KALDOR'S GROWTH THEORY

BY

NANCY J. WULWICK

I. INTRODUCTION
The last decade has seen an outburst of growth models designed to
replace the conventional Solow growth model, with its exogenous trend of
technical progress, by more realistic models that generate increasing
returns (to labor, capital and/or scale) as a result of endogenous technical
progress. In contrast to the Solow model, the new models suggest that
policy interventions can affect the long-run rate of economic growth.
Nicholas Kaldor's growth model, designed in the late 1950s and early 1960s
to replace the Solow growth model, is a precursor of the new growth models.
Many of the new growth models are intended to rationalize the stylized
facts of growth established by Kaldor (Kaldor 1958, p. 178; Romer 1989, p.
54). Those stylized facts include the existence of:
(1) continued growth of labor productivity, with no tendency for afalling
rate of productivity growth;
(2) a continued increase in the ratio of capital to labor, however capital
may be measured
(3) steady capital-output ratios, or at least the absence of a clear long-
term trend in the positive or negative direction, once less than full
capacity utilization is taken into account;
(4) a constant rate of profit on capital;
(5) a constant share of investment in output and profits in income (which
follows from facts #3 and #4);
(6) no tendency for country growth rates to converge.
Some evidence has accumulated since the 1970s to suggest that the capital-
output ratio has not been constant (Haache 1979, p. 280), the share of
profits in income has fallen (Romer 1989, p. 63) and the growth rates of
countries have converged (Baumol and Wolff, 1988). However, the con-
sensus of opinion is that Kaldor's stylized facts are accurate broad generali-
zations.

Old Dominion Universily. The author benefited from the comments of two anonymous
referees. The Jerome Levy Economics Institute at Bard College sponsored the research.

Journal of the History of Economic Thought, 14, Spring 1992.

°1992 by the History of Economics Society. '
36
 KALDOR'S GROWTH THEORY 37

The Solow growth model can reproduce only five of Kaldor's six facts.
Exogenous technical progress can account for productivity growth (fact
# 1 ) . Although the capital-labor ratio is rising (fact # 2 ) , as long as labor is
measured in efficiency units the neoclassical model is consistent with
balanced growth (facts # 3 - # 5 ) (Solow 1970, pp. 2-3,33-38). However, the
Solow growth model cannot account for persistent differences in country
growth rates (fact # 6 ) (Lucas 1988, p. 4). Assuming a Cobb-Douglas-type
technology, identical capital shares, perfect capital markets and complete
mobility, the neoclassical model predicts that capital will flow from rich to
poor countries, where the rate of profit on capital is highest, and that the
capital flows are reinforced by the diffusion of technical progress from rich
to poor countries (Lucas 1990, p. 92). Moreover, the Solow model with an
exogeneous time trend cannot explain (as opposed to reproduce) long-run
growth, which renders the model ineffective for the analysis of the impact
of government policies on long-run growth.
Kaldor and many other economists in the 1950s and 1960s, including
Robert Solow himself, criticized the Solow model for its limited capacity to
reflect the stylized facts of growth (Solow 1959, p. 98). The growth
economists in the last decade who have repeated those old criticisms have
often adopted devices used in Kaldor's models—such as endogenous
technical progress, increasing returns, path-dependence (or "history mat-
ters"), the priority of the manufacturing sector, and the presence of
complementarities to avoid the limitations of the Solow model (Wulwick
1990, pp. 68-103). Indeed, the recent debate over how to model growth
reflects in many ways the Kaldor-Solow controversy of the 1950s. Kaldor,
who was from the University of Cambridge, and Solow, who was at MIT,
were embroiled in the Cambridge-Cambridge controversy over how to
model production. They, like many economists of the time, saw that
technical progress provided the main explanation for the rapid growth of
production, but their models of technical progress differed.
In Kaldor's model, technical progress was embodied in capital with the
result that the growth of demand for capital (up to a definite limit) would
induce technical progress. It was as if attempts to move along the neoclas-
sical production function led to shifts of that function. There were, in other
words, increasing returns to capital and scale (in the model, capital is a
complement to a growing fully-employed labor force). Kaldor thought that
this, by conflicting with the neoclassical "adding up" theorem for a perfectly
competitive economy, was "sufficient to cause the whole [neoclassical]
structure to collapse like a pack of cards" (Kaldor 1966b, p. 297). In light
of his model, Kaldor argued that demand management that promoted
capital investment promised to increase the rate of technical progress and
the potential rate of growth. Solow's models accepted that capital invest-
ment was needed for progress to occur, but he treated the potential rate of
38 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

