been preferred in this appendix to speak of rate of profits rather than the rate of interest as
does traditional theory, because Sraffa is reviving the classical approach and the classical
economists from Smith to Ricardo (and also Marx) who spoke of a rate of profits. Profits are
envisaged by these authors, and Sraffa and current surplus approach’s theorists, as a surplus after
having deduced from the social product the necessaries for the reproduction of the economy
(wages and replacements of capital goods). However, in this study we have used sometimes the
term ‘rate of interest’ (as used by marginalist theory) as a synonym of the classical rate of profits.
20
The only exception being the special case of “equal organic composition” noted by Marx (1961)
in Das Kapital. In this case relative prices are proportional to labour embodied and independent of
the distribution of income.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

154
where the terms L
j(t)
, L
j(t-1)
,... represent the dated quantity of labour one obtains with the
reduction. Sraffa then considers the case of two commodities whose reduction to dated
quantities of labour generates identical dated labour terms except for three:

One of them, ‘a’, has an excess of 20 units of labour applied 8 years before,
whereas the excess of the other, ‘b’, consists of 19 units employed in the current
year and 1 unit bestowed 25 years earlier.
21


In mathematical terms it means that the difference between both prices can be
written in the following way:

8 25
20 (1 ) [19 (1 ) ]
a b
p p w r w w r       (2.C)

The difference on both prices is zero when r=0, r=17% and r=25%, which is the
value of the maximum rate of profits (the wage rate is zero). The price of commodity
‘a’ rises relative to ‘b’ when 0<r<9% and when 22%<r<25%. Instead, the price of
commodity ‘a’ falls relative to ‘b’ when 9%<r<22%. Sraffa comments this result:

The reversals in the direction of the movement of relative prices, in the face of
unchanged method of production, cannot be reconciled with any notion of capital
as a measurable quantity independent of distribution and prices.
22


C.2. To understand Sraffa’s conclusion, imagine that corn and cloth are produced by
means of labour and heterogeneous capital goods. Then Sraffa’s result means that, as
the wage rate rises (and the rate of profits decreases), the relative price of cloth (which
is ‘more’ labour intensive) relative to corn first increases and then decreases. Hence
cloth is more labour-intensive for certain values of the rate of interest, but more
‘capital’ intensive for other values. Hence there is no monotonic relation between
changes in distribution and changes in the prices of consumption goods (or costs of
production). This explains why there does not exist any measurement of ‘capital’
intensity production method independent of changes in distribution. Thus, ‘reverse
capital deepening’ may arise even when nothing technically changes.

21
Sraffa (1960), p. 37.
22
Sraffa (1960), p. 38. For a first echo of Sraffa’s discovery (in particular, of his striking
conclusion quoted in the text) cf. Harrod (1961) and Sraffa (1962).


Appendix D
‘Reverse capital deepening’ without ‘reswitching’
1

D.1. Even in the absence of ‘reswitching’ there might be a ‘perverse’ relationship
between ‘capital’ and the rate of interest. That will be the case when the value of capital
per worker (k) increases when r increases too. For instance, consider a family of wage
curves such that all of them are concave. While the ‘jumps’ through the switch points
from technique to technique goes in the neoclassical behaviour – as r increases k
decreases at those switch points (Cf. in Figure D.1 below: k in B at r=r
2
is lower than k in
A at r=r
1
, and r
1
<r
2
) – there are, on the other hand, increments in k as r increases i) from
0 to r
1
, ii) from r
1
to r
2
, iii) from r
2
to R
¸
.

D.2. There is no general reason to presume that the ‘jumps’ could more than offset k
associated with values of r in between switch points. The figure below illustrates the case
in which k more than offset the ‘jumps’ in between the switch points.




Figure D.1. Concave wage-curves and the employment function of ‘capital’

1
Appendix to Chapter 3, Paragraph 3.4.


Bibliography
Ahmad, S. (1998). Cambridge Controversy on the Theory of Capital, in Kurz, H., Salvadori, N.
(eds.). The Elgar Companion to Classical Economics. Edward Elgar, Cheltenham and
Northampton.
Arrow, K. (1989). Joan Robinson and Modern Economic Theory: An Interview, in Feiwel, G.R.
(ed.). Joan Robinson and Modern Economic Theory. Macmillan, London.
Arrow, K., Chenery, H., Minhas, B.S., and Solow, R. (1961). “Capital-Labor Substitution and
Economic Efficiency”. Review of Economic Studies, 42, pp. 225-250.
Arrow, K., Hahn, F. (1971). General Competitive Analysis. Holden Day, San Francisco.
Asimakopulos, A., Harcourt, G. (1974). “Proportionality and the Neo-classical Parables”.
Southern Economic Journal, 40, pp. 481-483.
Backhouse, R. (1985). A History of Modern Economic Analysis. Basil Blackwell, Oxford.
Baldone, S. (1984). “From Surrogate to Pseudo Production Function”. Cambridge Journal of
Economics, 8, pp. 271-288.
Barro, R., Sala-i-Martin, X. (1995). Economic Growth. McGraw-Hill, New York.
Bellino, E. (1993). “Continuous Switching of Techniques in Linear Production Models”. The
Manchester School, 61, pp. 185-201.
Bellino, E. (1998). Reswitching and Reverse Capital Deepening, in Kurz, H., Salvadori, N. (eds.).
The Elgar Companion to Classical Economics. Edward Elgar, Cheltenham and
Northampton.
Bellino, E. (2005). “Review of Capital Theory edited by Bliss, C., Cohen, A., Harcourt, G., 2005”.
History of Economic Thought Newsletter, 75, pp. 23-26.
Bhaduri, A. (1966). “The Concept of the Marginal Productivity of Capital and the Wicksell
Effect”. Oxford Economic Papers, 18, pp. 284-288.
Bhaduri, A. (1969). “On the Significance of Recent Controversies on Capital Theory: A Marxian
View”. Economic Journal, 79, pp. 532-539.
Bharadwaj, K. (1970). “On the Maximum Number of Switches Between Two Production
Systems”. Schweirische Zeitschrift für Volkswirtschaft und Statistik, 104, pp. 409-429.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

158
Birner, J. (2002). The Cambridge Controversies in Capital Theory: A Study in the Logic of Theory
Development. Routledge, London.
Birner, J. (2004). “The Cambridge Controversies in Capital Theory”. Bulletin of Advanced
Reasoning and Knowledge, 2, pp. 109-117.
Blaug, M. (1968). Economic Theory in Retrospect. Heinemann Educational Books, London.
Blaug, M. (1974). The Cambridge Revolution: Success or Failure? A Critical Analysis of
Cambridge Theories of Value and Distribution. Institute of Economic Affairs, London.
Bliss, C. (1970). “Comment on Garegnani”. Review of Economic Studies, 37, pp. 437-438.
Bliss, C. (1975). Capital Theory and the Distribution of Income. North-Holland, Amsterdam.
Bliss, C. (1976). “Controversy in Capital Theory: A Subjective View”. (Unpublished manuscript)
Barcelona.
Bliss, C. (1987). Equal Rates of Profit, in Eatwell, J., Milgate, M. and Newman, P. (eds.). The New
Palgrave: A Dictionary of Economics. Macmillan, London.
Bliss, C. (2005). The Theory of Capital: A Personal Overview, in Bliss, C., Cohen, A. and
Harcourt, G.C. (eds.). Capital Theory, 3 vols., Edward Elgar, Cheltenham and
Northampton.
Bliss, C., Cohen, A. and Harcourt, G.C. (eds.) (2005). Capital Theory, 3 vols., Edward Elgar,
Cheltenham and Northampton.
Böhm-Bawerk, E. (1891). The Positive Theory of Capital. G.E. Stechert, New York.
Braun, O. (ed.) (1973). Teoría del Capital y la Distribución. Tiempo Contemporáneo, Buenos
Aires.
Brems, H. (1975). “The Capital Controversy: A Cambridge Massachusetts View of Cambridge
England”. De Economist, 123, pp. 369-384.
Bruno, M., Burmeister, E. and Sheshinski, E. (1966). “Nature and Implications of the Reswitching
of Techniques”. Quarterly Journal of Economics, 80, pp. 526-553.
Burmeister, E. (1980). Capital Theory and Dynamics. Cambridge University Press, Cambridge.
Burmeister, E. (1987). Wicksell Effects, in Eatwell, J., Milgate, M. and Newman, P. (eds.). The
New Palgrave: A Dictionary of Economics. Macmillan, London.
Burmeister, E. (2000). The Capital Theory Controversy, in Kurz, H. (ed.). Critical Essays on Piero
Sraffa’s Legacy in Economics. Cambridge University Press, Cambridge.
Carver, T.N. (1904). The Distribution of Wealth. Macmillan, London.
Caselli, F., Feyer, J. (2007). “The Marginal Productivity of Capital”. Quarterly Journal of
Economics, 72, pp. 535-568.
Bi bl i ography
159
Champernowne, D.G. (1953). “The Production Function and the Theory of Capital: A Comment”.
Review of Economic Studies, 21, pp. 112-135.
Ciccone, R. (1996). “Possibilità e probabilità di comportamento ‘perverso’ del capitale”. Studi
Economici, 58, pp. 41-73.
Clark, J.B. (1891). “Distribution as Determined by a Law of Rent”. Quarterly Journal of
Economics, 5, pp. 289-318.
Clark, J.B. (1965). The Distribution of Wealth: A Theory of Wages, Interest and Profits. Kelley
Reprints, New York [1899 1st ed.].
Cobb, C., Douglas, P. (1928). “A Theory of Production”. American Economic Review, 18,
pp. 139-165.
Cohen, A. (1993). What Was Abandoned Following the Cambridge Capital Controversies?
Samuelson, Substance, Scarcity, and Value, in De Marchi, N. (ed.). Non-Natural Social
Science: Reflecting on the Enterprise of More Heat than Light, Duke University Press,
London.
Cohen, A. (2005). “Book Review: ‘The Cambridge Controversies in Capital Theory: A Study in
the Logic of Theory Development’”. History of Political Economy, 37, pp. 755-756.
Cohen, A., Harcourt, G.C. (2003). “Whatever Happened to the Cambridge Capital Theory
Controversies?”. Journal of Economic Perspective, 17, pp. 199-214.
Cohen, A., Harcourt, G.C. (2005). Introduction Capital Theory Controversy: Scarcity, Production,
Equilibrium and Time, in Bliss, C., Cohen, A. and Harcourt, G.C. (eds.). Capital Theory, 3
vols., Edward Elgar, Cheltenham and Northampton.
Collard, D. (1973). “Leon Walras and the Cambridge Caricature”. Economic Journal, 83, pp. 465-476.
Debreu, G. (1959). Theory of Value: An Axiomatic Analysis of Economic Equilibrium. Yale
University Press, New Haven and London.
Dewey, D. (1965). Modern Capital Theory. Columbia University Press, New York.
D’Ippolito, G. (1987). “Probabilità di Perverso Comportamento del Capitale al Variare del Saggio
di Profitto. Il Modello Embrionale a Due Settori”. Note Economiche, 2, pp. 5-37.
Dixit, A.K. (1977). “The Accumulation of Capital Theory”. Oxford Economic Papers, 29, pp. 3-29.
Dobb, M. (1970). “The Sraffa System and Critique of the Neo-Classical Theory of Distribution”.
De Economist, 4, pp. 347-362.
Dobb, M. (1973). Theories of Value and Distribution since Adam Smith. Cambridge University
Press, London.
Donzelli, F. (1986). Il concetto di equilibrio nella teoria economica neoclassica. Nuova Italia
Scientifica, Roma.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

160
Dorfman, J., Samuelson, P. and Solow, R. (1958). Linear Programming and Economic Analysis.
McGraw-Hill, New York.
Eatwell, J. (1974). “Controversy in the Theory of Surplus Value: Old and New”. Science and
Society, 38, pp. 281-303.
Eatwell, J. (1990). Walras Theory of Capital, in Eatwell, J., Milgate, M. and Newman, P. (eds.).
The New Palgrave: Capital Theory. Macmillan, London.
Eatwell, J., Robinson, J. (1973). An Introduction to Modern Economics. MacGraw-Hill, London.
Felipe, J., Fisher, F.M. (2003). “Aggregation in Production Functions: What Applied Economists
Should Know”. Metroeconomica, 54, pp. 208-262.
Ferguson, C.E. (1969). The Neo-Classical Theory of Production and Distribution. Cambridge
University Press, Cambridge.
Ferguson, C.E. (1971). “Capital Theory up to date: A Comment on Mrs Robinson’s Article”.
Canadian Journal of Economics, 4, pp. 250-254.
Ferguson, C.E., Hooks, D.L. (1971). “The Wicksell Effects in Wicksell and in Modern Capital
Theory”. History of Political Economy, 3, pp. 353-372.
Fisher, F.M. (1969). “The Existence of Aggregate Production Functions”. Econometrica, 37,
pp. 553-577.
Fisher, F.M. (1971). “Aggregate Production Functions and the Explanation of Wages: A
Simulation Experiment”. Review of Economics and Statistics, 53, pp. 305-325.
Fisher, I. (1930). The Theory of Interest. Macmillan, New York.
Friedman, M. (1962). Price Theory. Frank Cass, London.
Gallaway, L., Shukla, V. (1974). “The Neoclassical Production Function”. American Economic
Review, 64, pp. 348-358.
Garegnani, P. (1958). “A Problem in the Theory of Distribution from Ricardo to Wicksell”.
Unpublished Ph.D. dissertation available at the Cambridge University Library, Cambridge
(UK).
Garegnani, P. (1960). Il Capitale nelle Teorie della Distribuzione. Giuffrè, Milano.
Garegnani, P. (1966). “Switching of Techniques”. Quarterly Journal of Economics, 80, pp. 554-567.
Garegnani, P. (1970a). “Heterogeneous Capital, the Production Function and the Theory of
Distribution”. Review of Economic Studies, 37, pp. 407-436.
Garegnani, P. (1970b). “A Reply”. Review of Economic Studies, 37, p. 439.
Garegnani, P. (1976a). On a Change in the Notion of Equilibrium in Recent Work on Value and
Distribution (A comment on Samuelson), in Brown, M., Sato, K. and Zarembka, P. (eds.).
Essays in Modern Capital Theory. North-Holland, Amsterdam.
Bi bl i ography
161
Garegnani, P. (1976b). “The Neoclassical Production Function: Comment”. American Economic
Review, 66, pp. 424-427.
Garegnani, P. (1979). Valore e Domanda Effettiva. Einaudi, Torino.
Garegnani, P. (1983). “Scuola di Cambridge” in Supplemento al Dizionario Enciclopedico
Treccani.
Garegnani, P. (1984a). “Value and Distribution in the Classical Economists and Marx”. Oxford
Economic Papers, 36, pp. 291-325.
Garegnani, P. (1984b). “On some Illusory Instances of ‘Marginal Products’”. Metroeconomica, 36,
pp. 143-160.
Garegnani, P. (1990). Quantity of Capital, in Eatwell, J., Milgate, M. and Newman, P. (eds.). The
New Palgrave: Capital Theory. Macmillan, London.
Garegnani, P. (1997). Comment, in Caravale, G. (ed.). Equilibrium and Economic Theory.
Routledge, London.
Garegnani, P. (2000). Savings, Investment and Capital in a System of General Intertemporal
Equilibrium, in Kurz, H. (ed.). Critical Essays on Piero Sraffa’s Legacy in Economics.
Cambridge University Press, Cambridge.
Garegnani, P. (2003). Savings, Investment and Capital in a System of General. Intertemporal
Equilibrium, in Petri, F., Hahn, F. (eds.). General Equilibrium: Problems and Prospects.
Routledge, London.
Garegnani, P. (2005). “Capital and Intertemporal Equilibria: A Reply to Mandler”.
Metroeconomica, 56, pp. 411-437.
Garegnani, P. (2008). “On Walras’s Theory of Capital (Provisional Draft 1962)”. Journal of the
History of Economic Thought, 30, pp. 367-384.
Gårdlund, T. (1958). The Life of Knut Wicksell. Almqvist and Wiksell, Stockolm.
Gibson, B. (ed.) (2005). Joan Robinson’s Economics: A Centennial Celebration. Edward Elgar,
Cheltenham and Northampton.
Hague, D.C., Lutz, F.A. (1961). The Theory of Capital. Macmillan, London.
Hahn, F.H. (1972). The Share of Wages in the National Income: An Enquiry into the Theory of
Distribution. Weidenfel and Nicholson, London.
Hahn, F.H. (1974). “Back to Square One”. Cambridge Review, 96, pp. 34-37.
Hahn, F.H. (1975a). “Revival of Political Economy: The Wrong Issues and the Wrong Argument”.
Economic Record, 51, pp. 360-364.
Hahn, F.H. (1975b). “Comment”. Cambridge Review, 96, p. 92.
Hahn, F.H. (1982). “The Neo-Ricardians”. Cambridge Journal of Economics, 6, pp. 353-374.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

162
Hahn, F.H., Matthews, R.C.O. (1964). “The Theory of Economic Growth: A Survey”. Economic
Journal, 74, pp. 779-902.
Harcourt, G.C. (1967). “Review”. Economic Journal, 77, pp. 359-361.
Harcourt, G.C. (1969). “Some Cambridge Controversies in the Theory of Capital”. Journal of
Economic Literature, 7, pp. 369-405.
Harcourt, G.C. (1970). “Review”. Journal of Economic Literature, 8, pp. 809-811.
Harcourt, G.C. (1972). Some Cambridge Controversies in the Theory of Capital. Cambridge
University Press, Cambridge.
Harcourt, G.C. (1974). The Cambridge Controversies: The Afterglow, in Parkin, M., Nobay, A.
(eds.). Contemporary Issues in Economics. Manchester University Press, Manchester.
Harcourt, G.C. (1975a). “Decline and Rise: The Revival of (Classical) Political Economy”.
Economic Record, 51, pp. 339-356.
Harcourt, G.C. (1975b). Capital Theory: Much ado about something. Thames Papers in Political
Economy, London.
Harcourt, G.C. (1976). “The Cambridge Controversies: Old Ways and New Horizons - or Dead
End?”. Oxford Economic Papers, 28, pp. 25-65.
Harcourt, G.C. (1986). “On the Influence of Piero Sraffa on the Contributions of Joan Robinson to
Economic Theory”. Economic Journal, 96, pp. 96-108.
Harcourt, G.C., Laing N.F. (eds.) (1971). Capital and Growth: Selected Readings. Penguin Books,
London.
Harrod, R. (1961). “Review”. Economic Journal, 71, pp. 783-787.
Hennings, K.H. (1990). Capital as a Factor of Production, in Eatwell, J., Milgate, M., and
Newman, P. (eds.). The New Palgrave: Capital Theory. Macmillan, London.
Hicks, J.R. (1932). The Theory of Wages. Macmillan, London [1963, reprint].
Hicks, J.R. (1937). “Mr. Keynes and the ‘Classics’; A Suggested Interpretation”. Econometrica, 5,
pp. 147-159.
Hicks, J.R. (1946). Value and Capital. Clarendon Press, Oxford [1939 1st ed.].
Hicks, J.R. (1960). “Thoughts on the Theory of Capital – the Corfu Conference”. Oxford
Economic Papers, 12, pp. 123-132.
Hicks, J.R. (1965). Capital and Growth. Clarendon Press, Oxford.
Hicks, J.R. (1974). “Capital Controversies: Ancient and Modern”. American Economic Review,
64, pp. 307-316.
Bi bl i ography
163
Hogan, W.P. (1958). “Technical Progress and Production Functions”. Review of Economic
Studies, 40, pp. 407-411.
Huth, T. (1998). Controversies in the Theory of Capital, in Kurz, H., Salvadori, N. (eds.). The
Elgar Companion to Classical Economics. Edward Elgar, Cheltenham and Northampton.
Jevons, S. (1879). The Theory of Political Economy, M. Kelly, New York.
Johnson, H.G. (1974). “Cambridge in the 1950s”. Encounter, 42, pp. 28-39.
Kahn, R.F. (1959). “Exercises in the Analysis of Growth”. Oxford Economic Papers, 11, pp. 143-156.
Kaldor, N. (1937). “Annual Survey of Economic Theory: the Recent Controversy in the Theory of
Capital”. Econometrica, 5, pp. 201-233.
Kaldor, N. (1955). “Alternative Theories of Distribution”. Review of Economic Studies, 23, pp. 83-100.
Kaldor, N. (1957). “A Model of Economic Growth”. Economic Journal, 67, pp. 591-624.
Kaldor, N. (1966). “Marginal Productivity and the Macroeconomic Theories of Distribution”.
Review of Economic Studies, 33, pp. 309-319.
Kaldor, N. (1972). “The Irrelevance of Equilibrium Economics”. Review of Economic Studies, 82,
pp. 1237-1255.
Kalecki, M. (1954). Theory of Economics Dynamics: An Essay on Cyclical and Long-Run
Changes in Capitalist Economy. George Allen and Unwin, London.
Keynes, J.M. (1936). The General Theory of Employment, Interest and Money. Macmillan,
London.
King, J.E. (2002). A History of Post Keynesian Economics Since 1936. Edward Elgar, Cheltenham
and Northampton.
Knapp, J. (1973). “Economics or Political Economy”. Lloyds Bank Review, 107.
Knight, F.H. (1950). “Capital and Interest”, in Encyclopaedia Britannica, vol. IV, 779-801
[reprinted in Fellner, W., Haley, B. (eds.) (1950). Readings in the Theory of Income
Distribution. Allen and Unwin, London].
Kregel, J. (1971). Rate of Profit, Distribution and Growth: Two Views. Macmillan, London.
Kurz, H. (1987). Capital Theory: Debates, in Eatwell, J., Milgate, M. and Newman, P. (eds.). The
New Palgrave: A Dictionary of Economics. Macmillan, London.
Kurz, H., Salvadori, N. (1997). Theory of Production: A Long Period Analysis. Cambridge
University Press, Cambridge.
Laibman, D., Nell, E. (1977). “Reswitching, Wicksell Effects and the Neoclassical Production
Function”. American Economic Review, 67, pp. 878-888.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

164
Lerner, A.P. (1953). “On the Marginal Product of Capital and the Marginal Efficiency of
Investment”. Journal of Political Economy, 61, pp. 1-14.
Levhari, D. (1965). “A Non-Substitution Theorem and Switching of Techniques”. Quarterly
Journal of Economics, 79, pp. 98-105.
Levhari, D., Samuelson, P.A. (1966). “The Non-Switching Theorem is False”. Quarterly Journal
of Economics, 80, pp. 518-519.
Mainwaring, L., Steedman, I. (2000). On the probability of reswitching and reverse capital
deepening in a two sector Sraffian model, in Kurz, H. (ed.). Critical Essays on Piero
Sraffa’s Legacy in Economics. Cambridge University Press, Cambridge.
Malinvaud, E. (1953). “Capital Accumulation and the Efficient Allocation of Resources”.
Econometrica, 21, pp. 233-268.
Marshall, A. (1959). Principles of Economics. 8th ed. 1920. Macmillan, London [1890 1st ed.].
Marx, K. (1961). Capital, vol. I, Foreign Languages Publishing House, Moscow.
Mata, T. (2003). “Constructing Identity: The Post Keynesians and the Capital Controversies”.
Journal of the History of Economic Thought, 26, pp. 241-259.
McManus, M. (1963). “Process Switching in the Theory of Capital”. Economica, 30, pp. 109-121.
Meade, J.E. (1961). A Neoclassical Theory of Economic Growth. Allen and Unwin, London.
Meek, R.L. (1973). Studies in the Labour Theory of Labour. Lawrence & Wishart, London.
Metcalfe, J.S., Steedman, I. (1973). Heterogeneous Capital and the Heckscher-Ohlin-Samuelson
Theory of Trade, in Parkin, M., Nobay, A. (eds.). Essays in Modern Economics. Longman,
London.
Milgate, M. (1979). “On the Origin of the Notion of ‘Intertemporal Equilibrium’”. Economica, 46,
pp. 1-10.
Modigliani, F., Samuelson, P.A. (1966a). “The Pasinetti Paradox in Neoclassical and More
General Models”. Review of Economic Studies, 33, pp. 269-301.
Modigliani, F., Samuelson, P.A. (1966b). “Reply to Pasinetti and Robinson”. Review of Economic
Studies, 33, pp. 321-330.
Morishima, M. (1960). “Existence of Solution to the Walrasian System of Capital Formation and
Credit”. Zeitschrift für Nationalökonomie (Journal of Economics). 20, pp. 238-243.
Morishima, M. (1966). “Refutation of the Non-Switching Theorem”. Quarterly Journal of
Economics, 80, pp. 520-525.
Morishima, M. (1969). Theory of Economic Growth. Clarendon Press, Oxford.
Naqvi, K.A., Robinson, J. (1967). “The Badly Behaved Production Function”. Quarterly Journal
of Economics, 81, pp. 579-591.
Bi bl i ography
165
Nell, E.J. (1970). “A Note on Cambridge Controversies in Capital Theory”. Journal of Economic
Literature, 8, pp. 41-44.
Ng, Y.K. (1974). “The Neoclassical and the Neo-Marxist-Keynesian Theories of Income
Distribution: A Non-Cambridge Contribution to the Cambridge Controversy in Capital
Theory”. Australian Economic Papers, June, pp. 124-132.
Nuti, D.M. (1976). On the Rates of Return on Investment, in Brown, M., Sato, K. and Zarembka,
P. (eds.). Essays in Modern Capital Theory. North-Holland, Amsterdam.
Nuti, D.M. (1977). “Price and Composition Effects and the Pseudo-production Function”. Revue
d’Économie Politique, 2, pp. 232-243.
Parrinello, S. (2005). “Intertemporal Competitive Equilibrium, Capital and the Stability of the
Tâtonnement Pricing Revisited”. Metroeconomica, 56, pp. 514-531.
Pasinetti, L. (1962). “Rate of Profit and Income Distribution in Relation to the Rate of Economic
Growth”. Review of Economic Studies, 29, pp. 267-279.
Pasinetti, L. (1965). “A New Theoretical Approach to the Problems of Economic Growth”.
Pontificiae Academiae Scientiarum, Scripta Varia, 28, pp. 571-696.
Pasinetti, L. (1966a). “Changes in the Rate of Profit and Switches of Techniques”. Quarterly
Journal of Economics, 80, pp. 503-517.
Pasinetti, L. (1966b). “New Results in an Old Framework”. Review of Economic Studies, 33,
pp. 303-306.
Pasinetti, L. (1969). “Switches of Techniques and the ‘Rate of Return’ in Capital Theory”.
Economic Journal, 79, pp. 508-531.
Pasinetti, L. (1970). “Again on Capital Theory and Solow’s ‘Rate of Return’”. Economic Journal,
80, pp. 428-431.
Pasinetti, L. (1977). Lectures on the Theory of Production. Macmillan, London.
Pasinetti, L. (1978). “Wicksell Effects and Reswitchings of Technique in Capital Theory”.
Scandinavian Journal of Economics, 80, pp. 181-189.
Pasinetti, L. (1996). Joan Robinson and ‘Reswitching’, in Marcuzzo, C., Pasinetti, L. and
Roncaglia, A. (eds.). The Economics of Joan Robinson. Routledge, London.
Pasinetti, L. (2003). Lezioni di Teoria della Produzione. Il Mulino, Bologna [1976 1st ed.].
Pasinetti, L., Scazzieri, R. (1990). Capital Theory: Paradoxes, in Eatwell, J., Milgate, M. and
Newman, P. (eds.). The New Palgrave: Capital Theory. Macmillan, London.
Pasinetti, L., Fisher, F., Felipe, J., McCombie, J.S.L., Greenfield, L., Cohen, A. and Harcourt, G.C.
(2003). “Cambridge Capital Controversies: Comments”. Journal of Economic Perspective,
17, pp. 227-235.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

166
Petri, F. (1978). “The Difference between Long-Period and Short-Period General Equilibrium and
the Capital Theory Controversy”. Australian Economic Papers, 17, pp. 246-260.
Petri, F. (2000). “On the Likelihood and Relevance of Reverse Capital Deepening”. Quaderni del
Dipartimento di Economia politica dell’Università degli Studi di Siena, 279.
Petri, F. (2004). General Equilibrium, Capital and Macroeconomics: A Key to Recent
Controversies in Equilibrium Theory, Edward Elgar, Cheltenham and Northampton.
Petri, F. (2007). “Review of Capital Theory edited by Bliss, C., Cohen, A. and Harcourt, G.
(2005)”. European Journal of the History of Economic Thought, 14, pp. 597-607.
Phelps Brown, E. (1972). “The Underdevelopment of Economics”. Economic Journal, 82, pp. 1-10.
Quandt, R.E. (1961). “Review”. Journal of Political Economy, 69, p. 500.
Reati, A. (2006). “Review of Capital Theory edited by Bliss, C., Cohen, A. and Harcourt, G.
(2005)”. EAEPE Newsletter, 36, pp. 19-24.
Ricardo, D. (1951-73). Works and Correspondence (vols. I-XI), edited by Piero Sraffa with the
collaboration of M. Dobb, Cambridge University Press, Cambridge.
Robbins, L. (1930). “On a Certain Ambiguity in the Conception of Stationary Equilibrium”.
Economic Journal, 40, pp. 194-214.
Robertson, D. (1957). Lectures on Economic Principles. (vols. 1-3). Asia Publishing House,
Calcutta.
Robinson, J. (1953). “The Production Function and the Theory of Capital”. Review of Economic
Studies, 21, pp. 81-106.
Robinson, J. (1955). “The Production Function and the Theory of Capital: A Reply”. Review of
Economic Studies, 23, p. 247.
Robinson, J. (1956). The Accumulation of Capital. Macmillan, London.
Robinson, J. (1957). “Economic Growth and Capital Accumulation: A Comment”. Economic
Record, 33, pp. 103-108.
Robinson, J. (1958). “The Real Wicksell Effect”. Economic Journal, 68, pp. 600-605.
Robinson, J. (1960). Exercises in Economic Analysis. Macmillan, London.
Robinson, J. (1962a). Essays in the Theory of Economic Growth. Macmillan, London.
Robinson, J. (1962b). “Comment”. Review of Economic Studies, 29, pp. 258-266.
Robinson, J. (1962c). Economic Philosophy. C.A. Watts, London.
Robinson, J. (1964a). “Pre-Keynesian Theory after Keynes”. Australian Economic Papers, 3, pp.
25-35, [in Robinson, J. (1965). Collected Economic Papers, Basil Blackwell, Oxford, vol.
III, pp. 56-69.].
Bi bl i ography
167
Robinson, J. (1964b). “Solow on the Rate of Return”. Economic Journal, 74, pp. 410-417.
Robinson, J. (1966). “Comment on Samuelson and Modigliani”. Review of Economic Studies, 33,
pp. 307-308.
Robinson, J. (1969). The Accumulation of Capital. 3rd ed. Macmillan, London.
Robinson, J. (1970a). “Capital Theory Up to Date”. Canadian Journal of Economics, 3, pp. 309-317.
Robinson, J. (1970b). “Review”. Economic Journal, 80, pp. 336-339.
Robinson, J. (1971a). Economic Heresies: Some Old-Fashioned Questions in Economic Theory.
Macmillan, London.
Robinson, J. (1971b). “The Measurement of Capital: The End of the Controversy”. Economic
Journal, 81, pp. 597-602.
Robinson, J. (1971c). “Capital Theory up to date: A Reply”. Canadian Journal of Economics, 4,
pp. 254-256.
Robinson, J. (1974). History versus Equilibrium. Thames Polytechnic, London [reprinted in
Robinson, J. (1979). Collected Economic Papers, Basil Blackwell, Oxford, vol. V, pp. 48-
58.].
Robinson, J. (1975a). “The Unimportance of Reswitching”. Quarterly Journal of Economics, 89,
pp. 32-39.
Robinson, J. (1975b). “Letter to Editor”. Cambridge Review, 96, pp. 91-92.
Rowthorn, R. (1974). “Neo-Classicism, Neo-Ricardianism and Marxism”. New Left Review, 86,
pp. 63-87.
Salter, W.E.G. (1960). Productivity and Technical Change. Cambridge University Press,
Cambridge.
Salvadori, N. (1996). ‘Productivity Curves’ in The Accumulation of Capital, in Marcuzzo, C.,
Pasinetti, L. and Roncaglia, A. (eds.). The Economics of Joan Robinson. Routledge,
London.
Samuelson, P.A. (1947). Foundations of Economic Analysis. Harvard University Press, Cambridge
(Mass.).
Samuelson, P.A. (1959). “A Modern Treatment of Ricardo Economy: II”. Quarterly Journal of
Economics, 73, pp. 217-231.
Samuelson, P.A. (1961). A New Theorem on Nonsubstitution, in Hegeland, H. (ed.). Money
Growth and Methodology (In honour of Johan Akerman). CWK Gleerup, Lund.
Samuelson, P.A. (1962). “Parable and Realism in Capital Theory: The Surrogate Production
Function”. Review of Economic Studies, 29, pp. 193-206.
Samuelson, P.A. (1966a). “A Summing Up”. Quarterly Journal of Economics, 80, pp. 568-583.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

168
Samuelson, P.A. (1966b). “Rejoinder: Agreements, Disagreements, Doubts, and the Case of Induced
Harrod-neutral Technical Change”. Review of Economics and Statistics, 48, pp. 444-448.
Samuelson, P.A. (1973). Economics. 9th ed. McGraw-Hill, New York.
Samuelson, P.A. (1975). “Steady-state and Transient Relations: A Reply on Reswitching”.
Quarterly Journal of Economics, 89, pp. 40-47.
Samuelson, P.A. (1976) Interest Rate Determinations and Oversimplifying Parables: A Summing
up, in Brown, M., Sato, K. and Zarembka, P. (eds.). Essays in Modern Capital Theory.
North-Holland, Amsterdam.
Samuelson, P.A., Solow, R. (1953). “Balanced Growth under Constant Returns to Scale”.
Econometrica, 21, pp. 412-424.
Schlicht, E. (1985). Isolation and Aggregation in Economics. Springer-Verlag, Berlin.
Sen, A. (1963). “Neo-Classical and Neo-Keynesian Theories of Distribution”. Economic Record,
39, pp. 53-64.
Sen, A. (1974). “On Some Debates on Capital Theories”. Economica, 41, pp. 328-335.
Solow, R. (1955). “The Production Function and the Theory of Capital”. Review of Economic
Studies, 23, pp. 101-108.
Solow, R. (1956). “A Contribution to the Theory of Economic Growth”. Quarterly Journal of
Economics, 70, pp. 65-94.
Solow, R. (1957). “Technical Change and the Aggregate Production Function”. Review of
Economics and Statistics, 39, pp. 312-320.
Solow, R. (1963). Capital Theory and the Rate of Return. North-Holland, Amsterdam.
Solow, R. (1967). The Interest Rate and Transition between Techniques, in Feinstein, C.H. (ed.).
Socialism, Capitalism and Economic Growth: Essays presented to Maurice Dobb.
Cambridge University Press, Cambridge.
Solow, R. (1970). “On the Rate of Return: Reply to Pasinetti”. Economic Journal, 80, pp. 423-428.
Solow, R. (1983). Modern Capital Theory, in Brown, E.C., Solow, R. (eds.). Paul Samuelson and
Modern Economic Theory. McGraw-Hill, New York.
Spaventa, L. (1970). “Rate of Profit, Rate of Growth, and Capital Intensity in a Simple Production
Model”. Oxford Economic Papers, 22, pp. 129-147.
Sraffa, P. (1926). “The Laws of Return under Competitive Conditions”. Economic Journal,
36, pp. 535-550.
Sraffa, P. (1951-73). Introduction, in The Works and Correspondence of D. Ricardo, vols. I-XI,
Cambridge University Press, Cambridge.
Bi bl i ography
169
Sraffa, P. (1960). Production of Commodities by means of Commodities: Prelude to a Critique of
Economic Theory. Cambridge University Press, Cambridge.
Sraffa, P. (1962). “Production of Commodities: A Comment”. Economic Journal, 72, pp. 477-479.
Steedman, I. (1985). “On input ‘demand curves’”. Cambridge Journal of Economics, 9, pp. 165-172.
Steedman, I. (2006). “Long run Demand for Labour in the Consumer Good Industry”.
Metroeconomica, 57, pp. 158-164.
Stigler, G. (1941). Production and Distribution Theories: The Formative Period. Macmillan, New
York.
Stiglitz, J.E. (ed.) (1966). The Collected Scientific Papers of Paul A. Samuelson. MIT Press,
Cambridge (Mass.).
Stiglitz, J.E. (1974). “The Cambridge-Cambridge Controversy in the Theory of Capital: a View
from New Haven: a Review Article”. Journal of Political Economy, 82, pp. 893-903.
Swan, T.W. (1956). “Economic Growth and Capital Accumulation”. Economic Record, 32, pp.
334-361.
Tobin, J. (1973). “Cambridge (UK) vs. Cambridge (Mass.)”. The Public Interest, 31, pp. 102-109.
Tosato, D. (1997). Equality of Rates of Return in Models of General Economic Equilibrium with
Capital Accumulation, in Caravale, G. (ed.). Equilibrium and Economic Theory. Routledge,
London.
Uhr, C.G. (1951). “Knut Wicksell: A Centennial Evaluation”. American Economic Review, 41,
pp. 829-860.
Valiente, W. (1980). “Is F. Knight the Victor in the Controversy between the Two Cambridge?”.
History of Political Economy, 12, pp. 41-64.
Varri, P. (1998). Fixed and Circulating Capital: Historical Aspects, in Kurz, H., Salvadori, N.
(eds.). The Elgar Companion to Classical Economics. Edward Elgar, Cheltenham and
Northampton.
Walras, L. (1954). Elements of Pure Economics or The Theory of Social Wealth. 5th ed. 1926,
translated by W. Jaffé, George Allen and Unwin, London.
Weizsäcker, C. von (1971). “Ende einer Wachstumstheorie?”. Kyklos, 24, pp. 97-101.
Wicksell, K. (1934). Lectures on Political Economy, vol. I., Routledge, London [1901, Swedish
ed.].
Wicksell, K. (1954). Value, Capital and Rent. George Allen and Unwin, London [1893, German
ed.].
Wicksteed, P.H. (1933). The Commonsense of Political Economy and Selected Papers and
Reviews of Economic Theories. Routledge, London.
Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

170
Worswick, G.D. (1972). “Is Progress in Economic Science Possible?”. Economic Journal, 82, pp.
73-86.
Yeager, L.B. (1973). “The Keynesian Diversion”. Western Economic Journal, 11, pp. 150-163.




Index of Figures
Chapter 2
Figure 2.1. Labour employment curve........................................................................ 20
Figure 2.2. Capital employment curve........................................................................ 20
Figure 2.3. Employment function of labour with upwards-sloping shape ................... 25
Figure 2.4. Employment function of capital with upwards-sloping shape................... 25
Chapter 3
Figure 3.1. The wage-curve ........................................................................................ 45
Figure 3.2. The wage-frontier and ‘reswitching’ of techniques................................... 48
Figure 3.3. ‘Reverse capital deepening’...................................................................... 50
Chapter 4
Figure 4.1. A straight wage-curve and a family of straight wage-curves ................... 60
Figure 4.2. The ‘parable system’ ................................................................................ 61
Figure 4.3. Concave, convex and straight ‘envelopes’................................................ 64
Figure 4.4. q is not equal to S/L when the wage-curve is convex ............................... 68
Chapter 5
Figure 5.1. The productivity curve ............................................................................. 96
Figure 5.2. Robinson’s ‘pseudo-production function’................................................. 97
Figure 5.3. Wage curves and the ‘pseudo-production function’ (neoclassical case).... 98
Figure 5.4. The ‘pseudo-production function’ ............................................................ 99
Figure 5.5. Robinson’s ‘pseudo-production function’ amended................................ 100
Figure 5.6. Wage-curves (‘reswitching’) .................................................................. 100
Figure 5.7. The ‘pseudo-production function’ .......................................................... 100
Figure 5.8. The ‘pseudo-production function’ in terms of capital per worker ........... 101
Chapter 6
Figure 6.1. ‘Reverse capital deepening’.................................................................... 133
Appendix A
Figure A.1. The net marginal product curve ............................................................. 146
Appendix D
Figure D.1. Concave wage-curves and the employment function of ‘capital’ ........... 155
PP
R
e
v
i
s
i
t
i
n
g

t
h
e

C
a
m
b
r
i
d
g
e

C
a
p
i
t
a
l

T
e
o
r
y

C
o
n
t
r
o
v
e
r
s
i
e
s
:

A

H
i
s
t
o
r
i
c
a
l

a
n
d

A
n
a
l
y
t
i
c
a
l

S
t
u
d
y
A
N
D
R
É
S

L
A
Z
Z
A
R
I
N
I
Edizioni dell’Università degli Studi di Pavia
€ 22,20
Texts published by Pavia University Press in the series “Editoria scientifca” have been peer-reviewed
prior to acceptance by the Editorial Board. www.paviauniversitypress.it/scientifca
Tis book deals with the Cambridge capital theory controversies both from a historical and
from an analytical standpoint.
In the frst part the author addresses the basic analytical and methodological issues underlying
the neoclassical theory, with special emphasis on how the problem of a measure of capital
arises within the marginalist theory of value and distribution. Te second part surveys the
most relevant contributions to the Cambridge debates in the light of the salient results of
‘reswitching’ and ‘reverse capital deepening’, and shows how the implications of these results,
which touch the principle of factor substitution, brought about diferent strategies pursued
by neoclassical scholars to overcome these theoretical problems.
Te book concludes that since the results of this debate touch the foundations of the theory,
and hence are of a general character, it is hard to accept that the contemporary versions of
the theory are free of capital problems, while at the same time the Cambridge controversies
are absent in the current literature used for training economics students. Te book should
therefore be useful to the undergraduate and graduate students, as well as to scholars devoted
to the History of Economic Analysis.

Andrés Lazzarini, Ph.D. in Economics from Roma Tre University, is a Visiting Fellow
at the Institute of Advanced Study – Jawaharlal Nehru University, New Delhi. He has
been Visiting Assistant Professor of Microeconomics at the University of Alicante,
Spain, in 2010, and Research Fellow and Lecturer of History of Economic Thought at
the Department of Political Economy and Quantitative Methods, University of Pavia,
in 2008 and 2009.
His main research areas are: History of Economic Analysis, Methodology of Economics
and Economic Development in Historical Perspective.
E-mail: andres.lazzarini@eco.unipv.it / alazzarini@gmail.com
Andrés Lazzarini
Revisiting the Cambridge Capital Theory
Controversies: A Historical and Analytical Study
Editoria scientifica
Pa v i a Un i v e r s i t y P r e s s
ISBN 978-88-96764-17-6

Editoria scientifica

A NDRÉS L AZZARINI Revisiting the Cambridge Capital Theory Controversies: A Historical and Analytical Study .

Cambridge (UK). No part of this book may be reproduced or transmitted in any form or by any means. – 20. Economia – Teorie – Gran Bretagna – Sec.it> Printed by: Print Service – Strada Nuova. without written permission from the author. 2011 – Pavia ISBN: 978-88-96764-17-6 Texts published by Pavia University Press in the series “Editoria scientifica” have been peer-reviewed prior to acceptance by the Editorial Board.1 CDD-22 – Economia. 67 – 27100 Pavia . 19. sistemi. 2007 Publisher: Pavia University Press – Edizioni dell’Università degli Studi di Pavia Biblioteca Delle Scienze . including photocopying.paviauniversitypress. 330. www. : ill. Margarita Olivera.paviauniversitypress. Andrés Revisiting the Cambridge capital theory controversies: a historical and analytical study / Andrés Lazzarini. – Pavia : Pavia University Press. teorie © Andrés Lazzarini. recording or by any information storage and retrieval system. 2011. 6 – 27100 Pavia <http://www. scuole. 24 cm.Lazzarini.it/scientifica. ISBN: 978-88-96764-17-6 1. – VI. electronic or mechanical. .Via Bassi. The cover shows: The Mathematical Bridge. 172 p.

...................................... From ‘employment’ functions to ‘demand’ functions (I)......... The data of the Marginal Theory...Contents Preface and Acknowledgements ...10.......... Why a study on the capital theory controversies? 1................................... Understanding the foundations of neoclassical economics: The principle of factor substitution and the problem of capital 2...............3... Economies with more than one system of production.....................5 1............................................1 1............................... 17 2.....2... Supply-side role of capital................................................. 34 Chapter 3................................ 18 2....... The relevance of this conflict to contemporary economic analysis .............................2............................4.... ‘Reswitching’ and ‘Reverse Capital Deepening’: Introduction to the heart of the controversies and their implications 3..5...4........9................................................. Persistence character of the determining circumstances: The traditional method of ‘normal positions’ ............... From ‘employment’ functions to ‘demand’ functions (II)............................1.........................2...........4. 42 3.......................................... Radically different views on the meaning of the critique for the contemporary versions of the theory ..............3...................................................................................... Analytical and Methodological Issues in Neoclassical Economics: Setting the Ground to Understand the Capital Controversies Chapter 2... Economies with heterogeneous capital goods............................... Problems connected with the heterogeneity of capital goods........ Demand-side role of capital: Conclusion .....8.........6 1........... 25 2......6.................. 27 2..........6............................. 24 2......... A second phase in the capital controversies. Key role of the downward-sloping shape of factors’ demand curves ... 21 2.................. 39 3.................. 19 2....... The wage-frontier. The capital supply function .. Some premises........5.... Implications for the basic premises of the theory: Conclusion........................ Capital as a factor of production and the capital theory debates ........ 23 2.............. V Chapter 1..................7..................... 31 2... Sketch of this study . The indirect mechanism of factor substitution .......... 10 PART I...............................................3 1..................................................... 46 3.............1...............9 1.. A first phase in the capital controversies ............ The wage curve ..........................1...... 49 ..........................3... The direct mechanism of factor substitution .............