progress as exogenous, which left the neoclassical marginal productivity

theory intact. Recognizing the presence of market failures that led to
deficient aggregate demand, Solow supported policies to increase the share
of output invested. That would ensure that technical progress occurs at its
potential rate and would increase the rate of growth of potential GNP
(Solow 1959, 1962). But "the asymptotic rate of growth," according to
Solow, "was independent of the share of output invested in the model."1
Despite the similarities in intent between the Kaldor growth models and
the new approaches to modeling economic growth, present-day economists
rarely have cited Kaldor's growth theory, as opposed to his stylized facts of
growth. This essay seeks to explain why this has been so by reference to the
changes in the nature of economics as a discipline since Kaldor developed
his growth theory. Section II discusses changes in Kaldor's reputation and
interests during the transitional period in economics from a literary to a
mathematical discipline. Section III contrasts two historic types of increas-
ing returns models, the type accepted by Kaldor which implies a dynamic,
macroeconomic equilibrium, and the type accepted by more orthodox
economists, which is consistent with a static, microeconomic equilibrium.
Sections IV and V discuss some problems that Kaldor's contemporaries
encountered in his theoretical and empirical models of economic growth.
The conclusion sums up the character of Kaldor as a creative literary
economist during the postwar transition of economics into mathematical
economics.

II. THE TRANSITIONAL PERIOD FROM LITERARY TO

MATHEMATICAL ECONOMICS
Kaldor's name in 1940 ranked seventh on the list of citations of econo-
mists in the top ten journals and received half as many citations as J. R.
Hicks. The economics profession and the size of the journals expanded
between 1940 and 1960. Nevertheless, in 1960 Kaldor's name ranked
sixteenth on the citation list, along with K. Arrow, and ranked above T.
Koopmans (Quandt 1976, p. 754). But while Kaldor was a major economist
in the 1940-1960 period, after 1960 his name no longer appeared on the list
of leading citations. His growth theory, developed in the period after 1957,
has received little attention outside a coterie of supporters centered in
England and Hungary (Kaldor was an Hungarian by birth). What hap-
pened to change Kaldor and the profession's opinion?
Kaldor intended his research between 1930 (when he received a B.S.
from the London School of Economics) and 1940 to contribute to main-
stream economic theory. He knew little differential calculus, Paul Samuelson

1. Letter to the author from R. M. Solow, January 30,1991.

 KALDOR'S G R O W T H THEORY 39
commented, but when confronted with a model of the trade cycle or
production, he saw "how it had to go" (Samuelson 1988, p. 321). After
World War II, Kaldor's research changed tack. Its aim became to develop
an alternative model to the one used by mainstream economists. Kaldor's
experiences as an applied economist during World War II and the postwar
reconstruction promoted this change of interest. He served as Chief of the
Planning Staff of the US Strategic Bombing Survey (1945), worked on the
Hungarian economic plan and was Director of the Research and Planning
Division of the UN Economic Commission for Europe (1947-1949). Work-
ing with data for the first time convinced him of the senselessness of a priori
theorizing. He also met G. Myrdal, I. Svennilson and P. J. Verdoorn,
economists who later influenced his growth theory. At the ECE, Kaldor
became tangled in some "behind-the-scene-machinations" about the
Marshall Plan (Thirwall 1987, pp. 104-5). The Foreign Office subsequently
pressured L. Robbins who headed the economics department at the LSE to
terminate Kaldor's leave of absence. In response to Robbins's attitude on
this matter, Kaldor resigned his readership at the LSE and went to Cam-
bridge. This move encouraged the change in Kaldor's theoretical perspec-
tive (Kaldor 1989, p. 21).
During the postwar years, Kaldor continued his career as a policy adviser.
He designed several fiscal proposals to promote equity as well as economic
growth. His proposal of an expenditure tax in several underdeveloped
countries, the incidence of which would fall largely on the wealthy, caused
many an ugly political situation. Serving as part-time adviser to Chancellor
of the Exchequer under Wilson's first Government, Kaldor devised the
selective employment tax scheme (SET) that taxed the service sector while
subsidizing the manufacturing sector, a discriminatory measure that aroused
intense opposition. R. Harrod commented in his review of Kaldor's Essays
on Economic Policy that "Kaldor has recently become a household word. It
is probably known to Kaldor that in some households he has become a sort
of bogey man" (Harrod 1965, p. 794). In recognition of his continuing
contribution to economic policy-making, Kaldor was appointed as a mem-
ber of the House of Lords in 1974, where he played an active role.
Once Kaldor set out to criticize mainstream economics, he went about it
like a bull in what he saw as a china shop.2 His behavior in this matter
contrasted with that of Joan Robinson, another critic of neoclassical
economics from Cambridge-England, with whom he maintained a close
relationship, though not one free of argument and rivalry. Though Kaldor
visited the University of California-Berkeley department as a Ford Re-
search Professor for a year in the late 1950s, he never was graced with the
attention from the American establishment gained by Robinson.1 Robinson
2. Philip Mirowski suggested this description to the author.
3. This information was supplied to the author by R. W. Clower.
40 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

initiated her onslaught on neoclassical economics from a position of great

influence. In the minds of many economists in the 1950s, she was a virtual
icon, the creator of a model of imperfect competition and the first exponent
of Keynes. Robinson personally engaged Solow and Samuelson in abstract
debates, while Kaldor rarely addressed his opponents by name. Although
her papers used fewer equations than Kaldor's, as a critic Robinson
revealed a keener sense of what was topical and an eye for technical
problems.4 Thus Samuelson and Solow could solve to their satisfaction her
apparently fundamental criticism of orthodox theory, on the basis of the
illegitimacy of a scalar aggregate of capital, by technical developments
within the orthodox framework (Samuelson 1988, p. 328; 1989, pp. 137-38).
What drew the attention of the economics profession to Kaldor was his
insistence that economics start with a "'stylized' —i.e. non-rigorous but
suggestive —description of a modern economy" (Samuelson 1963, p. 197).
This prescription, for a discipline ostensibly modeled after physics and
intended to have a practical outcome, offered a welcome relief from the
view of Robbins and Friedman that economic assumptions are untestable
and the realism of these assumptions is unimportant. A philosophical
realist, Kaldor was eager to explain reality by means of a plausible model
and could not countenance the reliance of orthodox economists on math-
ematically tractable models of optimization (Lawson 1989). He was critical,
in particular, of Solow's custom of taking the line of mathematical least
resistance. It was this custom that led neoclassical growth theory to
maintain the simple deductive framework imposed by the Cobb-Douglas
production function, despite the obvious cost in terms of explanatory power
of making technical progress exogenous. Kaldor believed that the superim-
position on economics of the differential calculus carried out by "the
brilliant minds of M.I.T.," - their use of convex production functions that
were "linear-homogeneous and 'well-behaved' (with isoquants asymptotic
to the axes)," made it impossible to deal with the stylized facts of growth
(Kaldor 1966b, pp. 81, 83).
Kaldor belonged to the last generation of economists whose credentials
did not require mathematical expertise. As his biographer, A. P. Thirlwall,
wrote, during Kaldor's undergraduate education "he was...weak in math-
ematics and had to take private tuition before taking his exam a second time.
After passing he wrote with apparent relief to his private tutor that he could
'now enjoy the glorious prospect that I have nothing to do with mathematics
in the future'"(Thirlwall 1987, p. 19). Three decades later Kaldor would
return to mathematics in order to offer an attractive alternative to neoclas-
sical economics. His mathematics resembled Keynes's in style, though
Kaldor lacked Keynes's training in the field. Much of Kaldor's mathematics
4. This was Ihe opinion, for instance, of A. S. Eichner (conversations with the author,
American Economic Association meetings, December 28-30,1988).
 KALDOR'S GROWTH THEORY 41