..........................5............................................... There is no heterogeneous capital goods model in Samuelson’s (1962) ...................................... 80 4................................................... 1982) and Bliss’s (1975) views on the question of capital: A first critical excursus ..................... 103 5........................8........... Walras’s way-out to the inconsistency...... 55 4.............6....... 1956) and Swan (1956) ............................................................................... 105 5...Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies PART II......... 68 4........1................4............................................. 119 II ....... Samuelson’s (1962) model with heterogeneous capital goods ..... 78 4.. 112 6........... The neoclassical ‘parable’: ‘Surrogate’ capital and the ‘Surrogate’ production function .......................14............................................................ Neoclassical reactions: Samuelson (1966a) .4....... Introduction ............. Introduction .................7.............. 70 4............................... 109 Chapter 6...........................................................................................4.... A first phase in the controversies: Samuelson’s 1962 intervention........................................... Robinson’s critique: Aggregate production functions as her main target ...... A second phase in the capital controversies and radical changes in the neoclassical theory 6................. 75 4..... Equilibrium and expectations ...................... 74 4....................................................8................ Rehabilitation of Walras’s treatment of capital as physically heterogeneous goods and Walras’s inconsistency ............... 63 4............ 91 5....................... Understanding Robinson’s way: The ‘Pseudo-production function’ and the ‘curiosum’ ..... 95 5..................2...... The meaning of the 1966 Symposium: A turning point in the controversies...3..............................................................................1...... Further neoclassical reactions after the Symposium: Ferguson (1969) ................ The Capital Theory Debates in a Historical Perspective and the Transformation of Dominant Theory Chapter 4........ Garegnani’s 1970 critique of the marginal theory..5............ 57 4...................... the 1966 Symposium........ 91 5................. Bliss’s 1970 intervention...... 59 4............ 93 5................. Implications of the Walrasian specification of capital .....3........................... 101 5... Conclusion .......... Samuelson’s ‘special sub-class of realistic cases’..13...... 111 6............................................................ Champernowne’s (1953) contribution ...... 73 4.. Garegnani (1966)....... neoclassical reactions and the highest point of the debates 4........... The zenith of the debates ........................................................... 116 6................. 87 4..9.......... Levhari (1965) and the 1966 Quarterly Journal of Economics Symposium ........... 85 4......6................................5................... Robinson rejoinders... 60 4........ Prelude to the shift to neo-walrasian equilibria and the beginning of a second phase in the debates.....................3.....2......7.........................................1..................2........... 89 Chapter 5.......... A step back in the debates: Joan Robinson’s role in the controversies and first neoclassical reactions 5.....................................12......................................... 118 6........................................... Hahn’s (1975a...............10... Introduction ...11.......... Further replies to Robinson: Solow (1955.....................15......................................... What did Samuelson actually demonstrate?.......................... Pasinetti (1966a).................................................................

................................................................ 125 6...................1..7..................................................................... 124 6..7................................. Conclusion: some obscurity in the debates .................... 153 D........................... Confrontations .............. Discussion of the ‘supply-side’ of capital in Wicksell’s general equilibrium system ........................ 132 6............................................ Commodity own rates of interest ............................. 124 6... The curve of the net marginal product of labour ..6............ 132 6.........6................ Conclusions ..........6..6............7.................................................. 171 III .................................. 147 C.............................................. 155 Bibliography ............3.....................................2...... 131 6................. 157 Index of Figures.............................. Introduction ..... ‘Reverse capital deepening’ without ‘reswitching’ ...2........ 139 Appendixes A........................................................................................ Further discussion of Hahn (1975a) ....Contents 6............ 124 6.....................6.......................................... Do ‘reswitching’ and ‘reverse capital deepening’ bear theoretic problems on intertemporal general equilibrium? Two views: Bliss and Garegnani.3.... Sraffa’s landmark in the critique of the marginal theory ... 143 B...... Garegnani ................................................ Bliss........ Uniform rates of return on supply prices in intertemporal equilibria .............. 127 6..................................................... 136 Chapter 7.................................................. 134 6......7...........1.......................................... Uniform profit rates and intertemporal prices................................ Hahn’s ‘special case’ .........6...........................................4................5........

.

and in fact the present book has dealt with most if not all of them in our reconstruction. That there existed a controversy between Cambridge (UK) and Cambridge. which I eventually achieved in April 2008. is the different – and sometimes we may say radically different – positions not only between the two camps of the debate but also within the critical side of the controversy. but which have not been thoroughly dealt with by previous surveys and which deserve due attention to better understand the present state of the controversy between the neoclassical authors and the critical side. But. were visibly recognised in the last contributions of a first phase of the debates. Massachusetts (US). Despite the fact that the critically oriented (Cambridge. does not follow a chronological survey of the contributions to the Cambridge controversy but rather focuses on the analytical arguments necessary to understand the implications of the phenomena of ‘reswitching’ and ‘reverse capital deepening’. and which also deserves due attention for understanding the evolution of the controversy. Yet the recognition of its most relevant results is still lacking even though they touch upon the core of the basic economic principles underlying the core of mainstream neoclassical economics. unlike many of the reviews available. In order to understand why these phenomena affect the very core of neoclassical economics. first. in particular during its later phase. after more than fifty years since the first exchanges in these debates many articles and books have been written about this theoretical conflict in the economic theory. These arguments. our reconstruction attempted to present an account which.D. UK) side agreed that the main results of the controversy had put the foundations of neoclassical economics at serious risks – and hence the latter had to be replaced with an alternative economic theory – differences in . The so-called Cambridge controversy in the theory of capital took place between the beginning of the 1950s and the mid-1970s. though arguably it got its heyday after the publication of Sraffa’s 1960 book. Roma Tre University. we present and discuss some contributions of a later phase of the controversy in which the treatment of capital among the data of the theory had already been shifted to its Walrasian specification. Since then I have presented and discussed several works drawn from my original dissertation which immensely helped to improve the present version for publication at the Pavia University Press. we have thoroughly examined the ultimate reason of their theoretical existence: the specification of capital in value terms among the data of the theory in order to determine equilibrium variables through supply and demand forces which ultimately rely on the factor substitution principle. to obtain my Ph.Preface and Acknowledgements This book is the result of my research on the post-war Cambridge capital theory controversies conducted at the Department of Economics. from a chronological viewpoint. Another issue examined in the present study. Certainly. Secondly. in Economics (Dottorato di Ricerca in Economia Politica). could hardly be ignored by any practitioner of the discipline.

Roberto Ciccone from Roma Tre University for his supervision. his guidance at the initial stages of my research and his enlightening criticisms. and who no doubt made many technical discussions here examined far more interesting. and Denis Melnik. I have to acknowledge the Alan Office (scholarship no: E04D040397AR) for providing me with the financial support to undertake my Ph. January 2011 VI . Scott Carter. Of course. I would also like to thank my family and friends from Argentina for their emotional support over these years in my efforts to pursue the Dottorato di Ricerca in Rome. I should like to acknowledge Prof. and Frederic Lee who provided me with very valuable material in the last phase of completion of this study. Prof. Stefano Finamore for our long talks and travels in Italy. Pierangelo Garegnani for his commitment to science. I am the only responsible for all remaining errors contained in this work. First of all.D. Ariel Dvoskin. and Prof. friendly and endless support and reception that my friends Luca Mantovan. I have also to thank all my colleagues from Roma Tre University. Gianni Vaggi. his encouragement and his invaluable critiques and observations on my research. Piero Mella and an anonymous referee for the possibility to publish my research at the Pavia University Press. Special gratitude goes to my partner. Andrés Lazzarini Pavia. Juan José Pradelli for long discussions (in Rome and elsewhere) on economic theory. Enrico Bellino. This book would have never existed if I had not got the warm. I would like to thank Dr. who not only helped me to complete and edit this book and shared her constructive comments on my work but also supported me with her endless love during many years in our European life experience. I am indebted to many people who in one way or another helped me in the endeavour to study capital theory and the ensuing controversies. studies in Roma Tre University from 2004 up to 2008.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies the method among the critical side (and also the silence on the part of the marginalist authors in the second phase as to accepting the controversy’s results) helped to prevent that alternative from surfacing at that time. Fabian Amico and Alejandro Fiorito. Eduardo Crespo. But I am also indebted to a number of colleagues and friends (from different corners of the world) with whom I have been discussing over the last two years or so. Amit Bhaduri and Irina Eliseeva. Last but not least. Margarita Olivera. Annalisa Doneda. In particular I would like to thank Professors Geoff Harcourt. Ariel Wirkierman and Nadia Garbellini provided me in my many recent visits to Pavia. Valeria Abbonato.

This gave rise to a series of exchanges between scholars associated with Cambridge. and the foundational premise underlying the dominant supply and demand approach: the factor substitution principle.1 The relevance of this controversy lies in that the criticisms of neoclassical theory raised by Cambridge (UK) concern both the theoretical illegitimacy of measuring ‘capital’ as a single magnitude in value terms to determine prices and distribution. 9. as this study will argue.1. discovered by means of the analysis of the relation between prices and distribution in economies with heterogeneous capital goods. See also n.Chapter 1 Why a study on the capital theory controversies? mainstream / 'meInstri:m / n [the+S] the main or most widely accepted way of thinking or acting in relation to a subject Longman Dictionary of Contemporary English (1991) 1. and. England. 2 In this book we shall refer to the neoclassical or marginal approach interchangeably – though the author should like to emphasise that the label ‘neoclassical’. Massachusetts. Paradoxically. This debate is broadly known in the literature as the ‘Cambridge capital theory controversies’. below. Capital as a factor of production and the capital theory debates Beginning in the mid-1950s and for the following twenty years or so. neoclassical theory is the only This label was christened in print by the first survey written by Harcourt (1969). a debate concerning the neoclassical treatment of capital turned apparent in the discipline. is a misfortune in the history of economic ideas. this chapter. 1 . current economics has put up this historical debate in a hermetically sealed compartment of its teaching programme. This result. has been revealed as theoretically irrefutable. and Cambridge. concerns the hard core of the neoclassical or marginal theory 2 both in its traditional (capital measured in value terms) and in its contemporary formulations (capital as a set of heterogeneous goods). (US). As can be seen by leafing through any undergraduate or graduate economic theory textbook. used in the literature to refer to a theory that is essentially anti-classical. In the controversies it has been shown that this principle cannot in general be posited to explain the distribution of the social product in terms of supply and demand.

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies one theory being taught. entailed a criticism of the whole marginalist analysis. Chapter 2. thus revealing that until those years capital as a single magnitude appeared to be the central notion. we are sometimes informed by the mainstream of economics that the Cambridge debates would only be relevant to an ‘old’ or incomplete version of the theory that treats capital in value terms but not to the contemporary versions. in particular p. Hayek. below. 4 Although this study has not systematically dealt with the general situation of neoclassical capital theory before the Cambridge controversies. capital in value terms not only is needed in order to allow for factor substitutability. 114. traditional theory had to resort to explain the distribution of the social product between wages and profits (and rents) in terms of supply and demand – in other words. and Chapter 4 below. which specify the factor capital as a set of heterogeneous goods. on the other hand. Problems associated with the treatment of capital in value terms started to be recognised by neoclassical economists. 424-427. it has been due to the implications of the debates that dominant theory shifted from the traditional to the contemporary treatment of the factor capital. under conditions of free competition. Dewey (1965) Chapter 8. and Kaldor. mainly involved Knight. we shall assume. Malinvaud 1953). Neoclassical capital theory has been one of the most controversial areas of economic science since its inception in the discipline in the last quarter of nineteenth century. while the second one. in spite of the increasing reservations arisen in the period referred to as to the legitimacy of this notion. who argues to dispense with the traditional notion of capital altogether (see Garegnani 1976a).3 At the time of the Cambridge controversies. Although serious misgivings regarding this notion of capital started to arise in the late 1930s and over the 1940s. 3 2 . Huth (1998). but also to determine a uniform rate of profits on the supply prices of the several capital goods. It is interesting to recall that Knight’s entry “Capital and Interest” in the Encyclopaedia Britannica. At best. because. as we will argue. section 5. and Robertson (1957).g. the traditional quantity of capital.B. still the centre of economic theorising treated capital as a single magnitude in value terms. According to Kurz (1987) the two former controversies within the neoclassical theory involved J. the Cambridge controversies. in the main. While former controversies arose out from different viewpoints on the supply side of capital within the marginal approach.4). the results drawn from the controversies are also relevant for a better understanding of the present state of neoclassical capital theory. As we shall see in the following chapter. Böhm-Bawerk at the turn of the twentieth century. pp. For the different controversies on capital theory within the neoclassical approach in the twentieth century see Kurz (1987). Indeed controversy in capital theory was not unusual in the neoclassical school.4 Capital as a single magnitude has been the treatment on which. Cf. and Kaldor (1937). However. 1990) and Petri (2004. by relying on Garegnani (1976a. in the 30s. Clark and E. though. which considers capital as a single magnitude expressed in value terms. that the notion of capital shared by the centre of economic theorising was. was reprinted in 1950 by the American Economic Association (see Knight 1950). by applying the factor substitution principle. The condition of a uniform rate of profits on the capitals’ supply prices was as universally accepted a condition by the founders of marginal analysis as it was by the classical political economists since Adam Smith. Satisfaction See Garegnani (1970a). such as Hicks (1946). (See e. capital as factor of production was still conceived as a single magnitude in value terms among the data of the theory.

6 Mainly in The Quarterly Journal of Economics. while the British one as the non-neoclassical or the critical side. p. (cont.8 Hardly could any practitioner of the discipline deny the fact that over more than twenty-five years the controversy in the theory of capital occupied a prominent magnitude in contemporary history of economic analysis. See also Bliss (1975. 1) and Robinson (1970a. For that determination it is crucial that the theory accounts for downward-sloping demand curves for factors. Burmeister (1980. Chapter 6. 2005). 9 Though the Cambridge-Cambridge divide is not strictly accurate to represent the division of the two different approaches in the history of economic analysis behind these controversies. Petri (2004. More recent assessments from the critical side also include Cohen and Harcourt (2003. 1. and Ahmad (1998). Chapter 4. that treatment of capital5 allows determining a normal position of the economy’s quantities and prices characterised by a uniform rate of profits on the capitals’ supply price under free competition both in consumption-goods’ and factors’ markets. 7 For instance the oral debates between Paul A. Robinson (1970a). Samuelson and Pierangelo Garegnani in 1961-62 in Cambridge.) 3 5 . This oral exchange later gave rise to Garegnani (1970a). the Cambridge controversies have shown that the belief on an inverse relation between demand for factors and their rewards cannot in general be postulated. still it is acknowledged in the literature that the American group may be regarded as the neoclassical side of this debate. 8 See Harcourt (1969. See Chapter 4 below and also Samuelson (1962. 1972.Chap te r 1 – Why a study on th e capita l th eory con tro versies? of this condition ensures that the equilibrium determined by neoclassical supply and demand forces could conceivably be taken as a long period equilibrium in that the persistent nature of the conditions of the economy can guarantee that the theoretical variables become gravitational centres of the actual ones (trends or average) over time. Garegnani (1976a) for assessments from the critically oriented sides of the debates and Blaug (1974). As we shall see in this study. after more than This treatment of the capital as a single magnitude in value terms is the way in which capital was traditionally conceived as a factor of production. Stiglitz (1974). This is so because. 2007). Chapter 15). The relevance of this conflict to contemporary economic analysis The importance of this controversy for the discipline is indubitable. The Review of Economic Studies. p. while for the neoclassical side we should refer to Hahn (1982). Birner (2002) and the entries provided by Kurz (1987). Hahn (1972. Capital in value terms as traditionally conceived is analysed in Chapter 2. Endless misunderstandings in the controversies were created as to whether the critique only bears on versions of the theory using aggregates production functions. Massachusetts (US) during a research visiting period of the latter. 1975a). Chapter 4. 193. 2003 “Appendix II”). See bibliography below. 2000). Garegnani (1990. 1976). Bliss (2005). Thus. Bhaduri (1969). The Journal of Economic Literature and The Economic Journal.9 However. On this issue cf. Ng (1974) for neoclassical surveys. Hicks (1974). below. But it will be seen below that this conception has nothing to do with the use of ‘aggregate’ production functions for the determination of distribution. To the vast amount of contributions which were published in that period by many scholarly journals on economic theory 6 – not to mention the oral debates 7 – we have to add the ensuing reviews and surveys from both sides assessing the significance of this remarkable episode. 311). n. and perhaps it might have been seen by some scholars as an aggregate factor.2.

Nicholas Kaldor.C. or “Anglo-Italian critics”. key neoclassical participants. Solow (this one still is) were affiliated to the MIT (US). Stiglitz. Pasinetti. On the other hand. 12 Cf. unlike his contemporary peers. Hahn’s (1975a. where we deal with this author’s view of the controversy. without even showing the slightest effort to inform students of this episode12 that is the capital theory controversies. Ng (1974). to either the neoclassical or the critical side. Harcourt (1972). in fact neoclassical scholars deem the theory of supply and demand relying on a Walrasian capital as the one and only theory. 13 See Felipe and Fisher (2003) who have drawn attention to why economists still use aggregate functions in theoretic and empirical research despite the capital controversies. But. 4 . UK (in Bliss’s case at least part of the time). P. for instance in some recent neoclassical growth literature like Real Business Cycle and Endogenous Growth theories. Krishna Bharadwaj. among others. which. Arrow. neo-Walrasian authors argue that the debates would have no relevance to the mainstream of economic thought. 1960s and even 1970s hosted some critics of orthodoxy. Piero Sraffa. Ferguson. Robinson 1962b. In this work we shall refer to critics as the ‘critically oriented side’ or. Harcourt. “neo-Marxist-Keynesians”. According to this view. A view similar to Hahn can be found in Sen (1974). the Cambridge debates would have petered out once neoclassical theory started to rely on a physical specification of capital (‘Walrasian’ treatment)11 thus rendering. were affiliated to Cambridge. we can see today. chiefly A. 11 It has been preferred to use the term ‘Walrasian’ treatment of capital because Walras. the ‘critical side’. in terms of theoretical membership. below. 364) global assessment of the debates: “Great charade”. G. it is worth noticing that the label ‘neo-neoclassical’ was sometimes used by some participants (especially J.10 we might perhaps be asked by a marginalist scholar something like “Why to rekindle this debate to-day?” Well-trained in mainstream ambiences he would remind us that the completely developed theory of capital is the one which treats ‘capital’ as a set of physically heterogeneous goods. as we shall see in this study. Hahn. Chapter 4 below. See Chapter 5 below. apparently. Both Christopher J. J. treated ‘capital’ as a collection of physically heterogeneous capital goods. Thus. 10 There is almost complete agreement that Joan Robinson’s (1953) article kicked off the debates. This reveals that a part of the discipline 13 accepts the supply and demand approach treating capital as a single magnitude as if nothing had broadly speaking. Also. C. F. like Paul A. Bliss and Frank H.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies fifty years since the first exchanges between Cambridge (UK) and Cambridge (US). Joan Robinson. D. 1971a) to refer to the fourth generation of neoclassical economists that not only includes the ones referred to above but also K. “PostKeynesians”. Chapter 2 and n. Barro and Salai-Martin (1995). Garegnani. on the other hand. the supply and demand approach immune from capital theoretic problems. Luigi L. Bliss. more simply. King (2002). See Chapter 6. p. who became the most active neoclassical participants during the 1970s. We keep to the geographical divide suggested by the literature in this study. Bhaduri. 100. There have of course been many other contributors that worked elsewhere at that time but were associated. Samuelson and Robert M. Note also that some literature refer to the critically oriented side as the “neo-Keynesians”. Hahn. the English school in the 1950s. See Chapter 5 below. that ‘aggregate production functions’ are still used by the bulk of mainstream economists working in growth and aggregate investment issues. Levhari. Cf. cf. C. have serious implications on the contemporary versions of the theory. and not capital as a single magnitude. Bliss (2005).

Harcourt’s 1972 survey.3. and concludes by summarising the Cambridge alternative distribution theory as represented by the works of Kaldor (1955). From a historical standpoint.14. then it continues by presenting the ‘reswitching’ debate. it is due to Robinson’s (1971b) suggesting title “The End of the controversy”. 17 Beginning with Joan Robinson’s kick-off 1953 article criticising the use of the aggregate production functions. As will be seen later (Paragraph 4. A first phase in the capital controversies The answer that will be attempted here suggests considering a breakdown of the period of the debates into two phases. our identification of two phases in the controversies goes along with a breakdown in the development of the debates as has been documented by the literature of the critical side of this theoretical conflict. 16 Harcourt’s (1972) book is an extended version of his original survey article Harcourt (1969). nobody talks about it any longer? 1. The second phase (1970/1-1976) 15 regards a situation which might be characterised. vintage models. unclear chapter in the history of the debates where many misunderstandings damaged communication between the two parties.14 after the first exchanges between the two Cambridge much centred on the use of aggregate production functions in economic growth models. it seems. the traditional long period supply and demand equilibrium theory. which. starting from Joan Robinson’s 1953 kick-off. Chapter 4 and Chapter 6).16 This work has spread the controversies in the discipline and is the most frequently cited work by the bulk of practitioners when referring to the first phase of the debates. The first phase (1953-1970/1) regards the period when salient discoveries like ‘reswitching’ and ‘reverse capital deepening’ started to gain notoriety in the discipline. Robinson (1962a) and Pasinetti (1962). namely. as an obscure. follows a chronological account of the different contributions to the controversies. Regarding the first phase it has been natural for the present author to take into account one of the most frequent references to the Cambridge controversy in the literature. So a very important question arises: What happened to the Cambridge debate that today. we have considered 1970/1 as the end of the first phase because in 1970 Bliss (1970) replied to Garegnani (1970a) on the grounds of a different notion of equilibrium and thus misunderstanding Garegnani’s criticisms of the theory at stake which was. following Joan Robinson’s target See Chapters 3 and 4 below. Harcourt’s account continues by reviewing later replies that Robinson’s article stirred up. the debate on the rate of return involving Solow and Pasinetti. As to 1971. In Harcourt (1972) we find an all-inclusive review of the main contributions to the debates from both sides.Chap te r 1 – Why a study on th e capita l th eory con tro versies? ever occurred in the recent history of economic analysis. In this second phase neoclassical capital theory had already shifted from the traditional treatment of capital in value terms to the Walrasian specification of it and this helps explain why some of the exchanges in this phase were at cross-purposes.17 Harcourt (1972). On the other hand 1976 has been chosen as the ending date of the second period for reasons that are described in the text. Though we shall deal with Robinson and her (cont. for reasons we shall analyse below.) 15 14 5 . technical progress and economic growth models. and is. His survey focuses on issues regarding aggregates production functions.

3) states that the aggregate production function “is indeed by far the most common version” of the neoclassical theory. along with Robinson. below. it is yet not clear. in Harcourt (1972). as we shall see later. But. xii). of these theoretic results on the versions of the theory relying on a Walrasian treatment of capital. as we shall see later.19 But. 18 Harcourt’s concern about the aggregate production function is particularly clear when Harcourt (1972. Indeed our contention can be verified by some different views in this regard in some relevant literature of the critical side. a Walrasian treatment of capital) or to another reason. below where we deal with Robinson’s critique. Then. see also Chapter 6. even if Harcourt points out that ‘double switching’ and ‘capital reversal’ damage the foundations of the marginal approach. Wicksell’s system does not use aggregate production function and yet has to resort to a quantity of capital in value terms. Note also that neoclassical authors have almost always taken Robinson’s target of criticisms at aggregate production function in order to contend the critics. p. 1. On this.e.18 Moreover. a second phase in the controversies started being recognised both by direct participants and commentators. the consequences. This is the position that Bliss also maintains in his (1975). We shall argue later (Chapter 6) whether Bliss’s contention is due to a supposedly ‘more complete’ analysis which takes as data the many capital goods there are in the economy (i. A second phase in the capital controversies If we now turn to Harcourt’s updated account of the Cambridge controversy in 1976 we shall notice a very different opinion regarding whether the results of ‘reswitching’ and ‘reverse capital deepening’ drawn from the first phase of the controversies bear on the immediate contestants in Chapter 5 on issues related to the debates on capital theory. 19 See Chapter 5.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies of criticisms. the problem of capital does not uniquely entail using aggregate production functions. Harcourt draws methodological implications in that both these authors argue that equilibrium comparisons are not good tools for an analysis of changes. See our discussion of Wicksell’s general equilibrium system in Appendix B. the so-called Cambridge theory of distribution is not considered by us here. thus bringing about important reactions from the neoclassical side that tried to save the basic principles underlying that theory. focuses on the so-called ‘aggregate production functions’ rather than on the problems for the theory owing to the treatment of capital in value terms it needs rely on. It is the opinion of the present author that when these arguments started to be raised by neoclassical participants in the debates. For instance. Thus. 20 Bliss (2005. p. 20 counteractions and replies from the critical side were not unified. from the criticisms centred on the aggregate production function. if any.4. when neoclassical writers started to argue against the critics by contending that a “general equilibrium model of capital does not require aggregation”. phenomena like ‘reswitching’ and ‘reverse capital deepening’ cast serious doubt on the robustness of theory itself independently of whether it uses aggregate functions or not. 6 .

Chap te r 1 – Why a study on th e capita l th eory con tro versies? versions which rely on a Walrasian treatment of capital. and see Bellino (2005). p. of the conception of capital it relies on. 22 Similar positions can be found in Cohen and Harcourt (2003. 26). p. Chapter 4. however. Cf. and. Thus. n. In fact. 23 Garegnani (1976a. See also the protestations raised against Cohen and Harcourt (2003) by Pasinetti et. p. 207. while Harcourt (1976) suggests that the neo-Walrasian theory “may emerge logically intact” from capital theoretic problems. we can find in Harcourt (1976) a quite different assessment of the relevance of the debates: We now conclude: if the arguments of this article are accepted. 21 7 .”23 What is important for our purposes to notice here is that the consequent abandonment of the long period equilibrium method has not been due to weaknesses of the method as such. (2003). al. while general equilibrium may emerge logically intact as ‘the theory of inter-temporarily efficient paths and their price duals’ [Hahn 1972. 3] it is not the theory which is relevant for the issues raised in the Cambridge controversies. with which is allied the problem of measurement of ‘capital’. 66. On the meaning of an intertemporal general equilibrium. 24 Garegnani (1976a. is disaggregated. and in particular. below. in Garegnani (1976a) we can find that his appraisal of the controversies is quite different from Harcourt’s comments. p. Reati (2006) and Petri (2007) for comments on Cohen and Harcourt (2005). needs a quantity of capital in value terms for the determination of equilibrium (not an ‘aggregate production function’ which is what Harcourt seems to suggest in the passage). 2005. emphasis in the original. Garegnani argues that after Hicks’ (1946 [1939]) Value and Capital an increase proportion of practitioners persuaded in the neoclassical approach started to shift to a “short period general equilibrium. xxxiv). This breakdown in different versions between an ‘aggregate’ and ‘general equilibrium’ one is misleading since it does not take care of Wicksell’s system which is a general equilibrium. Harcourt means the latter to be the neo-Walrasian approach of simultaneous equations. but rather to weaknesses of the dominant theory of distribution. underlines added).24 Harcourt (1976. 26). Harcourt splits the theory in two versions: the aggregate versions (“the other versions of neoclassical theory” in the passage) and the ‘general equilibrium’ one. and the (more fundamental) criticisms that stem from the distinction between comparisons and changes.21 This different appreciation of the results of the debates on the ‘intertemporal general equilibrium’ theory as viewed by Harcourt (1976) 22 has induced the present author to delve into this second part of the controversies and place at the centre another view on the relevance of the critique from the critically oriented side. p. The attempts to use the other versions of neoclassical theory flounder both on the results of the reswitching and capital-reversing debates. p. 58. Thus. see Chapter 6.

p. 38).).” Echoing Robinson’s concerns. with all its methodological implications. 28 Among other problems of the theory which relies on a Walrasian treatment of capital. whatever the different reasons.” Note that ‘disaggregated neoclassical models’ are meant as the neo-Walrasian equilibrium theory. 1990) and Milgate (1979). 58). there does not seem to be agreement in the critical literature on whether the results of the Cambridge debates would have forced the theory – in its attempts to preserve the supply and demand approach and to overcome the deficiencies of capital – to adopt a method which is radically different from the traditional one and hence whether the debates bear on the new versions of the theory. It is apparent that in this work Harcourt has given more importance to the “criticisms that stem from the distinction between comparisons and changes. cf. Although in Garegnani (1976a) “the question of whether the two kinds of analysis [intertemporal and temporary equilibria] have in fact overcome the deficiencies of demand-and-supply theory with respect to capital [is left] aside”.29 Indeed Harcourt (along with Cohen) have recently argued that the shift from ‘aggregate’ versions to a theory relying on a Walrasian capital would 25 Garegnani (1976a. due to the Walrasian specification of capital the marginalist theory was forced to adopt a short period general equilibrium method as a last refuge from the results of the controversies regarding the treatment of capital which the long period equilibrium neoclassical theory need rely on.26 But. I submit. then that shift would be representing the very strength of the critique. 8 . is the result of the capital theory debates. it is then no surprise that Harcourt has regarded the ‘reswitching’ and ‘reverse capital deepening’ debates as a “Dead End” (id. then it is hard to accept that the new versions of the theory may emerge logically intact. These different views also point to a gap within the critical side with respect to whether ‘reswitching’ and ‘reverse capital deepening’ are the “more fundamental” criticisms in the debates. 328) who says: “the implications of reswitching for disaggregated neoclassical models [of the Arrow-Debreu type] were still not clear. p. it is interesting to note that if the controversies compelled the theory to adopt a method of short period general equilibrium. the shift to Walrasian theory. See Chapter 6 below. Thus. as Harcourt (1976) seems to conclude. 27 Cf. Garegnani’s arguments are briefly dealt with in Chapter 6 below. Garegnani (1976a. p. Thus. 2000. 25 once we take into account that a radical shift in method undergone by the theory has taken place.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies According to this view. in particular Robinson (1975a). 27 which lacks the necessary persistency of the equilibrium28 and was in the main confined to a minority group of mathematical economists. Garegnani (1990. it is not the case that Walrasian theory is not the version of the theory which is relevant for the issues raised in the Cambridge debates. 29 Harcourt (1976. 2003) has taken up this issue and has shown that the same kind of difficulties regarding the conception of capital in the traditional versions of the theory relying on a notion of capital in value terms also reappears in the new versions in the investment and savings markets. 26 This apparent halfway situation of the controversies is also recorded by Backhouse (1985.

the 2nd ed. 34 For instance. 2004). by disseminating misleading ways to treat the issues. Garegnani’s (1976a) insights into the issue at stake.32 One of the aims of this book is to clearly show how the Cambridge capital theory controversies endanger the specification of capital as a single magnitude on which the traditional marginal theory relies. 9 . as Petri (2007. which relies on a Walrasian treatment of the factor capital.’ 30 They thus neglect other critics’ views such as Garegnani’s (1976a). Bliss (1970. Dixit (1977). xxii). in order to contribute for a better understanding of the results of the capital debates and their implications for the neoclassical theory nowadays. the works cited in n. 34 have put the debates in a quite obscure chapter of contemporary history of economic analysis. above. 1982). 1975. 33. Cohen and Harcourt (2005 p. p. that the sole ‘disaggregated’ analysis would appear to be the new versions of neo-Walrasian theory. 31 In this book we have considered Hicks (1946). Wicksell’s analysis is as disaggregated as were analyses conducted by Böhm-Bawerk (1891). this chapter. at the same time.g. thus broadcasting a misleading idea. 1975a. xxiv) suggest that the “characterisation of simultaneous equations [in disaggregated analysis of neo-Walrasian theory] is correct”. But this view seems not to have considered that e. 1. of Hicks’ (1939) Value and Capital. below. 604) has recently remarked. 1975b. however. Eatwell (1990). we shall discuss the economic meaning of Walras’s system of capitalisation. 1974. p. the misleading way in which neoclassical economists received the critique was as if it concerned the so-called aggregate production functions. 2005). and see also Chapter 6. neoclassical economists took advantage of the nonunified position with respect to the implications of ‘reswitching’ and ‘reverse capital deepening’ on the new versions of the theory and that these neoclassical authors.Chap te r 1 – Why a study on th e capita l th eory con tro versies? have represented a shift to ‘more disaggregated analysis. that is. which suggests 30 This idea is implicitly suggested by Cohen and Harcourt (2005. In the present author’s opinion.5. Cf. neoclassical authors started to oppose the critique by contending that capital theory need not have recourse to any aggregation. gave rise to a series of contributions attempting to clarify some issues of the controversies in capital theory in its second phase. See: Petri (1978. the critical side did not immediately provide a unified rejoinder to the above neoclassical contention. In addition. without however discussing the economic meaning underlying those formal systems. In Chapter 6 below. Milgate (1979). Radically different views on the meaning of the critique for the contemporary versions of the theory At the turn of the 1970s. 32 See Chapter 6 below. who argues that the key to understand the shift to a Walrasian specification of capital is to realise that the theory had to adopt that way – which had already been provided by Hicks’ (1946)31 – due to the deficiency of the traditional notion of capital as revealed by the implications of the results of the debates. 33 thus suggesting that the contemporary versions of the supply and demand approach would not be subject to theoretic capital problems as in the traditional versions (capital in value). how far neoclassical theory relying on a Walrasian capital is immune from capital theoretic problems. or by Hicks (1932). 33 This is the position shared by Hahn (1972.

the method adopted in this study embraces. 1. from a chronological viewpoint. below.6. Of course. the one that confuses the neoclassical treatment of capital as a single magnitude in value terms with the illegitimacy of using aggregate production functions is arguably the most dangerous when assessing the importance of this theoretical clash. As will be seen later. Since we will argue that the capital controversies touch the basic premises of the neoclassical theory. in that we deal with historical aspects of this conflict in the discipline. And this methodological decision has been made by the present author for a very important purpose: to rescue the true nature of the critique produced in the controversies in order to contribute to a better understanding of the current situation of neoclassical capital theory. misleading ways used by some participants in the second phase of the debates have fogged communication and might have therefore detained a full appreciation of the strength of the implications of the controversies that touch the general principles of dominant theory. misunderstandings included. 10 . a historical standpoint.”36 Thus. Of the many misunderstandings created along this controversy. essential notions of the marginal approach are introduced in the first part of this study. Thus. on the one hand. Garegnani (2005). I submit. 411. In order to understand the main issues debated over the capital theory debates.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies regarding this period as the second phase of the Cambridge controversies. p.35 this obscurity in the debates. the chapter aims at providing the basic analytical elements to understand the importance of 35 36 See Chapter 6. corn-capital and labour economy). on the other hand. Following the description method developed in Garegnani (1990). However. were brought about by marginalist authors in their attempt to reverse the sign of the critique by introducing misleading interpretations of previous results (first phase) and reinforcing further misunderstandings that “have marred the controversies on capital. we have not chosen to proceed in a chronological order: we have rather preferred to place at the centre the analytical arguments necessary to understand the implications of the phenomena of ‘reswitching’ and ‘reverse capital deepening’ which. Sketch of this study This study is divided in two parts: while the first part is devoted to introduce the analytical and methodological issues in neoclassical economics. were visibly recognised in the last contributions of the first phase of the debates (particularly after Sraffa’s 1960 book). the confusion between steady state equilibrium and long period equilibrium) were due to historical aspects which are not completely covered by this study. which is explained in simple terms for the sake of exposition (corn-consumption. the second part analyses and discusses the historical aspects of the Cambridge controversies in the light of the analytical elements provided in the first part. in the next chapter (Chapter 2).g. many other misunderstanding (e. one of the first key notions in neoclassical theory examined is the factor substitution principle.

Thus. we shall see that 11 . issues and episodes developed in the first phase of the controversies. The chapter also introduces the meaning of the persistent circumstances underlying the method of the long period equilibrium. In the second part of this study – which consists of three Chapters (4. attempted to minimise or downplay the relevance of the results produced in the capital debates by invoking the ‘empirical’ way-out. Ferguson (1969). Then. neoclassical scholars. (Chapter 2. it is worth recalling an important question in the aftermath of the Symposium: while the heart of the controversies have referred to the traditional. who tried to find out the conditions to rule out ‘reswitching’. Rather. student Levhari (1965). In fact in economies with heterogeneous capital goods a problem of measurement of capital arises. Chapter 3 introduces the meaning of the most relevant results of the Cambridge controversies: ‘reswitching’ and ‘reverse capital deepening’. by surveying some key marginalist authors’ views on the matter. While Chapters 4 and 5 deal with the themes. as we shall argue. In this connection we shall face that problem from two angles: we distinguish a supply-side or supply-aspect of the problem – which is related to the inescapable need to treat ‘capital’ as a factor in value terms – from a demand-side or demand-aspect of the problem. we turn to economies with heterogeneous capital goods and the treatment of capital measured in value terms is examined. and the neoclassical replies that the latter gave rise to.10). Yet. because they subsequently gave rise to a famous Symposium on Paradoxes in Capital Theory held by The Quarterly Journal of Economics in 1966. the implications of these phenomena which were at the centre of the most salient exchanges between the two sides. Once the basic notions of the theory are laid down. more importantly. 5. we have preferred to firstly analyse and discuss the group of contributions that provide the essential nature of the criticisms. The purpose of this chapter is to introduce the reader to the concepts and. Paragraphs 2. the importance in the development of this conflict is analysed by bringing to the fore the interventions raised in that Symposium by Pasinetti (1966a) and Garegnani (1966). These interventions were very important from a historical perspective of the controversies.7-2. which is related to the possibility to conceive of that factor’s demand such that owing to changes both in the methods of production and in output the quantity of that factor demanded could be conceived to be inversely related to its rate of remuneration. The chapter then goes on to analyse the attempts made by Samuelson’s Ph. We thus begin with (Chapter 4) some of the most important contributions in these debates by surveying Samuelson’s (1962) intervention and the limits of his particular defence of the neoclassical theory. Chapter 6 considers the questions and problems of the second phase of the debates.Chap te r 1 – Why a study on th e capita l th eory con tro versies? the factor substitution principle for the neoclassical approach so as to grasp the notion of a downward-sloping demand curves for ‘factors of production’ which is key to derive a plausible explanation of distribution in terms of supply and demand (Chapter 2. though recognising theoretic difficulties with respect to capital for the whole approach. long period equilibrium versions of the neoclassical theory.1-2. e. Paragraphs 2.6). the equilibrium notion that was at the heart of the controversies. Moreover.D. and 6) – we turn our attention to the salient historical aspects of the capital controversy.g. The present author should like to emphasise that Chapters 4 and 5 do not follow a chronological order. that is.

we shall examine in this chapter the problems related to the issue of such different rates by discussing Walras’s system of capitalisation37 and the critique provided by Garegnani (1958. without however pointing to the general problems entailed by a treatment of capital in value terms for the derivation of supply and demand functions which are essential to the marginalist approach – although. thus starting to create misunderstandings and confusions which did not help communication and which in fact characterises this second part of the debates. we present and discuss the particular reaction from the neoclassical side that has tried to avoid the criticisms by returning to a treatment of capital in physical terms (Walrasian treatment). short period general equilibrium versions of the theory represent a radical transformation of the dominant approach in the wake of the capital controversies. The following chapter (Chapter 5) turns back in our historical reconstruction of the controversies: we survey Robinson’s (1953. 1956) first interventions in the Cambridge controversies and her immediate contestants Champernowne (1953). It will be argued there that over the first exchanges between 1953 and 1960 the debate rather centred on problems regarding measurements of ‘aggregate’ capital in ‘aggregate’ production functions. 1956. Since one of the most frequent misunderstandings is the confusion between uniformity of the rate of profit on the capital goods’ supply prices (the traditional rate) and uniformity of commodities-own rates of interest. we shall see how these neoclassical authors took refuge from the criticisms by confining the analysis to short-run general equilibrium and pinpointing that the Cambridge critique only bears on the ‘aggregate’ versions (read it as the ‘aggregate’ production function) but not on the modern versions of the theory. 2008). following Garegnani (1976a). Thus. that is why the present author has deemed appropriate to briefly deal with Bliss’s first intervention in 1970 against Garegnani (1970a). in Chapter 6. or the traditional method of long period equilibrium has to be abandoned. some scholars had tangentially faced the issues of ‘reswitching’ and ‘reverse capital deepening’ at that time. Solow (1955. 1957). as it will be argued. 1990. Swan (1956). Thus. the last word will be devoted to a provisional examination regarding whether or not ‘reswitching’ and ‘reverse capital deepening’ bear on the new 37 Walras (1954). Because the relevant literature regards Bliss and Hahn as the most notably exponents of this type of reaction we shall deal with how both these authors faced the results of the controversies. So far the study devotes to what the present author has above referred to as the first phase of the debates. as we shall see. in order to point out the way adopted at that moment by neoclassical authors to overcome the problems concerning the treatment of capital.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies some reactions on the part of some neoclassical scholars actually relied on a short period general equilibrium. 12 . However. Finally. Also. it will be argued that the problem of the non-uniformity of the rates of return poses the problem of which way-outs could be pursued by dominant theory if the theoretical framework of supply and demand were still to be maintained: either capital is expressed in value terms.

38 The confrontation is very provisional and relies mainly on Bliss (1975) and Garegnani (2000.Chap te r 1 – Why a study on th e capita l th eory con tro versies? versions of intertemporal and temporary general equilibrium models. where some conclusions are advanced. two positions will be confronted: Bliss (who has attempted to overcome some of the implications of ‘reswitching’ on the intertemporal theory) and Garegnani (who has extended the critique to these new versions). 2003).38 The study finally ends in Chapter 7. Accordingly. 13 .

.

PART I Analytical and Methodological Issues in Neoclassical Economics: Setting the Ground to Understand the Capital Controversies .

.

Since the data are of an exogenous nature. it is worth considering firstly the foundations on which the principle of factor substitution relies. It must now be seen how this determination comes about. The endowment of each ‘factor of production’ available in the economy. tastes and preferences are rather studied by psychological and sociological theories). To begin with. the marginalist theory can determine the relative prices. By using these data. ii. output. it will be explained how the two mechanisms of factor substitution. This simplifying assumption could be replaced by the one establishing that the composition of the output – actually composed of many goods – is given. however. which underlie the whole marginal approach. depend crucially on those data. The direct mechanism of factor substitution To understand the basic features of the critique of capital theory. operate. by introducing a very simple economy. their determination falls outside the purview of analysis of the theory (e. The data of the Marginal Theory. The endogenous variables to be determined. Alternative techniques of production (methods of production available) at a certain attained state of technical knowledge. Paragraph 2. . output and distribution. In what follows.g.1 The three groups of data taken as independent variables by marginal theory are: i.Chapter 2 Understanding the foundations of neoclassical economics: The principle of factor substitution and the problem of capital 2. therefore the data must not change while the endogenous variables are being determined.1. Let us assume an economy where the only consumption good is corn and the only capital good in it is corn too. and which roles the above data play in it.9. it can be useful to list the three groups of the data taken as independent variables – as ‘given’– by the marginalist theory in order to determine prices. Consumers’ tastes and preferences. below. iii.2 We further assume that corn is produced by means of 1 2 Cf. this chapter. and distribution simultaneously.

first. since any fall in the wage rate will make the price of ‘corn’ fall relative to ‘cloth’. each capitalist is faced with a curve of the net ‘marginal product’ of labour. We know that. Now it is consumer’s choice through group (i) of the data that plays a similar role to producer’s choice and marginal products in providing the basis for a decreasing ‘labour-employment’ schedule. that is: K cloth / Lcloth  K corn / Lcorn . 2. Therefore. with ‘cloth’ requiring the higher proportion of corn-capital to labour. p.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies labour and corn-capital in continuously variable proportions.g. Let us see how it works. which results when the variable quantities of labour she may hire is applied to the given quantity of corn-capital she possesses. cost-minimising firms tend to adopt production methods that utilise a higher proportion of labour relative to corn-capital. 337): “The notion of the marginal employment of any agent of production implies a possible tendency to diminishing return from its increased employment. therefore diminishing returns – by keeping fixed one of the factors – must prevail within each firm of the industry). To simplify matters. both productive factors – labour and corn-capital – are assumed to be in fixed supply. not implying therefore to use ‘all’ the given quantity of corn-capital – and then decreasing (otherwise. ‘cloth’. We now however assume that each of the two goods is produced with only one method of production (fixed coefficients of production). by allowing group (ii) of the data to work. after introducing consumer’s choice.g.3 (See Appendix A) The net marginal product of the labour curve indicates the entrepreneur how many labourers it would be profitable to hire with her capital at any given real wage. There is free competition. the increment of output due to hiring one more unit of labour would always be higher. mechanism of factor substitution. which are given by the group (ii) of the data. by Marshall (1959. by adding up the individual ‘labour employment’ curves of each capitalist it is can easily be obtained a ‘labour employment’ schedule for the whole economy at any given real wage. If distribution changes (e. given the quantity of corn-capital. it will follow that more corn will be The importance for the marginal theory of the diminishing returns principle is expressed e. then. we have illustrated the first. land is free (so we disregard the issue of rent). which is also produced with labour and corn-capital. The indirect mechanism of factor substitution Now let us introduce a second consumption good. and entrepreneurs are the capitalists. direct. the net ‘marginal product’ curve of labour will be.2. Then. In fact. constant – because the entrepreneur applies the exact amount of corn-capital per unit of labour so as to optimise the combination of both inputs in order to obtain the maximum profit. the wage rate falls). All transactions are assumed to occur in terms of corn. Excessive applications of any means to the attainment of any end are indeed sure to yield diminishing returns in every branch of business. Under this very simplified economy. Thus this curve becomes a ‘labour employment’ curve.” 3 18 .