was algebra, with each algebraic term having an observational counterpart.

His mathematical arguments started off with an identity, which he elabo-
rated into a causal relation, in the same way that Keynes's investment
multiplier can be derived from the GNP accounting identity. Even with the
help of D. G. Champernowne (Kaldor 1957), K. Mirrlees (Kaldor 1962),
and F. Hahn and L. Pasinetti (Kaldor 1958), Kaldor's lack of experience in
solving mathematical problems inevitably meant that he would fail occa-
sionally to see the ramifications of his own mathematical models, which got
him into serious trouble.
The 1950s, when Kaldor initiated his growth theory, was a period of
debate about the character of economics as a formal discipline. Harrod and
Austin Robinson, then joint editors of the Economic Journal, requested that
the mathematical formulations in submissions to the journal be translated
into "ordinary economic language" (March 1954, pp. 1-2). The editor of the
Review of Economics and Statistics, Seymour Harris, faced by a plethora of
mathematical submissions that neither he nor most subscribers could read,
initiated a symposium on the pros and cons of mathematical economics. Of
the several economists invited to oppose the formalization of economics
only D. Novick was willing to do so, while nine mathematical economists
stated arguments generally in favor of formalization. Several of those
economists remarked that Novick suffered a disadvantage as the critic of a
technique which he had not mastered and thus could not fully appreciate.
Kaldor like Novick was a nonmathematical economist who criticized the
formalization of economics while supporting the use of mathematical
terminology to state hypotheses that originated with observations. Kaldor
also shared the view stated by his collaborator, Champernowne, at the RES
debate that "the ability to judge the relevance of an economic theory and its
conclusions to the real world is but rarely associated with the ability to
understand advanced mathematics. An important article on economic
theory is therefore likely to be wasted unless it can be set out in prose
supported by the most elementary mathematics," including diagrams
(Champernowne 1954, p. 369). Unfortunately, in attempting thirty years
ago to use elementary algebra and geometry to model mathematically how
endogenous technical progress generates economy-wide increasing re-
turns, Kaldor lit on a crucial problem with which not only he but most
economists then were mathematically ill-equipped to deal (Hahn 1989, p.
47).

III. THE AMBIGUITY OF THE CONCEPT OF INCREASING

RETURNS
Postwar growth models typically have generated growth by assuming the
presence of technical progress. When models make technical progress
42 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

endogenous, the result typically is increasing returns to scale, because of the

special contribution of labor or of capital, depending on the model in
question. There have been two main types of models with endogenous
technical progress and increasing returns, namely models that assume a
staticequilibrium and those which do not. Kaldor thought that the existence
of increasing returns prevents a static general equilibrium. The models of
increasing returns of P. Romer, who is a former student of R. E. Lucas and
major growth theorist, assumed the existence of a static general equilibrium
(Romer 1989).
Those two conflicting treatments of increasing returns date at least as far
back as the first three chapters of Adam Smith's The Wealth of Nations
(1776). Alfred M&rs\\&\Ys Principles of Economics, which developed Smith's
theme of economic development, identified two sources of increasing
returns, namely internal economies dependent on the resources of the firm
and external economies that accompanied the general progress of industry.
The firm's supply curve depended on both internal and external economies,
which meant that marginal cost varied with scale and historical time.
Though Marshall used the ordinary supply curve to analyze the benefits of
a subsidy to increasing returns industries, he cautioned that "the uses of the
statical method in problems relating to very long periods are dangerous"
(Marshall 1898/1926,Bk5,Ch.5:8). The problem posed by Marshall of how
to model increasing returns led to a debate in the Economic Journal during
the interwar period. Piero Sraffa (1926) developed Marshall's argument
against the use of the firm's supply curve to analyze increasing returns.
Responding to Sraffa, A. C. Pigou as Marshall's protegd used the static
model of supply and demand to analyze a subsidy for the competitive,
increasing returns industry (Pigou 1927).
Kaldor learned about Smith's and Marshall's concept of increasing
returns from a third contributor to that interwar debate, A. A. Young, a
former Harvard professor who was head of the Department of Economics
at the LSE during 1926-1928 and Kaldor's "first real teacher in economics"
(Kaldor 1989, p. 12). Young, an avid reader of Marshall's Principles,
identified two external economies which involved the use of machinery to
transform a group of complex tasks into a series of simpler ones. What were
internal economies for the firm were external economies from the point of
view of the firms with which that firm dealt. But "not all of the economies
which are properly to be called external," Young explained, "can be
accounted for by adding up the internal economies of all the separate
firms.... [The] operations [of the firm] change in the sense that they are
progressively adapted to an increasing output, but they are kept within
definitely circumscribed bounds" (Young 1928, p. 528). Interacting with
changes at the firm level, the division of labor—the development of
specialized products and industries —occurred at the macro-level. Accord-
 KALDOR'S GROWTH THEORY 43

ing to Young, those macro-level external economies, while consistent with

competition (involving as they did "large production" as opposed to "large-
scale production"), were inconsistent with a static general equilibrium
(ibid., p. 531). As Young stated,

the size of the market is determined and defined by the volume of

production.... Modified, then, in the light of this broader concep-
tion of the market, Adam Smith's dictum [that the division of labor
depends on the size of the market] amounts to the theorem that the
division of labour depends in large part upon the division of labour.
This is more than mere tautology. It means...that the counter forces
which are continually defeating the forces which make for economic
equilibrium are more pervasivc.than we commonly realise. Not
only [are] new or adventitious elements, coming in from the
outside, but...every important advance in the organisation of
production...alters the conditions of industrial activity and initiates
responses elsewhere in the industrial structure which in turn have
a further unsettling effect. Thus change becomes progressive and
propagates itself in a cumulative way (ibid., p. 533; emphasis
added).
Kaldor, following Young's teaching, studied those increasing returns which
resulted in a moving (as opposed to a static) equilibrium at the
macroeconomic level.
In contrast, Romer, who read Young's 1928 essay as a graduate student,
concluded that "in an article published in 1928, Allyn Young gave a verbal
discussion of economic growth driven by increasing returns resulting from
specialization. Much of this discussion is problematical, but he seems to
have understood...that externalities could permit the existence of a com-
petitive equilibrium" (Romer 1983, p. 1). Romer adopted Young's idea of
internal economies that are other firms's external economies. The "Marshall-
Young-Romer model" was based on an increasing returns to scale produc-
tion function with labor and resources devoted to producing a range of
specialized inputs, which acted like a technological externality (Romer
1986b; 1989,pp. 108-13). Like models with external effects, Romer's model
generated a general equilibrium that was suboptimal, with the set of
specialized inputs liable to increase, given suitable subsidies.
Young's 1928 essay inspired both the Romer growth and the Kaldor
growth models, which each served as the basis for government intervention.
Nevertheless, the approaches of the two models are different. Romer's
basic model incorporated an externality into the static production function
of the firm, an approach that fit well with standard economics. Kaldor's
model of production with increasing returns concerned only a dynamic
equilibrium at the macroeconomic level (Kaldor 1966a, 1972). It is difficult
44 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

for most contemporary economists, given an education that stresses

microeconomics and excludes the history of economic thought, to under-
stand the latter approach.