2. corn needs more labour for any quantity of corn used as capital than cloth does.e. So. Paragraph 2. The important point to notice here is that even when there is no alternative methods of production available (i. Accordingly. it is assumed that workers will tend to bid down their wage so long as some of them are unemployed. and therefore the interest rate. 6. as we shall see below 4 – a higher proportion of it – of that constant corn-capital – will be employed in the production of corn. this chapter. This helps us understand why both marginal productivity and marginal utility are both key concepts in the foundations of neoclassical theory. as will be examined in more detail in the following paragraphs.4.5 That said.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics consumed. since. group (ii) of the data only provides one fixed-coefficients production method). let us meanwhile go on assuming that supply of corn-capital is fixed. it can be envisaged that the concepts of marginal utility and marginal productivity will reinforce each other in ensuring the downward sloping labour-employment schedule. From ‘employment’ functions to ‘demand’ functions (I) To use the ‘factor-employment’ curve as a basis for demand functions we should find a justification for assuming that the amount of corn-capital employed tends to remain constant. because the substitution among consumption goods implies an indirect substitution among the factors employed in their production. as Garegnani has pointed out: we need establish that the amount of ‘corn-capital’ employed tends to vary in accordance with its ‘supply function’ as the wage rate. Therefore. Garegnani (1990).3. What we need. vary. p. in other words. given the quantity of corn-capital employed in the economy – a crucial assumption to derive demand functions. by assuming fixed production coefficients. substitutability among consumption goods can give rise to decreasing employment curves for factors. If we reintroduce the assumption of variable coefficients of production. a further condition to ensure that the quantity of labour ‘demanded’ and ‘supplied’ are equal at any level of the wage rate and the interest rate (whatever those levels may be) is to assume some kind of flexibility on both variables. whereas capitalists unable to 4 5 Cf.. below. 19 . is that the shape of those ‘employment functions’ will not be in contradiction with the conception of free competition. Or. analogous to the employment factor curves derivable from technical substitution. more precisely. with a consequent increase in the quantity of labour employed in the economy as a whole.

because the existence of unemployment u (that is u=L*–L’) will push workers to bid w down. from the economic theory’s viewpoint. which will result in unemployment and a fall in wages through the competition of the unemployed (…) The final wage must therefore lie somewhere between the marginal product of the sixteenth and that of an imaginary seventeenth labourer. the number of labourers employed is L’.2. Again.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies hire the quantity of labourers they want will bid the wage up. tend to gravitate around the equilibrium wage rate w* and interest rate i*. Capital employment curve 20 .1-2. 6 Figure 2.2 below). the tendency of capitalists to bid the wages up. as assumed by free competition. if wages are higher than the marginal product. L*). which is lower than L* (the quantity of labourers seeking employment). the assumed fall in the wage rate will lead to such a situation that the quantity L* will be employed at w*. By the same token. at an initial wage rate at w’’. Now. pp.” 7 Consider Figure 2. emphasis in the original): “So long as wages are materially lower than the marginal product of the sixteenth labourer [the last]. and consequently their endeavour must result in a rise of wages. but even more importantly. therefore. As Wicksell wrote (1934. See Appendix A below. 113-114. when they are unable to hire the workers they want. will lead to the point (w*.7 By referring to those functions – and particularly to their properties referring to their shapes – not only we can plausibly support how free competition works or tends to work (as shown in Figures 2. we have arrived at the resulting explanation of distribution of the social product in terms of the forces of supply and demand.1. Suppose that the market wage rate is w’. it will be to the advantage of every landowner [being the entrepreneur] to employ more than sixteen labourers [the maximum]. The economy will. But all the landowners cannot simultaneously succeed in this object. each of the landowners will content himself with less than sixteen workers.1. Labour employment curve Figure 2. It is actually this gravitational process which variables undergo that allows us to establish that demand and supply schedules determine both w* and i*.6 This implies that the ‘employment-curves’ are downward-sloping.

1. in order to derive the resulting increase in the labour employed and.1. Due to competition. w drops) the quantity of corn-capital employed should not change with the wage. the marginalist theoretical framework is not complete. since corn production is more ‘labour intensive’ than cloth production.) 21 8 . since more corn produce is needed a higher proportion of corn capital will be employed in the corn sector (we keep on assuming fixed coefficients of production). that the quantity of capital changes in the way expressed by its supply function (compatible with the labour employment schedule) is plausible because the entrepreneurs are the capitalists. or.g. because the production of a unit of cloth requires more corn-capital than labour relatively to a unit of corn output.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics 2. But. a consequent increase in the quantity of labour employed in the economy will be reached.9 In diagram 2. Upwards arrows indicate ‘increase in’. and i to be the interest rate of the market – that is the interest rate which the capitalists would obtain by supplying corn-capital in case the workers (or a third party) were the entrepreneurs. KDcloth and KDcorn are the demand for corn-capital from the cloth industry and from the corn industry respectively. As the wage rate falls the production of cloth becomes more expensive relatively to corn production. corn (in its role as consumption commodity) will be demanded in a higher proportion relatively to cloth. w is the wage rate in terms of corn. to arrive at an explanation of equilibrium by the forces of supply and demand through the direct and indirect substitution mechanisms. if (cont. i(r) = i in equilibrium. Pcorn and Pcloth are the relative prices of corn and of cloth respectively. LDcorn is the demand for labour from the corn industry. or change in accordance with the labour-employment schedule. From ‘employment’ functions to ‘demand’ functions (II) Yet. Labour market when w falls corn  L Not only the assumption of constant capital employment. Through consumer choice corn is demanded in a higher proportion to cloth. more generally. CornY is the output of corn. 9 In other words. e.4.g. Production of corn must be increased to satisfy the new consumer demand.8  w  Pcorn  Pcloth  Corn D  CornY  K D cloth  K D corn  LD Diagram 2. suppose i(r) to be the interest rate that capitalists as entrepreneurs earn for each unit of corn-capital employed. But this assertion makes sense so long as the corncapital employed in the economy remains constant. and L is the quantity of labour employed in the economy. In fact if we assume that the quantity of corn capital employed in the economy is kept constant as the wage rate falls. Even if workers (or a third party) are the entrepreneurs that assumption is still plausible because a similar mechanism to the one corresponding to labour market will also be at work in the corn-capital market. hence. CornD is the demand for corn as a consumption good.. therefore. while downward arrows ‘decrease in’. It has been said above that when w varies (e. Otherwise. therefore. This is the (indirect) substitution mechanism derived from consumers’ tastes and preferences. The following diagram illustrates our argument.

KDcloth is the demand for corn-capital from the cloth industry. K corncapital. It follows. then he will receive interest. Suppose P to be corn-output. accordingly.) 22 . then he must receive a wage like any other worker. the marginal productivity of the last unit of each factor. If the share. Therefore. or. Then suppose a 1% increment on each factor: 101L  100 K  P  PL and 101L  101K  P  PL  Pk . For this. of labour in the product is different according as which party acts as entrepreneurs (that is w(r)  w). If. As is well known. (that is (1 + %. in equilibrium. Competition prevents this contradiction from taking place. constant returns do not prevail at the level of the industry) the levelling influence of competition could no longer be invoked by the theory. further. as Wicksell had put it. the profits of entrepreneur must be nil.e. Capital market In diagram 2. since corn-capital. on the other hand. then every capitalist would find it more profitable to act as an entrepreneur than to supply corn-capital to firms managed by workers or a third party. 129. would not be supplied. p. on the contrary. the total product also increases exactly proportionally”. he might obtain a share of product not based on either corn-capital or labour. then everybody would like to enter this ‘easy’ business. as Wicksell (1934. consider that an increment of 1% on each factor will yield an increment of P higher than 1%. Suppose. P With constant returns. for the marginal contribution of the last unit of each factor – (cont. puts it that “large-scale and small-scale operations are equally productive. in this case. it is “the law of constant returns”10 that ensures that both the wage and the interest rates will not be affected by which person acts as entrepreneur. hence.2. and symbol  stands for ‘produce’). In fact. and L labour. If this person also employs corn-capital in production.2. and K is the quantity of corn-capital employed in the economy. Consider the following diagram. the latter being distributed between both factors according to their marginal contribution to production. the difference will soon be cancelled out by competition. 10 Wicksell (1934).. And this reasoning is similar for the case of corn-capital. that the rates of remuneration to pay for 100 100 cannot be paid from the total output. 11 The profits of the entrepreneur must therefore be zero. if the entrepreneur devotes his human capacities to the management of production. were it otherwise (i.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies In fact. Now. Consider the following example. when all factors are increased in the same proportion. that one unit of P can be obtained with hundred units of each factor: 100 L  100 K  P . 126). But this would contradict our hypothesis of the existence of a corn-capital market. ClothD is the demand for cloth. ClothY is the output of cloth.  i  Pcloth  Pcorn  Cloth  Cloth  ( givenL )  K D Y D cloth  K Diagram 2. we have PL  PK  which immediately implies that 100 100 PL  100 PK  P so that the wages of 100L and interests of 100K together exactly correspond to the whole product. (Symbol  stands for ‘together with’. say. with >0: P P 101L  101K  P   . Both labourers and capitalists are free to employ their labour or corncapital on their own account or to hire it out to others. i is the interest rate in terms of corn. equals the respective rates of remuneration (both expressed in terms of P). Constant returns can therefore be seen as the necessary condition for an explanation of distribution according to marginal productivity principles. so that. we need the condition of constant returns.11 i(r) > i. p.

difficulties which the marginalist theory might come across. due to competition. and so to ensure that the labour employed in the economy will be increased. Furthermore. a fixed one.14 which governs the remuneration’s equilibrium levels – is higher than the product: 100 PL  100 PK  (1   ) P . to pay for PL and PK. (Similarly. vary. so if the wage rate in terms of corn decreases.1 above) and the downward-sloped demand schedule of corn(cont. can we a priori establish that the supply function of capital will always be compatible with the downward-sloped employment schedule of labour? When w falls (and i rises).) 13 23 . constant returns to scale imply F ( L. More generally. L labour. Hence. though plausible. more generally. e. Consequently. also with variable coefficients does the interest rate increase as the wage rate falls. the entrepreneur would earn extra-profits). In the case with fixed coefficients that inverse relationship can be easily seen. the entrepreneur would lose (-). as distribution changes (e. however. in this regard. The same reasoning can then be applied to the compatibility condition between the labour supply function (L*L*’ in figure 2. However. The point here is to ensure that the supply function of corn-capital be not in contradiction with the demand schedule for labour. If that were the case. the downward-sloped shape of the demand curves would guarantee the equilibrium stability.5. we are not concern with these further. we need the amount of corn-capital employed tend to vary in accordance with its ‘supply function’ as the wage rate. in equilibrium. the interest rate increases. it would be possible for the theory to encounter some difficulties in that.12 what we need is a supply function that allows us to relocate a part of that constant corn-capital from the cloth industry to the corn industry. let Y=F(L.K) be the production function. Moreover. dFL dL  dFK dK  0 so dFL and dFK are of opposite sign and therefore. K corn-capital. the supply function of capital should not be less negatively inclined than is the demand curve at any level of the interest rate i. with Y corn-product. For this to be so. 14 Apart from both an upward-sloped supply and a fixed supply function. let FL and FK indicate the marginal products of labour and corn-capital respectively equal to the wage rate and the interest rate.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics 2. The capital supply function To use the labour employment curve as a basis to derive the demand function for labour. a very important question arises: is the assumption on constancy of capital employed coherent? Or. and with it the interest rate.. the capital supplied varies in such a way that the quantity of labour employed is not increased – even though a huge number of labourers seeking for a job can be assumed to exist. constant returns to scale ensure that the wage rate and the interest rate will not be affected by which person acts as entrepreneur. As we have argued in the text. With alternative methods of production. if the supply function of corn-capital is.g. Then dF  FL dL  FK dK . with decreasing returns. K )  FL L  FK K and by 12 differentiation of both sides: dF  dFL dL  FL dL  dFK dK  FK dK . The capital supply function must then be pre-determinable. The compatibility of the capital supply function with the downward-sloped employment schedule of labour will be ensured. factor payments will exhaust the product in equilibrium.g. the supply schedule of corn-capital must not be in contradiction with that equilibrium process.13 Moreover. one may assert a negative answer to the question posed above. Nevertheless. the wage rate falls).

and since this holds true for either factor. 2. what is the same. So. a tendency toward the full employment of the first factor (labour) will be at work. in general. it only requires that the second factor’s employment be given. it can be argued that the economy will be able to reach the full employment of both factors. Then. 15 We shall see below in Chapter 3 how. And this risk would affect the conception of a unique and stable equilibrium that is so important for an analysis of changes in the permanent conditions of the economies. we could say that.e. in deriving the ‘employment curve’ for a factor of production (first labour as in the example) the reasoning does not require that the second factor (corn-capital) be fully employed.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies To sum up. the explanation of the distribution of income between wages and interests in terms of supply and demand would be at serious risk. if the shapes of the demand functions were upwardssloping). (The analysis we have so far undertaken can be extended to any number of factors of production). what is the same. the shape of the demand function for capital may not be downward sloped and therefore. capital.15 This is why if the shape of those demand functions were not inversely related to its factor price (or.. The downward-sloping shape of such curves do in fact play a key role both in the definition of free competition as assumed by the marginalist approach (i. However. since at whichever level the second factor’s employment (corn-capital) is given. Key role of the downward-sloping shape of factors’ demand curves So far it has been shown the role played by the factor substitution principle to derive a plausible explanation of the downward-sloping factors’ demand functions. when one wants to determine the price (wage) of the first factor (labour) which competition will tend to establish.6.g. actually. if there is some unemployed labourers. that unemployed labourers would bid wages down indefinitely so long as they remain unemployed – a basic definition of free competition. the tendency towards full employment of productive factors in an economy depends crucially on the negative-sloping shape of the demand functions. under free competition. Moreover. they will bid down their wages) and in the determination of a supply-and-demand explanation of distribution of the social product between wages and interests. Thus. for particular ranges of values of the rate of interest. the supply functions of factors must reflect their relative scarcities in accordance with their demand functions. the marginal product curve of a factor (or. since the assumption of constancy of one of the factors when distribution changes is already entailed by the full employment of such a factor. it would no longer be possible to make the assumption e. the assumption that the second factor (corn-capital) is fully employed comes out to be the correct assumption. 24 . the ‘employment’ factor curve) derived under the assumption of full employment of the other factor can then be seen as a demand curve.

Turning the analysis to more realistic economies it must no longer be assumed that capital homogeneous both in the alternative processes of production and in the production of heterogeneous consumption goods.6 above.1-2. The implication is that two fundamental pillars of the marginal approach to explain the determination of distribution in real economies through the forces of supply and demand are: (i) the downward-sloped shape of the demand for labour vis-à-vis the real wage rate. Problems connected with the heterogeneity of capital goods In the simple economy described in Paragraphs 2. therefore.3. a problem of measurement cannot arise since the units in which capital is expressed are the same in both its roles: a role describing the alternative methods of production [group ii) of the data] and the role defining the ‘factor endowment’ [group iii) of the data].7. 25 . Accordingly it is now worth providing the three basic characteristics of those commodities normally treated in the literature as capital goods: (a) They are reproducible goods – unlike land (or to labour) since either of these original factors cannot be reproduced.4. capital consisted of a single good.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics Figures 2. (ii) the downward-sloped shape of the demand for capital vis-à-vis the interest rate. (b) They are utilised for further production both of consumption and of capital goods.-2. which are decisive to be borne in mind to assess the Cambridge controversies in the theory of capital. and. 2. Employment function of labour and capital with upwards-sloping shapes A very important implication can now be drawn from this concise analysis of the basic mechanisms at work underlying the marginal theory.

a contradiction between the roles pointed out above arises. However. say. those kinds of labour can always be dealt with without resorting to value magnitudes. on the one hand. it can be assumed that.B] in the appendix already referred to). cf.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies (c) They wear out within a sufficiently short period of time. economically speaking. p. What we call ‘labour’ is actually composed of different kinds. by treating each as a factor to which there correspond separate supply and demand functions. emphasis added). different kinds of ‘land’ can also be dealt with in the same way as we have seen for the case of different kinds of labour when the wages of those kinds cannot be taken as constant. the capital goods which enter in the argument of the individual production functions defining the methods of production are (and must) be expressed in technical units to define the alternative methods of production. 16 26 . capital as a ‘factor of production’ has been conceived in value terms. etc. emphasis added). as it can be tracked down in the history of marginalist economic analysis. This contradiction lies in the fact that. on the border line between capital in the strict sense and land […] [T]he operation of the laws of capital depends upon the assumption of a constant adjustment of concrete capital goods in an endless repetition of the same process of investment and production. equation [1. 18 Note that the identification of factor’s endowments can be easily satisfied by the factors traditionally grouped as labour and land. on the one hand. Similarly. as we shall see in the remainder of the present chapter. for milk.” Wicksell (1934. Likewise. 118.) In that system. expressed into the production function in technical units (saved-up labour and saved-up land used in previous periods for the production of capital goods necessary for the production of the consumption good. the hierarchy of wages relating to the different kinds of labour does not steeply change.18 Traditionally. 186. Nowadays one might compare the depreciation that. some pages later we find in Wicksell’s Lectures the following distinction: “nearly all so-called fixed capital (houses. A farmer who breeds cattle for meat. p. Moreover. durable machinery. But even if wages of different kinds of labour cannot be taken as constant. The definition and specification of the endowment of capital One can better regard the long-life capital goods as rent goods. capital is.) are. the endowment of ‘capital’ in the economy. there is a serious problem for the theory in that the ‘factors of production’ should be measured – and hence known in advance – before knowing distribution. But this is only of practical importance in capital investments of relatively short duration. which is the aim of the theory. together with the equilibrium of supply and demand functions for original factors (labour and land expressed in technical units) are the supply and demand equilibrium functions of capital. as Wicksell (1934. (…) must pay rent for his pasturage for many years before he can employ the animals in question”. On the other. a lathe would undergo through its utilisation life with the depreciation of a dam or a windmill in order for the former to be replaced through a “constant adjustment in an endless repetition”. 17 Both these roles of capital can be seen in Wicksell’s general equilibrium system developed in his Lectures (See Appendix B to this chapter. If this is so. has been expressed not in technical units but in value terms. puts it: “capital in many cases also advances rent. so one can regard the different kinds of labour as different powers of a standard labour. for long periods of time. in the explanation of supply and demand equilibrium functions for productive factors capital has been specified among the data in traditional general equilibrium analysis as a single magnitude in value terms.17 while on the other.16 When capital consists of heterogeneous goods bearing the properties just listed.

p.that is “capitalistic production” as he called it (Wicksell 1934. however. Jevons. a ‘supply-side’ role of capital. below. Carver.N. 10. and why capital was so treated in the specification of the endowment?20 Since the Cambridge critique is intrinsically connected with the role of capital as a ‘factor of production’ by the marginalist theory. 22 These examples constitute the core of the traditional neoclassical approach which can be reassumed by taking as datum of the equilibrium the ‘factor’ capital in value terms. which allows us to determine distribution and relative prices only when we know the methods of production and the factor endowment of the economy in question. each single capital good as a component of the endowments. a ‘demand-side’ role of capital. are analysed in Chapter 5 below. emphasis added) has pointed out: “if we cannot define the quantity of a factor before knowing distribution and relative prices.19 A very important question arises: What are the consequences of treating capital in value terms when it is used as part of the argument to derive supply and demand equilibrium functions for the productive factors. K. Along this work we shall return to this topic. The question is.” 20 The problems. of capital. Marshall (among others) conceived of capital as a single magnitude in value terms susceptible to changing form without however changing its value. as we did before when capital consisted of a single good? If one goes back in time to see how traditional economists had considered capital among the data of equilibrium. On the one hand. T. which arise out by treating capital in value terms as arguments of the production function. 2. J. and hence to measure.22 As Garegnani (1990. Wicksell.8.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics is not independent of distribution. and. 19 27 . 21 See Appendix B where the supply-role of capital is treated in some more detail by considering the general equilibrium system which Wicksell put forth to analyse production in economies with capital goods. E. which is associated with the identification of the capital endowment. p. BöhmBawerk. however. where Joan Robinson’s participation in the Cambridge controversies is analysed. the given ‘capital’ as an endowment independently of distribution in case it does not consist of a single good? Should we regard. on the other.B. 166). Clark. we shall see. how to specify. thus a contradiction between those roles arises. equation (4. in a world of heterogeneous capital goods. or a supply-role. it might be useful to approach the difficulties associated with the question raised above from two angles. Supply-side role of capital21 Some difficulties associated with the measurement of capital may immediately be envisaged from a ‘supply-aspect’. we shall clearly be unable to use the theory. For the treatment of capital by Pigou see Hennings (1990. that is why these marginalist authors conceived of capital in value terms to define the equilibrium of supply and demand – and not as a vector of the heterogeneous capital goods which we may come across in an economy as is the case of Walras’s treatment of capital. or A. which is associated with the conditions required to derive demand functions of capital. p. that economists such as S.B) of Wicksell’s system. Appendix B. 144). Cf.

[A]ll these [different capital goods] […] have only one quality in common. money. 144-145. regarded ‘capital’ (the endowment) as a single factor in value terms susceptible to adopting any physical form. still the problem would evidently be there for all other capital goods. 25 If the value of a particular capital good were made constant by using the capital good in question as its own unit value. at that rate of interest.23 On the same page. below. too. 28 . Marshall’s ideas. […] This. of course. 443. or also in terms of a capital good)25 is taken as datum in the determination of output. capital. this “single sum” is taken as given in the group (iii) of the data. when he deals with “the concept of Capital”. In fact.3 (Appendix B). emphasis added). is the commonly accepted sense of the term. pp. a certain amount of the medium of exchange. a quantity measured in a single unit (a consumption good for instance. Cf.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies For instance we have the case of Marshall who in his time-honoured Principles of Economics stated: the chief trouble [regarding ‘the general rate of interest’] comes from the fact that the income derived from capital […] invested in particular things […] is properly a quasi-rent and can be regarded as interest only on the assumption that the capital value of the investment has remained unaltered. however.24 Both authors – along with other representative traditional economists mentioned above – conceived of capital the endowment as a substance in value terms. were not isolated. also Paragraph B. he says: Capital also includes the raw materials […] and other commodities which must be saved-up. prices and distribution in terms of supply 23 Marshall (1959. p. is equal to the aggregate stock forthcoming there at that rate. The adjective which I chose to use in the text refers to the several reprints of Marshall’s book. Thus. 24 Wicksell (1934. Wicksell. the following conclusion is raised by the renowned English economist: thus then interest […] tends towards an equilibrium level such that the aggregate demand for capital in that market. namely that they represent certain quantities of exchangeable value so that […] they may be regarded as a single sum of value. emphasis added).

in other words. ‘capital’ is a kind of ‘stuff’ which can change form without changing its total quantity in value.2). Walras (1954). imagine for a moment a non-equilibrium situation. see also Garegnani (1960. a change in a pattern of consumption or a change in the quantity of people seeking for jobs due to a process of immigration): in that equilibrium process the physical composition of that substance would be endogenously determined. capital was specified as a single factor in value terms which could adopt the physical form according to the equilibrium conditions of the economy. The determination of the physical form of the capital stock was envisaged as a process through which a uniform rate of profit (interest) on the supply prices of the several capital goods could be ascertained. 26 What was the justification of such a measure in marginal theory? Although this measurement does not satisfy the requirement of independence from changes in prices and distribution. the uniform 26 The outstanding exception among traditional authors is Walras. in the production of new capital goods. For example. Thus the new capital goods are produced by employing the same economic resources of the economy which produced the hitherto capital goods and. In a world of heterogeneous capital goods. Therefore the analysis is thought to apply symmetrically to factors of production. which in turn were thought to be sufficiently persistent so as to regard the theoretical variables determined by the theory as gravitational centres of actual variables.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics and demand functions.g.27 In this way. We shall deal with this author later (Chapter 6) and see the problem arising out from such a conception of capital (Paragraph 6. before and after some change in the conditions (e. who conceived of ‘capital’ as a physically heterogeneous capital goods vector existent in the economy. 2008). given a sufficient long period of time for adjustment. which must go in accordance with the factors’ substitution principle. in particular to labour and ‘capital’ (assuming free land) such that both labour and ‘capital’ can be seen as receiving their marginal product. In fact. which may had been previously used to produce the previous capital goods no longer profitable to the new conditions of production (and consumption). and it is re-interpreted as evidence of the two substitution mechanisms we (cont. labour. establishing an inverse relation between the relative quantities employed of a productive factor and its rate of remuneration.) 29 . 27 The clue to such an explanation provided by marginalist authors (that is the justification of the principle of factor substitution) can be traced in the attempt at a generalisation of the principle of intensive rent applied by Ricardo however to ‘land’ and ‘labour and capital’ to explain profits: the generalisation was then applied to any number of factors. would have the same total cost of production and hence the same total value as the no longer produced capital goods. a change in the ‘form’ of the single factor ‘capital’ was thought to be brought about by the employment of factors. if. The idea underlying this process is the conception that the several capital goods can embody different amounts of the substance ‘capital’. the theory would somehow determine distribution according to the factors substitution principle. Cf. namely – labour and ‘capital’. and ‘capital’. the specific physical capital goods to be produced corresponding to the new conditions of the economy could be produced by the most profitable use of the available resources (techniques). it was traditionally believed that owing to group i) and ii) of the data. part II. therefore. we can envisage an ‘old’ and a ‘new’ situation. land.

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

rate of profit (interest) on the supply prices of the several capital goods is ascertained. This process was thought to work somehow through the sufficiently persistency of the data of equilibrium. As noted above, capital in value terms not only ensures the uniformity of the rates of profits (interest) on the supply prices of the capital goods, but also it allows for the substitutability of the factors of production. 28 In the wake of any change on the permanent conditions of the economy (e.g., new quantities available of some resource), the economy would reach a new position of equilibrium by the supply and demand forces which work through the data of the theory. The new position to be reached implies a process where production, demands, prices, quantities are actually in disequilibrium; however, after certain period of time the supply and demand forces would put the economy again in balance. For such a result, it is key to assume that the capital ‘substance’ be expressed in a single value magnitude susceptible to change form but not changing in quantity in order to let the supply and demand functions work throughout. Yet that ‘quantity’ in value, like any other value magnitude, depends on the variable which theory attempts to determine – the rate of interest, prices, etc.; thus a problem, which we may identify as the ‘supply-aspect’ problem, regarding the measurement of capital as the data consistent with the logic of the theory arises.

have considered in the text. Accordingly, were it not for conceiving that ‘capital’ in constant value terms susceptible to adopting any physical form as the economy conditions change, an explanation of distribution relying on ‘factor’ substitution – from which in turn we can derive the equilibrium supply-and-demand functions – would not be possible. Were it otherwise, this would entail the impossibility to derive demand functions for factors from employment factor’s curves, as we have already argued in our simple economy’s example (cf. Paragraph 2.2-2.4 above.) This, as we argue in the text, also explains why the physical composition of the stock of capital is left to be determined endogenously, while the value of that stock does not change. For an analysis of Ricardo and the classical political economy, see in particular Garegnani (1960, part I; 1984a). 28 See Appendix B where the issue of the uniformity of the rate of interest of capitals’ supply prices is discussed within the framework of Wicksell’s general equilibrium system.

30

Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics

2.9. Persistence character of the determining circumstances: The traditional method of ‘normal positions’29

We have seen above the meaning of the data taken as given by the theory to determine the dependent variables output, distribution and prices simultaneously; how the forces of supply and demand work in the factors’ markets have been explained in the first part: those forces are supposed to work in real economies, since they are thought to be founded on the basic principle of factor substitution. We have finally seen that the factor substitution can be brought into analysis both through the alternative techniques of production (direct mechanism of substitution) and through consumers’ tastes and preferences (indirect mechanism of substitution). The resulting equilibrium (let us assume it to be unique for the sake of exposition) of supply-and-demand (i.e., the level of the dependent variables, at which it is arrived through the working of those forces) is ensured to be stable – provided the ‘right’ shape of the demand curves for factors. But to apply this approach to real-world economies the data through which one arrives at equilibrium must account for some basic features: the data must be thought of as having a persistence nature through time, in that they are supposed to be those variables on which the working of the mechanisms of factor substitution rest. In actual markets, however, prices, quantities, the level of the wage rate are all magnitudes that depend on circumstances that are at work due to many causes that can change on a daily (or hourly) basis. A specialised journalist, say, in agriculture products’ prices, does every morning try to provide his audience (in general perhaps investors) with the level of those prices due to the variables he considers the most relevant at that particular moment in that particular place. They run from the drought of agriculture fields in American prairies to the level of the oil price, on which the production costs of corn depend and which in turn is conditioned by the political situation in some oil exporter’s countries. 30 But these particular considerations this

29

The term ‘normal positions’ has been adopted by Garegnani (1990) in order to refer both to the marginal long period equilibrium of supply and demand, which defines a ‘long period position’ of supply and demand, and to the ‘normal’ or ‘natural’ positions of the classical economists (Smith; Ricardo), which are not derived from supply and demand forces, as in marginal theory. The latter school of economic analysis, which has been revived by Sraffa (1960) and the successive works undertaken by the surplus approach, has grounded the explanation of normal or natural position of the economy by relying on the key notions of a subsistence wage (determined by social and political circumstances) and the existence of a surplus, which arises after the necessaries of the workers and the replacement of means of production are deduced from the social product. (See Garegnani 1984a) Though in this study we shall deal with the marginal approach, it is worth reminding that, nevertheless both theories have relied on entirely different forces to explain distribution of income, they have shared the method of normal positions – although the marginal theory has since recent decades shifted to a different notion of equilibrium. We shall argue that this shift has been one of the consequences of the debates on capital theory, which is analysed in Chapter 6. 30 Cf. the following argument raised by Marshall (1920, p. 291): “The actual value at any time, the market value as it is often called, is often more influenced by passing events, and by causes whose action is fitful and short lived, than by those which work persistently. But in long periods these (cont.) 31

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

journalist is concerned with are not, at a theoretical level, thought to be persistent circumstances on which those prices ultimately depend. So, in general, supply-anddemand equilibrium prices (and quantities, and distribution), as envisaged by the theory, are not daily prices, but rather average, ‘normal’, ‘natural’ ones. The actual variables, or better still, the average of actual variables, if the theory were correct, would follow the trend described by those theoretical equilibrium values. But to arrive at such averages does take time: it is thanks to the persistence of the conditions of economic phenomena that ensures the gravitation process, where divergences between the values of the theoretical and actual variables are cancelled out over time, so long as repetition of transactions in markets is guaranteed;31 (following the analogy, temporary droughts in American prairies are not a persistent causes in the determination of corn prices). Thus an explanation of distribution of the social product in real economies in terms of equilibrium of supply and demand is further grounded on the persistent nature of the economic forces meant to be at work. From the foregoing discussion on the persistent nature of economic conditions, a very important question can be posed: Does persistency imply that the data of the theory must not change? Traditional marginalist authors did obviously not deny that the data might change. However, thanks to the persistence of the causes of economic phenomena, they envisaged that their changes were generally either pretty slow enough so as to be negligible relative to the presumable speed of adaptation of demand and supply for factors and products, or, if they were rapid, changes were assumed to be once-and-forall changes, whose effects could be dealt with through comparative statics, that is by assuming a tendency to the corresponding new equilibrium in the new situation. In other words, the same persistent nature of the causes allows the theory to deal with changes by relying on the method of comparison of long period equilibrium positions.32 Let the present author explain it a little further, if only briefly. Imagine an economy where a process of immigration is taking place. This process will certainly bring about a change in the quantity of labour supplied. Before the newcomers got into the labour market, there was an equilibrium in such market, say (w*, L*), to which there also corresponded a determined set of equilibrium prices and

fitful and irregular causes in large measure efface one another’s influence so that in the long run persistent causes dominate value completely.” 31 The variables that the theory determines cannot be the actual magnitudes observable moment by moment in the market, which are influenced by many accidental factors, since those factors could not be known in advance. See, Garegnani (1990, p. 46). 32 The usefulness of the method of long period equilibrium lies in the fact that the theory provides a logic framework to analyse changes through an analysis of comparison of different ‘normal positions’ by means of the mechanisms pointed out in the text. It is however very important to bear in mind that for this method to be used the capital must be conceived in value terms, thus an illegitimacy is entailed. Yet we will return along this work to these uses of the theory, in particular, when we come to approach the ‘new’ versions of the general equilibrium theory in its intertemporal or temporary versions.

32

there of course is a ‘quantity of capital’ in value terms which is slowly changing.34 As the immigration process takes place. more ‘labour-intensive’ processes of production are adopted – given the technical knowledge of our imaginary economy – since they are the most profitable.. 34 A logical necessity of the traditional method – which attempts to explain actual magnitudes as oscillating around centres of gravitation – is that these gravitational centres must not be affected by the ‘day-by-day’ disequilibrium production and consumption activities undertaken by the economic agents. as those forces are pushing the system towards equilibrium. cf.7. In between. For a detailed methodological analysis. How then is the equilibrium reached? As the population at the age of getting a job changes. However.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics outputs. they take place by thinking at the same time of a tendency to a future equilibrium position. in particular equation (3. at a general level. 6). Once the uniform rate of return on the supply prices of the capital goods is reached. the composition of that capital will also be changing. that is. imagine that workers’ wages remain at w*. §1. as expected by the theory. see Schlicht (1985. transactions. However. and the quantity employed rises. p. the period required by a ‘new’ arrived person (or family or group) to get a job was short due to specific circumstances (e. In that very process the theory assumed that quantity to remain constant. As time goes by. As the immigration sets off. That period of time must be a long one. s/he or they was/were given a job in advance before arriving in situ) and so on. the composition is endogenously adjusted. the period of time must be long. See Garegnani (1990. the supply of labour will change (some of the data of the theory change). it was thought that in the definition of long period equilibrium of supply and demand. 33 The theory can give an appropriate picture of the process if the data are stable in comparison to the speed of adjustment. In that process. other ones go elsewhere. when adjustments are taking place.33 Accordingly. 51). even though. the allocation of savings among capital goods could not appreciably influence the equilibrium position.2. both new and old labourers are modifying the labour supply: however. 33 . individually perhaps. population is (re)adapting itself to the new conditions in the labour market: some new arrived people go to some firms to seek jobs.g. that is – as the physical composition of that ‘quantity’ is being adapted to the new conditions. we would have reached a new long period equilibrium. prices’ changes could be ignored. Appendix B below. which turn out to be almost imperceptible) but. as the more ‘labour intensive’ methods of production come to be adopted in that very process of adjustment (a time consuming process which runs from a disequilibrium situation towards an equilibrium one). When the composition of ‘capital’ is then adjusted to the new conditions of the economy. thus ensuring the persistence of the equilibrium position itself. contracts. however. since the immigration sets off but before reaching the new equilibrium. take place at non-equilibrium values (with prices changing very gradually. it is necessary time for that supply to arrive at a determined level compatible with demand and free competition. Traditionally. p. in order to let the physical composition of capital be adjusted. The argument for such a thesis is the assumption that.B) regarding Wicksell’s treatment of savings. But for this we have to assume that the quantity of ‘capital’ employed remains constant (in fact in full employment). the wage rate falls. population at the age of getting a job is changing. But. etc.

This sufficient substitutability – and 35 With the outstanding exception of Walras as noted earlier.3 and part II below). 34 . seek the most profitable capital goods wherein to invest their capital. 2. First. i. in general. This demand side concerns two related aspects. investors will. can therefore be seen as an equilibrating process in that the several goods will pay off a uniform return over costs. the demand function for capital is inversely related vis-à-vis its remuneration rate – the rate of interest.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Under conditions of free competition in factors’ markets.10. In the wake of some change in the demand conditions. and due to the properties of the capital goods. production of the specific capital goods (investment) depends upon the demand conditions of the economy. in turn allowed by free competition among investors throughout the different industries and capital goods. This process. The substitution of factors when capital consists of a family of heterogeneous capital goods entails. In fact. 35 if a process involving disequilibrium adjustments is taken into account: in general those adjustments need long periods of time. The resulting equilibrium capital goods’ prices will tend towards the level of their costs of production (supply prices). we have said that once the composition of the stock of capital is endogenously determined a new long period equilibrium will be reached. but a change in the kinds of the capital goods due to the adoption of new methods of production.e. and on b) outputs. capital could be treated as any other factor of production to ensure that the substitution mechanism work in the ‘right’ direction. we may understand this basic reason why in traditional writers the endowments of the several capital goods cannot be taken as part of the data of the equilibrium. independently of the particular physical capital good concerned. This is the demand-role or demand-aspect of capital. whether an economy would produce particular kinds of capital goods will depend on a) the methods of production available in the economy. In other words. which in general are commodities with a short utilisation period. whilst others to buy other capital goods. thus the rates of return over capital goods’ costs will tend to be uniform: the differences in profitability among the several capital goods will cause some investors to sell their capital. in the wake of some change in demand conditions. which are very important to bear in mind when we shall come down to the results of the controversies in the theory of capital (ch. Thus provided. that it is not a change in the proportions of ‘factors’ that allows theory to arrive at a new equilibrium. to allow the demand for ‘capital’ to be a decreasing function relative to its ‘price’ – the interest rate. The conception of capital of variable form in turn makes it possible to assume the existence of a sufficiently substitutability between capital and labour (or land). marginal theory still needs a treatment of ‘capital’ as analogous to that of any ‘factor of production’ in the direct and indirect factor substitution mechanisms. Demand-side role of capital: Conclusion In the example about immigration above. by satisfying the condition of uniformity of the rates of profits on the capitals’ supply prices.

this incompatibility between both compositions is resolved itself precisely in the investment savings market: what is underlying the capital supply curve is that the physical composition of the stock appropriate to the equilibrium will be assumed by the (cont. and can take the best form adapted to demand conditions: changes in demand due to changes in consumer patterns. actually. in Hicks’ 1932 The Theory of Wages.. when not only the quantity. at each moment of time. The importance of this sufficient substitutability between the factors of production can be traced e. This is the idea of a ‘capital’ as a single factor susceptible to changing form.. different. capital is ‘crystallised’ in specific capital goods. 20-21. This linking is what allows us to consider the investment demand as a manifestation of the demand for capital. Thus. etc. in the long period. for single capital goods: In the short period it is reasonable to expect that the demand for labour will be very inelastic. pp. Thus ‘capital’ – the value substance – embodied in the old capital goods becomes free to be re-invested in the new. since the possibility of adjusting the organisation of industry to a changed level of wages is relatively small.36 In fact that variability in the form of capital cannot be conceived as an instantaneous process. where the author argues that such a substitutability would be impossible to conceive. On the other hand the physical composition of the capital supply function cannot be but what is commonly known as the stock of the economy. It is very doubtful if this conception can be given any precise meaning which is capable of useful application. That is to say: the supply and demand functions for ‘capital’ as a fund were envisaged by traditional authors to analyse the forces operating in the investment and savings market – where capital appears as a flow. little advantage seems to be gained from the attempt which is sometimes made to define a ‘short period marginal product’ – the additional production due to a small increase in the quantity of labour. above) can manifest itself in investment in the new plants: the flow (investment) of free capital would be met with the labour ‘freed’ by the closure of the old capital goods’ plants.6. In general. of the co-operating capital is supposed unchanged. both compositions will differ. technical progress.) Since the whole conception of marginal productivity depends upon the variability of industrial methods.g.37 Then 36 37 Hicks (1932.1-2. As it was shown in the text the physical composition of the capital demand function will be determined by the production methods and outputs and hence are unknowns of the problem.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics here we come to the second related aspect referred to above – is compatible with a conception of a ‘quantity of capital’ as a single magnitude in value terms but incompatible with a specification of the physically heterogeneous capital goods among the data of the theory. emphasis added). which are physically different from the depreciating ones.) 35 . To understand this it is sufficient to think about the traditional supply and demand functions for capital the single factor. ones. But that change in ‘form’ is gradual: it occurs once the existing capital goods are depreciated and the reinvestment of these allowances into new capital goods takes place. but also the form. (. the ‘marginal product of capital’ (see the basic scheme in Paragraphs 2.

we shall introduce some main results drawn on the capital controversies in order to see whether the marginal theory can determine distribution on the grounds of an inverse demand curve for factors in economies with heterogeneous capital goods. in order to analyse the “substitution of factors in a purer form”. thus being coherent with the traditional method of a uniform rate of profits on the capitals’ supply prices (see the case of Wicksell in Appendix B). Therefore. We shall see the link between the savings-investment market with the fund capital market in more detail later. What is key to keep in mind is the role of the downward-sloping demand curve for capital as it was traditionally conceived – a single magnitude in value terms – existing stock over a period of time when part of the capital goods has to be replaced by reinvesting the depreciating allowances into new capital goods. however. the quantity of capital per worker in the economy had to be inversely related to the rate of interest.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies if this connection is admitted. it is capital conceived in value terms that allows the theory to leave the physical composition of the stock be endogenously determined. since it is that the datum from which we should depart in order to arrive at the equilibrium values of output. the circularity in the reasoning is not avoided. 38 This is the most evident contradiction which arises from the measurement of capital in value terms for the marginal theory. 36 . To conclude. it would seem that the traditional authors’ focus was on analysing the forces of the fund concept of capital rather than the flow. since in general the arbitrary composition of capital in his treatment will prevent the rates of return of the several capital goods from being uniform on the supply prices of that goods. See Chapter 6. In the next chapter. and. 395-396) has argued. Later in this work we shall also present the other treatment of capital namely that of Walras who. As Garegnani (2000. Chapter 6. and hence the very distribution. then the way to the explanation of distribution in terms of supply and demand through the mechanisms of substitution of relative ‘scarce’ factors would be achieved. even sharing with his contemporary peers the traditional method. This ‘belief’ regards the demand-aspect of the problem: that aspect resolves itself in the necessity for the theory to have downward-sloped demand curves for the productive factors. considered the capital endowment in the data of equilibrium as a vector of heterogeneous capital goods of the economy: this led Walras to contradict his own general equilibrium system.38 it was once believed by the founders of the neoclassical approach that. pp. the interest rate could then be seen as the price bringing into equality the demand for and supply of ‘capital’ the flow. This allows the theory to apply the method of comparison of normal positions for an analysis of changes. If downward-sloped demand curves for factors of production can be rigorously built on. in order to ascertain an equilibrium position as a result of applying supply and demand general equilibrium functions. while leaving the endowments of the several capital goods to be endogenously determined by the tendency toward a uniform rate of return on the supply prices of the capital goods through factor substitution. The introduction of a ‘quantity of capital’ (a single magnitude) into the general equilibrium supply and demand functions makes it possible to arrive at long period equilibrium positions. by means of changes in demand associated with groups i) and ii) of the data of the theory. prices. provided the ‘right’ shape of the demand curve for capital. Notwithstanding the supply-side problem regarding the illegitimacy of conceiving of capital in value terms to determine distribution.

The role of the demand curve’s slope is decisive for uniqueness and stability of the equilibrium and hence for the plausibility of the traditional supply-and-demand theory of distribution. 37 .3 and 2. (Cf.Chap te r 2 – Und e rstanding th e founda tions o f neocla ssica l econom ics and with it the supply and demand aspects of the problem of capital.4 above). Figures 2.

.

the discovery of the phenomena of ‘reswitching’ and ‘reverse capital deepening’ in the Cambridge controversies has put that neoclassical belief at stake. This is the demand side of the problem. e. the bulk of well-known marginalist authors have conceived of capital as a single magnitude in value terms susceptible to adopting the physical form of the several capital goods – hence their quantities as well – determined by the equilibrium condition of a uniform rate of profit on the supply prices of the capital goods. It was believed that owing to changes in demand the factor substitution principle could be also taken for granted even when factors were measured in value terms so as to derive a monotonic relation between changes in distribution and factors’ intensities.1.e. an inverse relation between the rate of interest and the demand for ‘capital’ could in general be postulated. As we shall see in this chapter. particularly after Sraffa’s (1960) famous book. by considering changes in demand due to changes in the production methods and outputs – associated with groups i) and ii) of the data –. in a labour and corn (both as consumption and as capital-good) economy. However. This principle can univocally establish. can we derive the same kind of relations when one of these factors is ‘capital’? As sketched above. i.g. to derive a negative relation between ‘capital’ and the rate of interest. ‘Reswitching’ and ‘reverse capital deepening’ started to reach prominence in the Cambridge controversies in the 1960s. Nevertheless the illegitimacy of the notion of capital in value (the supply-side aspect). Some premises In the previous chapter it has been introduced the key role of the downward-sloping demand curves for productive factors that can be derived according to the factors’ substitution principle. which was followed by a series of contributions that started to spread these discoveries .Chapter 3 ‘Reswitching’ and ‘Reverse Capital Deepening’: Introduction to the heart of the controversies and their implications 3. an inverse relationship between factors’ intensities and factors’ rates of remuneration. the founders of the neoclassical approach somehow believed that.

But it was also believed that before a further rise in r the old technique could not be adopted again because that technique would have been categorised by the theory as a ‘more capital’ intensive one. it has been preferred in this part of the study to bring to the foreground the meaning and implications of ‘reswitching’ and ‘reverse capital deepening’. 1 Rather than following a chronological description of the episodes around the Cambridge debates. Bruno et. minimise the cost of production of the final good. it could be the case that at a higher r the technique that had been used at lower values of r could be chosen again – why? Because this technique might also minimise costs at the new situation of prices and distribution. which could have never again been chosen at a higher rate of interest. 3 Or a given bundle of final commodities. however. to derive a univocal relation between capital intensity and changes in distribution? Neoclassical traditional theory certainly took on board such question and traditionally provided its solution by putting forth the belief that. Since capital goods are reproducible and also require further capital goods in their production. Garegnani (1966). in particular in Robinson (1953) and Champernowne (1953). Samuelson (1966a).g. that the phenomena of ‘reswitching’ and ‘reverse capital deepening’ had been noticed in the literature in the controversies before 1960. more importantly. 109 where the Cambridge author regards the ‘curiosum’ as an “intricate piece of analysis which is not of great importance. (1956) p. This theoretical possibility has been known in the debates as ‘reswitching’ of techniques. Among the several contributions to the ‘reswitching’ debate we should cite: Sraffa (1960) and the articles published in the 1966 Symposium: Pasinetti (1966a). this chapter provides the basic ground through which further exchanges between the two sides evolved. it would See Appendix C. at a low rate of interest (r) and from a spectrum of alternative techniques (say A and B both using heterogeneous capital goods). The aim of this chapter is therefore to introduce the reader to the meaning of such phenomena and. however these authors gave these phenomena little prominence at the time (Cf: Robinson (1953) p. Note. That Symposium was the result of the intervention from the neoclassical side of Levhari (1965). the first work where the terminology of ‘switching’ commenced to be used was – to the present author’s knowledge – McManus (1963).”) Nevertheless we shall deal with these authors in Chapter 5. and also Robinson. 1 40 . it was also believed that a rise in r might lead producers to change (switch) to a ‘less capital’ intensive technique (B) – so long as this technique minimised costs – leading the economy to account for a lower quantity of ‘capital’ per worker. Thus.2 Is it possible. despite that these phenomena started being acknowledged only in a later stage of the first phase of the controversies. because these results touch the hard core premises of the marginal theory.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies over the whole decade. Levhari and Samuelson (1966). The content of the present chapter is mainly drawn from Samuelson (1962) and Garegnani (1970a). We deal with the historical aspects of the Symposium in the next chapter. (1966). their implications for the marginal theory. of a single final consumption good 3 – would be led to use a ‘more capital intensive’ technique (say A) so that this technique would. Finally note that. In the Cambridge controversies it was shown that that possibility cannot be excluded a priori. Morishima (1966). Thus. in competitive long-run equilibrium. Further. 2 See Chapter 4. al. 106 where ‘reswitching’ is treated as a “curiosum”. producers – e. after Sraffa (1960). in a heterogeneous capital goods economy. below.

which is key to the marginal theory. hard core premises of the approach are at stake. and hence of the factor substitution principle in general. then there would be no theoretic support to think of an inverse relation between factors’ demands and their rates of remuneration.5 the value of capital per worker in the economy may rise with a rise in r as the economy ‘re-switches’ from one technique to the one that had been previously used at lower values of r. then there will not be any monotonic relation between changes in distribution and ‘capital’ intensities. for a change in distribution entails a change in the way in which – if such way existed – those techniques were ordered according to their relative factor intensities. the possibility that the demand for capital be in a direct rather than in an inverse relationship to the rate of interest.Chap te r 3 – ‘ R eswitching’ and ‘Reverse Cap ita l D e epen ing’ seem. this notion is central for the whole approach (cf. Chapter 2). this chapter. Garegnani (1970a. UK. side of the debates. below. The idea of an ordering of techniques according to ‘more’ or ‘less’ capital intensity disappears with changes in distribution. a technique a priori considered as ‘more capital’ intensive is chosen for a range of values of r and also is chosen for other range of values of r. due to changes in distribution it might be the case that the value of capital per worker changes in the opposite direction as traditionally conceived. 4 41 . ch. lies in the fact that even if the illegitimacy of the supply side is put aside. then also used by Hicks (1965) and fully studied and criticised by Garegnani (1970a). 4 If this is so. In fact. see Sraffa (1960). serious difficulties would arise in order to derive downward-sloping demand curves by relying in demand changes. with ‘reswitching’. even without ‘reswitching’. as we shall see below in more detail. System of production involves two methods of production: the one for the consumption good and the other for the capital good. In what follows the demonstration of the results of ‘reswitching’ and ‘reverse capital deepening’ is introduced by using very simple graphical analysis as used by some of the participants in the debates. because this demand aspect is undermined by ‘reswitching’. 417-421). For more general models. Thus.6). 5 See Paragraph 3. 7 For the present exposition we shall follow Samuelson (1962) and Garegnani. revealed by the Cambridge. ‘Reverse capital deepening’ indicates. and by taking into account the very premises of the theory. But. 6 See Appendixes C and D. Pasinetti (2003. But also. for reasons already explained. ‘reverse capital deepening’ may arise. while another technique is chosen in between.4. Why? Thanks to ‘reswitching’.6 The novelty of these results. if that were the case. Hence it turns out to be impossible to derive any measure of capital in value terms in a univocal direction with changes in distribution.7 Or that technique may be ‘more capital’ intensive for some ranges of values of r and ‘more labour (or other factor)’ intensive for other ranges of r. in economies with production involving heterogeneous capital goods. It is assumed that for each system the capital good is produced by itself and labour which is a very simple case chosen by Samuelson (1962) in the controversies. (1970a). pp. but.

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies 3. • • Let us first consider production with only one system of production. (ca). As usual. w.8 The capital good enters the production of both commodities. The consumption good has been considered as our numéraire. Capital is circulating capital. The reader may be referred to Sraffa (1960). 408. Each system consists of two ‘methods of production’. The wage curve In order to show the results of ‘reswitching’ and ‘reverse capital deepening’ it is assumed an economy with the following characteristics. w is the wage rate in terms of the integrated industry of the consumption good and r is the interest rate in the same terms. hence the proportion of the capital good entering the production of the capital good itself must be lower than unity (i. cc<1). apart from homogenous labour that is used by the different systems. 1. For an analysis of the historical aspects of fixed capital and its distinction from circulating capital see Varri (1998). No scarcity of natural resources. 8 Lower than one because we assume capital goods to be circulating.) We shall not deal with fixed capital since our aim is basically to show these results of the debates as clear as possible. and the other method to produce the capital good by means of fixed quantities of labour (lc) and of the capital good itself (cc). 42 . that is. and then both w and pc will be univocally determined for any given value of r. A. Some positive amount of labour is required directly or indirectly to produce both commodities. by means of fixed quantities of labour (la) and of the capital good.e. 9 Other systems of production involve.9 Positing the equilibrium conditions of a uniform rate of interest and a uniform wage we shall be able to write the following two price equations: pa=1= law+ca pc (1+r) pc=lcw+cc pc (1+r) (3. Constant returns to scale. hence pa=1. The wages are paid at the end of each productive cycle. There exists a surplus obtainable over the pure replacement of machines: that is to say. r). the economy is viable. There are two equations and three unknowns (pc . p. The degree of freedom will allow us to take one of the distributive variable as independent (r from now on). Economies with heterogeneous capital goods. one method to produce the consumption good. n. (See Garegnani 1970a. different capital goods for the production of the consumption good A. • • • • • • Yearly cycles of production of the consumption good and of the capital good.1) where pa and pc are the consumption good’s and the capital good’s prices respectively.2. C.

proving by contradiction we may consider pc be negative. This imaginary industry is such that the proportion. K. equation (3. 12 Likewise. where D is the determinant of the productive coefficient . there corresponds a unique value of w. while the other elements remain constant. which is equivalent to w  matrix. So pc >0 for every 0  r  R .2’) and (3.3) we can obtain pc  1 cc  1 ). then 1    1 hence D  la cc and hence finally lc ca  0 . 11 Cf. Then if D  0 hence pc > 0 for r  0 . we obtain (3. In fact.Chap te r 3 – ‘ R eswitching’ and ‘Reverse Cap ita l D e epen ing’ From this system it can be obtained the following relation between the wage and the interest rate corresponding to the current system of production: w 1  cc (1  r ) [la  (lc ca  la cc )(1  r )] (3. This determinant is central to the entire analysis. 16.11 A rise in r (the rate of interest).1) for (3. in this case denominator of (3. then pc > 0 too.12 To compute the value of capital per worker.2). hence (3. And if D  0 . one of the distributive variables is negatively related to the other. for the graphical representation of the relation between w and r which allow us to easily grasp many of the results of ‘reverse capital deepening’ and ‘reswitching’. Moreover. it may be useful to introduce the integrated industry of the consumption good. and for any value of r.2) is a function. this chapter. this industry must re-integrate the quantity of the capital good necessary for producing the consumption-good directly (ca). of the capital industry relative to the consumption good industry is the one which ensures the replacement of the capital goods consumed in the integrated industry. In fact from lc la  D (1  r ) (3. that is D  1  cc (1  r ) la  D(1  r ) lc la cc ca (3. 43 . in particular.2’).4) should be negative (since lc is positive throughout).2) fall.2)10 From equation (3.9) n. but this is impossible since lc and ca are both positive. Then r 1 cc la D  1 but we know that r  R so R   la D la D  1 .4) .1) we obtain pc  lc w 1  cc (1  r ) (3. it can be verified that the price of the capital good (pc) is positive for every 0  r  R (where R is the maximum rate of profits obtainable when w=0 then R  (3. Accordingly. Then by substituting pc on the Right Hand Side (RHS) of the first equation of (3. will make the numerator of (3. whereas the denominator will rise: consequently w (the wage rate) will fall. below.3). and the quantity which is 10 From the second equation in (3.3).2).

can be measured as follows: q 1 la  lc K (3.7).13 q. In formal terms: K=ca+ccK Then.8). by replacing (3.5) in (3.6).7) But. whenever r=0.2’). Hence: w(r  0)  W  1  cc q la (1  cc )  ca lc (3. Therefore.6) Then.8)14 If r is zero then the segment 0W measuring the maximum wage rate (see Figure 3. (3. Since w  14 1  cc (1  r ) [la  (lc ca  la cc )(1  r )] with a zero r becomes equation (3. and equation (3. 44 . the value of capital per worker can be measured in the following way: 13 Consumption-good production is normalised to one so q is one unit of net A divided by the quantity of labour employed both directly and indirectly in the production of such a unit of net physical output.2) or to (3. the net physical product (of the integrated industry) per worker.8) is equal to equation (3.7) will be equal to (3. we obtain: q 1  c la   a  1  cc   lc   1  cc la (1  cc )  ca lc (3.5) Equation (3.1 below) will also measure the net physical product of the integrated industry per labourer.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies necessary for producing itself (cc) – that is the proportion of the capital good consumed in the production of the proportion of the capital good necessary for producing the consumption good. where W is the maximum wage. K ca 1  cc (3. If the wage rate is w1 and the rate of interest r1. Recourse to the integrated industry is necessary if the net product (that is the product to be divided between wages and interest) is to consist of a physical quantity of the consumption good.5) provides the quantity of capital of the integrated industry of the consumption good.

(See Figure 3.1. and this is the result at which we wanted to arrive.1. Garegnani 1970a. tan()=tan (w1PW) = interest per w1 P 0 r1 worker / rate of interest = value of capital per worker. e. The wage-curve: 0W is the net physical product per worker. Figure 3.2’): 15 dw( r ) dr 17  cc [la  D (1  r )]  [1  cc (1  r )]D [la  D (1  r )] 2 = cc la  D [la  D (1  r )] 2  la ca [la  D (1  r )]2 (3.2) is shown by its graphic representation. Some formal (and economical) considerations are in order: if. ca) are both positive. and is drawn in Figure 3. the value of the interest per worker (w1W).g. while in the vertical axis. In fact. r=0 then w=W. And this is so because of the assumptions here assumed for the economy. the way to obtain it is by measuring the trigonometric tangent of the angle  –of the triangle formed by points P. and 0r1 on the other hand indicates the level of the interest rate. correspondingly. as r rises from zero. tan() is the value of capital per worker (k) when the wage rate (w) is 0w1 Mathematically. and then. Samuelson (1962) named it “factor-price frontier”.9) which is negative provided that the production coefficients (la .1). as we have already seen. the trigonometric tangent of angle  will give us: wW wW tan( )  tan( w1 PW )  1  1 (P is point w1r1). then the value of capital k can be obtained as the ratio between:15 k = Interests per worker / Rate of interest The system of production gives the relationship between r and w. Then. 16 The inverse relationship between w and r in (3.Chap te r 3 – ‘ R eswitching’ and ‘Reverse Cap ita l D e epen ing’ Since w1W indicates the quantity of consumption good received as profits for each worker employed in the integrated industry. 17 whose graphic can be obtained from equation (3.2). w falls reaching zero when r reaches R – its maximum value. Other authors have called this relationship in different ways. the inverse relationship between w and r can be shown by differentiating (3. which ensure that w be a continuous differentiable function of r. 45 . W and w1.16 We shall label this relationship as the wage-curve. We have in the abscissa the interest rate value 0r1. On the same assumptions.

On the other hand Harcourt (1972. pp. 46 . If D=0. Each system involves two methods of production – one for the consumption good.10 is negative and will indicate that the price of the capital good per worker increases as the rate of profits falls) was called in the debates a negative ‘price Wicksell effect’. it has been assumed D<0. In fact. one for the capital good. or (3. ca) and (lc .19 In other words. even when the physical quantities there involved vary in a direction in accordance with the theory (A1 rises. the interest rate. Pasinetti (1978). then the function (3. 180-181) where the Swedish economist argues against von Thünen’s theorem that holds that the ‘marginal productivity of capital’ equals the interest rate. cc). would define a straight line. and the form of the wage curve will be convex.e.10). on the relative factor coefficients in the production of each of the goods).2). and the form of the w curve will be concave.) These problems regarding changes in the value of capital due to changes in distribution are discussed by Wicksell in his Lectures (1934. Economies with more than one system of production Thus equipped with the fundamental relationships between the wage rate. Lastly. 149-passim and pp. that is cc and ca are proportions of the same capital good C (the capital good associated with system ).(3. Robinson (1958). For example if cc/lc>ca/la. by differentiating equation (3. which actually depends on the value of the production coefficients of each system.3) with respect to the rate of interest we obtain: 18 dpc dr  lc D [la  D (1  r )]2 …. since the capital good used in both sectors is the same. The case D>0 (equation 3. or lower (or equal) than (to) the absolute decrement in interest and hence the value of capital (on the RHS of the equation) may increase or decrease (or remain unchanged) as the rate of interest fall. Note that we can confront both of the relations referred to.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies In Figure 3. etc. cc). the price of the capital good will rise as the rate of interest rises.2’). the price of the capital good will fall as the rate of interest rises. it may be considered two systems of production. Wicksell shows in those pages that due to “unproductive absorption” of the increased output von Thünen’s theorem is false.18 3. for instance  and . pp. if D=0. The wage-frontier. if D<0. ca) and the ones regarding production of the capital good are (lc . See also Swan (1956). Bhaduri (1966).. that is a rise in interest cost affects the production of the capital good more than that of the consumption good – then D<0. This was called a positive ‘price Wicksell effect’. The point is that the absolute increment in wages may be higher. 39-45) has instead preferred to call a negative price ‘Wicksell effect’ the case when D<0 and vice versa. Garegnani (1984b). Instead. The terminology here adopted is that used by Burmeister (1987).1. For the case of Beta system we have (la. The connection between changes in the value of capital when distribution changes with the name of Wicksell can be seen in terms of the RHS of equation 4.B in Appendix B. the form of the function will crucially depend on the sign of determinant D (i. 19 In Alpha system the coefficients regarding the methods of production of the consumption good are (la . The Wicksell effect was a term coined during the controversies by Robinson (1953). The curvature of each wage-curve corresponding to each system crucially depends on the value of the determinant D.3. the price of the capital good does not change as the rate of interest changes. the value of capital and the net output per worker. If D>0. and the form of the wage curve will be a straight-line (a neutral ‘price Wicksell effect’). who actually picked it up from Uhr (1951). for instance if w rises – assume that the rate of rent does not vary for the sake of exposition – then the interest rate must fall.

and pa<pa is verified.12) pa=law+capc(1+r) pc=lcw+ccpc(1+r) The last two equations of system give us the estimated cost of both the consumption and the capital goods with system  but under the prices and wage of system . (cont. since the cost of producing the consumption good is lower than the current price pa. If. and pa<pais verified. we have w and pc. The cost of the consumption good will in turn depend on the price system which happens to be in use. since paand pc. the costs of the consumption good will be different according to which system happens to be in use. if we consider each system separately. If system  is in use. So it would seem to appear that. given two methods of production. then pc and pa can be calculated at that r and at w (prices of capital good and consumption good using  technique at prices and w of system ). if technique  is in use.  is in use instead of  at that level of r. Assume system  to be in use: System  pa=1 pa=law+capc(1+r) pc=lcw+ccpc(1+r) (3. for instance system . However. In other words. we shall have four independent equations to solve for five unknowns. However it can be shown that producers’ tendency to switch from one system to another more convenient (cheaper) one in the existing price situation will bring them to the one giving the highest wage rate at the given rate of interest. then there would exist a convenience to change to the method of producing A given by system . is . for a given r we have w and pc. The producer knows that if system  is in use. that is the system underlying unknowns paand pc. as such. producers’ plan as to which technique20 s/he might choose depends on the costs associated with the production of the consumption good. not The same applies to both pa and pc in case we consider system . are qualitatively different from the remainder unknowns (pc. on the other hand. the choice will depend upon where the producer happens to be initially. then pc and pa can be calculated at that r but at w. In case  is in use we have: System  pa=1 pa=law+capc(1+r) pc=lcw+ccpc(1+r) (3. w) because the system used to produce both A and the capital good. then he would change technique  for technique .11) pa=law+ca pc(1+r) pc=lcw+cc pc(1+r) The last two equations of system give us the estimated costs both of the consumption and of the capital goods with system  but under the prices and wage of system . The problem here is that we cannot determine the value of those unknowns.21 20 21 The term technique is used as synonymous of system involving two methods of production. a priori.) 47 .Chap te r 3 – ‘ R eswitching’ and ‘Reverse Cap ita l D e epen ing’ Under long period conditions. On the other hand.

(See Figure 3. Note that in the process of switching. no more extra profits can be obtained.2. Pasinetti (2003). we have: w/w=pa/pa (3.13) and by discarding each of the first equation of both systems. At r=r2 producers switch back to technique  Even though we cannot determine the value of the unknowns. See Chapter 4 below.15) Whenever the ratios (3. 23 This envelope could also be generated by intersecting a family of systems of production. it can be generated an outer envelope (Northeast envelope) by intersecting wage-curves of the two systems of production considered in the present exposition. By considering the relations between the unknowns: pa : pc : pa ( ) : pc ( ) : w  pa  : pc  : pa (  ) : pc (  ) : w  (3. 412).e. since both systems are structurally identical.2). This is the objective method.14) and (3. we can determine the ratios between them. the cheaper system will be the same at both wage rates and price systems. and will represent the relation between r and w. But when the economy has already adopted the cheapest method. 198). Kurz and Salvadori (1997).Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Since producers will change systems according to this rule. and allows at the same time to have a higher wage. Thus if w>w .14) w/w=pa/pa     (3. p.23 Figure 3. p.15) are equal. 48 . Garegnani (1970a. 22 Samuelson (1962. which has a ‘corner’ where a switch occurs (‘switching point’). See: Garegnani (1970a). then pa>pa i. entrepreneurs who first happen to switch to the most profitable technique might get some extra profits. This curve may be labelled as the wage-frontier. and in the corner itself the two systems can co-exist (“adjacent”22 systems). The wage-frontier (thicker line) and ‘reswitching’ of techniques. system  is cheaper for producing the consumption good than .

consider two goods (corn and cloth) and decide. Then. In the controversies many examples have been built (cf. as assumed) is chosen. of the average ‘capital’-intensity of the demand for consumption goods. if r increases further such that r>r2 . 25 The purpose pursued by Samuelson (1962). then as r rises further such that r1<r<r2 technique  (‘less capital’ intensive) is chosen. there will be one or more ranges of values of the interest rate. However. whichever the more ‘capital’ intensive consumption-good. We shall also see in that chapter.4. D=0). Bharadwaj (1970). implying a shift of the composition of consumption in the direction of an increase. 27 In the text it has been illustrated ‘reswitching’ undermining the ‘direct’ factor substitution mechanism. the ordering itself will change with changes in distribution.e. Cf. Assume that. 21. it will be found that the ‘more’ – and not the ‘less’ – capital intensive technique is chosen according to the rule governing the choice of techniques. 24 49 . As to the ‘indirect’ mechanism. or concave and convex curves if each system involves one or more capital goods in such a way as to give rise to more than one intersect between different wage curves. Symposium 1966: Pasinetti 1966a. namely: the higher the rate of interest. Implications for the basic premises of the theory: Conclusion With the aid of the wage-curves and the wage-frontier the intuitive explanation of ‘reswitching’ introduced in this chapter may be reinforced by analysing the relations between the net physical product. whichever the criterion adopted. above). Assume that  gives the highest net product and is a straight-line curve (i. D<0). so far neoclassical long period technical choices go in direction as hypothesised by the theory.Chap te r 3 – ‘ R eswitching’ and ‘Reverse Cap ita l D e epen ing’ 3. Garegnani 1966 and see also Levhari 1965) in order to generate concave or convex. 28 In the following chapter we shall survey the views of critics and defenders of the theory at stake on these results.25 while the second technique.28 It is also possible to consider w as the independent variable. which consumption-good is more ‘capital’ intensive. and the results will be as similar as those reached when we regard r as the independent variable. 26 Then. 27 (See Figure 3. n. yields a lower net product and intersects system  at two points. even if a criterion could be found so as to order ‘intensities’ according to different levels of distribution. where an increment of r makes the more ‘capital’ intensive good relatively cheaper.2) Thus. that Samuelson’s (1962) attempt at rescuing neoclassical theory implies the fact that each wage curve intersects each other only once. which is concave in its entire trajectory (i. this chapter. . Appendix D where it is also illustrated ‘perverse’ behaviour when there is no ‘reswitching’. Assume that a method is ‘more’ capital intensive because it is the method yielding the highest net product per worker of the integrated industry.e. rather than a decrease. and the value of capital per worker as the rate of interest changes. are analysed in Chapter 4 below. who introduced this case. 26 On the issue of the number of switch-points cf. as required by the theory. which is derived from the group (ii) of the data. the lower the ‘capital’ intensity. when 0<r<r1 technique  (‘more capital’ intensive. according to the method explained above (cf. an a priori criterion to establish which technique is more (less) capital intensive might be chosen.24 For our purposes here it will be sufficient to consider what happens to the value of capital per worker as the rate of interest changes. the wage rate.

At that point both techniques can co-exist. along the wage frontier there is a switch point. Figure 3. 31 For 0  r  r1 there is a vertical curve indicating a constant value of capital due to the straightline wage curve ( is in use.3) which corresponds to values of Figure 3. suffice it to use the graphical device below (Figure 3.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies In the foregoing analysis capital-inputs have been measured in technical units independent of distribution. above.2) which corresponds to these levels of the rate of interest. Chapter 2. as r increases from r1 up to r2. 30) have argued that the form of the demand curve with ‘reverse capital deepening’ and the discontinuities in the values of k is due to discrete technologies. Figure 3.) 50 . However. Then. system  is in use and the value of capital per worker gradually increases because the value of each successive tangent along the wage curve increases as r increases until r2. the ‘employment curve’ for capital (a basis to derive the demand function for capital) is no longer inversely related to the rate of interest. Cf.3. 30 However.2.g. ‘Reverse capital deepening’: k (capital in value) is not inversely related to r As can be seen from Figure 3.31-32 29 30 Measured in terms of the consumption good of the integrated industry. Paragraph 2. thus raising the claim that the rate of interest might bring investment (capital demand) into equality with full-employment savings (capital supply). At r=r2. as r rises from r2 up to R the value of the capital is constant. 32 Some (e. some might argue that the substitution mechanisms work ‘in the right direction’ when capital is measured straightforwardly in value terms. Then.10.3. there is ‘reswitching’ of techniques. p. in the light of ‘reverse capital deepening’ and ‘reswitching’. and that is why the value of capital assumes different levels at that rate. since the  wage curve associated to that range of values of r is a straight-line – hence  is being used again. when r=r1. Then. yet there is no guarantee that changes in the rate of interest will cause changes in the opposite direction in the quantity of capital per worker 29 employed in the economy. cf. hence both systems can co-exist (adjacent systems) and the value of capital per worker assumes different levels at that rate. Solow 1967. hence ‘reswitching’. To show that the demand for capital might be directly related to r. there seems to be no guarantee to derive a downward-sloping demand curve (or employment curve for capital). so if (cont.

not necessarily negatively sloping. r and q. and which do not necessarily entail a neoclassical relation between k. by using the analysis of production theory (minimisation of costs). if e. ca. supply and demand forces were conceived by the founders of the marginal theory to operate in markets under free competition by relying on the belief of the factors’ substitution principle. la. 51 . then it would not be so difficult to envisage the scope of the implications of the Cambridge (UK) critique based on ‘reswitching’ and ‘reverse capital deepening’. But. also Paragraph 4. as Joan Robinson has continuously insisted. if we are reminded of the key role played by the downward-sloping demand curves so as to provide a plausible basis for the theory to postulate a stable and unique long period equilibrium. pp.g.33 The results described in this chapter cast serious doubt on the issue concerning how the distribution of the social product between wage-earners and profit-earners can be explained. On the contrary. ‘reswitching’ and ‘reverse capital deepening’ undermine the demand aspect so that it turns out at least doubtful to believe in an inverse relationship between the rate of remuneration and the quantity of the factor employed. 523) and Garegnani (1970a.Chap te r 3 – ‘ R eswitching’ and ‘Reverse Cap ita l D e epen ing’ So far it has been seen that even if the illegitimacy of the supply aspect of the problem of capital (the necessity to conceive of capital in value terms to explain long period supply-and-demand equilibrium. See Chapter 5 below where Joan Robinson’s role in the Cambridge debates is surveyed. without however implying changes in those coefficients. in economies with heterogeneous capital goods the demand curves could be of any form. expressed. demand curves are not downwards-sloping? Can we rely on the factor substitution principle in order to determine the distribution of the social product between wages and interests? As we shall see in the second part of this study. through the equilibrium condition of the uniformity of the rates of return on the capital goods’ supply prices) is put aside. etc. also Pasinetti (1969. that distribution might be explained by forces other than the supply and demand ones – the explanation of which we have seen in Chapter 2 above. We thus conclude this first part of this study with the following question: Could we then expect from this theory predictions and forecasts on real economies. 412-414. On the grounds of supply-anddemand theory.6.) 33 The author would like to highlight that the negative implications of these results for the theory have nothing to do with the utilisation of any ‘aggregate’ production function. switch from a technique to another would imply an infinitesimal change in the production coefficients (la. As we have also seen in this chapter. Cf. Cf. in the capital market. it has been shown by Bellino (1993) that the same wage frontier – which entail ‘enveloping’ wage curves and would have ensured changes in the coefficients so as to keep neoclassical relations among variables – could also be generated by other wage curves. But these results do not arise out from outside the neoclassical assumptions.) so as to ensure such a continuous relation between k and r (and q) along the technical frontier (the wage frontier) that the neoclassical relations could be sustained. below. p. the critical side in the arguments between the two Cambridge has shown. the theoretic possibilities of ‘reswitching’ and ‘reverse capital deepening’ have serious implications for the the number of techniques becomes larger and larger. However.

this author’s participation together with the replies and counteractions it gave rise constitute. such as Samuelson. from a historical standpoint. which one may appreciate as soon as it is realised why notable neoclassical scholars. the heart of the Cambridge controversies in its first phase of its development. As the following chapter will discuss. got in the capital debates. 52 .Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies dominant theory.

PART II The Capital Theory Debates in a Historical Perspective and the Transformation of Dominant Theory .

.

of course. it is fair to say it. the way in which Joan Robinson sparked the ensuing controversies in her famous. . Sraffa’s publication date) may be characterised as a ‘pre-reswitching’ period of the debate – although. the phenomenon had already been noticed by Robinson (1953.2 In particular.1 As we shall see. but little importance was given to it by these and other authors at that time.1. 196. has been to introduce the reader to the analytical grounds on which the controversies evolved. Samuelson (1962) p. See Chapter 5 where we treat Robinson’s role in the controversies as well as Champernowne’s contribution. serious doubts concerning its theoretical tenability had started to arise. 1953 article much characterised the debates that immediately followed it. 4. 2 Robinson (1953). namely: ‘reswitching’ of techniques and ‘reverse capital deepening’.e. neoclassical reactions and the highest point of the debates “No alchemist can turn one capital good into the other” P. one of the most relevant results of the controversies has been introduced. Introduction In the previous chapter. Although by the beginning of the 1960s the most salient arguments raised in the debate were not clear with respect to the traditional role ascribed to ‘capital’ as a factor of production in deriving the equilibrium supply-and-demand functions for that ‘factor’. 1956) and Champernowne (1953).Chapter 4 A first phase in the controversies: Samuelson’s 1962 intervention. the 1966 Symposium. the first critiques in the controversies were mainly concerned about the use of aggregate production function as an instrument to determine distribution by the partial derivatives 1 Eventually the period previous to 1960 (i. so as to understand the arguments raised by the participants in the successive debates in the light of those results. The aim of the previous chapter. We have seen that these phenomena seriously concern the core foundations of neoclassical theory: it cannot in general be postulated a negative relation between demand for capital and the rate of interest.

it had not been discussed in a clearer ground the consequence for the marginal theory of distribution the fact that. 4 Samuelson (1962). though possible at the level of single 3 Champernowne (1953) tried to save the production functions from being a multi-valued relation due to value measurement of capital by the ‘chain index’ measurement of capital.6 below. this chapter deals with the reactions of neoclassical economists to the criticisms raised in the 1966 Quarterly Journal of Economics Symposium on paradoxes in capital theory (Pasinetti 1966a. it turns out to be impossible to specify. who. because the very classification of techniques according to ‘factor intensities’ changes as the distribution of income between wages and profits changes too. 26. Garegnani 1966). we shall turn back to Robinson’s criticisms and her immediate contestants of the 1950s in the following chapter). which came into being thanks to the intervention of one of Samuelson’s go-between at that time: Levhari (1965). a measurement which.2-4. Samuelson’s article appeared soon after Sraffa’s book. while. is also dependent on distribution. This problem. independently of distribution. we submit. which finally reached readers somewhat eight or nine years later in (1970a)5 (Paragraphs 4. in particular Paragraph 5. The aim of this chapter is therefore to put at the centre the salient exchanges that surrounded the phenomena of ‘reswitching’ and ‘reverse capital deepening’ since Samuelson’s intervention in the controversies in 1962. however.6). See Chapter 5. Also interestingly is to note that Samuelson’s article triggered off a critique by Garegnani. Samuelson laid the ground for the discussion by presenting systems of production with heterogeneous capital goods – though from a particular viewpoint as we shall see below.3 Yet. which is however essential to the traditional theory. the marginal theory relies on the factor substitution principle. having visited MIT at that time. this chapter. due to the phenomena examined above (Chapter 3). 4 In the wake of Samuelson’s article the debates followed an interesting path both from a historical as well as from an analytical viewpoint. As to the first aspect. But. it is impossible to talk about ‘more’ or ‘less’ capital-intensive production processes independently of distribution. where a manuscript of Samuelson’s article circulated. Indeed. Also. It was there that Garegnani developed his criticisms. below.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies of that construct. as to the latter aspect. rather than having adopted a chronological standpoint onto the discussions. who pretended to have shown that ‘reswitching’. is totally independent of the aggregate production function and shows that a negative relation between ‘capital’ and the rate of interest cannot in general be postulated. as Sraffa demonstrates in his 1960 book. had criticised the former in an oral intervention around 1961-62 in MIT. 56 . a theoretical principle that should indicate the cheaper or dearer production processes of the net output according to ‘relative factor’s intensities’ as distribution changes. 5 See n. That is why the present author has deemed more appropriate to deal with the first phase in the capital debates by discussing the contributions which appeared in the 1960s decade. which production process is more or less ‘capital’ intensive among the alternative productive processes. thus casting serious concerns on the determination of a ‘quantity of capital’. (However.

where the important result of ‘reverse capital deepening’ emerged out of the analysis of the relation between prices (choice of techniques) and distribution.15). p.114. n. For the way in which discussions evolved over the first stages of the debates see Chapter 5 below. A similar view can be found in King (2002) who writes: Ironically the first to expose the depth of the Sraffian critique were defenders of the neoclassical theory.Chap te r 4 – A first phase in the con tro versies industries. 4.8 Samuelson (1962. 192. where we have introduced the ground for the forthcoming discussions in these debates. Samuelson’s (1962) model with heterogeneous capital goods Although Samuelson’s contribution is a direct response to Robinson’s concerns about the theoretical tenability of the production function – in fact Samuelson dedicates his paper to Joan Robinson’s “memorable 1961 visit to MIT” 6 – it is important to emphasise the fact that in Samuelson’s contribution one can find the discussion of the relation between distribution and demand for factors in heterogeneous capital goods economies on a ground which is clearer than the ground on which Robinson (and her immediate contestants) had put it. 95).2. 8 King (2002. (See Appendix C below. See Chapter 3 above. Some successive reactions (Samuelson. Bliss) after the Symposium and Garegnani’s 1970 critique are analysed (Paragraphs 4.10) This was contested in such a detail and accuracy by the critical side that ‘reswitching’ turned out to be issue for the discipline. P.7 Indeed that ground was the one on which Sraffa had already laid the argument of the choice of techniques in his 1960 book. and that in Samuelson’s surrogate note there is the subtle message that heterogeneity of capital goods could not bring big headaches to the traditional theory. Ferguson.7-4. 7 6 57 . 1). Garegnani] that initiated the Cambridge Controversies. […] [It was] Samuelson’s attempts to solve [Sraffa’s results on reswitching and reverse capital deepening which were pointed out by former PhD student in Cambridge. p. cannot arise for the economy as a whole (Paragraphs 4.) That we have decided to deal with Samuelson (1962) in this chapter (which is the first of the second part of the study) is justified by the fact that in that article it is possible to find one of the most serious attempts from the neoclassical camp to deal with heterogeneous capital goods without any recourse to a production function.

9 The attempt pursued by Samuelson is to show that. 194). quite similar to that present in Sraffa (1960) though there are quite important differences which are worth mentioning.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Samuelson starts his note by pointing out that capital theory can be rigorously developed without any recourse to a ‘Clark-like’ concept of aggregate capital. in order to determine distribution. apparently.… 999 or anything else. 11 Samuelson (1962. Concepts like ‘Clark-Ramsey parable’. Then he assumes that each ‘Clark-like’ concept of ‘capital’ refers to the idea that the production of net output of an economy can be represented by a ‘production function’ whose argument contains homogeneous labour and ‘capital’ (in terms of output) such that relative rates of remuneration can be determined by the partial derivatives of such a function with respect to each factor. ‘Clark parable’ and the like refer to the same idea. In Samuelson’s model there are concrete different capital goods that can produce an output composed of a consumption-good and of a capital-good in fixed proportions with homogeneous labour. while each particular capital good is produced by itself and labour. Samuelson generalises the fixed-coefficient model for any system of production. beta. so that the consumption good can be produced with homogeneous labour and a particular capital good associated with each system of production. the “surrogate production function”. Samuelson begins with the introduction of a heterogeneous-capital-good model: I begin with a concrete model in which there are a great variety of capital goods: call them: alpha.10 Thus. Samuelson’s model is. In Samuelson’s model there is a fixed coefficient production process for each component of output. he claims that by using the surrogate production function and the associated “surrogate capital” one can predict exactly how “complicated heterogeneous capital models will behave”. Moreover. 10 Samuelson (1962. p.11 and in the same page some rows below he explains (emphasis in the original): One need never speak of the production function. but rather should speak of a great number of separate production functions. which is supposed to consider the heterogeneous capital goods involved in the different techniques of production. 9 58 . p. Both production processes encompass the system of production of the economy where constant returns to scale and perfect competition prevail. economists can use a “new concept”. 194).

The fact that in Samuelson we have switch points in his very special cases is due to the existence of only one good common among techniques.) (cont. Cf. It must be noticed that not only is Samuelson’s model quite arbitrary because of the equality of the capital-labour ratio in each industry. Samuelson’s ‘special sub-class of realistic cases’ To be coherent with the hypothesis of production with strictly heterogeneous physical capital goods. also a very unrealistic case. on this issue see Baldone (1984). 1). Different wage-curves may or may not intersect. Then the value of capital is constant and independently of changes in distribution. 4.Chap te r 4 – A first phase in the con tro versies capital good does not enter the production of any other capital good except for the production of itself. Samuelson finally extended his restrictive assumption to all systems of production in the economy (“book of blueprints”). p. 408. the maximum wage rate for the process at zero profit rate. pp. the points on the ‘north-east frontier’ Samuelson’s model is similar to that discussed in Chapter 3 with the difference that we have considered only circulating capital whereas Samuelson considers fixed capital.12 As is well known. 15 Then. 196-197. As the labour-means of production ratio is assumed to be equal in both sectors of the economy.) 59 12 . p.) of the outer portions of the straight-lines will depict the most profitable techniques at any profit rate. 14 Samuelson (1962). though his treatment of fixed capital evades the problem of joint production.3. then. In this regard Samuelson makes a highly restrictive assumption: that the proportion of labour to the means of production for the system of production concerned must be the same for all goods produced. n. as already explained. The case of the straight-line wage curve is D=0. (See Figure 4. Samuelson puts forth a “special sub-class of realistic case”13 in which every capital good is different according to the different systems of production. different production systems will in general imply different wage-curves or “Factor-price frontiers” as Samuelson has put it. but also because he assumes that every activity needs just only one particular capital good – a very unrealistic case. 15 Thus one can plot in the wage-profit rate layout the different straight-line “factor price frontiers” according to each system of production so that. The straight line connects two points. p. and the ratio of technical surplus over replacement to means of production when the wage rate is zero. and enters the production process in fixed proportions with labour. that the “factor-price frontier” built for a single capital-good technique where the capital-labour ratio is constant as distribution changes must result in a straight line.14 No doubt. 198. i. namely. if a range of production systems (each involving only one capital-good) is considered. 13 Samuelson (1962). Chapter 3.e. for the final consumption good and for the capital good (la/ca=lc/cc). 196. Now we must discuss a crucial assumption in Samuelson’s model. the “envelope” (Samuelson 1962.1. depending crucially upon the relative production coefficients. there will only be one possible capital-labour ratio in the economy as a whole. each with either a higher proportion of ‘capital’ to labour. above. See Garegnani (1970a.

(Cf.e. for a finite number of techniques. are of equilibrium.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies envelope. The neoclassical ‘parable’: ‘Surrogate’ capital and the ‘Surrogate’ production function Did Paul Samuelson succeed in inventing the ‘right fairy tale’? Samuelson believed in having found a rationalisation of the ‘Clark parable’ because the envelope generated by the latter (see Figure 4. (Cf. 201. 60 . i. The successive step that Samuelson took was to associate the Clark-Ramsey parable with his ‘more realistic’ heterogeneous capital-good model.c) is generally similar to the envelope derived from his supposedly ‘real’ heterogeneous capital goods model. we can come as close as we like to duplicating the true blue-print reality in all its complexity. B. Because of the particular assumption made by Samuelson this envelope. Figure 4. emphasis added. the wage-frontier. maintains that the envelope generated by straight-line wage curves is “generally similar” to the function relating the wage rate and the profit rate derived from the “Clark-Ramsey parable”.1) The switch points on Samuelson’s envelope have the basic neoclassical property of being all ‘forward switches’. points A.1.17 4.2.1) Thus. i.16 Samuelson then claims that if we invent the right fairy tale. Samuelson (1962). A straight wage-curve and a family of straight wage-curves 16 17 Samuelson (1962). after having drawn the envelope corresponding to the book of blue prints (see Figure 4. or C of the right-side graph in Figure 4. Samuelson. as the profit rate rises the capital-labour ratio falls.b above). p. he Figure 4. p. has straight segments (when one technique is the one most profitable) as well as corner points (‘Switch points’.e. 201.4.1. ‘adjacent techniques’) where two techniques can coexist.

p. 19 which replicates the parable. which consists of consumption goods or of net capital formation “the two being infinitely substitutable on a one-for-one basis” 20 Samuelson (1962. That is why Samuelson believes that his “new concept”18 of Surrogate production function.1’) where y is net output per worker and k is capital per worker. as Samuelson assumed. the profit rate and the net output per worker that prevail in the ‘real’ economy by analysing the properties of the surrogate production function. p. The ‘Clark parable’ entails using a production function. p. once wages and profits are determined by the marginal productivities of each factor respectively. L) (4. with its associated Surrogate capital. 201. which represents the whole techniques of production of the net output that is produced with labour and ‘capital’ homogeneous with the output.” 19 18 Figure 4. 201) thinks that ‘K’ is surrogate capital “that is to go into the Surrogate Production Function that will predict all behaviour.Chap te r 4 – A first phase in the con tro versies pretended to have supplied a demonstration that the formal properties of the ‘parable system’ are the same as those derived from his “quasi-realistic complete system of heterogeneous capital goods”. from this information the family of the wage-curves can be derived. 200) assumed that labour and homogeneous physical capital (jelly) produce a flow of homogeneous national product.1) because of constant returns we may re-write the function as follows: y  f (k ) (4. The ‘parable system’ Samuelson pretended to analyse the relations between the wage rate.2. Then.) 61 . Samuelson (1962. It is correspondence between the formal properties of both constructs – that is an ‘imaginary’ surrogate function (or parable)20 and the ‘real’ economy of straight wagecurves – that quickened Samuelson to conclude: Samuelson (1962).3) If we derive both of these equations with respect to k we obtain: (cont. In formal terms we have: Y  f ( K . could be a useful tool for analysis of heterogeneous capital models.2) (4. 200). Samuelson (1962. so that: w  f ( k )  kf '( k ) r  f '( k ) (4. p.

in actual fact. As we well know. one could analyse the relations between w. and q (i. every single wage-curve is a straight line.2.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies [T]he Surrogate Homogeneous Capital gives the same result as does the shifting collection of diverse capital goods in our more realistic model. 22 Even more. 23 Samuelson (1962). 62 . Then. p. 202 where Samuelson develops the surrogate function.1 and cf. then the ‘envelope’ will be a smooth convex envelope (see Figure 4.2 above.23 dw dk   kf ''( k ) (4. apparently. by studying the formal properties of the surrogate production function.21 (Again. Samuelson would have dealt with heterogeneous capital goods economies and then would have shown that. (Cf.5) If we divide both sides of (4. it is worth looking at Figure 4. these apparently similar formal properties of the ‘parable system’ and of the ‘real’ model seem to have made Samuelson maintain that [the surrogate function] is not an approximation but a rigorous equivalence. 203. p.6) which is the slope of the envelope (family of wage-curves) and represents the capital-labour ratio of the economy. Figure 4. This justifies the Surrogate Production Function as a useful summarising device. Samuelson.) So. 1962. 22 In fact Samuelson derived from equations (4.5) a functional relation between w and r.c). p.e. 201.4) and (4.4) and (4. Samuelson concluded that the envelope is a convex curve and therefore could be compared with the wage-curves derived from Samuelson’s ‘quasi-realistic’ cases where.) 21 Samuelson (1962). as output and surrogate capital are homogeneous. the net physical consumption output of the ‘real’ economy).5) we finally get: dw dk  dw   k dr dr dk (4. r.4) and dr dk  f ''( k ) (4. which would be the wage-frontier (envelope) representing the whole competitive techniques. if we consider an infinite number of straight wage-curves.

p. In the discussions with Samuelson. See Paragraph 4. Samuelson’s analysis assumes the existence of a ‘real’ economy.12 below. one feature of which being an envelope function relating the wage rate Robinson (1970a).24 Five years later Robinson provides further information on the historical aspects surrounding Samuelson’s participation in the controversies: For several years.3-4. from the critically oriented side of the controversies. Joan Robinson: At first the neo-neoclassicals were happy to accept this parable [the surrogate production function] […] For some years they remained cooped up in this position. We shall explore in the following why this is so. 37. 24 63 . and n. 203) of the ‘real’ heterogeneous capital goods model. 311. Robinson is thinking of Ferguson (1969) when she asserts that the neo-neoclassicals “were happy” with the surrogate function.Chap te r 4 – A first phase in the con tro versies 4. yet that did not prevent Samuelson from asserting that the Surrogate function is a “rigorous equivalence” (Samuelson 1962.4). p. 536. As we saw (Paragraphs 4. 62. Garegnani already pointed out the extreme assumptions on which the surrogate function is based. This issue was finally settled in Garegnani’s 1970 article. 26 Garegnani (1970a). n. as is recognised e. below. by Joan Robinson. The article finally published in 1970 is an extension of a formerly accepted paper submitted to the editors of The Review of Economic Studies in April 1963. in the period immediately previous to the publication of Samuelson’s ‘Surrogate’ article (1962).25 Joan Robinson’s appraisals suggest that among the critical contributions to the Cambridge controversies one may find in Garegnani’s (1970a)26 article a critique of Samuelson’s construct. 1). What did Samuelson actually demonstrate? Before answering that question. it is interesting to note to what extent Samuelson parable had been accepted by the discipline at that time. The issue under dispute has been precisely how far that ‘real’ model actually represented production with heterogeneous capital goods. p. everyone (except Piero Garegnani) was somewhat baffled by the surrogate production function.g. where he started writing the manuscript finally published in 1970. in the history of the contemporary capital debates. this chapter. Bhaduri (1969) p. Garegnani was a visiting Rockefeller Fellow at MIT in 1961-62. 25 Robinson (1975a). Thus Garegnani’s (1970a) became a very important reference. In this regard it is worth citing the following views from a well-known protagonist of the controversies. p. 202. where we deal with Ferguson (1969). Though Samuelson acknowledges Garegnani for having saved him “from asserting the false conjecture that my extreme assumption of equi-proportional inputs could be lightened and still leave one with many of the Surrogate propositions” (Samuelson 1962. that is to say.5. On the influence of Garegnani’s criticisms of Samuelson cf. which had been accepted subject to revision.