IV. KALDOR'S FORMAL MODEL O F INCREASING RETURNS

Most developments in economics are incremental and based upon previ-
ous work. Kaldor's theory of increasing returns was part and parcel of the
growth economics of the 1950s. Kaldor, like most economists of that
decade, believed that technical progress provided the source of productivity
growth and mainly was embodied in capital equipment. Applying standard
tools to formalize his theory, Kaldor's algebra was based on Harrod's well-
known equation, and his graph shared the axes of the steady-state economic
growth models of the time (Hahn and Matthews 1964). In addition, Kaldor,
who presented several alternative growth models, had the reputation of
changing his mind more often than most economists. Because of his
dependence on the mathematical skills of his colleagues, Kaldor's models
sometimes said things he apparently did not mean and later would alter.
The simplicity of the mathematical models also meant that Kaldor could not
include complex ideas he later emphasized. Thus Kaldor never produced
a formal model of cumulative causation, though his more literary papers
which discussed it have recently inspired mathematical economists (Boyer
and Petit 1990, pp. 488-92).
We shall use the following notation for the variables pertaining to
Kaldor's model of increasing returns:
A a scaling factor r = rate of exogenous
technical progress
c = the constant in the V = Verdoorn elasticity
Verdoorn "law"
I = dK X = labor employed

K * capital stock , = ^ l o g (*»

dt
d(\og
v &\(K))
))
k s Y s output
dt
m s manufacturing v = —' ° ^ ' "
y
dt
P = labor productivity Y/X * = least squares estimates
L , •• •
=y -x n, a, b = output elasticities
 KALDOR'S GROWTH THEORY 45

Figures 1 and 2 present Kaldor's technical progress function T which,

Figure 2
Figure 1 Kaldor-s Figure 1 (1957)
Kaldor's Figure 5 (1957)

y-x

(y-x)*

ll ll ll (k-x)* k-x
K, K 2 K3

relating as it did the growth of output (per worker) to the growth of capital
(per worker), constituted his basic model of increasing returns. The
intersection of the technical progress function (the position and shape of
which would vary by country) and the 45° line represented the position of
long-run growth equilibrium at which the rate of profit on capital, the
capital-output ratio and the distribution of income were constant, in line
with the stylized facts of growth # l - # 5 . Kaldor stated the technical
progress function either by equation
y=Plk + al (1)
as represented by Figure 1, where oc,/(l - p,) is the equilibrium rate of
growth of output (Kaldor 1957, pp. 276), or equation
y-x=P2(k-x) (2)
written for convenience in a linear form, with cc2 / (1 - p2) as the equilibrium
rate of output growth (Kaldor 1958, pp. 215-16). Both equations 1 and 2
hold only if the constant a, = a 2 and the slopes p,, p2 = 1, a restriction that
Kaldor would not accept.
Kaldor intended the technical progress function as an alternative to the
model of economic growth offered by the Cobb-Douglas production func-
tion,
Y = 0< n<1 (3)
Assuming the existence of perfect competition, the output elasticities equal
the income shares of labor and capital. Normalizing for labor, taking
logarithms and differentiating with respect to time results in the equation,
y-x = r\(k -x) + r (4)
46 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