2 above. however small. if all the wagecurves are straight lines or convex ones the envelope will be convex.3. 194. p. Is that true? Following Garegnani. 1). Accordingly. 29 This smooth envelope entails avoiding switch points. The assumptions underlying the present discussion are those present in Paragraph 3. causes a change of system. but its curvature can be of any shape. Concave. and may therefore alternate convex with concave segments. and ii) that one system only is in use at each possible level of r.30 As it was assumed that at any possible value of r a single system is in use. and the wage-curve of each system contributes only points to the wage-frontier and not stretches. Samuelson (1962. dr In general this function envelops several wage-curves of different curvatures. to come to grips with the particular defence of traditional theory attempted by Samuelson with his Surrogate function we need set out a more familiar scheme of traditional theory: it will be necessary to assume that each system of production can change continuously with r. and determining a continuous function between the wage rate and the rate of interest.27 and showed that the results derived from this ‘new’ device are the same as those obtained from the ‘realistic’ model.7) As a matter of fact. a given physical net output and physical capital per worker corresponding to each system can 27 28 30 The ‘as if’ aggregate production function.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies and the profit rate. the family of systems 28 does not show any cornerpoint. whereas the latter will be concave or straight when the wage-curves are concave. it is guaranteed that this wage-frontier becomes a smooth envelope29 enclosing the whole family of the wage-curves from above. convex and straight ‘envelopes’ 64 . that is w = e (r) (4. Figure 4. Then he introduced the “surrogate production function”. n. derived from the alternative systems of production. Thus. By this it is meant i) that any change in r. For example. we perfectly know that the higher the rate of interest the lower the wage rate ( de(r ) < 0).

9). In analytical terms. which must be equal to L S ) of the ‘real’ economy at the given r corresponding to J ii) S/L = q(r = S ) J (4. then heterogeneous capital-goods might be expressed as quantities of homogeneous capital. defined for the ‘parable’. are the Clark system and the heterogeneous capital-goods model.10) The marginal product of labour ( the ‘envelope’ function e(r = that equilibrium. The core of Samuelson’s attempt was to apparently determine the relations (in the ‘real’ economy) between r. and q through a production function. S ) must be equal to w. as Samuelson so decidedly claimed. as Samuelson hypothesises.L) (4.9) with S as the quantity of net product.Chap te r 4 – A first phase in the con tro versies therefore be found for any given r. by using Euler’s theorem: S S S J  ( ) L L J L (4. is therefore function of r: q=q (r) (4. and must satisfy the following two conditions: i) S S = e( ) L J (4. J that of capital. the imaginary parable.8) It is important to notice that equations (4. and L that of labour. It is useful to re-write equation (4. This is the ‘surrogate function’. must be equal to the function q for the real economy defined above.11) The net product per worker. which belong to the imaginary economy (the ‘Clark parable’).12) 65 . and must have. Since according to Samuelson both the ‘Clark system’ and his heterogeneous capital-goods model are ‘rigorously equivalent’. This net product per worker. a parallel with equations (4.8) reported above belong to the heterogeneous capital goods economy (the actual ‘real’ economy).2) and (4. we are asked to find a production function homogeneous of the first degree: S=S(J.3) above. It is a confrontation between the meanings associated with both groups of equations that will actually allow us to see how ‘rigorously equivalent’.7) and (4. w. q.

15) establishes that the slope of the ‘envelope’ curve measures the capital-labour ratio of the economy. equation (4. J/L would rise with r = S as long as the envelope is concave. 66 .13) So far we have obtained functions expressing the net output per worker in the imaginary economy. S S S S J J L L ) d( ) ) d( d( d( ) J J J J S 32 Cf.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Since in the imaginary economy the equilibrium profit (interest) rate must equal the ‘marginal productivity of capital’. Therefore.15) Function (4.14). from which 0   L  that is equation (4. by using again condition i).12) can be written as follows: S J  e( r )  r L L (4. In effect we know that in any real economy the envelope may be concave to the origin in parts or throughout if the wage-curves are concave.12) with respect to S J we immediately obtain S S J d( ) d( ) d( ) J L  L   L  S . by using condition i) noted above. by multiplying and dividing the LHS of this equation by d ( J ) S S S J L L ) d( ) ) d( d( J J J S S J J d( ) d( ) d( ) d( ) J S J L  L   L then .14). equation (4. Now let us turn to the capital-labour ratio of that imaginary economy. is: J/L = – e’(r) (4. If we differentiate function (4. but this implies that the marginal J 31 If we derive (4. 32 But now let us compare this result with the information we can trace from our real economy thanks to knowledge of the curvature of the envelope.12) with respect to S ) J L  S L d( ) J d( S we obtain:31 J (4.6) above.14) But expression (4. then.

for the ‘marginal products’ would change when the ratio of ‘capital’ to labour does not. combined with result (4. When the ‘envelope’ is a convex curve we have two alternatives. Furthermore. should we consider some straight-line part of the ‘envelope’ – which is theoretically admissible when we have concave wage-curves – the contrast with the ‘Clark-Ramsey parable’ would be even more striking. In fact we must rule out concave wage-curves because they imply that. and accordingly we would have that J/L increases when r rises. it is necessary to re-write equation (4. The foregoing considerations are sufficient to dismiss the existence of such a surrogate production function.16) underestimates q as can be seen from the following graph. either the wagecurves are convex or they are straight lines. the value of capital increases as the profit rate increases.12). Function (4.15). even when there is only one system of production from which a concave wage-curve is derived.35 So the only one case in which function (4. p. In this case that function will over-estimate the real value of q. (cont.) 67 33 . “the function S could not be a production function”34 because equilibrium in the ‘imaginary’ economy requires an inverse relation between ‘marginal product of capital’ and the capital-labour ratio.16) will not equal q when the wage curve is convex either. This would also imply that condition i) above cannot be satisfied with any differentiable homogeneous function as it is impossible to talk about marginal changes when factor’s proportions do not vary (they cannot vary by definition of a straight line envelope). But S/L will equal q=q(r) of the real economy only when the wage curve is a straight line but not when the wage curve is a convex one. 34 Garegnani (1970a). but it is still of some interest – as Garegnani points out – to consider the case of convex envelopes. becomes: S  e( r )  r[ e '( r )] (4. 415. 35 To see that a convexity relation between the wage rate and the profit rate does not ensure the existence of the surrogate function.16) correctly estimates q. therefore. a behaviour that goes in contradiction with the marginalist premises establishing an inverse relationship between demand for ‘capital’ (investment) and the interest rate. while In concave parts of the envelope we have e’’(r) < 0.16) L In fact. function (4. which.Chap te r 4 – A first phase in the con tro versies product of capital rises as J/L (the ratio of ‘capital’ to labour) rises33 and.

37 Since heterogeneity of commodities can be defined only as differences in the production conditions (that is different systems of productions involving different capital goods). In this regard Garegnani clears the ground: [B]ut for the arbitrary choice of the capital-good unit.4: the elasticity on point P is 0w1/w1A.36 4. 415). For example. q is not equal to S/L when the wage-curve is convex It is important to note that Samuelson’s way of measuring the relative shares of labour and capital through the ‘elasticity’ of the envelope is incorrect for general cases (i. for convex and concave curves).6. now. is when the wage-curves are all straight lines. 37 Garegnani (1970a. There is no heterogeneous capital goods model in Samuelson’s (1962) Drawing on our knowledge of the systems of production involving heterogeneous capital goods.e. whereas the relative share actually is 0w1/w1W. consider the convex wage-curve of Figure 4. a straight wage-curve actually means that the proportion of capital goods to labour is the same both in the consumption industry and in the capital good industry. No doubt. It is in this particular case that the ‘marginal product of capital’ (the partial derivative of the production function with respect to ‘capital’) would measure the exact level of the rate of profit.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies function (4. 36 68 . the input coefficients of the two industries are identical.4. and only in this particular case does a surrogate function exist. p. a straight wage-curve does actually mean that the consumption good is produced Figure 4.9) satisfies conditions i) and ii). that Samuelson has pretended to generalise the ‘parable’ to apparently ‘real’ heterogeneous capital goods models by precisely turning some particular heterogeneous models into the parable system.

can exist. 416. therefore. p. Moreover. 421).) 39 38 69 . linear and hence homogeneous. whose existence in such an economy no critic has ever doubted. or surrogate production function.Chap te r 4 – A first phase in the con tro versies by itself and labour. since each technique is single-valued in the capital-labour that it contains (the value of capital is always independent of changes in distribution). If. emphasis added. i. regarding whether and how far the parable system had been accepted by most of the practitioners of the discipline at that time: “Samuelson’s ‘surrogate production function’ is thus nothing more than the production function. whose input coefficients are equal in both sectors. there are many consumption goods as well as capital goods. So Samuelson’s system (i. Samuelson’s quasi-real model. In a word. that the ‘parable’ provides the same Garegnani (1970a. linear. 1972a. because if there is only one consumption commodity in the economy. there is no need to ‘aggregate’. Samuelson would have demonstrated. Garegnani also noted 38 that the aggregate production function is arrant nonsense. the aggregate production function does not exist either. Samuelson’s claim – that establishes that to deal with heterogeneous capital goods economies there is no need to recourse to a ‘quantity of capital’ 39 – is absolutely dependent upon the highly restrictive assumptions of equi-proportionality of inputs. partial derivative) interpretations. Therefore. in Samuelson’s model the switch points would all be ‘forward’ switches.e. with the help of Robinson’s views. p.” (Garegnani. as Garegnani has shown it. a ‘rigorous equivalence’ must take into account the general cases (convex.e.e. a straight-line wage-curve representing the production methods of the consumption good and of the capital good) becomes indistinguishable from one where the consumption good is produced by itself and labour. neoclassical switches establishing a lower value of capital per labourer as the profit rate increases. Samuelson (1962. the factor prices can still be given various long-run marginalism (i. 200) wrote: “An interesting point […] is that even in our discrete-activity fixed-coefficient model of heterogeneous physical capital goods. concave wage-curves) not only a very particular one: indeed both ‘worlds’ turn out to be equivalent because Samuelson disguised the ‘Clark parable’ into his ‘quasi-real’ model – though. in our opinion. so that the capital per labourer demanded is independent of changes in the profit rate. This jelly world is the only one world in which an aggregate production function. is not more real than the jelly world he believes in.. actually. p. 40 The following conclusion raised by Garegnani seems to go in accordance with what we have suggested above. because it can only be built if and only if the national net output is composed of only one commodity produced by itself and homogeneous labour. on the other hand. the ratio J/L and the ‘slope’ of the envelope is at the same time the ‘slope’ of the wage-curve tangent to the envelope at that r. where is the ‘rigorous equivalence’ between the ‘Clark system’ and the quasi-real heterogeneous capital-good model that Samuelson would have demonstrated? The point is that. since there is no more than a single commodity in that economy and hence there is no aggregate production function. And all this without our ever having to pretend there is any quantitative aggregate homogenous ‘capital’ that itself truly produces anything” (emphasis added). because at any given r.40 In Samuelson’s jelly world the demand aspect problem of capital (see Chapter 2 above) in the context of the traditional method of long period equilibrium or normal positions is totally saved.

42 Levhari (1965). not because there are physically heterogeneous capital goods. Samuelson (1966a). but.10).7. would not have any work to do at all. Levhari’s purpose has been to demonstrate that 41 Pasinetti (1966a). Levhari (1965) and the 1966 Quarterly Journal of Economics Symposium In 1966 the MIT-based Quarterly Journal of Economics gathered together a series of contributions which conclusively showed that ‘reswitching’ and ‘reverse capital deepening’ are phenomena which can in general exist either in economies with indecomposable or decomposable technology matrices. Samuelson’s 1962 model actually avoided dealing with heterogeneous capital-goods economies. if existed. al (1966). 41 In terms of Sraffa. in 1965. it is true that in Samuelson’s world “No alchemist can turn one capital good into the other” (1962. since there exists only one capital good homogeneous with the consumption good.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies properties of a heterogeneous capital-good model if and only if there had existed only one good produced by itself and labour but then there is no heterogeneous capital good economy at all. Morishima (1966). 70 . The ‘Samuelson parable’ is not less unrealistic than Clark’s. Bruno et. Levhari’s analysis is briefly analysed here below. Garegnani (1966). This story regarding the possibility of perverse behaviour with indecomposable or decomposable technologies is due to the claim carried forward by Levhari. To conclude. 4. It is logical to follow our visit to the controversies by looking at some neoclassical reactions and critical counteractions over the period 1965-1970. 196). The alchemist would therefore find that his métier in that world is no longer profitable. Thus in Samuelson’s model the crucial problems of ‘reswitching’ and ‘reverse capital deepening’ cannot occur by definition: the consumption good is produced by itself and labour thus the systems of production under use (each using heterogeneous capital goods) are indistinguishable. and will resign himself to pass his mysterious task on neoclassical economists. because in that world the alchemist. Levhari and Samuelson (1966). on the other hand.8-4. while the important criticisms raised in the Symposium by Pasinetti (1966a) and Garegnani (1966) in the wake of Levhari’s attempt will be analysed later (Paragraphs 4. 42 published an article in which he pretended to have proved that ‘reswitching’ is not possible at the economy level for indecomposable techniques. ‘reswitching’ is possible in economies where all commodities are basic (an indecomposable technology matrix) as well as in economies where not all the commodities are basic but for at least one. who. p.

it is not possible in an economy as a whole where all commodities are basic.44 Because Levhari’s article gave rise to the Symposium. 103). for a given rate of profit. p.L) Levhari (1965. Levhari (1965. that is w=1. p. the technique will be chosen according to which one provides the lowest price (analogous to the highest wage). and because the issue of ‘reswitching’ became publicly acknowledged in this Symposium. and prices are expressed in terms of labour commanded. Levhari himself acknowledges this conjecture to Samuelson. Wage-curves represent different systems of production in terms of real-wages (w/p=1/p) and the rate of discount =1/1+r). In terms of Sraffa that means that all commodities are basic. Levhari assumes an economy where all capital is circulating and is in stationary equilibrium. 71 . The input-output matrix representing this type of economies is indecomposable. 5). in what follows the main argument carried out by Levhari will be very briefly reproduced. this cannot happen with the whole matrix.43 His ‘whole matrix’ refers to a technology wherein every commodity enters directly or indirectly in the production of all the commodities of the economy. Each technique differs from the other in at least one activity (method of production). To simplify matters assume that there are only two techniques. we can write the price equations as follow: pa  a0 ( I  a ) 1 pb  b0 ( I  b) 1 The condition for a to be chosen is that pa  pb . 103 n. Row-vectors a0 and b0 refer to the labour input associated with each technique respectively. then switch to another technique and. as we reduce it further. So while Levhari recognises the possibility of ‘reswitching’ for single industries. so a0 ( I  a) 1  b0 ( I  b)1 then a0 ( I  a) 1 ( I  b)  b0 and. a and b which produce n basic commodities. return to the old technique. So. all commodities are basic in the sense of Sraffa (so technological matrices are indecomposable). As is well known. with some manipulation we can finally get: a0 ( I  a) 1 (a  b)  b0  a0 43 44 (4.Chap te r 4 – A first phase in the con tro versies though it is quite possible that we could use some technique of producing good i at a high rate of interest.

L’) Levhari (1965. an issue we are not going to deal with in this study. or we are indifferent.). then this inequality must continue to hold. that either technique a is in use and b is never to be chosen ((a-b)x  0) or that the economy switched from technique a to technique b but a is never to return ((a-b)x  0). In Levhari’s article these inequalities are written as strong inequalities. 46 Levhari (1965. 511). pp. p. let us introduce the last part of Levhari’s argument. 72 . Levhari believes. Levhari attempted to prove this by finding sufficient conditions such that a0 ( I  a ) 1 (a  b) be a monotone non-decreasing function of . there exists a semi-positive vector x such that either (a-b)x  0. 105).L) – the inequality showing that system a is at least as profitable as system b – by the semi-positive vector x*: a0 ( I  a) 1 (a  b) x*  ( b0  a0 )x* 45 (4. If a is not to return. for two positive indecomposable matrices. p. This Levhari has already demonstrated in his proof of the nonsubstitution theorem. Levhari’s next step was to multiply both sides of (4. then pb  pa or what is the same a0 ( I  a) 1 (a  b)  b0  a0 . the sign of (a  b) . 100-102.. or (a-b)x=0. 104-105). (a-b)x  0. As “all the elements of a0 ( I  a) 1 are monotonic decreasing functions of ”45 the elements composing the difference (a  b) must be negative or zero to ensure a0 ( I  a) 1 (a  b) to be a monotonic non-decreasing function. This crucially depends on the sign of the difference between matrices a and b.e. i.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies If techniques switch from a to b. Levhari then introduces a key condition for his whole argument: Now. Before discussing the validity of this conjecture. as speculated by Levhari.46 The economic meaning of the above condition is expressed by Levhari in the following way: there exists some activity level x such that we need more circular capital of all goods either with a or with b. 47 Levhari (1965. p. which has been precisely the issue of the Symposium. but we have here introduced them as reported by Pasinetti (1966a. See Levhari (1965 pp. who had been told by Levhari that that had been a misprint. 104).47 The above mathematical condition would ensure.

49 The first part of the article is again devoted to his counter-example which invalidates Levhari’s lemma. See Pasinetti (1966a. n. 49 Pasinetti (1966a. as Pasinetti himself recalls in his article. is function ().8. 540-510). p. and hence ’() must be negative at the switch point. 73 . 51 Pasinetti (1966a. p. since the example which “satisfies all the requirements of Samuelson-Levhari analysis can yield results that contradict their conclusions”. therefore. Pasinetti also provides a “direct logical criticism” of that construct. we need that () be monotonic decreasing in  so as a not to return. his illustration “is sufficient to disprove their theorem”. whereas if (a  b) x*  0 . If we differentiate this function with respect to  we will get:  '( )  a0 ( I  a) 2 (a  b) x* Hence if (a  b) x*  0 . ‘reswitching’ would. had already presented a refutation of Levhari’s lemma at the Rome Econometric Society meeting by providing a counter-example which rebutted the ‘nonreswitching’ theorem.Chap te r 4 – A first phase in the con tro versies The left side of the above inequality is called by Levhari function (). So if Levhari’s argument were true. *). 48 Finally if (a  b) x*  0 . p. 4. () is monotonic increasing. take place only for single goods production but not for an economy as a whole. () is a constant. Now we turn to see whether Levhari’s argument is valid by visiting some key works gathered in the Symposium. who. Note. 50 Pasinetti (1966a. Pasinetti (1966a) Levhari’s – and Samuelson’s – conjecture was firmly contested by some participants of the controversies in the already introduced Symposium. 510). Thus Levhari attempted to minimise the critical strength of ‘reswitching’ and. Levhari then argues that in such case where system a has been used at a given  and then at a higher  (lower r) the economy switches to system b. once more. pp. actually zero. 503. In the article Levhari (1965. in 1965. to rescue received theory. 105) makes a mistake when he establishes that ’() is either monotonic increasing or decreasing according to whether (a-b)x*>0 or (a-b)x*<0. What is relevant. instead.51 Albeit this would have sufficed to show that Levhari’s argument is totally unfounded. The first participant in the capital theory debates that replied Levhari was Pasinetti. 50 Pasinetti concludes in the first part that. Pasinetti’s contribution to the Symposium is an improved version of that paper. () is monotonic decreasing. however. p. 48 Every element of  a0 ( I  a ) 2 is negative. that in Pasinetti’s example the underlying technological matrices are not indecomposable. 505).

when the switch takes place. Pasinetti (1966a. If the column responsible for the switch – which we have referred to above – is multiplied by the zero elements of x*.53 Therefore the exercise of comparison between two series of prices after eliminating the components which actually make them differ is totally absurd. And if precisely that column is eliminated which contains all the differences between the two matrixes. which are multiplied by the non-zero column of (a  b) . we may. 511). it will become impossible to tell which of the two is more profitable.9.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies In the discussion about Levhari’s mathematical construct. This is the one trivial case for which Levhari’s argument would stand. Then “the sign = would apply to all the relations”54 in (4. “Thus”.52 However in order to get Levhari’s theorem right.L) must hold at least for one sector. and to the false conclusion * that matrix a and b are indifferent for all levels of the rate of profit. the matrix elements which are responsible for making all prices change. Pasinetti explains. and again the sign = would apply to all the 52 53 Pasinetti (1966a. which will contain both positive and negative elements. because of having multiplied inequality (4. it is necessary that the elements of the semi-positive vector x*. we have seen that. 512). But the implications of such a trivial case are quite strong.L’). therefore. Garegnani (1966) Garegnani (1966) demonstrates that if technique a is more profitable than b then the strong inequality in (4. p. then this will simply eliminate the column. as Pasinetti argued. Pasinetti shows that Levhari’s theorem is false. 54 Garegnani (1966. be all zero. have multiplied by zero one or more relations in (4. p. But. The same reasoning can be applied when it is technique b at a higher  that is the most profitable. the elements of the matrix corresponding to the difference between both of them are all zeroes except for those located in one column.L) by a semi-positive vector x*. 558). 4. hence getting 0=0 in their place.L’) to which the strong inequality previously applied. are all contained in that nonzero column. […] Levhari’s device would thus lead us to the straight equality ( a  b) x  0 . for two given techniques that differ in only one activity. p. 74 .

after multiplying the matrix containing the elements of the differences of both techniques (matrices a and b) by a semi-positive vector x*. 58 What Garegnani shows is that. 59 See Garegnani (1966. p. it has to be admitted that there is no logical ground on which supporting Garegnani (1966. at r’ or r’’. Thus. and Bruno et. 57 Garegnani (1966. for two given techniques (a and b) which produce. Garegnani uses function (r) instead of ().10. p.L’). pp. a zero constant) does not exclude that producers may switch back to A as r rises beyond r '' . 561. that nonzero elements of vector x* be “applied to at least one of those industries for which. proved wrong. Then. thus contradicting Levhari’s lemma i. therefore. or (a-b)x=0. Remember that Levhari uses =1/1+r.Chap te r 4 – A first phase in the con tro versies relations in (4.e. p. they conclude. but at r=r’’ b is so. 4. Samuelson’s conjecture – as Levhari recalls in his article – is. as well as in any single industry. assuming that r’<r’’ and that at r=r’ a is the most profitable technique.” 55 75 . the resulting vector (a  b) x*=y* can exhibit elements such that one of them is negative and the other positive. al. 58 Garegnani (1966.55 So the validity of Levhari’s theorem requires “stricter conditions”. 9). [since]  ( r '')   ( r ') and the fact that (r) is a monotonic function (in this case. 558 n. which establishes that there exists a semi-positive vector x such that either (a-b)x  0. Garegnani in addition points out which Levhari’s “main error” is. once more.56 namely. Garegnani concludes. once it is shown that those phenomena can exist either in decomposable or indecomposable technologies. 558). emphasis added). (a-b)x  0. 543). where these authors give Solow credit for having correctly “pointed out that one cannot in general find a semipositive vector x such that either (a-b)x  0. cannot “be proved since it is false”. (a-b)x  0. The meaning of the 1966 Symposium: A turning point in the controversies Both Pasinetti and Garegnani end their respective contributions to the Symposium by drawing attention on the implications of ‘reswitching’ and ‘reverse capital deepening’ for the marginalist theory. 558-589. two commodities. then. false: ‘reswitching’ (with all its implications) can exist in economies where techniques are either indecomposable or decomposable. 559).59 Levhari’s argument is. 56 Garegnani (1966. the methods of systems A and B are not indifferent”. or (a-b)x=0. p. (a-b)x  0. emphasis in the original). or (a-b)x=0. (1966. as did Levhari. p. say. (a-b)x  0. In fact Garegnani shows that Levhari’s proposition. and both techniques differing in two methods (activities).57 Despite that these results would have sufficed to invalidate Levhari’s conclusion.

Pasinetti (1966a. An example is precisely the case of the oral controversy between Garegnani and Samuelson then crystallised in the 1970 paper. that order may change as prices and distribution change.62 Both authors refer to Hicks (1965). However. 565). Nevertheless. p. Birner (2002. By the way. on the other. which would then be taken as the measure of capital intensity: the ‘return’ of a technique shows that any such measure. Both authors raised very strong concerns about the admission of these results by the hitherto literature. once we correctly recognize how the value of capital goods depends on distribution.) 61 60 76 . Thus. 2) writes that when he asked Paul Samuelson (in an interview in (cont. whether we consider any single industry or the economic system as a whole. 26. 516). in the literature on the controversies. the very possibility of ordering the techniques according to the proportions of capital to labour slips from our hands. p. above). this chapter. 62 Garegnani (1966. p. even if it could be found.61 Garegnani. which establishes that there is a negative relationship between the rate of profit and the quantity of capital per labour. n.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies the traditional premise of marginal theory. this does not mean that previous to that date ‘reswitching’ has not been used by critics as a powerful instrument of critique of the marginal approach nor does it mean that neoclassical economists have not realised their implications. 564. switches of techniques due to changes in the rate of profit do not allow us to make any general statement on changes in the ‘quantity of capital’ per unit of labour. there does not seem to be total agreement on this. or it may require a higher ‘quantity of capital’ per unit of labour. who argues that “the whole theory of capital seems to have been caught in the trap of an old mode of thinking”. p. We have written ‘long gestated’ reply to Samuelson partly because the reasons we have already seen (see n. p. In the same direction Garegnani (1966. at any given state of technical knowledge. The new technology may require a lower ‘quantity of capital’ per unit of labour. that importance arose in the wake of the 1966 Symposium. 107. 154. 514) concluded: The general conclusion is that. Recently. Pasinetti (1966a. emphasis added) thus concluded: In reality. p. would lead to contradicting the principle of an inverse relationship between rate of interest and capital intensity. concludes that those results were admitted only “sporadically” and that. For the reasons we explain in this chapter.60 On the one hand we have Pasinetti. as he also pointed out in his long gestated reply to Samuelson’s Surrogate note. whether ‘capital’ is measured in terms of value or in terms of any chosen physical unit. And no way-out of this difficulty can be found by relating capital per man to some magnitude independent of distribution. the mainstream approach to economics has ever had the tendency to regard those possibilities as “exceptions”. the present author would like to call attention on the issue about when ‘reswitching’ became publicly important in the debate.

except that this is the way economists have always been accustomed to think. Cf. as is Samuelson. It is hard to go along with Birner’s position once it is realised that. 77) of Samuelson.5 above for the consequences that Samuelson’s surrogate would have created among the discipline and how that situation was viewed by Joan Robinson). 516-517). p. despite the fact that “reswitching is only mentioned in a footnote of [Garegnani’s] manuscript” (Birner 2002. by relying on the support of Samuelson’s answers to Birner’s questionnaire.. preface) clearly points out the classical political economy had already been ‘submerged’ by the marginal revolution at the end of the nineteenth century. because on that principle has been erected the dominant theory of distribution. p. p. 64 The word frankly was italicised for reasons that will appear clearer immediately below (See Samuelson’s reactions). also Keynes (1936) p. pp. 66-83) is that.63 On the other hand Garegnani points to the problematic consequence that traditional theory would undergo had those negative results been admitted. as Birner (2002.e. however.65 April 1989) whether or not the latter had read Sraffa (1960) in the time of Garegnani’s visit to MIT in 1961-62 and whether or not Garegnani had orally used ‘reswitching’ in his critique of Samuelson. which had already been sketched in the 1961-62 MIT manuscript. precisely in the Symposium of 1966 owing to Levhari’s 1965) is not under dispute (see this chapter). in the criticisms raised by the Italian economist against one of the most important representative of the marginal school in the world. 65 Garegnani (1966). are far-reaching. 77) – which is. But he was rather emphatic in his recollection that ‘reswitching’ had not come up in Garegnani’s oral criticism”. The “way economists have always been accustomed to think” to which Pasinetti refers is the marginal theory.Chap te r 4 – A first phase in the con tro versies Pasinetti affirms that the traditional principle (inverse relationship between demand for capital per worker and the profit rate) has been taken for granted without any justification. This may allow us to envisage what could happen to the discipline if the problematic issues like ‘reswitching’ and ‘reverse capital deepening’ had been frankly admitted:64 [T]he consequences of admitting this. (See Paragraph 4. Actually what Birner wants to argue in that passage. 411) of section II of (1970a). the only one reason other than Samuelson’s answers pointed out by this author for justification of his thesis – the ‘reswitching’ case is. That ‘reswitching’ became publicly important some years after 1961-62 (i. 77) conjectures. actually illustrated by Garegnani at the outset (p. 565. viii. just to argue in the same terms as does Birner. ‘reswitching’ “did not play a major part in Garegnani’s criticisms” (p. But this is radically different from saying that ‘reswitching’ “did not play a major” role. as well as in the chapter referred to the controversy between Garegnani and Samuelson (Birner 2002. 63 Pasinetti (1966a pp. “Samuelson replied that he did not want to say that (at that time) he understood all the things that were in this cryptically written book. As Sraffa (1960. by the way. 77 .

recently. The problem of capital undermines even the idea of thinking of that building (and re-building) of those columns because there is no appropriate – even less definable – material for such a construct. whatever its empirical likelihood.11. 227).66 4. 2003. the 1966 Symposium. who argues that the Symposium “was the central point of the controversy and which is inexplicably ignored by the authors” (Pasinetti in Pasinetti et. Reswitching. emphasis added. Neoclassical reactions: Samuelson (1966a) The 1966 Symposium conclusively showed that the traditional principle establishing a negative relation between the rate of interest and the ‘quantity of capital’ could not be postulated in economies with heterogeneous capital goods. indecomposable or decomposable. 582. 67 Samuelson (1966a). BrunoBurmeister-Sheshinski. However. Morishima. It is therefore hard to conceive of an explanation of profits determined by the scarcity of capital considered as a factor of production. Defeat was then admitted by defenders of marginalism.67 One of these possibilities for Samuelson is that there would no longer be any guarantee to ensure an inverse relationship between consumption per capita and the interest rate. as represented by Samuelson. there exist serious problems for the theory. Thus Samuelson (1966a). p. However.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies What Pasinetti and Garegnani unveil in their respective interventions in the Symposium is that the columns upon which the edifice of the dominant theory has for so long been built are not so strong. In fact the lack of this issue in their review immediately provoked the reactions of Pasinetti. the importance of the Symposium in the Cambridge controversies is indubitable. p. Samuelson totally minimises the relevance of ‘reswitching’ and ‘reverse capital deepening’ by pointing to a problem of a ‘dynamic stability’ in steady- For the reasons we have here examined. This recent exchange on issues which took place around forty years ago reveals the current importance of setting in perspective both the Cambridge debates in general and. specifically. as it was once believed. al. Cohen and Harcourt (2003) seem to have forgotten the turning point of this episode by completely ignoring it in their accounts of the controversies. 66 78 . does alert us to several vital possibilities. An authoritative neoclassical view. admits that ‘reswitching’ is a theoretically possible phenomenon: Pathology illuminates healthy physiology. admitted this by pointing out that in the light of the results of the Symposium. Let us see now how this admission has appeared soon after the Symposium. in the well-known “Summing up”. Pasinetti. Garegnani merit our gratitude for demonstrating that reswitching is a logical possibility in any technology.

which is balanced by a subsequent perpetual reduction (rather than increase) of consumption as a result of the drop in interest rate. as it is the theory relying on factor’s substitution what is shown to be unfounded. would be brought about: in the presence of ‘reverse capital deepening’.g. according to supply-and-demand analysis. when referring to the consequences of ‘reverse capital deepening’. 69 68 79 .Chap te r 4 – A first phase in the con tro versies state equilibria. actually tried to judge the validity of the criticisms. 310. as Samuelson stated to the practitioners of the discipline? As we shall see below. when Samuelson writes: “[Due to the possibility of reverse capital deepening] after sacrificing present consumption and accumulating capital goods.70 Did successive theorists appraise and respect “the facts of life”. 68 Quite incomprehensibly. none the less. p. (See Chapter 3. a situation showing savings exceeding investment would need a rise in the interest rate so as to increase investment to the level of the former. p.69 Samuelson of course is taking for granted that accumulation is created by “sacrifice of current consumption”. emphasis in the original). the facts of life. increasingly relying on the empirical ‘importance’. we must remind ourselves that scholars are not born to live in an easy existence. A similar position can also be found on p. Samuelson’s position is not isolated: a similar attitude can be found in e. about how the accumulation of capital. and a surplus (rather than sacrifice) of current consumption. in the light of ‘reverse capital deepening’. This argument is what we have illustrated above by introducing the meaning – and their importance in economic terms – of ‘reverse capital deepening’ and ‘reswitching’. his admission does not mean abandoning the theory. Samuelson (1966a. Ferguson (1969) is a good example (see below). the new steady-state equilibrium can represent a rise in interest rate!” Interestingly. rather than to respect them – let alone to accept them. Then it is no wonder the way in which the famous MIT economist receives the criticisms: If all this causes headaches for those nostalgic for the old time parables of neoclassical writing. 70 Samuelson (1966a. so as to minimise their relevance. Samuelson writes: going to a lower interest rate may have to involve a disaccumulation of capital. should be to lower that rate. 581. that is by full employment savings. and even Samuelson himself. 579. Burmeister (2000). and appraise. 578). whereas the tendency in the savings-investment market. 583). p. Despite the fact that Samuelson admits the problems for the theory as a result of the definitive critiques raised in the Symposium. some Samuelson’s disciples. He does not seem to be worried. p. We must respect. Samuelson (1966a.).

258. xvii.g. below. placing reliance upon neo-classical economic theory is a matter of faith. have had the common tendency to associate those results. 574. 80 . that marginal 71 72 Samuelson (1966a) p. on the contrary. it is a matter of persistent misunderstanding that the damaging implications of ‘reswitching’ on the conception of capital as a single magnitude (which in a sense might be seen as a problem of ‘aggregation of capital’) – which is what actually is relevant for the long period supply and demand theory – is misread as damaging the ‘aggregate’ production function. Despite the fact that this version of the theory has clearly been shown to be theoretically secondary in Garegnani’s critique of Samuelson as far as the building of the roots of received theory is concerned. when admitting the possibilities of irrefutable discoveries like ‘reswitching’. Ferguson recognises that the Cambridge criticisms are valid. not a theoretical. These misunderstandings were mostly created by neoclassical economists – though sometimes owing to responsibility to some critical economists like Joan Robinson. Further neoclassical reactions after the Symposium: Ferguson (1969) We deal with Ferguson (1969).71 a term that in the controversies was associated almost exclusively with the aggregate production function ‘version’.12.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Needless to repeat that Samuelson. Chapter 5. nevertheless. as if the results of the controversies would have only affected a ‘less rigorous’ version of the theory. p. Later on the book Ferguson also claimed: “The crucial point to emphasize is that the validity of neoclassical theory is an empirical. Blaug 1974). But the point is. but its importance is an empirical matter that depends upon the amount of substitutability there is in the system. At the outset of his book. This does not mean. which undermine the ‘aggregation’ of capital. Until the econometricians have the answer for us. Ferguson 1969. that he has not attempted to minimise the significance of the criticisms of the theory he tried to shield: [The validity of the] Cambridge Criticism is unquestionable. question” (Ferguson 1969. because he reduces the problem of the basic principle on which the theory relies to an ‘empirical’ matter. and others (e.72 4. with a problem of ‘aggregate’ production function: in fact they characterise these results as elements raising ‘suspicion’ of the neoclassical ‘parables’. emphasis in the original). Firstly. because this author has been one of the most notable representatives of the “neo-neoclassicals cooped up by the surrogate function” to whom Robinson refers in some above citations. Cf.73 The terms in which Ferguson disputes the relevance of the Cambridge criticism are hard to accept. 73 Ferguson (1969) p.

the ‘Symposium on Capital Theory’ appeared in the Quarterly Journal of Economics. 76 Ferguson (1969. since what matters is the empirical probability to derive ‘well behaved’ demand curves. not just a problem of “amounts of” as Ferguson believes. we must look forward to the “econometricians’ answer” in order for us to assess whether or not one can rely on the traditional supply and demand functions. Ferguson dodges the problem at all. 74 81 . pp. because. and it then became quite apparent that the Cambridge Criticism. purely empirical regularity”. However. p. p. What is behind the consideration of the two facts mentioned above regarding the data of the theory. 72) pointed out.76 These facts (alternative methods of production and tastes and preferences) allow deriving logical deductions resulting in inverse relations between quantities of ‘factors of production’ used and their rates of remuneration. D’Ippolito (1987). Although we do not deal in detail with this drastic consequence that the theory would undergo if the ‘empirical’ foundation were adopted. a basic principle that. The ‘empirical issue’ as viewed by these authors is not examined in this study. hence in the downward-sloping demand curves. had long been taken for granted since the inception of marginal economics at the turn of the twentieth century (see Chapter 2 above). When the theory meets insurmountable theoretical flaws some of its defenders appeal to the empirical way-out. where experiment is impossible and the material is so highly complex as generally to limit considerably the weight one can attribute to any observed. Yet this extreme opinion is quite different from some other Ferguson’s views on the debates: Shortly after this [Samuelson’s ‘surrogate’ article] was completed. 75 Thirdly. 75 The ‘empirical’ route in the wake of the controversies was not only taken by the neoclassiclal side. To try to ascertain the probabilities of ‘perverse’ behaviour would mean to try to ascertain the probabilities of ‘non-perverse behaviour’ that is of downward-sloping demand functions for factors. This general principle is what is so necessary for the theory rather than the ‘empirical probability’. 2004. this attempt would imply a “drastic change in the nature of the theory”. as Garegnani (1990. see Ciccone (1996) and Petri (2000. must be accorded more careful consideration. etc. it is worth noting that the problem is the “substitutability” itself. still the downward-sloping demand functions were basically derived from logical deductions.Chap te r 4 – A first phase in the con tro versies economic theory is not grounded on the probabilities of having downward-sloping demand curves for factors. in turn derived from self-evident facts. The substitutability principle is a basic premise of the theory. being necessary for the theory. because “that shift of basis would occur in a field like economics. 252-255). For a discussion of these works. See Mainwaring and Steedman (2000). is that these data might directly and/or indirectly guarantee the substitution of factors as distribution changes. as I call it. since we have preferred to focus on the neoclassical reactions against the critical Cambridge. We also have to note that some critically oriented authors in the controversies somehow attempted to ascertain the ‘probabilities’ of ‘reverse capital deepening’ in some heterogeneous capital-goods models. In the meantime the author proposes to “have faith” in the neoclassical theory. it is worth noticing that all these attempts appeared soon after the controversies in capital theory. xvii).74 Secondly.

p. which are what Samuelson means when this author speaks of “bidding in markets”. p. This passage is reproduced in Ferguson (1969. “a fall in the rate of profit would lead to a shift in technique such as […] the technique [chosen] must become more capital intensive. 251-270. 80 Ferguson (1969). But the critiques of marginal theory precisely showed how groundless is to postulate a general principle by relying on ‘factor substitution’ when one of those ‘factors’ is measured in value terms – capital. On the point Ferguson introduces. Hicks (1965. 250.81 Ferguson (1969. 81 Cf. To logically derive such functions it is indispensable the factor substitution principle. but Ferguson does not provide any convincing argument in order for economists to have faith. or neoclassical. That principle is the ultimate foundation on which the determination of “rewards in quasi competitive markets” relies – namely ‘well behaved’ supply and demand functions. 259. 79 Ferguson (1969). Ferguson (1969). Ferguson’s purpose is to show that “a small deviation from Samuelson’s [surrogate] model” will result in a “slope of the frontier [which] does not give the capital-labour ratio [of the economy]. which would act through the data of the theory. it is hard to go along with Ferguson here as well. in cases where both factor proportions are equal and by comparing two straight-line wage curves. but the absolute magnitudes of the coefficients differ. xviii. 444-445. Ferguson later introduces “Samuelson’s authority” again in his book 79 when he attempted to give the criticisms “more careful consideration”. namely: that the sole way to envisage a process of “bidding in quasi-competitive markets” is by invoking marginal. We are not arguing that basing economic principles on ‘faith’ is unfair. 153) pointed out that. Clark models. Samuelson (1966b). I shall adhere to generalized neoclassical approximations (…) somewhat limited factor substitution can fail to have some of the simple properties of the idealized J. However. p. 259-260. pp.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Anyone who happens to come across Ferguson’s reaction grounded in terms of ‘faith’ would probably have expected for another kind of ‘more careful consideration’ of the ‘Cambridge Criticism’. Ferguson refers to Hicks (1965) pp. and his only attempt to that purpose is “to invoke the weight of Samuelson’s authority as represented by [the following words]:”77 Until factors cease to have their rewards determined by bidding in quasicompetitive markets. Hicks had emphasised this case “so strongly”. since Hicks’ chapter Ferguson refers to actually treats those cases in which the relative factor proportions in the ‘corn’ and in the ‘tractor’ industries may differ. This is called by Ferguson the “Hicks case” because. pp. vi) as a ‘flyleaf’ and in p.” Hicks justified that by arguing that as the profit rate falls the technique (cont. according to Ferguson. 148-159.”80 The small deviation alleged is that the relative factor production coefficients are the same in both sectors (Samuelson parable).) 78 77 82 . supply and demand functions and not any other alternative. emphasis added). pp.78 These words reveal an embedded belief of neoclassical economists.B. p.

since this technique is depicted by a linear curve. 83 . if the absolute magnitudes are the same in both sectors. then. Of course. now if a second linear wage curve is introduced. However this is a more general condition. though apparently involving a heterogeneous capital good. a straight-line wage curve actually means production of ‘corn’ by means of ‘corn’ and labour. the consumption chosen must account for a higher “capital-capital coefficient” and a lower “labour-consumption good coefficient” – in other words. emphasis in the original).82 Ferguson’s conclusion is that “the factor price frontier may not yield the neoclassical relation between factor proportions and relative factor prices even in the absence of reswitching. assume an economy where cc lc  ca la 4 c c K .”83 Of course ‘reswitching’ cannot arise in Ferguson’s model. 260. to which is of course equal both to cc c and to a . then the capitallabour ratio of the economy will be equal to the relative coefficient and. again.” 82 For the sake of the argument. And this is valid for any number of techniques whose wage equations are linear. In fact since in Samuelson’s model la L the relative factor production coefficients are equal in both sectors. Hicks (p. equal to the relative ratios of quantities of labour in both industries. will be constant as distribution changes because the price of the capital per labourer is constant throughout. as we have seen above with regards to Samuelson’s parable and the criticisms raised by Garegnani. as r falls. dr la ca K  . lc L which is equal.Chap te r 4 – A first phase in the con tro versies In Samuelson’s parable the wage-curve is w  1 cc (1  r )  la la hence  dw cc  . will be equal to cc  la    – that is a physical magnitude multiplied by la  lc  a value magnitude. A linear wage curve actually implies that the consumption good is produced by itself and labour. Then c la  a lc L la la and from this equation we can obtain: 83 cc la  ca lc la la so that c l K 4 c a . Hicks here does not realise that he is actually dealing with homogenous systems since. The capital-labour ratio of the economy. cc/la must increase. in Samuelson’s model. For linear wage curves there is no heterogeneous model at all. 154) considers cases in which relative factor proportions are not equal in both productive sectors but minimised the relevance of the possibility of ‘reswitching’ by pointing out that “such cases [could be regarded] as exceptional. Now Ferguson by introducing his “Hicks case” argues that cc K  . which is. (In the example it is being considered lc la only circulating capital). L la lc Ferguson (1969. p. moreover. when absolute magnitudes differ. In fact. it is actually implied that the same consumption good is being produced and therefore.