It was Kaldor's view that equations (3) and (4) artificially separated the
growth of the capital-labor ratio from the rate of technical progress which
made productivity grow. "The use of more capital per worker," Kaldor
explained, "inevitably entails the introduction of superior techniques which
require 'inventiveness' of some kind," while most "technical innovations
which are capable of raising the productivity of labor require the use of
more capital per man" (Kaldor 1958, p. 264). Solow's (1957) pathbreaking
estimation of the exogenous rate of technical progress r in equation (4),
Kaldor remonstrated, was "purely circular." The slope of the curve (q) was
"supposed to determine the share of profits in income, the share of profits
is taken to be an indication of the slope, and the residual is then attributed
to the shift of the curve! There could be no better example," he concluded,
"of post hoc ergopropter hoc" (ibid., p. 33).
Yet, shortly after the publication of the linear technical progress function
several economists informed Kaldor that the technical function (as stated
in equation 2) was formally identical to the Cobb-Douglas production
function in growth form (equation 4, p2 = q, r = a j . Virtually every survey
on growth economics has emphasized that fact. Of course, mathematical
oversights of all sorts abound in the economics literature (Tinbergen 1988,
p. 225). Solow's (1956) presentation of neoclassical growth theory was
replete with a mislabeled graph (Figure 1) and errors of calculus, which he
later acknowledged (Solow 1969, p. 87). Those errors, however, hardly
diminished the clarity of the author's argument. Kaldor, recognizing that
the mathematical oversight threw his whole argument into question, re-
sponded that the representation of increasing returns required a function
which took into account the distribution of capital and labor between
sectors and the rates of growth of output and labor, so that there is no unique
relation between the levels of output and factor input (Kaldor 1958, pp. 44-
45, n. 2). Several years later Kaldor broached such a function in the form
of an econometric model of increasing returns. The coherence of Kaldor's
growth theory rested on that econometric exercise.

V. KALDOR'S EMPIRICAL MODEL OF INCREASING RETURNS

Applied economists of the interwar period recognized the importance of
increasing returns in studying economic growth. The notion appeared in
the works of Young's pupil G. T. Jones (1933) and Jones's former research
assistant C. Clark (1940), as well as in the League of Nations study papers
(Svennilson 1954, p. 57). S. Fabricant (1942) at the National Bureau of
Economic Research found the rank correlation coefficient between changes
in the levels of output per unit of employment and output (Scott 1989, p.
361). Svennilson (1945) of the University of Stockholm regressed the
growth of labor productivity on the growth of output in Swedish industries.
 KALDOR'S G R O W T H THEORY 47
Finally, Verdoorn, an economist at the Dutch Central Planning Bureau,
who worked under Kaldor when Kaldor was Director of Research and
Planning for the Economic Commission for Europe, studied the relation
between industrial productivity growth and industrial output using cross-
country data for the period 1924-1938 (Verdoorn 1949).
Thirlwall and Dixon (1975, p. 209) showed that the Svennilson and the
Verdoorn models could be derived from Kaldor's technical progress func-
tion, as follows. Define the constant in the technical progress function in
equation (2) as disembodied technical progress, which includes an autono-
mous component r, and K. Arrow's (1962) "learning by doing" dy,
<x2=rx+dy 0 < dl < 1 (5)
Technical progress is embodied in capital which is partly autonomous r2 and
partly induced d2y,
k-x =r2 + d2y 0<d2<l (6)
Substituting (5) and (6) into the technical progress function in (2) yields (7)
relating the growth of labor productivity to the growth of output, without
the term for capital which according to Kaldor (1960) could not be mea-
sured consistently over time,
p = c + vy (7)

c = rl + a/2 v = dx + a/l2 0< v< 1

Kaldor estimated equation (7) using cross-country data for the manufac-
turing sector, arrived at an estimate of v close to the result of Verdoorn's
regression analysis and dubbed the estimate (on Arrow's suggestion) the
"Verdoorn law" (Kaldor 1966a, p. 106).5 Having cited the writings of Smith,
Marshall and Young, Kaldor explained that the law arose as a result of
"increasing returns, which causes productivity to increase in response to, or
as a by-product of the increase in total output" (ibid.). The Verdoorn model
accounted for differences in countries' growth rates, or stylized fact #6.
Kaldor presented his estimates of the Verdoorn "law" as part of a larger
model in his inaugural address just when he was leaving his advisory
position at the Treasury and accepting a chair at Cambridge (Thirlwall
1983). Because of Kaldor's political position, his lecture received more
publicity than the usual academic lecture.6 Kaldor gave the lecture soon
after the enactment of the national selective employment tax scheme (SET)