”84 Ferguson recognises this. Is there. 85 84 84 . but manifests his concerns by pinpointing “that economists may be unable to make any statement concerning the relation of production to competitive input and output markets. This was the way chosen by Ferguson to rebut the criticisms of capital. neoclassical economists. the problems for the theory not only involve ‘reswitching’ but also ‘reverse capital deepening’. p. 1969. that is. on which Ferguson is relying his ‘faith’? The point is that the criticisms destroy or invalidate the ‘simple relation’ from which “all neoclassical results follow”. his analysis fails from its very beginning. p. 269). p. Independently of Ferguson’s explicit minimisation of the ‘reswitching’ results for the theory. as Ferguson himself put it: “the lower the rate of interest. and. p.87 As noted. he justifies his claiming on his own faith. “until the econometricians have the answer for us. 87 It is worth noticing that the ‘empirical’ way out was emphasised by some commentators from the neoclassical side such as Blaug (1974). 265). which does not seem to disturb so much Ferguson. once again. Ferguson like many. but the latter is “an econometric question susceptible of resolution in a probabilistic sense. 39-40.88 Ferguson (1969. a more robust or more sophisticated version than the ‘fairy tale’. p. the greater the capital intensity of production. 1969. as he thinks of having dealt with heterogeneous capital goods economies with systems of production described by linear wage-curves. 269). a problem. Ferguson believes that economists can still use the theory under attack. 88 See n.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies good is produced by itself and labour. 270). 252). if not most. Yet. pp. as we noted above. above. this chapter. show that “the Clark parables may not hold” (Ferguson. as we have previously seen. “point up a definite potential weakness of simple neoclassical theory” (p. downplayed the relevance of ‘reswitching’ by incessantly pointing out that the Cambridge criticisms only “invalidate the simple Clark fairy tale” (Ferguson. Ferguson’s reactions are very extreme. as it is pretty evident the great deal of work done in econometrics relying on the traditional theory – and all this as if nothing had ever happened in the theory of capital after the post-war controversies. 270).” 86 Hence. And yet it would seem that many practitioners of the discipline have taken up this route. 74. 86 Ferguson (1969. perhaps. He argues that all that is needed to establish the neoclassical parables is the existence of a sufficient substitutability establishing capital intensity uniqueness. Ferguson (1969.”85 However. placing reliance upon neoclassical economic theory is a matter of faith”. and that “simple neoclassical results concerning the relation between production and input and output markets may not hold” (p. 269).

see Barro and Sala-i-Martin (1995).93 The fact that the factor substitution principle is undermined by the phenomena of ‘reswitching’ and ‘reverse capital deepening’ implies that. they can only exist if and only if the net consumption output is produced by itself and labour. emphasis added. And this problem is quite independently of considering the systems of production as represented by an aggregate production function. so this single commodity would be produced by itself and labour. while the other one being labour. below.13. 422. hence the question which arises is: What kind of representation of different systems of production – involving heterogeneous capital goods – can these individual production functions be? Yet these striking conclusions seem not to disturb the authors who are devoted to use aggregate production functions in e. p. 425. 90 89 85 . that is to say. modern growth theory.Chap te r 4 – A first phase in the con tro versies 4. Thus. because of the same reasons of why it cannot exist an aggregate function for the economy as a whole. with ‘reverse capital deepening’ there would not be any support to the claim that it is only by postponing See Paragraph 4. has been a “feature of only some versions of the traditional theory. The basic premise on which the traditional theory ultimately relies is the ‘belief’ that a fall of the profit rate cheapens the more ‘capitalintensive’ method either by means of the alternative methods of production available (direct factor substitution mechanism) or by changes in consumer substitution in favour of the ‘more capital intensive’ goods. in systems of production with heterogeneous capital goods.g. It must be said that nor can individual production functions exist such that they could represent the production of a single commodity by means of labour and ‘capital’.92 In addition it is interesting to note how Garegnani uses the implications of ‘reswitching’ and ‘reverse capital deepening’.”91 In fact what is being argued by Garegnani is that marginal theory reduces its core explanation of distribution to the functions of supply and demand for productive factors. Garegnani’s 1970 critique of the marginal theory Garegnani’s (1970a) article not only provided the analytical elements necessary to definitely dismiss Samuelson’s surrogate function that “somewhat baffled everyone” – as Robinson pointed out89 – but also shows that expressing the production conditions in terms of a production function (either aggregate or individual)90 with ‘capital’ as one of its arguments. 91 Garegnani (1970a) p. a fall in the rate of profit may well bring about a fall rather than a rise in the capital-labour ratio of the economy. when he argues that [Theoretical] examples do not seem to indicate that the conditions in which a fall of r results in a relative cheapening of the less capital-intensive productive processes are any less plausible than those in which the opposite would be true. 92 The particular target of attack at production functions pursued by Robinson in the controversies is analysed in Chapter 5. 93 Garegnani (1970a).5 above. This appears to undermine the ground on which rests the explanation of distribution in terms of demand and supply for capital and labour.

Paragraph 4. 426. and therefore the forces “in order to explain distribution”. the variables determined by the theory would have served as theoretical guides of actual ones. had one ever detected in any real economy. emphasis added. In this regard. the consequences for the traditional marginal theory of having upward-sloping demand curves for productive factors. 86 . as Garegnani concludes no such instability of an economy’s wage. Garegnani (1970a).94 We have seen above (Chapter 1). the ‘belief’ being ‘supported’ by two groups of the data: the alternative methods of production. This ‘belief’ is what is behind the demand aspect of the problem of capital. In the light of these results it is clear that the traditional belief does not work. ‘Belief’ is here taken to mean that. Garegnani’s article also provides a clear-cut example 95 which conclusively demonstrates that “a change. we must rely on forces other than ‘supply’ and ‘demand’ […] [T]he theory becomes implausible once it is admitted that this principle [that a fall in r would always raise the proportion of ‘capital’ to labour in the economy] is not always valid. independently of the supply aspect of the problem of capital (that is the illegitimacy of conceiving of capital in value terms). as 94 95 Cf. as theoretical possibilities. p. the idea of an inverse relation between demand for ‘capital’ and the profit rate would somehow have been taken for granted by the theory. as supply and demand for each respective ‘factor’ will not be brought into equality. however small. 424. 96 Garegnani (1970a). and it is this belief that is totally destroyed by ‘reswitching’. The point is that. an infinite value of. extreme values of distributive variables.and interest-rates has ever been observed. in order to explain distribution. Garegnani asks. and explanations in terms of supply and demand would have resulted in a plausible theory of distribution. because as the profit rate falls it is a less ‘capital intensive’ method which is chosen (because it is the most profitable at the new distributive situation) with a lower value and not a higher ‘quantity of capital’ per worker. a zero wage rate with unemployment. with ‘reswitching’ and ‘reverse capital deepening’ we would have to admit. in the ‘supply’ or ‘demand’ conditions of labour or capital (saving) may result in drastic changes of r and w. p. p.11 above. labour resulting in a zero wage rate. Samuelson’s reactions in the Symposium. The natural conclusion is that. and demand for. 426). But. and tastes and preferences of consumers. for example. the interest rate – owing to an upward-sloping demand function for ‘capital’ – would imply zero wages: it is hard to conceive of any explanation establishing an equilibrium in terms of supply of.97 What. then to an economic theory which cannot describe any actual economy? In a word. say.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies consumption (that is by saving) that investment can be brought about.”96 These drastic changes in distribution actually mean that the profit rate might fall to zero while wages to infinite or vice versa. Garegnani in this passage talks about the existence of a neo-classical ‘belief’ in the traditional principle. 97 Garegnani (1970a.

In fact Bliss takes Garegnani’s argument as if the latter would have referred to a short-period. 101 Bliss is referring to the well-known work of Debreu (1959). of Walras). to a completely different notion of the equilibrium owing to the not satisfaction of the condition of a uniform rate of returns on the supply prices of the capital goods. according to Bliss. 100 Bliss (1970) refers to a “momentary equilibrium”. Cf. in a Debreu’s economy. pp. Chapter 6. who in a brief comment99 completely misunderstood the theoretical background in which Garegnani’s argument relied – the conception of capital in value terms.98 must be sought in a theoretical approach alternative to marginal theory. who.101 But. below. the temporary and intertemporal equilibrium versions associated with the names of Arrow. the data regarding capital as the heterogeneous vector of the several capital goods (the ‘Walrasian’. 87 . would have considered the same required conditions as did Garegnani for the existence of equilibrium. We call ‘neo-Walrasian’ equilibrium the notion of equilibrium professed by the modern versions of the theory (temporary/momentary equilibria or intertemporal equilibria) which. In Debreu’s words: 98 99 Garegnani (1970a).e. as Walras did. as we shall explain in Chapter 6 below. he shared. but the latter was contradictory with the conception of capital as a collection of heterogeneous goods. there appeared the sole reply that Garegnani’s article ever received for the record from the neoclassical camp. Bliss) conceived. These conditions are continuity of demand functions and convexity of technology. because it indicates that latest versions of the theory (i. Bliss (1970). take as datum of the equilibrium the physically heterogeneous capital goods of the economy instead of the single magnitude ‘capital’ as it was traditionally conceived (with the exception. particularly Paragraph 6. Garegnani 1976a): this Walras was very concerned on. Bliss’s 1970 intervention. Debreu. while ‘neo’ refers. Hahn. instead. It was the case of neoclassical Christopher Bliss. where the author after having criticised the traditional method of long period equilibrium proposes a ‘temporary’ equilibrium by relying in a Walrasian conception of capital.100 Bliss argues that one of the most striking Garegnani’s conclusions – the possible non-existence of equilibrium in the savings-investment market due to ‘reverse capital deepening’ – is contradictory with Debreu’s results. thus renouncing to the traditional long period (or normal position) method.14. which had characterised received theory until recent decades (Cf. Prelude to the shift to neo-walrasian equilibria and the beginning of a second phase in the debates In the same issue of The Review of Economic Studies where Garegnani (1970a) published his critique of Samuelson. ‘arm’ of that label). The label ‘neo-Walrasian’ is not misleading.. 424-427. capital is conceived as the heterogeneous capital goods vector of the economy. because despite he had conceived of capital as a heterogeneous vector. It is important to notice that Debreu and other economists of mathematical orientation pursued the research programme commenced by Hicks in Value and Capital (1946[1939]). as is well known. neo-Walrasian equilibrium.Chap te r 4 – A first phase in the con tro versies Garegnani strongly stressed. so to speak.1. on the other hand. 4. the traditional method of the normal positions.

emphasis added). 104 Bliss points out that “Garegnani’s error” would have been due to an overlook of “expected changes [of prices] over time” (Bliss 1970. On the one hand. mineral deposits. In this situation. They include the capital of the economy at the present instant. 107 Garegnani (1970b. p. 439.105 But. now existing.107 Unfortunately Garegnani’s arguments were not replied by Bliss or by any other marginalist scholar at that moment. p. as it has been described in the present chapter. Chapter 2.. Paragraph 2. Although it cannot be assumed that communication between the different sides of the debates has been an easy matter for the participants. 438). However.e. 102 103 Debreu (1959).Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies The total resources of an economy are the given quantities of commodities that are available. there are short-period analyses where the attempt to avoid these inconsistencies (by referring to heterogeneous capital goods) leaves room for the doubts about the fruitfulness of the whole approach. equipment. pointing out that he had never treated in his argument the discussion of such short-period equilibrium and argues that: the demand and supply explanation of distribution seems faced. 105 Bliss (1970. 74-75. i. the background of the controversies was completely another one. 438). price changes “over time” is a matter that the traditional approach of long period equilibrium took into account. On the other hand. 104 Bliss’s misunderstanding is finally completed by his conclusion: the major fault in [Garegnani’s] argument is the illicit importation of long-period equilibrium theory into the analysis of a short run situation. 438.9 above. all the land. 88 . p. Cf.102 Then. Bliss goes on arguing that “[w]ith more than one capital good current demand for investment goods will depend not only upon present prices but also upon future expected prices.” 103 Evidently Bliss is taking Garegnani’s argument regarding investment as if it were a short-run problem of expectations. buildings. by two alternatives. the temptation might be strong to use the short-period approach to provide some rationalisation of results which derived their plausibility only from a long-run analysis. pp. p. at the present moment. All these are a legacy of the past. they are a priori given. Bliss (1970). 106 Garegnani (1970b). there is a long-run analysis undermined by the inconsistencies of the notion of capital it must use. In fact Garegnani immediately replied 106 Bliss.

Counterexamples were soon produced and reswitching became an issue instead of a curiosum. In this regard. On the other hand. The zenith of the debates It has been seen at the end of the previous chapter the theoretic problems surrounding the notion of capital for the theory. it is worth mentioning Samuelson’s influence in stimulating the debates: on the one hand we have seen how Samuelson’s Surrogate article resulted in what we can here call one of the cornerstones of this controversy – namely Garegnani’s criticisms of the neoclassical theory. In this chapter. Since the notion is false.109 108 109 See Paragraph 4. Robert Solow: Samuelson had the notion that universal diminishing returns. as Garegnani points out. These kinds of exchanges reflect what then turned out to be an increasing attitude by defenders of the neoclassical theory soon after “more careful consideration” – to use Ferguson’s wording – the phenomena of ‘reverse capital deepening’ and ‘reswitching’ raised in the debates was “accorded”. on the other hand Bliss is pointing to a way-out to those negative results. 89 . we have again Samuelson – though indirectly – trying to defend dominant theory but at the same time bringing about one of the most notorious results of these controversies for the discipline as a whole – the 1966 Symposium. it may seem.15. the proof contains a slip. yet this kind of neoclassical reaction. “the temptation to use the short-period approach”. Solow (1983) p. is raised for a purpose. But sometime in 1964 Samuelson planted it in the mind of David Levhari. would rule out the paradoxes of reswitching even in models with many capital goods. in the light of Samuelson’s view after the Symposium)108 of the results of the controversies.11 above. we have seen how defenders of neoclassical theory reacted against the threats of ‘reswitching’ and ‘reverse capital deepening’. It is interesting to look at a retrospective view on Samuelson’s role in the controversies from another authoritative voice from MIT. in 1965 Levhari published a proof of the notion.Chap te r 4 – A first phase in the con tro versies because communication is per se a question with many ramifications. Solow’s participation in the controversies is analysed. With such a gardener it would be hard not to grow something. in addition. like the Giffen good. not respect. the reason of which Garegnani has already envisaged clearly in his brief rejoinder. 4. As we shall see in Chapter 6 in much more detail. was in fact the way out adopted by the neoclassical approach to the inconsistencies of the traditional theory unveiled in the controversies by the critics. 184. […] [T]he notion is false. In Chapter 5 below. together with the assumption that every line of production involves every commodity directly or indirectly. While it is worth noticing the misunderstanding surrounding Bliss’s appreciation (though. and his struggling for an alternative. which no doubt obscures the argument for an otherwise better communication.

hitherto sporadically. The debates were gaining momentum and defeat from the neoclassical side had to be admitted – though. 90 . Then. In the next chapter we make a step back in the history of the debates. as noted. The lesson to be learned from the Symposium is that ‘reswitching’ and more generally paradoxes in neoclassical capital theory cannot be ruled out in any one of the imaginable worlds proposed by neoclassical authors. Chapter 6 returns to the consequences of the tensions created by the implications of ‘reswitching’ and ‘reverse capital deepening’ in a second phase of the controversy. We shall discuss whether and how the question of capital appeared in the literature prior to 1960. The critically oriented side of the controversies was winning the debate in the theory of capital by showing how groundless the idea of conceiving of capital as a factor of production is.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies The Symposium in paradoxes in capital theory definitely revealed that the seeds planted by Samuelson – to follow Solow’s analogy – were totally sterile to fence in the traditional theory from the critical Cambridge.

1. 1 . Lastly some neoclassical reactions raised against Robinson critique in the mid-1950s (Champernowne 1953. Now. It has been said above that we have preferred to place at the centre the results of ‘reswitching’ and ‘reverse capital deepening’ and their implications because these phenomena allow us to clearly grasp the relevance of the debates. Solow 1955. Introduction1 Joan Robinson was a renowned participant in the Cambridge controversies for various reasons. Robinson’s critique: Aggregate production functions as her main target Joan Robinson starts her critique of the neoclassical capital theory in her vividly written article of 1953. While the first point is focused on why Joan Robinson might have chosen the aggregate production function as her main target of attack.2. Harcourt (1972) chapter 1.Chapter 5 A step back in the debates: Joan Robinson’s role in the controversies and first neoclassical reactions 5. 5. In that article Robinson accuses the neoclassical economist of not We have considered for the part regarding Robinson the following works: Robinson (1953). which has been considered as the kick-off for the socalled Cambridge debates in capital theory. One of them is her famous and provocative article published in 1953 by the Review of Economics Studies. Swan 1956) will be reviewed. and (1962a). See also Naqvi and Robinson (1967). The object of this chapter is not to provide an exhaustive account of all the contributions of the Cambridge economist but rather to critically focus on two points that we have deemed the most relevant for the problems of capital. (1956) chapters 8-10 and “Diagrams”. in the light of the implications of the critique. we turn to the analysis of the first arguments raised in the Cambridge controversies by Robinson and her immediate contenders. 1956. the second one is to explain as simple as possible Robinson’s analysis in order to understand the kind of criticisms she raised and how far those criticisms actually touch the heart of the problem of capital as we have discussed above (Chapters 3 and 4).

when she says that in equilibrium the value of capital could be measured either in terms of costs or current purchasing power or future earnings but to do so. Cf. But Robinson does not focus her critique on the supply and demand mechanisms as such.C). 5 Robinson (1970a) considers that neither Wicksell nor Walras have a theory of distribution at all. whereas the main purpose of the production function is to show how wages and the rate of interest are determined by technical conditions and the factor ratio. and that the sole neoclassical theory (‘neo-neoclassical’ in her terms) is that represented by the aggregate production functions.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies making clear.e. Robinson believes that the production function is a very poor instrument to analyse process of accumulation occurring in real life because the “pseudo-production function”4 (and hence the neoclassical theory. together with labour. 106). p. “in what units C is measured. and indeed theoretically secondary. 81. “the relative prices of the factors of production [i.” 2 The aggregate production function Robinson criticises is a very particular. above. p. 4 Originally Robinson labelled the production function as a “pseudo-productivity curve” (Robinson 1953. we have to begin by taking the rate of interest as given. below. 6 Robinson (1953.6 The citation makes it clear that for Robinson it is only in the aggregate production function representation that the problem of valuation might arise. 3 Within this construct it is assumed that the physical composition of capital is instantaneously adjusted by the supply and demand mechanisms.” In fact. according to Robinson)5 could be only used in conditions of long run equilibrium. representation of the neoclassical long period equilibrium theory. of an aggregate production function. Relying her arguments on Keynesian premises. but this view is devoid of a critical examination of the problematic issue entailed by a value measure of 2 3 Robinson (1953). Joan Robinson’s critique of capital has almost always been associated with the problem of considering this ‘factor’ as an argument. p. Robinson is criticising the fact that in real life the composition of capital cannot be instantaneously adjusted due to problems of uncertainty and expectations underlying problems of effective demand. See Paragraph 5. when he teaches students the production function O=f(L. In fact Robinson sees the problem of capital only limited to this representation of the long period equilibrium. Chapter 5. in which. So the use of aggregate functions already entails accepting the supply and demand approach to value and distribution.4. according to this economist. which already implies that expectations about the uncertain future must be fulfilled. 81). where both output and capital per worker are assumed to be homogeneous. but rather on the long run equilibrium conditions which must prevail – always according to Robinson – in order to use that production function. 92 . the wage and interest rates] are exhibited as a function of the ratio in which they are employed. Samuelson’s construct in Chapter 4.

7 On the other hand. below and Chapter 2. Using Robinson’s words. also. are irreconcilable. above. as we have seen above. See Appendix B. equilibrium of realised expectations in a world where uncertainty prevails implies that the capital stock composition was adjusted to the conditions expected by investors in the past. long-period equilibrium system has illustrated this issue. expectations and uncertainty. then it will be also equal to the value of capital in terms of future earnings. and all we can do is botch up some conventional method of measuring capital that will satisfy neither of them.8 According to Robinson. 8 Robinson (1953). is expected to prevail in the future. p. and of the man of words making observations about the past. Equilibrium and expectations Robinson conceives of equilibrium as a situation where the composition of the physical capital stock is instantaneously fully adapted. 7 93 . In this regard. But the corollary of Robinson’s message would be that in a state of realised expectations the theory would not be faced with any problem regarding the issue of Our visit of Wicksell’s general dis-aggregated. 5. In fact.3. but. it would seem that the strength of her arguments crucially lies in her notion of long-run equilibrium. the “equilibrium rate of profit” today (which will equal the interest rate) can be used to value capital in terms of current purchasing power. and the expectation of future profits obtaining today was expected to obtain today.Chap te r 5 – A step back in the debates the endowment of capital for the theory independently of the aggregate production function. In a world where unexpected events occur which alter values. long-period equilibrium theory. because that rate of profits prevailing today. in terms of the costs of production because that rate “was expected to be ruling today” when investment (in the past) was made. the point of view of the man of deeds. The core of her arguments on equilibrium lies precisely in those Keynesian premises encompassing issues such as investment. Thus the value of capital in existence today is equal to its supply price calculated in this manner. which characterised all her further works on the Cambridge controversy: Equilibrium requires that the rate of profit ruling today was expected to be ruling today when investment in any plant now extant was made. The heavy weight which this method of valuing capital puts upon the assumption of equilibrium emphasizes the impossibility of valuing capital in an uncertain world. when investment actually was made. 96. it is neither necessary nor sufficient the existence of any such construct to realise the necessity of a value measure for capital in marginal. making investment decisions about the future. it is worth quoting her definition of equilibrium.

and cause the additional capital to be consumed. It entails […] that the human beings in the situation are expecting the future to be just like the past. there is no difficulty in measuring the stock of capital within any one economy. in equilibrium capital can be given a meaning: Under the protection of our assumptions [of equilibrium]. This is the definition of equilibrium. 82. The problem of having one of the data in value terms for the determination of equilibrium – outside the one-commodity aggregate production function – is totally got rid of by Robinson’s analysis. as Robinson also informs us in her 1956 book. Rather. 117.9 These citations suggest that the theory would meet the question of capital (the illegitimacy of conceiving of capital in value terms to explain prices and distribution in terms of supply and demand. 9 Robinson (1953). p. 10 94 . p. 11 Robinson (1953). on the other hand the neoclassical relation between accumulation of capital and the rate of interest seems not to be put under dispute by the English economist. Moreover. according to Robinson: In a position of equilibrium […] there is a quantity [of capital] which can be translated from one number to another by changing the unit.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies conceiving of capital in value terms for the determination of a neoclassical equilibrium.11 Nevertheless Robinson criticises the neoclassical theory on the grounds that it does not provide the appropriate tools for an analysis of accumulation owing to the Keynesian problems such as expectations and uncertainty. Robinson (1956).10 Under realised expectations (and free competition) Robinson believes that the traditional adjustment mechanisms of supply and demand work properly: an accidental increase in the stock of capital above the equilibrium quantity would depress the rate of profit. 87. as it has been analysed in Chapters 2-3-4 above) only in a state of not realised expectations. p.

100: “In short. The production function.13 To show that the production function is ridiculous for an analysis of changes. To begin with it is necessary to introduce the ‘productivity curve’. labour. which has been used by Robinson since 1953. of inventions. of technical knowledge) running in historical time. As early as in her 1953 article.” 13 Cf.4. and it is expected to be at that level in the (infinite) future as well. 95 . where the profit rate is what it is because it has for a long time been so. pp. the comparison between equilibrium positions with different factor ratios cannot be used to analyse changes in the factor ratio taking place through time. we are informed by Robinson that this kind of exercises are not appropriate for an analysis of changes (of factor’s supplies. Robinson analyses the relation between output. Robinson thinks.12 As she persistently insisted. to compare two equilibrium positions on the curve due to a shift in some parameter does not tell us so much about how those positions were actually reached. has a very limited relevance to actual problems. with the difference that it is drawn for a given value of r.14 In fact it is only in equilibrium of full realised expectations and knowing distribution in advance that that construct can be built. it seems. the following critique extracted from Robinson (1974. the position of the function) is being changed. because as we pass from one point to another one. all the history (the form of the function.” 14 See Robinson (1953). 12 Cf.Chap te r 5 – A step back in the debates 5. is similar to a neoclassical one-good production function. It is in connection with the building of that construct – which is only useful to make comparisons but not for analysing changes – where Robinson’s analysis tangentially touches upon the question of capital. p. This curve. and it is impossible to discuss changes (as opposed to differences) in neo-classical terms. Outside equilibrium of full employment that construct is impossible to build.g. Understanding Robinson’s way: The ‘Pseudo-production function’ and the ‘curiosum’ The production function is considered by the Cambridge economist as her main target of attack because she believes that the analysis proposed by a production function is nothing more than an exercise of static comparison of long period equilibrium positions of the economy. That is why Robinson called it the ‘pseudo-production function’. To such constructs we now turn to. e. 85-87. distribution and technical knowledge in an economy under equilibrium conditions. 58): “The problem of the ‘measurement of capital’ is a minor element in the criticism of the neo-classical doctrines […] What they pretend to offer is nothing but an error in methodology – a confusion between comparisons of imagined equilibrium positions and a process of accumulation going on through history. capital. Robinson (1953) p.

while stretches on it represent combination of techniques. For the given r (1/0N. since Robinson is just measuring those magnitudes in a long-run equilibrium in order to compare the different positions on that curve at a given rate of interest (profits) – and hence a given wage rate.). for a given r.17 This generalisation is done by choosing the maximum rate of (net) output at the minimum costs (‘real capital’) for all feasible values of r. Any point lying both on each tangent and on the corresponding productivity curve represents equilibrium positions. The pseudo-production function is the result of joining all equilibrium segments for every r. represented by the slope of the tangent drawn) the equilibrium is reached only at the stretch where  and  are used. 15 96 . a determined productivity curve. 16 Robinson (1953. 1956) also calls it factor-ratio curve. Yet it is not a problem. etc. the other techniques being not eligible at that level of r. The productivity curve This curve represents. for each feasible value of r.15 For the case illustrated. The ‘pseudo-production function’16 Robinson draws. Notice that the two aggregates are expressed in different numéraire. or as Robinson preferred to call it. On this see Salvadori (1996) and Pasinetti (1996). 17 Once we have got these productivity curves corresponding to the whole possible values of r we must draw tangents from every level of the wage rate onto the former curves.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Figure 5. each corner on the curve represents the use of a determined technique (. Then by generalising the productivity curves for all values of r we may draw what Robinson calls the ‘pseudo-production function’.1. “real capital”. the relation between the (net) output per man (vertical axis) and the value of capital goods per man (horizontal axis) in terms of wage-units (labour commanded).

we are told by Robinson. reflecting the relation between output and capital per man. without showing points out of equilibrium in some stretches. p. on the “gestation period” and “length of life of equipments”. In fact Robinson in her analysis prefers to take the traditional premises for granted in that the value of capital (in terms of output) in the economy will be higher as the rate of profit falls due to the use of a ‘more mechanised’ technique. 2) critical view on Robinson. this depends. Again.” Cf. which supplies us with the relationships It is the measurement of capital in wage terms associated with non profitable techniques at a given r. But. Robinson (1953. The construct of the pseudo-production function crucially depends on the relation between one productivity curve and the next one as r is modified – a relation which depends crucially on the reaction of the cost of equipment to differences r. In fact by drawing on the analysis provided by the wage curves. on the other hand.2. In fact the stretches of the productivity curves that do not belong to the pseudo-production function18 do not bear any economic interpretation and so it is hard to understand them with the analytics of Robinson. that “it is reasonable to suppose that the most mechanised techniques are the most sensitive to the rate of interest. 19 Regarding the relation between one productivity curve and the next. her productivity curves are entirely arbitrary. My reinterpretation of Robinson’s construct in the light of the analysis of wage curves will allow us to draw the pseudo-production function.19 It is possible to overcome the deficiency around Robinson’s construction of the pseudo-production function. which might result misleading for readers otherwise. also Pasinetti’s (1978. so that the family of curves fans out laterally [to the right] as it rises. since in Robinson’s analysis the production conditions of capital goods associated with each technique are not specified.Chap te r 5 – A step back in the debates Figure 5. which becomes profitable for lower rates of profits. 186 n. Robinson’s ‘pseudo-production function’ It is the present author’s opinion that in the above analysis clarity is lacking. 18 97 . we can precisely use the analysis we have described in Chapter 3 regarding wage curves and the wage-frontier (or envelope). To such a purpose. 94) comments that “there is little to be said a priori” and she adds. p.

We shall analyse two cases: i) a typical neoclassical case. at some r just a little 98 . We know that at that r both and  are equi-profitable. and distribution. So when we have a switch point we have inclined stretches in Robinson’s pseudo-production function. All these techniques show the neoclassical property that the value of capital in terms of output corresponding to each technique increases as r falls. The pseudo-production function is built by joining each value of real capital to the corresponding level of output: for example the minimum value of real capital is associated with level D.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies between value of (net) output per man.) i) Neoclassical case. a pseudoproduction function can be rebuilt. Assume that equilibrium conditions entail r=r then we can univocally pass from our measure of capital in terms of output to the value of capital in wage units (real capital).3. Every production techniques have values of productive coefficients such that D>0. This is the example with which Robinson exemplifies her analysis. At that point the value of capital in terms of output varies from the value associated with  to the value associated with . Four techniques are considered.2 above) Figure 5. As we move from right to left onto the ‘envelope’ (from higher to lower values of r) we find that it is only the only profitable technique. (cf. As we pass to measure capital in wage units. and the maximum real capital at that rate of profit corresponds to C. capital per man. and ii) ‘reswitching’ (Robinson’s ‘curiosum’. which is higher than the former. Paragraph 3. Then.3 we draw four wage curves. Wage curves and the ‘pseudo-production function’ (neoclassical case) On the right hand side of Figure 5. real capital increases from the value associated with  to the one associated with at w=w: the value in wage units is changing due to a change in the composition of the stock of capital from  to  at that given r and w. depicting techniques .

Figure 5. then we can see that at an even lower r. For example the stretches 3-3-3 or 2-2 in figure 5.Chap te r 5 – A step back in the debates higher than r the value of capital in wage units decreases as w at which is now profitable is higher. thanks to this re-interpretation a ‘pseudo-production function’ can be built on the grounds provided by the analysis of choice of techniques in terms of the wage curves. the relevant curve is the following ‘pseudo-production function’ (and not Figure 5. What is interesting to notice is that Robinson. because they are arbitrarily drawn on the basis of having already assumed that a determined technique is cost minimising at a given r while others are not. So when we have dominance of a technique over the rest we have a horizontal bar in Robinson’s pseudo-production function. Robinson might have drawn them with a determined slope to show that those positions are not of equilibrium (and that the techniques associated with those points are not profitable) but no objective method can support that drawing.3) Thus. In fact this is the general case.2 are not drawn on an objective basis. Again we can see that ‘real capital’ at that r increases due to the change in the composition in the stock from capital-goods to capital-goods. is also profitable and so on. the value of capital in wage units varies from the value associated with to the value associated with . even until 1969 (Accumulation of Capital. that is C.2):20 20 The important thing to note is that we need not draw the productivity curves which are not competitive. At a lower r such that and are both profitable.) 99 . thus showing a horizontal bar joining different real capital levels to the corresponding level of output associated with . The value of capital goods (like the value of any other produced commodity) might not have a univocal relation with changes distribution.) when (cont. The ‘pseudo-production function’ (from Figure 5. This analysis may be extended to lower values of r where  is the only profitable technique. Because Joan Robinson tried to show the equilibrium positions. 3rd ed. technique continues to be the only profitable one and therefore the value of capital in wage units decreases. For higher w.4.

5.6-5. Cf. Robinson (1969) “Diagrams” p.7. Wage-curves (‘reswitching’) and the ‘pseudo-production function’ (from Figure 5. Robinson’s ‘pseudo-production function’ amended ii) ‘Reswitching’ of techniques (implying ‘reverse capital deepening’) or Joan Robinson’s “Curiosum”. continued to use the productivity curves as her main instruments.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Figure 5. There are two techniques such that D=0. D<0 (W>W and R>R for every r and w) Figures 5. 100 . 427. the ‘reswitching’ debate had got already started.6) When r<r1  is more profitable than and this is reflected by a horizontal bar measuring different values of capital in wage units corresponding to net output level A.

1. Conclusion The main question which emerges from our brief analysis of some main criticisms raised by Robinson is: Does her particular way of criticisms point to the heart of the Cambridge controversy. 106.21 This is what Robinson labelled a “Ruth Cohen’s curiosum”. When r=r2 we have ‘reswitching’. The form of the factor-ratio curve resembles a U-like line shape.5. but does not reach r2 yet. Robinson (1970a. 309) recognises the influence of the famous Italian economist. provided that entrepreneurs are changing the composition from  capital-goods to  -ones. When r increases further. Robinson (1970a.  is the only profitable technique. When r>r2 only  is the most profitable technique. Also. This phenomenon. the function will have a triangle-like form. At that rate of profit real capital falls. however. n. p. since the clues for her arguments had been “picked up from Sraffa’s Preface to Ricardo’s Principles.” 22 101 . k) move anticlockwise. ‘Real capital’ increases as a result of dividing a constant value of capital in terms of output by a lower and lower w. 21 Figure 5. p.22 5. Basically. real capital falls. we have co-existence of techniques. p. We have a horizontal stretch here representing a higher real capital as a result of dividing an increasing value of capital by a lower and lower w.Chap te r 5 – A step back in the debates Since the value of capital in terms of output when  is used is a constant value for any level of r so that as r falls. was ascribed to the agricultural economist Ruth Cohen by Robinson. since  is also profitable at this r and will continue to be so as that r increases further and further. 309) later revealed that it was a “personal joke” and recognised that the critical arguments raised in her article were a “preview” of Sraffa’s (1960). As the rate of profit rises from its minimum (r=0) to its maximum (Ra). points (y. When r=r1. as we have seen in above chapters? What emerges from our review of some Robinson’s works is that she might have chosen the aggregate production function as her main target of attack because she might probably have thought that it was a didactical way to spread the critique against this We could draw the pseudo-production function in terms of capital measured in terms of output by plotting in the output-capital (in output terms) diagram the values of capital and net output obtained from the wage curves. later known as ‘reswitching’.8. The ‘pseudo-production function’ in terms of capital per worker in terms of output Robinson (1953).

“The student of economic theory […] is instructed to assume all workers alike. even when Robinson herself faced what later became known as ‘reverse capital deepening’ and ‘reswitching’. the more mechanised the technique of production. though these phenomena rather appear to be general. See Pasinetti (1962).23 Lastly.” Robinson (1953) p. In fact the famous Cambridge economist seemed to downplay those phenomena when she speaks of a ‘curiosum’. 24 E. 1962a). in the hope that he will forget to ask in what units C [capital] is measured. 25 And this could also be appreciated by recalling the post-Keynesian literature which arose in the wake of the controversies.g. Moreover. In particular. if more capital had been accumulated in the past. Kaldor (1955). Robinson (1956. p. and to measure [labour] in man-hours. on the relation between ‘capital’ and the rate of interest Robinson says: The production function has a very limited relevance to actual problems and after all these labours we can add little to the platitudes with which we began: in country Gamma. the smaller the amount of employment would have been. he is told something about the index-number problem involved in choosing a unit of output. The necessity of capital in value terms to determine a uniform rate of profits and the key role of a downward-sloping demand curve for that factor are aspects that are not discussed by Robinson. In addition. her remarks reveal the intuitiveness of her arguments used to broaden the view of traditional theory by Cambridge (UK). 100.25 Notwithstanding Robinson’s harsh criticisms of the neoclassical production function for analysis of changes. 81. and so sloppy habits of thought are handed on from one generation to the next. and Hicks. the level of real wages would probably have been higher and the technique of production more mechanised. Hicks (1937). where the road builders use wooden shovels. it is hard to find in her writings an actual critical assessment of the supply and demand mechanisms which underlie the core of the theory she attempted to deny. The ‘curiosum’ case is regarded by Joan Robinson as unimportant and hence dismissed: Cf. given the amount of capital accumulated. and then he is hurried on the next question.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies particular representation of the long period neoclassical equilibrium. he has become a professor. and. 23 102 . Robinson’s arguments also reveal her revival of Keynesian economics at a time when some neoclassical scholars. Before ever he does ask.26 where the relation between ‘accumulation’ and the interest rate is taken by the critical economist on traditional grounds. such as Samuelson. relatively to labour available for employment. 26 Robinson (1953).24 had started to bring Keynes in the neoclassical theoretic framework. Modigliani. which destroy the belief on an inverse relation between capital and the rate of interest. in Robinson’s analysis it is hard to find the analytical roles of capital conceived as a single magnitude for a determination of equilibrium of supply and demand.

112. 5. The last possibility may lead to the “paradoxical result” that the relation between the output per capita and the capital per worker might be negatively related. p. p. 109. occurs when to identical values of ‘real capital’ there correspond different levels of net output per worker – hence the ‘pseudo-production function’ is actually not a function for some intervals of real capital. To some of those first reactions we turn in what follows. “[T]he use of [the chain index] removes the more glaring difficulties in the way of regarding aggregate output as a function of the amounts of labour and capital employed. we shall show that the rewards per unit of the factors are once again given by the partial derivatives of the aggregate production function if stationary state conditions with perfect competition are assumed. the first “anomaly” pointed out by Champernowne can be seen in the horizontal bars of Joan Robinson’s ‘pseudo-production function’.28 Champernowne wants to overcome Robinson’s puzzle in order for the production function to determine distribution. p. Champernowne’s endeavour was to provide a kind of measure of capital alternative to the one adopted by Robinson (‘real capital’). Champernowne’s (1953) contribution David Champernowne’s 1953 article was the first echo of Joan Robinson’s kick-off of the controversy. 28 Champernowne (1953). For such a purpose Champernowne built a ‘chain-index’ measure of capital.29 To form Robinson (1953). and ii) that the production function may not preserve a 1-1 relationship between output per capita and capital per capita. 106. if capital is measured in “Joan Robinson units” (in wage units). 115.27 Nevertheless this economist did not bestow much importance on this line of criticism.” Champernowne (1953). for two “anomalies” might arise.Chap te r 5 – A step back in the debates A good deal of exploration is needed before we can say whether the above [curiosum] is a mere theoretical rigmarole. In Robinson (1956) she totally dismisses the pursue of this line of critique: “[The curiosum is] a somewhat intricate piece of analysis which is not of great importance” (1956. it was due to Robinson’s vivid intervention that later on stirred up a series of replies from the neoclassical camp. In terms of figure 5. In particular. p. on the other hand. 29 Cf. 27 103 . where only one technique is being used but bear different values owing to different values of r and w. The second anomaly. The ‘chain-index’ is a device to measure the capital as an argument in the production function in such a way as to make it possible to determine the distribution of product between wages and profits through the partial derivatives of that function. are two possible phenomena: i) that the same physical capital equipment may have different values at two different ‘stationary’ positions due to the two different values both of w and of r associated with each position. 1).2 above. because he considered her measure as being “not convenient”. n. her preferred way of attacking neoclassical theory reveals weaker in comparison with the arguments of ‘reswitching’ and ‘reverse capital deepening’. or whether there is likely to be anything in reality corresponding to it. What Champernowne rejects.6.

and that b) of two techniques. unless we rule out cases in which a lowering of interest rates can cause the introduction of techniques with a lower productivity than those used up until then. what is interesting to notice is that. p. calculated at those w and r. Champernowne believes. In fact. he says that “intuition suggests that the excluded case is unrealistic” (p. (Cf. in order to construct the ‘chain-index’ for Champernowne’s purposes. there is no univocal relation between the costs of capital and changes in distribution. emphasis in the original. In other words the only way to keep the value of capital per worker to move in accordance with the traditional principles. Appendix D. Champernowne’s chain-index attempts to eliminate. could never reappear again at another value (or range of values). 32 Champernowne (1953). chapter 1). 33 Champernowne (1953). 31 Intuitively. 30 104 . he had to assume that:32 a) if any one technique. having been the most (or equi.31 But how sure was Champernowne that his device would have removed the “glaring difficulties”? As we may now speculate. 130. so as to achieve a one-to-one relation between output and capital per worker. This construct has the apparent advantage. The present author owes the ‘zigzag’ analogy to Harcourt (1972. of removing the effects of the interest rates on the costs. is by a priori assuming that phenomena like ‘reswitching’ (point a) and ‘reverse capital deepening’ (point b) not to arise. It is interesting to note that it is “intuition” what might have led the author to Champernowne regards the long period equilibrium positions as ‘sequence of equilibria’ or as historical sequences of equilibria. which are equiprofitable at a given r. However. pp. below) So. in general. Champernowne’s construct cannot avoid “anomalies”. and thus “paradoxical results” or “anomalies” would be prevented. 115-116.most) profitable technique at a given r (or range of values of r). the ‘zags’ (the first anomaly in Robinson’s pseudo-production function) and to diminish the slope of the successive ‘zigs’ (the second anomaly) in Robinson’s production function.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies such a construct in a historical sequence. 119). in the light of ‘reswitching’ and ‘reverse capital deepening’. he produced examples of “exceptional cases” which may lead to “paradox results” (a triple valued production function) and points out that formidable difficulties arise.33 Nevertheless Champernowne recognises that these cases cannot be ruled out “on logical grounds”.30 the index of capital is increased at each step of the sequence by the proportion in which the cost of capital at current w and r at the end of the step exceeded the cost of capital at the beginning of that step. so to speak. it is the one with higher output per capita and higher capital per capita that is more profitable (and hence chosen) at lower values of r.

37 On the issue of capital as an argument of the production function. my translation.Chap te r 5 – A step back in the debates believe in the basic. do not necessarily involve that the change in the rate of profit has to stop at the level in which two techniques are competitive. 31). But changes of techniques. both the same technique and the same physical quantity of output would be compatible with more than one quantity of capital. premise of the theory i. what is more. foundational. which needs a physical measurement of all factors. 5. emphasis in the original. as Petri (2000) has suggested. after measuring capital in value terms: given the quantity of the other factors of production. when economies change techniques (production systems). this construction as well as the so-called aggregate production functions. and so will the value of capital. Hence the chain index is not a “reliable measure of the quantity of capital”. that Champernowne’s chain index assumes that changes of technique involve the rate of profit stopping at a value in which two techniques are equiprofitable. seem to address the basic question on which the issues of the measure of capital turn: whether or not the marginalist supply-and-demand functions provide a logically valid explanation of prices and distribution. Garegnani (1960) has pointed out: “[If the] aggregate of capital goods is measured so as to appear as a variable ‘quantity of capital’ when distribution changes. (Petri 2000. p. the ‘form’ or kind of capital goods will in general change. p. 85. which should have resulted in the ‘marginal products’. 1956) and Swan (1956) Besides Champernowne’s response to Robinson’s first complaints are two further authors from the neoclassical side who contributed to the debates in the first exchanges. that “a gradual fall in the rate of interest would entail increases in the quantity of capital per head”34 namely – a belief that it has for long been taken for granted in the discipline at the time of the first blows of this controversy. In conclusion.e. Anyway the attempts of these responses have not been to solve for the problems that Champernowne (1953).” (p. did not give his “unrealistic” case (that is Robinson’s Ruth Cohen’s curiosum) much importance. it does not. its partial derivatives. are however an entirely different construction from the traditional meaning given to the ‘marginal products’ of the theory. it is worth noticing that Champernowne. 36 It might be also pointed out.7. As Champernowne calls the different long period equilibrium positions of the economy. And it turns out to be so. 35 34 105 . since the theory needs this ‘factor’ susceptible to changing form but keeping the quantity unaltered. the ‘chain-index’ measure of capital appears to be designed only to analyse a process of capital accumulation by constructing the path of sequence equilibria35 by already assuming that the neoclassical theory works in the ‘correct way’. assumes the explanation in terms of supply and demand to be valid.) Once a production function is so built. like Joan Robinson.36 On the contrary.37 Finally. in disequilibrium adjustment processes. Further replies to Robinson: Solow (1955. And. 118. it will no longer be possible to establish a univocal relation between the quantity of capital and the physical quantity of the product. however.

Solow (1955) considers that the process of production can be visualised as a process divided into two stages: first. So this is sufficient to say that Solow’s attempt does not answer to Robinson’s problems.C2) into the form Q=H(L. p. if the first function can be logically built. Solow (1955.g. and more important.38 We consider here Solow (1955 and 1956) and Swan (1956). there is no justification why C1. a priori. 106).C1.K). can be a function of the physically heterogeneous capital goods. in terms of real economies. p. this relation between the index and changes in distribution is not univocal. I have no explanation to offer”. In fact Solow himself commented on this: “[the condition of independence of the marginal substitution rate from the quantities employed of the other factors] is not satisfied [e.C2))=H(L. they may stand for amounts of original factors such as (different types of) labour or land both measured in physical units. Solow believes that “there is a whole class of situations in which the condition may be expected to hold […] For example. but merely to dismiss them altogether. p. 106) comments: “Apart from the comforting thought that at this late date no one should be surprised to find price-quantity dualities. say. an ‘index’ quantity of capital.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Robinson had raised. there are two points worth noticing: first. in the literature. However. p. for such a defence. Nevertheless. p. Solow justifies this procedure by resorting to Leontief’s weak separability condition (the substitution rates between the factors that one wishes to aggregate – here C1 and C2 – must be independent of the quantities employed of the other factors. Secondly. we can imagine that C1 and C2 to be two kinds of equipment for generating electricity which is then used in further production. he cannot overcome the difficulty entailed by the fact that any index of capital depends on changes in distribution. Solow (1955. Rather. if K=(C1. we might come across many instances in which “prices and quantities indices come in pairs” (Solow 1955. Secondly.(C1. if we keep in the long period tradition). Certainly.) 106 38 . the consumption good. is produced by means of labour and the services of the several capital goods through function F. 103. Solow’s attempt is nothing but a clear-cut defence of the validity of the marginalist propositions. which is necessary and sufficient for the representation of the production function of the form Q=F(L. 247. p. in the article. then Q will be a function both of labour and of an ‘index’ of capital. Q. cases like this are known as the so-called ‘duality problems’. 39 Solow (1955). To conclude. 106) recognises that his “quantity-of-capital-index” is also the result of the transformation of a “price index”. But the problem is well another one: if capital is expressed in value terms (and it has to be so. C1 and C2. However. C2 should be services of heterogeneous capital goods.”39 However. there is of course no guarantee to “justify [Solow’s] statement: that there is a perfectly definite and (cont.C2). 40 Actually. technical substitution possibilities will depend on the number of drivers available. Solow dismisses it by considering the problem raised by Robinson as “antiquated”. see Solow (1955).40 Moreover. 101 and Robinson (1955). K. it is very unlikely to find situations such that the marginal substitution rate of C1 and C2 are independent of the quantity of labour. 103). p.] for one-ton trucks and two-ton trucks. [E]lectric power would be an index of the capital inputs” (Solow 1955. To dismiss the problem that distribution cannot be determined by the marginal productivities of the factors. L).

1. is produced as if it were a single commodity which is produced by itself and labour. he regards capital as “the number of meccano sets”.41 Let us now focus on Swan’s (Swan. his belief in aggregate production function (and hence in neoclassical theory) is expressed in the following way: “Were the data available. One can at least hope that an aggregate analysis gives some notion of the way a detailed analysis would lead. actually composed of heterogeneous goods. 1956) reactions to Robinson’s 1953 kicking-off paper. We are in an economy just as the same as the one we have described at the beginning of this work in Chapter 2: the only way to save the neoclassical propositions. since it is produced only one commodity there is no indirect mechanism of factor substitution.” Solow (1957) p. there is no guarantee that capital employed in the economy (gross investment of fullemployment savings) be an inverse function of the rate of interest. In that work Solow represents an economy that produces a single output by means of itself and homogenous labour. provided a sufficient flexibility of the wage in perfect competition. investment and capital are all measured in the same technical units. In his long appendix to that article. (Cf. p. under these restrictive assumptions. no problem regarding the irreversibility of investment – another of Robinson’s concerns in the 1953 article).) 107 . In fact. n. is to assume that the output. Robinson’s concerns about capital are plainly dismissed. We have. where he attempts to apply the (1956) model to US data for the period 19091949. there is no problem regarding the measure of the marginal productivity and of the measure of capital. Of course. Savings determine investment: the consumption forgone is saved and hence to be added to the stock of capital.Chap te r 5 – A step back in the debates One of the most known works from the neoclassical side that indicates that the first exchanges much centred around aggregate production functions has been Solow (1956). 41 In Solow (1957). Full employment of labour is guaranteed. (There is. The only mechanism available therefore is the direct one: through technical conditions of production. Savings. Since technical conditions are assumed to be variable. 105). the discussion of Samuelson’s ‘surrogate’ function in Chapter 4 above. ‘capital’ can always adapt itself as to employ the whole quantity of labour of the economy. The commodity produced can be used either as consumption good or as an input of itself. This ‘meccano’ can consistent sense in which it can be said that the relative factor price ratio PK/PL is a decreasing function of the ratio of capital to labor K/L. it would be better to apply the analysis to some precisely defined production function with many precisely defined inputs. in economies with heterogeneous capital goods as between techniques of production for the consumption good. Swan (1956) is concerned with Robinson’s notion of capital as a factor of production which is accumulated over a period of time by savings.” (Solow 1955. 312. It is assumed continuous variability in the proportions of the factors necessary for producing the output. however. seen that even when the heterogeneous capital goods are measured in technical units according to the different techniques for producing the consumption good. His belief in aggregate production functions ultimately relies on the acceptance of the basic premises of the theory.

capital can be aggregated. Similar other names were used during this period in the literature concerned with economic growth models to refer to ‘capital’ in aggregate production functions such as steel. Swan (1956). 344.”45 This last statement can hardly be satisfactory. measured in that unit. p. that we can derive the direct factor substitution mechanism as a base for an explanation of the equilibrating forces of supply and demand. butter. this author considered that. However. but. p. Swan agrees with Robinson insofar as she was complaining that the neoclassical tradition contains no indication of how a technical unit may be devised in order to measure capital. they can directly eliminate them. Swan (1956). Swan takes the assumption of a “malleable”43 capital (‘meccano’ sets) as a “scarecrow” assumption in order “to keep off the index-number birds and Joan Robinson herself”. are produced as if it were a single commodity produced by itself and labour – that is Solow’s model. but he assumes that prices are constant when distribution changes. this does not overcome one of the criticisms raised by Robinson. But. Swan affirms that his device does not solve for Robinson’s problems. (Cf. is by assuming an economy where the commodities. 44 Cf. p. since Swan assumes that its price does not change. Were a technical unit measure of capital existed. according to economic profitability and technical progress. Swan (1956). does not depend on the existence of such a unit in which measure ‘capital’. This attempt too avoids the irreversibility of investment. which compose the output.) And also. whereas it is only in equilibrium that.42 Swan assumes the existence of more than one commodity. in equilibrium. On the other hand. leets. Nevertheless. p. 349. That is to say. Robinson assumes. equal to savings and investment44 – will exactly be the current output of ‘meccano’ sets. he concludes: the neoclassical theory. which concerns the nonsense of talking about marginal productivity in equilibrium. 350. in which case its marginal product equals the rate of profit. instead. and accordingly can always be measured in terms of any kind of consumption good. It is by drawing on this. the marginal product of capital. This is not very different from Solow’s (1956) model: the only way to assume that prices do not change when distribution does. when aiming at explaining the marginal variations about the stationary position. because of the problems that we have already seen when capital in value terms is considered as argument of production functions.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies adapt itself instantaneously and without costs to any form. accumulation – the level of which is. because the original meaning of marginal productivity is related to the description of the methods of production of the consumption good – measured in physical terms – according to the different combinations of the factors of production which must be measured in physical terms too. 43 42 108 . because the “natural unit of Capital is simply an equilibrium dollar worth regardless the physical variety of capital goods. 45 Swan (1956). would be equal to the rate of profit multiplied by the price (in output terms) of a unit of capital (in technical units). ectoplasm. 343. due to the assumptions on constant relative price (including that of the capital goods ‘meccano’). however. Chapter 2 above. capital could be also measured in value terms.

Chap te r 5 – A step back in the debates

To conclude on these authors we may say that the only way that neoclassical economic growth models can be rigorously formulated is in economies which produce either a single commodity by means of itself and labour (Solow) or a composite good redistributed without costs as between the different uses (Swan). Outside these instances, the shortcomings entailed by the dependence of the value of capital upon distribution cannot be shunned – nevertheless Robinson trifled with the role of a quantity of capital as a determining one for the definition of the supply-and-demand functions for productive factors.

5.8. Robinson rejoinders

On Solow’s reply: Robinson does of course not go along with Solow’s view. For her, the economist from Massachusetts does not deal with the problem of capital. She correctly asserts that his endeavour to collapse “C1 and C2” into K is rather a way to represent heterogeneous natural resources into a single index and not a way to represent capital goods because “nothing is said about the time which it takes to produce them.”46 In her reply to Solow (1955), as well as in her further interventions in the debates,47 there is no reference to the problematic attempt to derive inverse functions between factors’ remuneration and factors’ quantities (employed hence demanded), when capital is measured in value terms – inverse relationships that Solow of course takes for granted.48 On Swan’s reply: Robinson replied to Swan in 1957.49 As to the question of not having considered a single-commodity output as Solow did, but an output composed of many goods, Robinson does not accept the assumption that prices can be kept constant in Swan’s model, since, if distribution varies, prices will vary too: “Swan must come off the fence, and say which kind of good is taking as numeraire.”50 On the other hand, Robinson correctly accuses Swan of thinking of “the wage rate as being determined by the physical marginal product of labour, which in turn is determined by the amount of meccano in existence” without, however, stating the units which it is a quantity of.51 Therefore she considers Swan’s way as unsatisfactory to
Robinson indicates how the issue may be considered: “[C]ompare two machines exactly alike in all respects except that one is drawn from an economy with a higher product-wage rate. The value of the two machines is different, and the investment required to create them is different. A difference in value remains if we deflate them by the wage rate (obtaining capital in labour time), for in two economies with different wage rates the rate of profit and the rate of interest are different.” (1955) p. 247. Solow (1955) does not address his answer to that issue. 47 In particular cf. Robinson (1962b; 1970a; 1971a; 1971b; 1974; 1975a). 48 Solow (1955), p. 105. 49 See Robinson (1957). 50 Robinson (1957, p. 104, emphasis added). Cf. Samuelson’s very special case in Chapter 4. 51 Robinson (1957, p. 105). Then, some pages later, Robinson (1957, p. 108) scathingly concludes her reply to Swan in the following way: “My scheme of analysis [in The Accumulation of Capital, 1956] is intended to make it possible to discuss accumulation in spite of the fact that capital cannot (cont.) 109
46

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

answer the questions which she had raised in her 1953 paper. According to Robinson, equilibrium is obtained, for a given marginal productivity of labour, if the conditions of the amount and form of capital, the rate of investment, and the propensity to save are such as to make the wage rate equal to that marginal productivity. This is the kernel of her reply; a reply relying on Keynesian principles about investment and accumulation. In this chapter we have seen the particular characteristic that the debates took during the first exchanges between the two Cambridge. We have seen that discussions seemed to be related to find a measure of aggregate capital in order to be considered into aggregate production functions, which would in turn determine distribution. Swan’s attempt was to consider the malleability assumption (his ‘meccano’ sets). Solow, instead, plainly regards social output as if it were a single commodity produced by itself and labour. Champernowne’s attempt does not overcome the anomalies for which the ‘chain-index’ was built. In sum, in the first reactions to the issues raised by Robinson the analytical problems connected with the notion of capital for the marginal theories are still unclear, and that is why we have introduced them in advance (Chapters 3 and 4) without following the chronological path of the many issues debated in the Cambridge controversies. From a historical standpoint, our brief review over these first exchanges indicates us that participants in the controversies had to await Sraffa (1960) from the critical side and Samuelson (1962) from the neoclassical one, in order to find more clarity in the issues at stake, as shown in Chapter 4. The following chapter restarts by discussing a new phase in the capital debates, which was initiated, as I shall argue, by some neoclassical participants soon after the most salient results analysed in Chapter 4 were acknowledged. In particular we shall focus on the way in which these advocates of dominant theory received the critique; a preview of that way was indicated in Chapter 4 too, when Bliss (1970), in his reply to Garegnani (1970a), drastically changed the background on which the critique had hitherto been conducted (cf. Paragraph 4.14 above).

be measured in physical units. […] It is not really my fault that capital does not consist of pieces of meccano.”

110

Chapter 6

A second phase in the capital controversies and radical changes in the neoclassical theory

6.1. Introduction

In the previous two chapters we have seen the main discussions of the first phase of the Cambridge controversies. Especially, the attention has been focused on the results of ‘reswitching’ and ‘reverse capital deepening’, the implications of which touch upon the basic premises of the neoclassical theory of production, value and distribution (Chapters 3 and 4). The neoclassical reactions reviewed have shown reluctance to admit the consequences of the debates; in fact, advocates of the marginalist theory have tried either to find out the ‘conditions’ to rule out ‘reswitching’ or to minimise their relevance by pointing to the empirical issue. However, the first phase of the controversies clearly showed that phenomena of ‘reswitching’ and ‘reverse capital deepening’ are theoretically irrefutable as some neoclassical contenders were compelled to admit (e.g. Samuelson 1966a). In the aftermath of the Symposium and related contributions to the capital debates, dominant theory increasingly started to rely on a different notion of capital and hence of equilibrium.1 This indicates that a different phase in the controversies, in comparison with the first one, had begun. Actually, the main characteristic of this second phase was the fact that almost the majority of the core of neoclassical theorists had already embraced the theoretical path provided by Hicks’ (1946 [1939]) Value and Capital, who incorporated the Walrasian conception of capital into the supply and demand approach. The history underlying the incorporation of Walrasian capital into dominant theory cannot be developed in full detail in this book, as it exceeds the scope of the present research. However, by relying on related works,2 we have to bear in mind that both before and during the Cambridge controversies pure theoretical work related to the rehabilitation of a Walrasian conception of capital started to be further developed by neoclassical economists – most of them of mathematical orientation3 – and this is why a
1 2

Cf. Paragraph 4.14 above. Garegnani (1976a), Petri (2004, pp. 136-205). 3 e.g. Debreu (1959), Morishima (1960; 1969).

For this purpose.) is taken as a ‘factor of production’ in the same footing as ‘labour’ and ‘land’.7). from lathes to tweezers. in the following three paragraphs. might have damaged both the relevance of the capital debates and the communication between the two sides of this theoretical conflict. which was translated by Jaffé and published in 1954. Then every single capital good (from ploughs to screws. The aim of this chapter is to discuss the way in which the neo-Walrasian theorists received the critique centred around the notion of capital.6 and 6.5 Walras general equilibrium system would admit solutions whenever capital services are considered as original factor’s services. Walras treats capital as a physically heterogeneous capital goods’ vector. though still keeping the supply and demand approach to value and distribution as the only one economic theory. 2008). As has already been analysed in the literature. 1990. also Eatwell (1990). Cf. 1960. 5 Garegnani (1958. In this way. However these solutions do not take into account the fact that capital goods are reproducible and. 6.5.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies different version of the marginal theory was. 4 Walras (1954). In the wake of the controversies’ results raised in the first phase. Then we shall return to the historical aspects of the controversies in this second phase (6.2). 2008).3) and the implications for the approach of adopting the Walrasian treatment of capital and the transformation of the theory (6. available to those willing to use it and renounce to the traditional notion of capital. 1960. as we shall see in what follows. embodies a radical change in the nature of the theory. That edition was strictly alike the 4th and definitive edition of Walras’s Éléments published in 1900.4). In this paragraph we follow Garegnani (1958. 112 . But this shift. the former would only refer to exchanges and production without capital. neoclassical participants started to argue against the critics of marginal theory in terms of this version of the theory (which we call Walrasian capital). the suggested way-out to that inconsistency (6. in order to avoid the problematic issue of capital. we shall deal with Walras’s treatment of capital as a set of physically heterogeneous goods and the inconsistency with the traditional conception of equilibrium which is entailed by this treatment of capital (6. We shall henceforth refer to the 5th edition of 1926 (in French). as we shall argue. therefore. which no doubt put the dominant theory at stake. etc. Rehabilitation of Walras’s treatment of capital as physically heterogeneous goods and Walras’s inconsistency Because the modern versions of the supply and demand theory rely on a Walrasian treatment of capital we shall here briefly deal with Walras’s theory of capitalisation.4 As is well known. 6. so to speak. attempting at the same time to clear up some misunderstandings created in this phase which.2. 1990. neoclassical theorists increasingly started to rely precisely on a physical specification of capital.

e.  k 'k pk  k 'k ' pk '  . see equation Y) pk ic pk ' i  c' …  kt pt  .. Up until the 3rd edition of Éléments Walras regarded the raw materials as being consumed as soon as they were produced and not stocked in advance.)  Dk ' (k 't pt  . 14) and Sraffa (1960). Note that satisfaction of this condition implies determining a uniform rate of return on the capitals’ supply prices. .  k 'k pk  k 'k ' pk '  . i – which.. However.. where pk indicates the gross price of the i  c' ic capital good k (the price from which a net price is obtained after subtracting the amount corresponding to depreciation and insurance. must equal the demand prices of these goods (which are the capitalisation of the service prices pk).  k 'p pp  ..) (Y) (Note that the total amount of ‘capital’ i.. For the purposes of our present discussion we will use Walras’s original formula.  k k pk  k k ' pk '  . p. Further. Pk '  pk ' 113 . (Y) and (Z).  k p p p  . k’....)  Dk (kt pt  .. 1990..  k 't pt  . Pk '  k 't pt  . pk .  k p p p  .. In order to deal with the production of capital goods Walras introduced the price equations of each capital good (k.  kp pp  ....... In the 4th edition circulating capital is treated in a strictly similar way as fixed capital thus assuming for these capital goods a value of c=1. both expressed by Walras as fraction c>0 of Pk – the price of the new capital good). etc.. that is...  k ' p p p  .  k 'k pk  k 'k ' pk '  ....... Walras introduces their production into the analysis and hence the equilibrium condition of uniformity of rates of returns on capitals’ supply prices – a condition.  k ' p p p  ....  kk pk  kk ' pk '  . ). there will annually emerge out of the annual output a flow of savings (Fe(. Fe ( pb .. which. the savings (investment) is endogenously determined. etc..)) which must be equal to the investment flow namely – valued quantities of the new capital goods produced ( Dk Pk  Dk ' Pk '  .. All these conditions can be summed up in the following two blocks of equations... is crucial to the whole analysis.… ) reflecting the costs of production (supply prices). … … … (Z)6 6 Equations in (Z) are the result of equalising supply prices Pk  kt pt  .. .  kk pk  kk ' pk '  ..Chap te r 6 – A second pha se in the cap ita l controversies To be coherent with the fact that capital goods are reproducible.. We are not going to discuss the incorrect way in which Walras treats depreciation of fixed capital since his formula does not take into account the role of interest upon the very depreciation through time.. problems as to economically meaningful solutions of the system arise. as soon as production of capital goods is introduced into the general equilibrium system. that is.. part II. and demand prices Pk  .. For a clarification in this regard see Garegnani (1960. which is implicit in the equality signs of the price equations of those goods. in equilibrium. as will be underlined.

let us begin by imagining that. if only formally. (Note that the RHS of equations in Z reflect the costs of production of the several capital goods. it must account for general cases. if a theory boasts itself of being a general economic theory. owing to the arbitrariness of the endowment supply of. This being so. to understand the meaning. where the configuration of the relative scarcities of the several capital goods taken as data should have to be appropriate so as to be in accordance with the consumers’ tastes and preferences which had determined the prices of the productive services of the capital goods. because these equations establish the equilibrium condition of equality between demand and supply prices of the capital goods produced – hence uniformity of rates of return on the supply prices of those goods. To show as clear as possible that in this more general case (positive savings) there does not exist economically meaningful solutions either. k’. kk’ … k’t. k’’.… . If a uniform rate of interest were determined under savings being zero it would be a very particular case.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies where kt.7 This case. we may consider two cases.… . In this case. 7 114 . k’k’…are the quantities of productive services necessary to produce k. 2008). Garegnani (1958. must be all zero. which considers gross savings zero. Of course those effects will crucially depend on the relative Cf. kk. including also the case in which savings happen to be zero. of course. The only possible way that the rates of return could be uniform throughout the system lies in the influences of the quantities currently produced of capital goods on the prices of the capitals’ productive services. Then. would however be sufficient to argue that there is no general solution to that system because. k’p. be negative magnitudes. kp. 8 To re-establish equality between number of unknowns and independent equations let us further assume the following relation: R Dk ' Dk Since we assume i’ > i free competition of capitals’ bearers would make R increase so as to bring about effects on both the costs of production of the capital goods and the productive services’ prices. since they cannot. k’k. say. Dk’. the system is over-determined: in general Walras’s capital formation system is inconsistent and hence has no solution. according to (Y). let us assume that consumers’ tastes led up the yearly gross savings to be at zero.… . The issue which concerns us here is the economic meaning attached to equations (Z). represented in (Y) by Dk. part II. First. two distinct capital goods. 8 These transformed equations we call (Z’). we shall see that gross investment in k.) Although this system has solution. However it is interesting to consider the case when savings are positive. see below. But in general the arbitrariness of the initial configuration of the capital goods deprives their quantities from being the appropriate ones determined by consumers’ conditions. i’ > i in (Z) so that we have one more unknown. it is interesting to see whether it admits economically meaningful results. k and k’.

.9 Since those proportions in general differ. In our example. then an increase in R – an increase in production of k’ relatively to k – will make the relative scarcity of k’ rise (because k’ needs ‘more’ of itself than of the other capital good – this is the auto-intensive case k’k’/k’k > kk’/kk) hence the price of the service associated with k’ will tend to rise as well.  k k pk  k k ' pk ' ic  k 't pt  . and by keeping constant the gross prices of both capital goods.Chap ter 6 – A second pha se in the cap ita l controversies proportions of the capitals’ productive services in the production of themselves.. For the sake of completeness we can see that in the opposite case to the one referred to above (the hetero-intensive case when k’k’/k’k<kk’/kk). since it depends on the arbitrary given endowments of capital goods which are at disposal of producers in the economy..  k 'k pk  k 'k ' pk ' i ' c ' 11 Cf.  k ' p p p  . Walras (1954. we have assumed an increase in R – one may even imagine that rise further and further to the point where all the gross savings are invested in k’ alone while k approaches zero. If this is so. The ultimate result is therefore totally uncertain. as a matter of fact. If. p..  k p p p  .. At that moment. p. then the original disparity between both rates of interest will tend to widen. also Garegnani (1962. On the other hand the effect of a rise in R on the prices of the productive services of the capital goods is not univocal. in our hypothesised economy. Hence to reestablish equilibrium... and with it the productive services’ prices and hence the rate of return over the costs of production. so that the initial disparity between both interest rates will tend to narrow. 292) where he analyses the effects of the services prices of the capital goods on the costs of capital goods to be produced and their services’ supply.10 A rise in R will cause a rise in the relative costs of production of k’. Still. the initial disparity in the rates of interest will tend to narrow as we make R rise. for the effects referred to above would not exist at all (the relative scarcity of capital goods would not be modified as R changes).. 10 Let us see this point analytically: pk  kt pt  . 19). Cf. i’ will tend to fall while i will tend to rise. pk ' (Z’) 115 . the productive requirements of k’ (relatively to k) are higher in the production of k’ than in the production of k. the difference between both interest rates cannot be 9 The particular case in which those proportions might happen to be equal is devoid of meaning here. we have a further negative element in Walras’s system which undermines the equilibrium conditions he set out there – equilibrium conditions in accordance to the long period equilibrium method or normal position. Cf. In particular the sign of the effects on those prices will crucially depend on the technical conditions of production which characterise capital goods. a change in R will make relative scarcities of capital goods change. this compensation is not guaranteed. because the initial disparity between the different rates of interest might be of any width.11 Thus. Garegnani (2008).

If doubts. 282. it is clear that the data of equilibrium could not change drastically in the very process of time-consuming adjustment towards equilibrium. in lieu of “over a certain period of time [pendant un certain temps]” (as written in the 4th edition) we read.3 below. which are full of significance in this regard: “Thus equilibrium in capital formation will first be established in principle [through the tickets]. during which no change in the data is allowed. emphasis in the original). in economic terms. it is worth reading some further words from Walras. 12 116 . 1954.” 14 See Walras (1954). “pendant tout le temps que dureront nos tâtonnements” [for the entire period of time that the tâtonnements will last. and subsequently we shall suppose them to change in order to study the effects of such changes. 15 Walras (1954). Walras’s way-out to the inconsistency But Walras did not limit himself only to count the number of unknowns and equations in order to seek economically meaningful solutions of his system.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies reduced12 since the economy is run out of savings to be invested in k.) To conclude. reasonably. pp. If we regard the tâtonnements as a purely analytical device for an analysis of solutions which were to be stable. say. Then it will be established effectively by the reciprocal exchange between savings to be accumulated and new capital goods to be supplied within a given period of time. it is important to notice that in Walras’s system the data of the equilibrium were assumed not to change as the time-consuming process of adjustment towards equilibrium was taking place. (See Paragraph 6. Dk is negative. (my translation)].15 Formally one can get a solution if.3.13 6. In fact in the part regarding capital formation in the 4th edition of the Éléments Walras somehow went by intuition that equations (Z) might undergo some modifications. 282-283.14 Before discussing the meaning of such formal changes. That the capital good k would not be (re-)produced later has important implications that deserve our attention. to account for some ‘negative’ investment. 13 In the words of Collard (1973. p. the Walrasian system is prevented from having an economically meaningful solution and hence fails in its purposes. 286-295. but since Dk is gross investment it would be totally meaningless. This can be noticed when he describes the tâtonnements process (to grope towards equilibrium by trial and error): Our problem is to reach equilibrium in capital formation ab ovo […] we propose to start by assuming the arbitrary data of our problem to be constant over a certain period of time. 472): “If there is production of new capital goods [in Walras system] the approach is open to the charge of sterility [because] almost anything might happen. pp. p. namely – to substitute the inequality “  ” sign for the equality sign “=”. in the 2nd edition. Although the economy is becoming progressive. still remain. it remains for the time being static because of the fact that the new capital goods play no part in the economy until later in a period subsequent to the one under consideration” (Walras. Interestingly enough.

20-22. hence in that process of groping towards equilibrium. thus determining the rate of interest at its highest level. Marshall. Then as Walras indicates: all the remaining equations of system [Z] will be satisfied after the exclusion of those new capital goods which it was not worth while produce. Clark.17 The fact that a capital good’s demand price falls below its supply price (costs of production) actually means that such capital good is not to be produced thereafter. However. thus implying changes in prices and quantities in the economy throughout. p. Walras is coherent when he excludes from production the capital goods whose demand prices would fall below their supply prices. by conceiving of capital as a single magnitude. attempting at determining the economy’s normal position) then it is at least not clear the suggested way out put forth by Walras himself in the 4th edition too. Cf. Walras keeps the system within the method of the long period equilibrium characterised by a uniform rate of return on costs – and this is what is done when system (Z) is modified by letting only those capitals whose supply and demand prices equal at the highest rate. if the purpose of Walras was precisely to determine a long period equilibrium (i. The consequences of this change cannot be overlooked: it will of course entail a change in the composition of the (initial) capital stock of the economy. could endogenously determine the composition of such a stock. changes in the quantities of capital goods. In fact alteration in (Z) would imply that when the inequality sign applies then the production of the capital good(s) concerned will be zero because the rental gained from investment in such capital good(s) will fall short of rentals obtainable from investing elsewhere. the problem is that the notion of capital as the physically heterogeneous vector as a given of the problem is in contradiction with that method. 17 16 117 .e. namely.Chap ter 6 – A second pha se in the cap ita l controversies For Walras it is crucial that the data of equilibrium do not change over the course of the trial and error actions carried out by agents. can well be in contradiction with the assumed endowments of capital goods which the theory takes as given. Garegnani (1990) pp. Thus the equilibrium reached would not therefore bear the necessary Walras (1954.16 So Walras seems to suggest that a solution to his system – namely to determine a long period equilibrium of supply and demand in free competition – can be obtained only with regards to those capital goods which can be (re-)produced. But if this is so. which Walras of course shared with all his contemporaries peers such as Böhm-Bawerk. that the tâtonnements process must be seen as an analytical device for an analysis of an equilibrium is clear when Walras introduced the so-called “tickets” (bons) in the 4th edition so that they could allow for transactions to be provisional until equilibrium is reached. as to (meaningful economic) solutions of equations (Z). In fact that is the way of the traditional method which. Wicksell. 294). as a result of an equilibrating process in the wake of any initial disequilibrium position.

as Bellino (2005. who proposes to dispense with the notion of capital in value terms because that notion would have only considered steady states. that lack might prevent the repetition of the markets – transactions – which are however enormously necessary in order to ascertain gravitational centres such that deviations between theoretical and observable – trends or averages in time – variables could be compensated each other. 18 The latest versions of the neoclassical theory. because the equilibrium variables are prone to rapid changes owing to any initial disequilibrium situation such that. which have become dominant since comparatively recent decades. that the recent 3-volume collection of works on Capital theory edited by Bliss.4. Implications of the Walrasian specification of capital On acknowledging the results of the first phase of the Cambridge debates. now. marginal theory abandoned the notion of capital in value terms and adopted the Walrasian specification of capital. 24) and Petri (2007. p. 18 118 . when groping towards equilibrium. 6. in the wake of the results drawn from the Cambridge controversies. relying on a Walrasian treatment of capital and dropping the condition of the uniform rate of profits bear drastic implications on the method of the theory itself. the restoration of Walras’s conception of capital was not new since the path adopted after the controversies had already been laid by Hicks (1946[1939]). In fact. In fact this new notion of equilibrium could no longer refer to the normal positions of the economy. 19 have tried to overcome Walras’s inconsistency by dropping the equilibrium condition of uniformity of rates of returns on capitals’ supply prices. Chapter 1 above. But. 19 As Garegnani (1976a) has suggested. also the discussion of Paragraph 1. to keep analysis within the supply and demand framework. Thus. it is not worthless to remember that the theory started to increasingly rely on a Walrasian conception of capital. and therefore it would be hard to conceive of that equilibrium as a gravitational centre taken by agents in their groping towards equilibrium. the theory was compelled to abandon the long period equilibrium It is a misfortune. This leads us to the problems of lack of persistence of the data of equilibrium. important as it is for understating the contemporary versions of the neoclassical capital theory. Cohen and Harcourt (2005) does not include anything about Walras’s theory of capital. Cf.4. some of the data will also change. Moreover. 597) have both remarked in their respective reviews.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies persistence that it is however required so as to regard the theoretical variables as a gravitational centre of actual ones both prices and quantities. which is arbitrarily given and hence will not allow the necessary substitutability among the capital goods – particularly among fixed capital goods – when time-consuming adjustments take place so as to determine their amount required by equilibrium. p. however. the datum of capital in the definition of equilibrium by the theory relying on a Walrasian capital is the physically heterogeneous vector.

e. Hahn (1975a. The point which must be clear is that this new notion – short-period – of equilibrium has essentially to do with capital. rendering the very theory of doubtful usefulness. 23 See e. xii). Hahn and Bliss.23 the tendency to regard the issue 20 21 See Chapter 1. 2005). 1982) and Bliss’s (1975) views on the question of capital: A first critical excursus To side-step the critique concerning capital and the problematic consequences for the whole approach some neo-Walrasian theorists have entertained that rigorous theory does not need of ‘aggregates’.6). As we shall see below. One of these authors’ most noticeable reaction22 has been the claim that the results of ‘reswitching’ and ‘reverse capital deepening’ would only bear negative consequences only on the so-called ‘aggregate’ versions of the theory. up to-day. above. Harcourt’s views. by which these authors have meant the particular and theoretically secondary version of a representation of a long period theory of distribution of supply and demand which regards production of commodities by an ‘aggregate’ production function.g.7.21 In the wake of results raised in the Cambridge debates Hahn and Bliss reacted in such a way so as to minimise their relevance for the theory under attack. as Garegnani (1976a) has shown in his assessment of the results of the controversies. although provisionally. Chapter 1 above. on the shift to Walrasian theory: a ‘more disaggregated’ analysis. p.4 in Chapter 1 above. these authors have had. Bliss (2005). 1975. but now we attempt to clear up some misunderstandings raised by neoclassical authors in the second phase of the controversies. This reaction however reveals a misapprehension of the traditional meaning attached to the notion of capital. Dixit (1977). which may emerge ‘logically intact’ from the capital debates). and not with the long-period method itself. Hahn’s (1975a. as introduced in our introduction. 22 And others. Whether or not neo-Walrasian theory is free of capital theoretic problems is an issue we shall deal with in Paragraph 6. 1976.20 are the most notably representatives of this reaction. 6. (cf.5. The controversies put in doubt the whole approach and compelled the theory to adopt a short period general equilibrium method. while at the same time modern marginalist authors tried to reverse the sign of the critique by pinpointing that “general equilibrium does not require aggregation” (Bliss 2005.) Hence the Cambridge debates are relevant to the current theory because it forced the latter to a theoretic apparatus of very doubtful results for explaining actual economies.g. who argue that neoclassical capital theory “does not require aggregation” (Paragraph 6. von Weizsäcker (1971). (Cf. 119 .5) and that the uniformity condition of the rates of profit on capitals’ supply prices is not important (Paragraph 6. 1982). Paragraph 1. Bliss (1970.Chap ter 6 – A second pha se in the cap ita l controversies method.

120 .Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies of capital as a problem of ‘aggregation’ which would not be different from the issue of aggregation of original factors like land or labour. thus defining normal positions. But neoclassical economists have shown these special cases to be without interest. i. If you put it the other way round. Hahn then seems to suggest that ‘aggregation’ would be the substitution of the vector of endowments for that single number. sometimes and in very special cases. these authors have created a dense fog which did not allow scholars to come to grips with the actual critique. Therefore. as we have seen. the vector of the endowments of the capital stock. Capital as a single magnitude in value terms among the data is essential and indispensable. By thus arguing. To attempt to clear up some confusions surrounding the ‘aggregation’ issue we may begin with F. Hahn (1982). say. Yet. This would then suggest that the version that relies in the quantity of capital the single magnitude would not be interesting for the determination of the “neoclassical equilibrium” because it would be hard to believe that a “single number” may supply “the same information” as does. p. the traditional versions of the theory – long-period equilibrium or normal positions method – could. The author affirms that there is no “single number” that can occupy the place of the endowments of the several capital goods in the determination of that equilibrium. but endogenously determined.e. It would be surprising if there were a single number which gives the same information as the 108 dimensional vector. was at the centre of attack in the debates. by relying on a notion of capital in value terms. the latter are determined according to the former and the satisfaction of the uniformity of rates of returns on their supply prices. who have endeavour to counterattack the critics of neoclassical theory in the following way: In general. 369. in the traditional version of the theory the “single number” does not imply to do away with the endowments. The “108 components” are not erased in the traditional equilibrium framework. That “single number” is the quantity of capital the single magnitude in value terms which. However. 108 components.24 In this passage Hahn refers to a “neoclassical equilibrium” where the endowments of the several capital goods are taken as given by the theory. there does not exist a function from the vector of endowments to the scalars such that knowledge of the scalar (and of preferences and of technology) is sufficient to allow one to determine a neoclassical equilibrium. In fact. In general. endogenously determine the composition of the capital stock. according to Hahn. it is even more obvious. the kinds and quantities of the several capital goods endowments subject to the condition of the uniformity of rates of returns on capitals’ supply prices. so long as the variables to be determined by the theory must refer to normal values. which we have visited in previous parts of this study. the neoclassical equilibrium can be found given the vector of endowment which may have. 24 Hahn (1982). this surprising property holds.

this chapter. See also Harcourt (1975a). In effect. The point is. Hahn (1975a). 364. In fact. the theory has to rely on an unacceptable conception of capital in value terms. it is not a question of a simple replacement of the endowment by a single number – the quantity of capital. by the condition of uniformity of the rates of returns on the capitals’ supply prices. “aggregation of machines” on the other hand poses very different theoretic problems since those endowments of “machines” will change with changes in outputs and demand – hence the endogenous determination of the endowments of the stock. It may be 25 26 Hahn (1975a). Bliss. See Paragraph 6. p. his rather misleading view on the centrality of the notion of capital in value terms for the traditional approach is shared by the second spokesman of the supply and demand approach we are considering here: C. below. Hahn. Hahn’s position has not been isolated at the time of this phase of the debates which started somewhat at the turn of 1970s. who had actually published the core of his 1982 message against the “Neo-Ricardians” sevens years earlier in a Symposium organised by the Economic Record.25 wrote: Why do people balk only at aggregation of machines and not of people?26 where it is clear that the author refers to the question of capital by pinpointing that it would not be a distinct problem from the technical aggregation of different kinds of labour. While “aggregation of people” may conceivably be assumed since the endowment of each type of labour can be taken as given since they change pretty slowly with respect to the speed of adjustment in their respective markets.Chap ter 6 – A second pha se in the cap ita l controversies to achieve that result. 147. on the other hand. under free competition.27 But. 121 . which is what has been uncovered over the controversies. p. again. 27 Bliss (1975). But the issue of ‘aggregation’ of items of heterogeneous capital goods must not be seen as a problem analogous to that of aggregation of original ‘factors’. This point seems to be missed in Hahn’s argument. Bliss says that the problem of ‘aggregation’ he tries to work out refers to what is involved in replacing a heterogeneous collection of inputs by an aggregate. hence the essential need for a quantity of capital in value terms. a matter that in the traditional versions was by far clearer: the quantity of capital is crucial for the definition of the equilibrium in that this notion allows the stock to adapt the form required by the equilibrium conditions entailed.6.

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

of interest to remind what we have seen in Chapter 2 regarding the supply aspect and the demand aspect of the problem of capital. In the traditional version the ‘supply’ aspect is expressed by the quantity of capital in value terms which has to be taken as datum to determine the very value of the rate of interest, among the other ‘unknowns’. The value of the total stock does not change, although its composition of course does according to the equilibrium conditions. This is the only way in which the physical form of the several capital goods can be adapted such that their total demand prices can be made equal to the quantity of capital – a given amount of value; yet there will still be an endowment composed of heterogeneous capital goods in that equilibrium position. Bliss’s “heterogeneous collection of inputs” is not erased from the traditional neoclassical blackboard by an ‘aggregate’ but rather both, so to speak, live together so long as that ‘aggregate’ is taken to mean the quantity of capital the single magnitude in value terms. Of course, it would have been clearer for us to grasp the centrality of conceiving of capital in value terms as datum had Bliss provided the crucial reasons why “capital cries out to be aggregated”, as he himself wrote in the outset of his work dealing with the theory of capital. 28 However, not only did this author provide no hint at such statement but also, to repeat, takes the problem of capital as a problem of ‘aggregation’ analogous to aggregation of non-produced factors of production. In fact, Bliss has devoted a good deal of his celebrated 1975 book to broadcasting the misleading idea that ‘capital aggregation’ is of little importance for the theory of capital (read it – the marginal theory needn’t rely on the quantity of capital). His contention seems to be based in the acceptance of the framework of ‘aggregation’ introduced by Solow in the debates by relying on the so-called Leontief separability conditions29 and then he argues that:
[T]he investigation has provided no support whatsoever for the idea that the aggregation of capital is relatively difficult. The conditions for general capital aggregation are identical to the conditions for the aggregation of labour, or of output.30

Bliss (1975), p. 8. These conditions establish that the rates of substitution of the factors one wishes to aggregate must be independent of the quantity employed of the other factors, and hence allow technically to aggregate a number of factors into a ‘single’ factor in a production function. As discussed above, there is no reason to assume that the factors to be aggregated be services of capital goods. Cf. Chapter 5 above where we have seen the reactions of Solow (1955) to some Robinson’s (1953) complaints. 30 Bliss (1975, p. 147, emphasis in the original). Note, however, that what has been at issue in the controversies is not a question of more or less difficulty, as Bliss seems to argue; it is on the other hand a crucial question to the supply and demand theory when explaining distribution by factors’ substitution forces. See n. 18, in Chapter 2 above, where I have discussed the entirely different meanings of aggregation of different kinds of labour or land with respect to that of capital.
29

28

122

Chap ter 6 – A second pha se in the cap ita l controversies

So what Bliss is saying is that, since ‘capital aggregation’ involves the same nature of technical difficulties – which underlie the strong conditions – as does the aggregation of labour or land or output, then the ‘aggregation of capital’ is actually not a problem.31 And Bliss concludes:
We may thus conclude that the widespread belief that there is a notable, particular and distinct problem posed by capital aggregation is at best an ill-formulated idea, and at worst is based simply on ignorance.32

But, was actually that the real problem for traditional theory? The “notable problem” Bliss refers to is precisely the problem to define and to determine and hence to measure the quantity of capital in value terms so long as we keep in the long period equilibrium version of neoclassical theory. However it is worth mentioning that the way in which Bliss refers to the problem of a ‘capital aggregation’ is the way in which Robinson has in her huge collection of critical works raised her concerns about capital – attacking the aggregate production function. But, we have also seen that the question of capital as uncovered in these debates, which touches the core of the traditional approach to distribution, has not to do with an aggregate production function (see Chapter 4). The actual arena where we can appreciate the relevance of the struggling between the two sides of the controversies is not that associated with aggregate
31

It might perhaps be interesting to recall that the process of selection of the data by a theory is a matter of the highest importance, which might perhaps have not been entirely grasped by some authors who have argued that a notion of ‘capital’ in value terms is not ‘relatively difficult’ when it is taken as a datum of the equilibrium. Cf. the following extract from Schlicht (1985, p. 19, emphasis added) regarding methodological issues in economics: “In freezing some factors by means of the ceteris paribus clause, we transform them into the data of our analysis. The choice of these data is not, however, simply a matter of caprice if the isolation is intended to be of substantive relevance. A substantive isolation requires those factors transformed into the data by means of the ceteris paribus clause have to be sufficiently stable with regard to the process we want to explain such that the movements of the data do not destroy or supersede the relations we are studying.” If one confronts this view vis-à-vis Bliss’s, one may understand that, in the traditional versions of the marginal theory, to regard ‘capital’ in value terms as datum is an exercise of “substantive relevance” when studying the forces explaining equilibrium. On the other hand, the author of this study asks himself what the “substantive isolation” undertaken by neoWalrasians is, when taking as datum the physically heterogeneous capital goods’ vector present in the economy. 32 Bliss (1975, p. 147). It is interesting to notice that these authors have left a lamentable imprint in successive work carried forward by neo-Walrasian theorists when dealing with the controversy in mid-1970s. In particular it is noticeable that authors who followed that path concentrated much of their efforts on presenting the issue of the notion of capital as a single magnitude by confusing it with the issue of ‘aggregation’ and hence contributing to increase misunderstandings on all these matters. For example we have the opinion of Dixit (1977, p. 18), who, in his review of Bliss (1975), affirmed: “I for one would not wish to make too much out of the fact of the relative simplicity of capital aggregation.” However, it is not a convenient simplifying device, as Dixit pretends to show, but essential to that theory.

123

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

production functions, but the supply and, particularly, the demand functions for productive factors, since what has been conclusively shown by critics in the controversies is the impossibility to generalise, in heterogeneous capital goods’ economies, the factor substitution principle – the basic premise of the theory.

6.6. Uniform profit rates and intertemporal prices. Hahn’s ‘special case’ 6.6.1. Introduction

In what follows it will be discussed the meaning of own-commodity rates of profits and the (traditional) rate of profit on the capitals’ supply prices, because the uniformity of the latter has been at times confused with uniformity of the former.33 Once we have cleared up those different two meanings we shall discuss why this confusion brought “obscurity to the controversies”, as Garegnani has pointed out, 34 which, as we shall argue, has eventually marred the relevance of the debates (and that is why the present study aims at reviving the meaning and relevance of the controversy for the economic theory). In order to clarify the issue we have to define, to begin with, what an own-rate of interest means in an intertemporal general equilibrium.35

6.6.2. Commodity own rates of interest

We shall consider a two-period intertemporal economy (the first from t=0 to t=1, and the second from t=1 to t=2; for brevity we shall refer to these periods as t=0 and t=1 respectively) with two goods (a and b), each being both a consumption and a capital good (we shall assume them to be circulating capital).36 In addition, we assume that production is viable (i.e., there exists a surplus from the pure replacement of that capital), and wages (w) are paid ex-post (after the production cycle) in terms of one commodity.37

See Bliss (1987). Garegnani (2000), pp. 392-393. 35 We are not considering the case of a temporary equilibrium, which takes as datum the expectations functions of the agents, since the meaning of own-rates of interest does not change if we consider this version. 36 We are following the model present in Garegnani (2000; 2003) which is the same presented by Hahn (1982). 37 Actually we have in this economy four commodities. In an intertemporal economy there are nT commodities; while n refers to the number of heterogeneous goods, T does to the number of periods the whole economy contains. This has been developed by Debreu (1959), who distinguishes the commodity as to location and time; for example, bread in London in 1998 is different from bread in Newcastle in 1998, or bread in London in year t is different from bread in London in, say, year t+s. However we limit the analysis only to time and do not consider location.
34

33

124

In other words. in general.. that operation is equivalent to that where someone lends (in t=0) the amount Pb1 of b to some other agent. In fact. a very important question arises: which is the form that the condition of a uniform rate of return on the supply prices of the capital goods takes.e. and the rate of interest there involved is 10%: Pb1 (1  rb )  Pb 0 hence rb  10% So. but paying for it in the moment of its delivery (t=1). Following the foregoing example. when the equilibrium relative prices change over time? Now let us assume that one unit of commodity a in t=0 has a value of 2 (i. we have to face the compelling issue regarding the (traditional) uniformity of rates of interest. it is pretty clear that this operation might be regarded as a loan. the amount of commodity b disbursed in t=0 against the promise that this agent will receive one unit of that good. in equilibrium. In effect. all contracts happen in markets at the initial moment (t=0) of the first period. one unit of a is exchangeable by 2 units of b). it would cost 1. In other words. We have seen that 125 . let us now assume that Pb1=10/11. this agent has to give up 1 unit of b in t=0 to consume 1/Pb1 in t=1.6. a loan in terms of b has been made in t=0. this agent will disburse Pb1.8 (notice that we are assuming this price of good a to be decreasing. let us consider Pb0=1 (good b in t=0 is taken as numéraire). i. Imagine that someone wishes to buy one unit of that good in t=0 (when the contract takes place) but to be delivered in t=1. it will first be necessary for us to consider that basic condition: the rate of interest.Chap ter 6 – A second pha se in the cap ita l controversies In an intertemporal framework.e. the amount of commodity b to be disbursed in t=0 in order to receive in t=1 one unit of that good.3. who will pay that debt back at the following period by providing one unit of b. and to be undiscounted). namely.. own-commodity rates will be defined by: ra  rb  Pa 0 1 Pa1 Pb 0 1 Pb1 6. Uniform rates of return on supply prices in intertemporal equilibria Having defined the two own-rates of return in our economy. Accordingly. For instance. However. cannot be different for the different loans made in terms of the different commodities. But before going into that problem. This operation entails a rate of interest which we can call the own-rate of interest in terms of good b. The question that arises is how to determine Pa1 (the discounted price over time).

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

we can obtain with 10/11 units of b in t=0 one unit of itself in t=1, the rate of interest there involved being 10%.38 Therefore, such a rate of interest will have to be the same as in the case we could obtain 10/11 units of b by renouncing to the acquisition of commodity a in t=0, and with that quantity of b ‘saved’ (since the 10/11 were not disbursed) buying commodity a1 in that period. In that period (t=1) we know that one unit of a costs 1.8 of b, therefore, in order to know how much we must pay in t=0 for obtaining one unit of a (but to be delivered in t=1 namely – Pa1), we must discount 1.8 at the rate of 10%. Therefore: Pa1=1.636 This will be the value of commodity a in terms of b, namely – the amount of b obtained in t=0 by selling one unit of a which will be delivered in t=1. That amount of b could be lent to someone, then obtaining an interest rate of 10% when the borrower pays back in t=1 one unit of commodity b. On the other hand, since we know that ra=Pa0/Pa1 – 1 and that Pa0=2, then ra=22.2%. Clearly both own-rates of interest are different; moreover, the own-rate corresponding to the good a is higher than b’s, for the relative value of the former commodity to the latter is decreasing over time, as was assumed above. On the other hand, if the loan were made in terms of b, while its relative price increases over time, it will follow that to compensate those relative advantages (of the lender) and disadvantages (of the borrower) rb would have to be lower (and ra higher) as shown in the example. This is no more than a simple triangular arbitrage operation.39 The above example also shows us that both own-rates of interest would have the same value if and only if Pa0/Pa1=Pb0/Pb1, that is, when both intertemporal prices are equal, from which it can be deduced Pa0/Pb0=Pa1/Pb1 namely – when both relative contemporary prices are equal. In general, however, relative prices will tend to be different and so will the own-rates of interest. So, when equilibrium prices are assumed to change over time, the condition of a uniform rate of interest on the capitals’ supply price will in general entail different own-rates of interest. Accordingly it is convenient to call the former uniformity of rate of interest as the effective rate as Garegnani has christened it in his attempt at clarification on some misconceptions created in the second phase of the controversies:
Of course, since we have assumed commodity b in t=0 to be the numéraire (Pb0=1), then both the own rate of return of that good and the (traditional) uniform rate of interest will coincide. It is clear, therefore, that such equality will entirely vary according as the choice of the good taken as numéraire. 39 In any intertemporal economy there are as many own-rates of interest as are goods for each pair of ‘periods’. In our case discussed in the text we have two own-rates since we have two goods (a and b) for only one pair of periods, namely – the only pair involving t=0 and t=1. In general, we may account for own-rates of interest by relating to the rates for the succession of adjacent periods linked by the so-called ‘Hicksian condition’, that is, the connection of two generic own-rates by the following condition: (1+r i(t, t+1)) / (1+r j(t, t+1)) = (P it / P jt) / (P it+1 / P jt+1) where i and j are two generic commodities. Cf. Bliss (1975), pp. 53-56.
38

126

Chap ter 6 – A second pha se in the cap ita l controversies

The word effective by which we qualified the long-period notion of a uniform rate of return is meant to take care of the fact that the definition of such a uniform rate will entail a non-uniformity of the own rates of returns of the several capital goods, when changes in their relative prices are considered.40

Now, having clarified the different meanings between the uniformity of the effective rate of interest on the capitals’ supply prices and the uniformity of own-rates, it must be seen in this connection what the critical side of the debates unveiled regarding the non-satisfaction of the uniformity of the effective rate owing to the Walrasian specification of capital namely – to take as datum of the equilibrium the physically heterogeneous capital goods’ endowment, instead of the quantity of capital the single magnitude in value terms. In fact the Walrasian specification of capital is not compatible with the traditional method. Yet in Walras’s equations we can explicitly see that the equilibrium condition underlying the traditional method, namely the uniformity of the effective rates of profit on the capitals’ supply prices, is present as soon as we recognise in those equilibrium price equations the equality sign between demand and supply prices of those capitals.41 But, because of assuming capital in physical terms, Walras was then compelled to admit solutions with price equations bearing inequality signs, instead of equality signs, thus leading some capital goods not to be produced. ‘Walras’s inconsistency’, as started to be revealed in the years of this controversy (see 6.6.4 below), has been disregarded by the new versions of the theory, because the latter disregard the traditional meaning attached to the normal positions or long period method entailed by the uniform rate of profit on the supply prices. 42 The dropping of the uniformity of the effective rate of profit is apparent when we recognise the inequality sign in the price equations of those versions of the theory.43

6.6.4. Further discussion of Hahn (1975a)44

What do all these digressions have to do with the Cambridge controversies? In the course of the debates the criticism of Walrasian theory as contained in Garegnani
Garegnani (2000), pp. 392-393, n. 2. In fact, as soon as capital goods’ demand prices are equalised with the corresponding supply prices it is assumed both i) production of all capital goods and ii) the introduction of the traditional condition of uniformity of the rate of profit on those supply prices. Cf. Garegnani (2003), p. 151. 42 Of course, it is worth noticing that this path was increasingly adopted by latest versions of marginal theory in the wake of the discoveries of ‘reverse capital deepening’ and ‘reswitching’ in the course of the controversies. 43 See e.g. Hahn (1982). 44 Part of this paragraph is based on Garegnani (2003), pp. 149-157.
41 40

127

Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies

(1958; 2008) started to be spread by Robinson (1970a).45 Robinson’s views were then echoed by Harcourt (1975a) whose arguments became the target of Hahn’s (1975a) attack, as we shall see below. The issue, not to be forgotten, is the uniformity of the effective rate of return. The following Harcourt’s message – though literally echoing Robinson’s (1970a) wording – reflects Garegnani’s critique of Walrasian theory:46
the Walrasian model of general equilibrium has the implications that it is not possible to have a theory of the rate of profits.47

where, no doubt, the impossibility in the ‘Walrasian model’ referred to in that passage clearly means the non-uniformity of the effective rate of return on capitals’ supply prices. This is what Hahn (1975a) took up in his reply to Harcourt (1975a) – anyhow intended to reverse the sign of the critique – to argue that this issue would only affect a “Special case”48 of the theory under attack. Hahn’s contention can indeed be derived from what this author says in the following passage:

General equilibrium theory is general and so can discuss the equilibrium of an economy whatever its ‘initial conditions’, e.g. outfit of goods inherited from the past. For most such specifications it will not be the case that the equilibrium price of a good for future delivery in terms of the same good for current delivery will be the same for all goods.49

Also, before 1970, we can read in Robinson (1964a, p. 60): “there is a notorious difficulty about introducing a rate of profit into the Walrasian general equilibrium system.” 46 To assess the evolution and results of the controversies, it must be clear that Walras’s treatment of capital was exposed and criticised by Garegnani in his Cambridge Ph.D. dissertation (1958), and which some years later (1962) he took up again and drafted a manuscript only published very recently (Garegnani, 2008). The argument was known by Joan Robinson, who was the internal examiner of Garegnani’s thesis (and L. Robbins the external one), so that Robinson was well acquainted with Garegnani’s work. Also it must be noted that some of Robinson’s closest collaborators, in particular, John Eatwell, made the reading of Garegnani’s dissertation in the Cambridge Library almost compulsory for all his students. (I owe this information to professors P. Garegnani and G.C. Harcourt) Therefore, Robinson’s claim that Walras does not have any theory of the profit rate at all can be taken as an echo of the inconsistency between the condition of the effective uniformity of rates of return on the supply prices of capital goods and regarding capital as the physical stocks. It might be noticed that perhaps the form chosen by Robinson to refer to the traditional authors who conceived of capital in value terms – ‘Walrasian leets’ – have misled some participants in the controversies, when it is well known that Walras does not use capital as a single magnitude. 47 Harcourt (1975a), p. 350. 48 Hahn (1975a), p. 360.

45

128

the issue of the impossibility to determine the effective rate of profit owing to a Walrasian specification of capital is misleadingly confined to a ‘case’ in which relative intertemporal prices are equal over time. 360-361) where it is clear that ‘Robinsonian leets’ refers to the homogeneous capital in value terms. While it is clear that. according to Hahn. In other words. Marshall. which was actually what Wicksell. Hahn (1975a). p. it is totally hidden in Hahn’s intervention that Walras’s inconsistency affects that ‘general theory’ independently of whether we consider changes in equilibrium prices over time. Jevons. Hahn attempted to reverse the sign of the critique by arguing that the effective rate of profit in a ‘Walrasian model’ is not an issue: The crudest empirical observations will convince one that there is no unique rate of profit to be observed in the economy. ‘general equilibrium theory’ is the neo-Walrasian theory which takes as datum of equilibrium the physically heterogeneous capital vector. Do we conclude from that that competition is functioning badly? Answer: No. 1975a. p. Consult any general equilibrium Hahn (1975a). 360. In addition there is a further issue. However. Clark had to work with in order to get the effective uniformity of rate of profit. 50 49 129 . pp. Nevertheless Hahn shows in those very passages awareness of the issue at stake. This reference to “Wicksell (and others)” in Hahn’s words is related to the way in which Hahn somehow acknowledges the results of critics in the controversies as raised by Robinson –echoed by Harcourt (1975a) – concerning the Walras’s inconsistency. 360. Yet the condition of uniformity will be present in the new formulations of the theory as soon as we reckon in the price equations of the capital goods the equality signs. yet he confuses both kinds of uniformities as the main passage quoted in the text shows it. emphasis in the original. To rebut criticisms in this second phase of the controversies. by recalling the example we have previously used. But Hahn’s hideaway can be easily discovered as soon as we recognise that the effective uniformity of the rate of profit of “Wicksell (and others)”50 is completely misunderstood by the uniformity of own-rates of profit due to constancy of prices. Böhm-Bawerk. Walras’s inconsistency would only affect the ‘case’ such that Pb1/Pb0=Pa1/Pa0 (“the equilibrium price of a good for future delivery in terms of the same good for current delivery will be the same for all goods”) namely – when the own-rates of profit happen to be equal owing to the constancy of equilibrium prices over time. According to Hahn. In fact we can read the entire Hahn’s passage: “to get a uniform rate of profit Wicksell (and others) had to work with Robinsonian leets” (Hahn. we have seen that the effective uniformity takes the form of non-uniform own-rates of profit when we admit changes in the equilibrium prices.Chap ter 6 – A second pha se in the cap ita l controversies It is thus apparent that Hahn’s “special case” is not but the result of a confusion involving the two kinds of uniformities of the rate of profit that we have previously discussed. then.

And the meaning – and the determination. when the price of the numéraire is changing. we may have a situation where the effective rate of profit is not ascertained.53 Hahn (1975a). Hahn (1975a). On the other hand. 361. as the founders of the marginal approach believed. p. and yet account for uniformity in the own-commodity rates (this case being when prices are assumed constant over time). as the critics revealed in the controversy. then. 53 The fact that Hahn has in this intervention confused the uniform effective rate of profit with the uniformity of own-rates of profit due to constancy of prices over time might perhaps be traced in the fact that in the traditional versions of the theory. own commodity rates will be entailing the same effective rate of profits expressed in different numéraires. which in turn is ensured by the effective (cont. however. could we ask more?51 Note that Hahn has apparently tried to show the existence of divergences in the own-commodity rates in reality. In fact Hahn seems to insist that his ‘general theory’ is quite safe and that a very extreme specialisation of a general model somehow shows the latter to be inapplicable requires the very summit of incomprehension. principally – of that “interesting unknown” was at the heart of the controversies.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies text. that is the Walras inconsistency due to the arbitrariness of the capital endowments. but. when determining the long period equilibrium. p. this divergence will have nothing to do with divergences of owncommodity rates because. what is clear is that the essential role of the uniformity of the effective rate of profit is totally minimised by Hahn because he believes that one could not “ask more” if “all relative prices” are known up to the end of the world – that which implies to know all the intertemporal prices and accordingly the own-commodity rates those prices would establish. the prices were assumed to be constant in that very process. while the effective rate of profit is being ascertained itself when the physical composition of capital is adjusted. However this method centred on the persistence of the equilibrium itself. so in that very process of adjustment when prices are assumed constant both uniformities coincide. Why is ‘the rate of profit’ an interesting unknown? Ask the neo-Ricardians.) 52 51 130 . For Hahn this would be a very special case of a general theory. which is taken by Hahn.52 where the “extreme specialisation” would be that associated with a uniformity of ownrates of profit. Notwithstanding Hahn’s confusions. but those observations would be actually indicating different rates of profits in different businesses compared in terms of money or other numéraire. 360. If we knew all relative prices from now to doomsday. So Hahn’s contention shown in the passage misinterprets the divergence between own-commodity rates: own-rates may diverge each other when prices change over time. to mean the effective rate of profit. The “rate of profit” is however “an interesting unknown” in order to determine long period equilibrium of the economy.

allows such an assumption. which treats capital in physical terms. as exemplified by Hahn.5. 54 In the neo-Walrasian authors’ ‘comprehension’ this extreme case would reveal itself as the constancy of equilibrium prices over time. have minimised the results raised by the critics in the debates by confining the issue of the effective uniformity of the rate of profit to steady-state equilibrium where relative prices are constant over time. Conclusion: some obscurity in the debates In the hope that the use of Walrasian capital might have overcome the deficiencies associated with the notion of capital for marginal theory. Thus a further misunderstanding was created as well. However. Cf. the neo-Walrasian theorists. by introducing such a misleading argument. n. 131 . unless the traditional notion of equilibrium be abandoned – that is what actually happened in marginal theory since recent decades. and the causes of the abandonment of the latter.Chap ter 6 – A second pha se in the cap ita l controversies With the lenses of modern neo-Walrasian equilibrium theory. The point to remark here is that Hahn and other authors. 392-393. This has had the result of obscuring both the necessity of the concept of a ‘quantity of capital’ for the traditional equilibrium. 55 Garegnani (2000). Not being acquainted with such a key notion of persistence might perhaps misled Hahn into confusing both kinds of uniformities. 6.6. In this connection – together with our excursus through Hahn’s misleading views on the issue of uniformity of the effective rate of profit – one can appreciate the following remark recently raised by Garegnani: A question which contributed to the opacity of the capital controversies of recent decades [has been] that inequality of own rates has often been confused with inequality of effective rates on the supply prices of the capital goods. p. it would only appear as a “very extreme specialisation” the case with which “Wicksell (and others)” had to deal with – namely capital the homogeneous magnitude. 32) and Robbins (1930).55 uniformity. the arbitrarily given physically heterogeneous capital goods vector cannot ameliorate the situation under which the marginal theory has fallen.54 However Hahn’s ‘general model’ cannot determine the relevant variables as intended not only by “Wicksell (and others)” but also by Walras – to determine long period equilibrium or normal positions of the economy. Garegnani (1976a. 2. which is instead due to the arbitrary initial composition of the capital endowment of Walrasian theory. in turn confused the traditional long period equilibrium or normal positions with equilibrium of steady state (constancy of prices). pp.

60 Bliss (1975). In fact. 59 Bliss (1975. we would not be guilt of slandering if we pointed out Hahn. chapter 3).1. in Bliss’s words) which are solution to a maximisation linear programming problem then any convex combination of those ‘weights’ p+(1-)p’. On visiting the foregoing exchanges in this second phase of the debates. Bliss According to Bliss (1975)57 the ‘paradox’ of ‘reswitching’ in an intertemporal general equilibrium concerns a key assumption underlying a maximisation problem of the linear programming type: 58 the possibility that two price systems could be solution to the maximisation problem. with ‘reswitching’. Bliss explains that if two prices systems maximise the present value of consumption. sometimes. Cf. Cf. the author also notices. n. 58 The linear programme maximisation of present consumption entails a dual minimisation of costs problem. Samuelson and Solow (1958). it was the neo-Walrasians who actually produced an endless series of misunderstandings on key questions for the theory. 5. Cf.”59 However. ‘obscurity’ per se is one of the worst appropriate grounds for debating.7. complete price systems) at each of which a development of the economy is a production-price equilibrium.”60 This phenomenon would undermine the convexity properties underlying price systems solutions of linear programming problems and it might probably have been this feature Bliss is concerned about. 6. obscurity might be the weapon used by some both to hide the issue at stake and to escape the critiques from their contenders. Hahn (1975a) ended his intervention by proposing to substitute one of Harcourt’s (1975a) appreciations of the debates – “The Great debate” – by “The Great Charade”. as the ones who actually contributed to foster such an image of the final results of the controversies that might have appeared as a sham. with 0<<1. in our well-known wage curves graphs. p. Dorfman.e. But. This could be understood as if the same technique. Yet the obscurity brought in by Hahn into the debates has had the further purpose to reverse the sign of the critique.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies As some reflection on communication might suggest. Stiglitz 1974. 364. The idea of a convex combination is the following: if p and p’ are both price systems (or “weights”. 57 56 132 . Bliss. 238. von Weizsäcker and many other neo-Walrasian adherents (such as J.7. 238. p. emphasis added). a “convex combination” of them should also be solution of the problem. or “discount factors”. will be solution as well. Dixit. 238. is in use for two disjoint intervals of the rate of profits. A similar position can be found in Hahn (1972). and therefore “it is not possible to have two disjoint sets of prices (i. Burmeister 1980). 8. nevertheless this author contends: Hahn (1975a).56 However. Do ‘reswitching’ and ‘reverse capital deepening’ bear theoretic problems on intertemporal general equilibrium? Two views: Bliss and Garegnani 6. two such price systems “may not be connected. as we have seen. p. p. Bliss (1975) p. Bliss (1975.

) 133 . Bliss would have re-established “the property of convexity in the set of solutions” and so their connectedness. 64 Yet it may be interesting to see that even without violation of the so-called connectedness property there is cause for concern when. phenomena like ‘reswitching’ may only arise when economic analysis is “confined to constant rate of interest price systems”. (See graph below.) Bliss could warn us of the fact that we are considering “the constant rate of interest price systems”.64 Bliss (1975. arose because it seemed that the set of values [of the rate of interest] at which a technique might be chosen was not connected. in the sense that the examples do not depend upon choosing very special values for the parameters. Note the way in which this author sees the long period rate of interest that is central in the controversies: a simple ‘constant’ rate. in the case of ‘reverse capital deepening’ without ‘reswitching’ on the ‘envelope’ (where one technique is in use only for one convex interval of the rate of profits). hence the ‘paradox’ would disappear in the intertemporal versions. to regard “complete intertemporal price systems” – these ‘paradoxes’ may not appear? 62 61 Figure 6. 238.”63 Thus. 239. how could one be sure that by widening the ranges of values of that rate – i. p. But this is an optical illusion. according to Bliss. emphasis in the original).” 62 Since. But even if in this ‘subset’ of values of rates of interest we may account for ‘perverse’ relations between distribution and net output per worker. p. we realise that when producers switch from technique to at r=r2 a higher rate of profits is associated with a higher net product per worker.e. 239. p. 63 Bliss (1975).Chap ter 6 – A second pha se in the cap ita l controversies Surely the excitement caused by the demonstration that double switching [reswitching] was a ‘common’ phenomenon. However apart from the fact that ‘reverse capital deepening’ is not accounted for in Bliss’s disputations. it would seem. by widening the range of values of the interest rates (implicitly in a “complete intertemporal price system”).61 The “optical illusion” Bliss refers to would have been caused because economists have had “the habit of working with a space of price sequences corresponding to constant rates of interest. Bliss (1975).1. ‘Reverse capital deepening’. the “illusion” would then disappear as soon as it is realised that “[firms choose their investment plans] in the light of complete intertemporal price systems. Each technique is used for convex intervals of r (cont. it is yet not clear why ‘reswitching’ would have disappeared in intertemporal general equilibrium theory when the set of values of the rate of interest can be enlarged and connectedness re-established.

this chapter. since each individual’s savings enters both the expenditureside in t as well as the income-side in t+1. Note that L0w becomes available in t=1 since w is paid in arrears. According to Garegnani the problems associated with capital are present in the current versions of neoclassical theory. t=1) where agents consume all output in t=1.66 … …t=0………… ……………t=1 Garegnani (2000) shows that despite savings and investment formally disappear in the so-called individual (or collective) wealth equations in the standard formulations of intertemporal general equilibrium. 65. In t=0 savings (S0) are the result after deducting from the initial endowments of commodities a and b (A0. both used as consumption and circulating capital) the respective consumption demand (Da0. To introduce the so-called individual wealth equations. In the intertemporal framework the excess of savings over investment entails an excess supply of commodities in t=0 and an excess demand of them in t=1. where savings for each individual have been cancelled out.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies 6. Savings and investment are realised in t=0 too. 436-439. and w is the wage rate (in terms of Pb1). 66 Garegnani (2000). it is appropriate to consider both the income-side and the expenditure-side in each period: y0=a0Pa0+b0Pb0=g0+s0 y1=l0w+s0=g1 By adding both equations. Db0): S0=Y0 – G0=(A0Pa0 +B0Pb0) – (Da0Pa0+Db0Pb0) S0=(A0 – Da0)Pa0+(B0-Db0)Pb0 where Y0: aggregate output in t=0 G0: aggregate consumption in t=0 In t=1 savings will be determined by the following relations: S1=Y1 – G1 = (L0w+S0) – (Da1Pa1+Db1Pb1)=0 where L0 is the quantity of labour employed in t=0 for producing output to be consumed in t=1. 2003) is one of the most notably representatives of the critical side for having brought the critique in the intertemporal marginal approach. that all markets other than savings and investment are in equilibrium. production will have sense only if it takes place in t=0. The crux of Garegnani’s argument lies in that these difficulties reappear in the savings-investment markets. Garegnani The critically oriented side of the controversies has started to reply to advocates of intertemporal general equilibrium theory by arguing that problems of ‘reswitching’ and ‘reverse capital deepening’ are also present in those versions. B0. we obtain: y0 + y1=a0Pa0+b0Pb0+l0w+s0= g0+g1+s0 hence: y0+ y1=a0Pa0+b0Pb0+l0w=g0+g1 Note that the highlighted equation above is the so-called wealth equation.65 How to see the problem? Let us assume. in the intertemporal economy explained in n. pp. In a two-period intertemporal economy (t=0. Garegnani (2000. such that savings exceed investment (S= S – I >0).2. savings and investment still are present in the intertemporal relations. 65 134 .7. Yet individual (and collective) decisions to save and to invest are present in the intertemporal versions.

and equilibrium may be therefore reached within the same period due to changes of contemporary relative prices caused by the excess demand of b0 relative to a0. capital goods in which investment is done are perfectly substitutable: it does not matter whether the capital good is a lathe or a loom or a plough: what matters is the expected rate of return that the investment might yield. i. That this adjustment takes place does not seem to meet insurmountable obstacles since it is in t=-1. If this is so. p. as Garegnani suggests. Garegnani argues that () represents the equilibrium of contemporary consumption but that it is equation () that represents the intertemporal adjustment of savings and investment. 437. [so there is] a perfectly consistent story that does not look any different from the story about choosing commodities today. 0.69 Garegnani (2000). we obtain the following equilibrium condition:  Pa 0 Da 0  Pb 0 Db 0 ( Equation () represents the case in which there is an excess demand of b0 relative to a0. 68 67 135 .e. labour and the given proportions of a and b composing the means of production. Now. Garegnani (2000). that is. in the case of the savers. as Garegnani argues. 1. the case of relative excess demand of contemporary commodities.68 three periods in the intertemporal economy. Some basic economic reflection may help us understand that from the consumers’ viewpoint. however formal as it seems. now consider. 69 It may be due to this kind of analogy. p. since the inputs needed to produce both consumption goods are similar. the quantity of capital – and its problems – will be also present in the latest versions of the theory. 439. If we consider the RHS of the first equality sign of equation () and equalise it to zero. the savers expect to obtain the highest rate of return on the value of the investment. in the case of the contemporary excess of supply of a0 relative to b0 (equation ) there would not appear to be strong difficulties to shift the labour and means of productions of t=-1 freed by excess supply of a0 in order to produce Db0. The root of seeing the problem of intertemporal allocation as if it were a problem of contemporary consumption goods’ adjustment might be realised by confronting two radically different principles guiding consumption and investment (savings). who demand for future net income. specific commodities satisfy specific needs (for instance it would be as futile to overcome hungry by coal as it would be to consume bread in order to get rid of the winter cold.e.) However.” (p. the variable are today’s prices for future goods. t= -1. i.Chap ter 6 – A second pha se in the cap ita l controversies S  ( Pa 0 Da 0  Pb 0 Db 0 )  Pa1Da1  Pb1Db1 () where D’s indicates the differences in the respective quantities of goods demanded as consumption between t=0 and t=1. Then. who wants to satisfy his needs. independently of the particular type of the capital good. Let us assume that a0 and b0 are produced with means of productions and labour of t = -1. that Arrow (1989) might have written that “[in any study of] intertemporal allocation. then. 155).67 In order to see the nature of the problem.

So far. the savingsinvestment adjustment. However. which directly produce Da1 and Db1. equation ). The difference lies in that the inputs required to produce the consumption goods in t= 0 – and to be consumed in t=1 – are physically heterogeneous from the inputs required to produce the means of productions (in t=-1) which.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies However. 206. But a rise in such a price is a fall of rb.7. 136 . the claim that “most observers consider [Bliss’s 1975 book as] the definitive neoclassical treatment of capital theory that ended the Cambridge controversies. ‘reverse capital deepening’ and ‘reswitching’ undermine the intertemporal versions of the theory no less than they do in the traditional marginalist theory. because b is the numéraire. Garegnani concludes. an excess supply in t=0 of a and b) may be directly employed in t=0 in order to produce in t=0 quantities Da1 and Db1. Confrontations The negative implications of ‘reswitching’ and ‘reverse capital deepening’ on the contemporary intertemporal version of the theory have not been utterly considered by many modern neoclassical advocates. Moreover. the nature of the problem is entirely different from that regarding the contemporary adjustment outlined above. this will in general not be possible. regarding the intertemporal adjustment (equation ).3. (cf. to re-establish equilibrium. i. Hence there should be a production of Ia and Ib (investment) in t = -1 by decreasing production of quantities Da0 and Db0. Accordingly. Consequently.e. p. the intertemporal equilibrium adjustment ought to be such that the ‘freed’ labour and means of production in t = -1 (due to. In this case. 6. we have seen how Bliss might have believed in overcoming capital theoretic problems in his framework without solidly arguing his point. Thus. are different from those ‘freed’ in t= -1 due to the excess supply of a0 and b0. without further inquiring. say. the only way to increase production in t=0 of Da1 and Db1 will be that of increasing the productivity of the already fully employed labour in t=0 by increasing the quantity of the means of production cooperating with that labour which directly produce a1 and b1. however. Assuming that the latter rate is also the effective rate of interest (r). as the capital controversies have revealed. are required to produce the consumption goods (in t= 0). the way to set off these increments in investment will be by means of a rise in the intertemporal price Pb0/Pb1 so as to make agents give up consumption in t=0 (Da0 and Db0) in order to obtain some extra in the future t=1.”70 70 Cohen and Harcourt (2003). we can no longer be sure that a fall in r will make investment rise. in turn. because labour and the means of productions in t=0. our brief assessment of Bliss’s treatment of the issue seems to show us strong reservations that make it hard to accept. emphasis added.

184. Thus.71 71 The interpretation which maintains that Bliss (1975) would have led the debates to an end is also supported by Solow (1983). p.Chap te r 6 – A second pha se in the cap ita l controversies On the other hand. suggesting that Bliss’s 1975 is a “definitive treatment of capital theory” that would have eventually led the controversies to an end – read it as a neoclassical end. 137 . in the light of this recent investigation. as Garegnani maintains. like the one quoted above. it is hard to go along with many voiced interpretations. the critically oriented side of the debates has recently started to bring the critique to the intertemporal land by strongly arguing that capital theoretic problems may reappear in the savings-investment markets.

.

iii) to minimise the implications of the results by pointing to the empirical way-out (Ferguson) – an extreme position that completely dodges the question raised by capital theoretic problems. In fact. they tried to find the conditions to rule out ‘reswitching’ (Levhari-Samuelson). neoclassical advocates did not remain silent when the serious threats started to be perceived by neoclassical theorists. as I argued. the range of strategies rehearsed by neoclassical participants have been i) to build up the so-called parables (Samuelson) – but this resulted in a totally unsuccessful theoretical device to save their theory. it will be realised that the implications of those results drawn from the Cambridge controversy undermine the factor substitution principle.Chapter 7 Conclusions The research presented in this book reveals the importance of the Cambridge capital theory debates for a better understanding of the present state of neoclassical capital theory. once the interdependency between relative prices and distribution is taken into full consideration. the book focused on the most salient and relevant phenomena of ‘reswitching’ and ‘reverse capital deepening’. Then what I thought to do in this research was to selectively reconsider some different positions in these debates by taking as the main line of argument the results and implications of ‘reswitching’ and ‘reverse capital deepening’. ii) then. As illustrated. the fact that advocates of the traditional approach – Nobel awardees included – have tried to rescue the supply and demand theory by recurring to these strategies indicates . once the negative implications started to be acknowledged. Among the many issues debated over the capital controversies. but this too was proved wrong altogether. namely: the general principle on which dominant theory (traditional and contemporary) founded an explanation of distribution in terms of supply and demand. and not simply as ‘curiosa’ as Joan Robinson remarked (cf. Chapter 5). Chapter 6) we find a shift to Walrasian theory and hence a shift in the notion of equilibrium. is that the counteractions from the neoclassical side arose out of the fact that the implications of ‘reswitching’ and ‘reverse capital deepening’ were perceived as very strong concerns. As we saw (cf. That is why we set out to deeply analyse and put at the centre these phenomena: to reveal how they undermine the marginal approach (cf. They attempted different strategies in order to preserve the neoclassical theory from the tension created by the capital theory debates. Chapter 3). As I showed in this book. Chapter 4). What this reveals. and finally (cf. and thus contributes to rescuing the relevance of the controversy from the apparently inconclusive state at which it might have been led in its second phase. These results cast serious doubts on the hard core premises of the neoclassical theory.

in the second phase. an outcome of my analysis was that the overemphasised critique of the aggregate production function pursued by some participants could have contributed to foster misunderstandings that still persist (cf. Far from debating. par. the critically oriented side of the controversies received neoclassical counteractions in different ways. Chapter 2). However. the neoclassical side started to try to escape the critique by claiming that the theory does not need ‘aggregates’. as I discussed (Chapter 6). Robinson’s passages in Chapter 5). Bliss’s 2005 views in Chapter 1). to an apparent inconclusive point. Chapter 4.1. Thus. In fact. thus supplying a defensive line to neoclassical advocates to totally shun the implications of the controversies. as I showed in the discussion of the Walrasian rehabilitation of capital (Paragraph 6. we have a word to say about the critical side as well.4). Moreover. in fact they diverted the central focus from the question of capital to a question of ‘aggregates’ as if the former question were a problem of considering an ‘aggregate’ capital in the same footing as labour or land (cf. but the role of the quantity of capital for the traditional theory is essential to the determination of a long period equilibrium (cf. the obscurity brought by neoclassical advocates arguing in misleading ways – together with a not rapid and unified reply from the critical side in the context of the Walrasian theory – might have contributed to lead the relevance of the Cambridge controversies. as we saw. However there was not a unified reply from the critical side in this second phase. However the disadvantages of this line of criticism turn evident when. rather than a problem of the traditional. Moreover.5). this study argued that there seems not to be reasons for taking the neo-Walrasian 140 . As I argued in Chapter 5. Thus.4). Paragraphs 4. Still this tension forced the theory to make all the possible efforts at their disposal to keep alive a theory whose basic premises cannot theoretically be supported on its own inner logic. Perhaps some critics might have believed that the critique centred on the aggregate production function would have supplied a ‘more evident’ line of argument against dominant theory (cf. In the effort to understand the reasons for the gap within the critical side. the phenomena of ‘reswitching’ and ‘reverse capital deepening’ have been confined to ‘dead end’ by some critical authors as well (cf. some authors might have preferred to bewilder the ground by ignoring the role of the ‘quantity of capital’ in the traditional versions and by entrenching the theory to short period general equilibrium so as to evade the problematic issue of capital (cf. In fact. regards the issue of the quantity of capital as a matter of ‘simplicity’ in presenting the theory.10 and 4. despite that the bulk of the critics placed themselves in opposition to the marginalist arguments raised in the first phase of the debates. in the second phase. But this confirms what critic economist Garegnani pointed out in (1976a): a new notion of equilibrium had become dominant in marginal theory due to inconsistencies in capital. the purpose of these neoclassical theorists in the second phase was far from debating. Chapter 1). long period method itself (cf.15). the historicalanalytical reconstruction here developed shows the relevance of these controversies.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies the tension caused by the Cambridge controversy. it must be pointed out that some critical participants chose to emphasise both in the first and the second phase a kind of critique that seems to be secondary in connection with the kinds of problems entailed by the question of capital. This view. Paragraph 6.

Rather the threat for the theory that problems regarding capital may reappear in the savings-investment market would indicate that neoclassical theory would have to face this issue – if that were the case. Bliss’s 2005 views) we are still far from that. communication between the two sides must be re-established. is intended to nurture future communication between the two sides on these very important matters at issue. what turns out to be relevant is that the results of the controversies touch the core premises of dominant theory. 141 . however. Although this book has partially dealt with the situation of capital theory in the new versions of the theory.3 and 1.4) On the contrary. Some would argue that the issues raised in the debates do not concern the modern versions of the theory like the intertemporal general equilibrium theory. on the other hand. advocates of these versions have not been able to show the opposite. given the present situation (cf. Paragraph 6. the provisional assessment of the recent critique of these versions of the theory could also be taken as a confirmation that there would not be grounds to dispense with the critique by neoclassical theory relying on Walrasian capital. however. Moreover. The clarifications made (Chapter 6) suggest that the fact that today nobody seems to talk about the capital debates is not because of the capital theory debates actually ended (cf.7). Harcourt’s 1976 passage and recent positions of this author in Paragraphs 1. as analysed in Paragraph 6. it is hoped.7. thus. our attempt at clarification. it is the critique that has to be relevant to the theory and not one or another ‘version’ of the theory for the issues raised in the debates. For this.Chap ter 7 – Con clusions theory as the theory that is not relevant for the issues raised in the controversies (cf.

.

The maximum gross average product of labour. which is four.5 of labour with 1.1 (whichever the level of corn-output produced). 1. the product may be increased with additional quantities of labour – that is the resulting area under that curve is larger the higher the quantity of labour employed. the limit of such a quantity will be determined both by the supply of labour and by the wage rate ruling in the market. Paragraph 2.1. as the negatively inclined part of the curve indicates. the competitive capitalist will minimise costs when hiring the number of labourers for which the net marginal product (an amount of corn) is equal to the corn-wage he finds ruling in the market. since. we know that the relevant part of the MgPL curve is the negatively inclined stretch (the second component of the above function).5 of corn-capital. the given capital is K = 2 units of corn. Given the quantity of the other factor.5 of corn-capital and so on up until labour reaches 2 units which is combined with 2 units of corn-capital.1 below. Lastly we assume that half a unit of labour is combined with 0. is the maximum value that can be reached by the gross marginal product of labour. To make the concept of a net marginal product clear. Of course. (See Figure A. This means that there will be a quota (K) of the given quantity of corn-capital which must be replaced for producing a given level of net output given by the technical conditions of production.1. All these combinations yield a gross marginal product of 4. let us imagine that the gross marginal product of labour (MgPL) is represented by the following function:  M g PL  4   2  M g PL  4  ( L  2)  if 0  L  2 if 2  L  4 For well known reasons. one unit of labour is combined with one unit of capital. and that the labour available in the economy is not higher than 4. But let us return to our construct of the net marginal productivity curve. . given the quantity of corn-capital. the rate at which the output increases is a decreasing one. Let us assume that for producing a unit of net corn output it is necessary to replace a quantity >0 of corn-capital.) 1 Appendix to Chapter 2. In a labour and corn-capital economy.Appendix A The curve of the net marginal product of labour1 A. We assume = 0.

Imagine.5.017. now. Then the net marginal increment in 35 3 175 output when the labour employed is approaching 3 units will be given by 3 3 175  2.5 7 4 hence L=3. Accordingly 2 2 K 12. as the labour employed is increased up until its maximum value – assumed to be 4 only for the sake of exposition. imagine that MgPL= 3. due to our 144 . amounts to 5  . and the net marginal increment in output when the 7 4  0. In this hypothetical situation the gross output is given by 8   4  ( L  2) 2 dL which is 2 3 equal to 35 3 .875 which is lower than 3 175 = 0. By subtracting that ‘share’. On the other hand. not net. This value. that the MgPL=3 hence L=3.0155  1. in order to see that this quota out of the gross output is a decreasing function of the quantity of labour employed (and hence of the output). which is a constant value too. K 35 3 3 . In this situation gross corn output is given by 1 8   4  ( L  2) dL which amounts to 12.875 = 5  0. It can be labour employed is approaching 3. But that marginal product is gross. from the MgPL function. so long as we move from north-west to south-east on the negatively inclined part of the gross marginal productivity curve.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Then what is relevant to our purposes here is to show that the ‘share’ of the amount of corn-capital to be replaced (K) out of the gross output will be diminishing as the output increases for the negatively inclined part of the MgPL curve – that is when the area under the MgPL curve is being increased by adding quantities of labour higher than 2. Now. for all relevant values of labour employed and given the quantity of the other factor.0155 12.7345 .5 units will be given by noted that. the respective levels of the net marginal product will be near the net marginal product of labour for the constant part of the gross marginal product curve. the net marginal productivity curve will be approaching the former.982 . we shall obtain the net marginal product of labour.875. for values of labour which correspond to the part of the gross marginal product curve near to its constant part. This ‘share’. To find the net MgPL it is 1 necessary to calculate the ‘share’ of the replacement of corn-capital when gross output is 35 3 . Let us make it sure we understand this construct by using the above function and the assumptions of the relevant variables.

net that is after deducing for any extra expenses that may be indirectly caused by the change” (Marshall. and when L=2. Then. who is the entrepreneur. emphasis in the original). He estimates as best he can how much net product (i. when L=1. is equal to 1 40  .1 8 2. resulting from. is constantly endeavouring to obtain a notion of the relative efficiency of every agent of production that he employs. pp. while the second one is applied when 2  L  4 .e. according to his energy and ability. Therefore we have built the net marginal product curve. as well as of others that might possibly be substituted for some of them.2.2 2 The notion of a net marginal product may be traced in Marshall’s Principles: “Every business man indeed. net addition to the value of his total product) will be caused by a certain extra use of any one agent. 336-337. will minimise costs when hiring the number of labourers for which the net marginal product (an amount of corn) is equal to the wage rate he finds ruling in the market. A. 145 . which provides the net marginal increment in output as an infinitesimal unit of labour is increased. K 8  0.1 4 1 . 1959.  K 4  0. the capitalist.App end ix A 1 40 1 40 example. Thus equipped we can derive the following net marginal productivity function of labour:  NetM P  1 g L  40  1   5  NetM g PL  l 2  8   4  ( L  2) dL  l  1 0 The first component of the function is applied when 0  L  2 . such that l0  l1 with 2  l0  4 and 2  l1  4 .

if the labour demanded were LL which is lower than L*.g. If the entrepreneur finds in the market w=w*. Were it otherwise.1. then the optimal quantity of L demanded should be L*. e. the entrepreneur would lose the shadowed area ACE 146 .Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies Figure A.

1934. 1960. which we illustrate in the present appendix and which may be useful to discuss with Joan Robinson’s way to deal with capital. as the vector of the heterogeneous capital goods.4 The consumption good is produced by means of current labour. like Walras and the ‘new’ versions of the theory). pp. B. 181) does so initially. Paragraph 2. p.2 Moreover. Cf. The crucial point seems to be to measure the factors of production before determining distribution. in order to obtain the quantities corresponding to any given method of production [group (ii)] and to any given physical output [group (iii)]. This appendix attempts to illustrate. current land. and later in the book the author (Wicksell. the arguments raised in Chapter 2 above concerning the supplyside or supply-aspect problem of capital.3 Let us not forget that the discussion of these problems meets those theorists who regarded capital in value terms as data (and not other theorists who considered capital. In both of these roles the factors should be measured – and hence known – before knowing distribution. and the saved-up (services of) land and labour: that is ‘capital’ in the form of past labour and land applied at different periods of time to the consumption good’s production Appendix to Chapter 2.2. it will be attempted to illustrate the other problem which arises with groups (ii) and (iii) of the data: in group (ii) the quantities of factors define the methods of production. as part of the data. in order to avoid circular reasoning. To begin with. Wicksell (1934). by using Wicksell’s general equilibrium system of his Lectures. 144-206. 196-206) proceeds to the case of more than one consumption good. pp. 4 Wicksell (1934. and on his works shall I rely for presenting the difficulties of capital associated with its supply-role. which has been associated to a discussion of ‘aggregate’ production functions. it will be useful to bear in mind how the problem of capital appears in Wicksell’s theory. 2 1 .1. (See Chapter 5). whilst in group (iii) those quantities appear in the role of a ‘factor endowment’. 1990). 3 Also.Appendix B Discussion of the ‘supply-side’ of capital in Wicksell’s general equilibrium system1 B. These problems have already been dealt with by Garegnani (1958. we shall simplify the exposition by assuming the production of a single consumption good in yearly cycles.8. that is the role on which the supply functions of each factor are based.

Or. A2. an economy where labour and land are saved up for no more than two years. p. Wicksell takes as given the quantities A. 158. (dated quantities of labour). The unknowns of the system are A0. A2: be the quantities of labour (labours-year) applied 2 years before Q emerges. B: be the acres of land available.” (Wicksell 1934. w. As Wicksell put it: “What we have to say in this connection can easily be extended to processes of production and capital investment over any period whatever. w: be the yearly wage rate in terms of Q paid in arrears. or two.B) Equation (1B) defines the production function of the year’s consumption good. Wicksell (1934). B0: be the acres of land applied ‘0’ years before Q emerges: that is current land applied.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies before the latter emerges. that is 5 E. therefore. B. and K. B0 . Anyway. A0: be the quantities of labour (labours-year) applied ‘0’ years before Q emerges: that is current labour applied to Q. land. and interest i of the saved up labour (A1. For that determination we have the following relations: Q  f ( A0 . B2. B1. i: be the yearly rate of interest in terms of Q. B0. B1 . 148 . In t. 6 Cf. B1: be the acres of land applied 1 year before Q emerges. is produced by means of labour. there is no loss of generality if the period of saved up labour and land is either one. K: be the value in terms of Q of the capital stock available in the economy. 5 It is here assumed. we reckon capital as the saved-up labour and saved-up land necessary to produce those ploughs. p.g. r: be the yearly rent rate in terms of Q. p. the consumption good Q. A1: be the quantities of labour (labours-year) applied 1 year before Q emerges. then the latter are regarded as the (past) labour and (past) land used in year t-1 to produce those ploughs (suppose to be of one-year duration). 29) reconstruction of the system. n=3 if the reader is familiar with Garegnani’s (1990. and ‘ploughs’. A1. If in year t. as Wicksell did. r.6 Let us consider the general system set out by Wicksell. (dated quantities of land). and i. – which is produced by means of current labour and land. 158). B2: be the acres of land applied 2 years before Q emerges. He further supposes the quantities supplied for productive use to be independent of the rates of remuneration. Therefore let: Q: be the yearly output of the consumption good. A: be the quantity of labourers available (they may be regarded as the absolute number). corn. Q. which in turn are used to produce corn in t. or n years. A2. Q. B2 ) (1. rents r of the current land (B0). and (saved and capitalised) resources during the two preceding years – and gives the combination which minimises the cost of production that the entrepreneur has to incur: wages w of the current labour (A0). A1 . A2 .

B) In fact. we would like to stress that. p. under r1  r r  i and competition. these equilibrium conditions (however not being used to determine equilibrium in the system) 7 indicate us that. both w1 and w2 will be determined too.B) are expressed in technical units.B) and (2. A caveat here is in order. and not simple interest. by the supply and demand equilibrium functions for factors. Firstly. B2. so: r2  r1 r1  i . one can determine r1 by r1  r (1  i ) . (r1 and r2 are the rent rates which corresponds to the quantity of land necessary to produce the capital goods one and two years ago respectively. uniformity of returns on the different capital goods is endogenously reached through equations (2. equation (1.B) is postponed later in the present Appendix. 156) assumed that. the factors’ quantities defining the methods of production in the arguments of equations (1.B) and (3. For investment of ‘labour’-capital for two years. In addition. so that the rate of interest will be w1  w w  i (w1 is the wage rate which corresponds to the quantity of labour applied one year ago necessary to produce the capital goods to be reckoned to-day). Discussion of equation (4. Wicksell (1934. that is through the relations expressed by.B)8 below. and on the other. It is worth 8 pointing out that those equilibrium equations entail a compound. we must assume that w1>w. we must have w2  w1 w1  i. Similarly. in equilibrium. on the one hand. Then we have the equations expressing the equilibrium relations between marginal productivity of each factor of production (measured in technical units) and their respective rates of remuneration. economic sense. Furthermore. Then: Q A0 Q A1 Q A2 7 w  w(1  i )  w(1  i ) 2 Q B0 Q B1 Q B2 r  r (1  i )  r (1  i ) 2 (2. 149 .App end ix B the labours-years of the two preceding years) and of saved up land (B1. and r2 by 2 r2  r (1  i ) . Secondly.B)-(4. the equilibrium conditions supplied by the role that the factors play in the definition of the methods of production. say A1. once i and r are determined. In other words. once w is determined.B) together with the following groups of equations do express ‘half’ of the problem which arises between the role of productive factors in both its roles: group (ii) & (iii) of the data. to give. and to be reckoned to-day). that is the utilised acres over the two preceding years). “all capital will receive approximately the same return”.

204. the expression K is obtained by bearing in mind that. capital goods. Then there are two equations. pp. p. that is. which Wicksell however introduced verbally. 12 The term wA2 on the RHS of (4.”11 Before putting forth the formal expression of Wicksell’s assertion. The importance of the quotation in the text. used in turn to produce corn (the consumption good Q). the labour embodied in coal produced two years before producing ploughs. under conditions of zero net accumulation. which Wicksell derives by assuming “stationary conditions”:9 year by year the total sum of the quantity of labour and of land annually consumed – saved up or current – must be equal to the supply of labour and of land annually available in the economy. two quantities of labour and land must be withdrawn from the production of the consumption goods during the current year and devoted (1) to production of goods which will only be used in the following year. 2 there will correspond j elements of capital. in our example. that is. a ‘second’ quantity A2 of labour paid for at the time of reckoning and embodied in the coal which is going to be worked upon by labourers producing ploughs. Cf. 10 150 . j= 1. while the term wA2(1+i) is the ‘second’ quantity of A2. also n. The complete citation runs: “[…] the total exchange value of the capital employed in the two industries together. above. and applied one year before the moment of our reckoning of capital. Chapter 3. which are required by the productive conditions of each of the “two industries”. besides the quantity A2 of labour embodied in ploughs. p. The missing equation. Hence: K  wA1  wA2 [1  (1  i )]  rB1  rB2 [1  (1  i )] 9 (4. each embodying Aj labour years. 18. is the realisation that Wicksell assumes a given capital in value terms as the endowment of the economy. Therefore: A 0  A1  A2  A B0  B1  B2  B (3. 204-205. above.” (Wicksell.Andrés Lazzarini – Revisiting the Cambridge Capital Theory Controversies This group of equations provide the equilibrium condition of uniformity of rates of returns over costs of the capital goods. This passage also provides insight into the issue of why the composition of the capital endowment must be endogenously determined. together with this complement. then we would have. as it has been argued in Chapter 2. pp. As Wicksell put it: “[I]f conditions are to remain stationary. is the one that it makes the total sum of the values of ‘labour’-capital and ‘land’-capital equal to “a certain given quantity – the total exchange value of the capital employed.B) is the ‘first’ quantity of A2 reckoned as capital (from our example: it is the labour employed in producing ploughs). 1934. For instance. express