5. Kaldor stated that Arrow originated the name "Verdoorn law" (letter to the author
from N. Kaldor, July 18,1983), but Arrow does not recall having done so (phone conversation
with K. Arrow, July 27,1983). He did not use the term in his 1962 essay.
6. Letter to the author from N. Kaldor, October 12,1983.
48 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

and shortly before the regionalization of the scheme (see Wulwick 1983).
Officially SET, which taxed employers in the service sector and subsidized
employers in manufacturing, was intended partly to raise Britain's rela-
tively low rate of economic growth.
Kaldor's regression analysis, like a lot of policy-oriented statistical re-
search was simple. The lack of sophistication also was a product of his time
and place. The Cowles Commission established econometrics in the United
States in the 1940s. Yet one could find virtually no econometrics course
taught in British universities in the 1950s. Students who wanted economet-
rics would spend time at an American university or try to teach themselves.
Still, notable applied econometrics work was being done at the Department
of Applied Economics (DAE) at Cambridge. Under R. S. Stone's direction,
the D A E developed the single errors-in-equation model. The criteria for
an acceptable model were correctly signed coefficients, a high R2, statisti-
cally significant coefficients, and a Durbin-Watson statistic within the
appropriate bounds (Gilbert 1991, p. 293). By the early 1960s, the errors-
in-equation model had become the standard method of estimation. Mean-
while the DAE shifted its interest to economic planning and the center of
time series analysis moved to the London School of Economics. The LSE
offered its first econometrics course in 1962 and introduced the M.Sc.
program in econometrics in 1967. At this time, econometrics came into
vogue in business and government (Johnston 1967, p. 3). Having received
the advance copies of Kaldor's inaugural lecture sent out by the press office
of the Treasury, The Economist remarked upon the "present Treasury
shibboleth..., the language of regression equations" (November 5,1966, p.
547).
Regression analysis at that time served mainly to estimate linear stochas-
tic models based on formal theory, with little attention paid to testing
alternative hypotheses. Kaldor, not unlike other policy-oriented econo-
mists, was prone to treat his least-squares estimates as measures of the true
model. Kaldor reported the R2, the standard error of the regression
coefficient and the standard error of the regression as a proportion of the
mean value of the dependent variable, but did not give these numbers a
statistical interpretation. His sample covered twelve of the twenty-two
OECD countries for the years 1953/54-1963/64, a period that he thought
was long enough to eliminate cyclical movements (Kaldor 1968, p. 390).
The data, which are available in Kaldor's publications, were based on
statistics from the recently developed data banks of the OECD and the UN
and were cross-sectional, with one exponential growth rate for each vari-
able for each country for the whole sample period. Like many economists,
Kaldor ignored the issue of the quality of the data. The manufacturing data
were fairly reliable, though the aggregate output and employment series
were not always comparable (Lomax 1964, p. 10). The service sector data
 KALDOR'S GROWTH THEORY 49

were unreliable because the indicators of real output of many (though not
the majority) of its activities were based only on employment data (Hall and
McGibbon 1966, p. 47). Like many international studies, Kaldor's study
neglected to explain its choice of sample, which included some but not all
of the industrialized OECD countries. The availability of data on produc-
tivity may explain Kaldor's choice of countries. Productivity was computed
in base year prices as the ratio of the index of net output to the index of labor
input at the industry level, with sizeable statistical adjustments made at the
aggregate level (OECD 1966, p. 19). Continuous OECD series usually
were limited to annual figures for 10-year periods and the OECD published
a productivity index only for twelve countries. Kaldor's sample covered
eleven of these countries, but excluded Ireland, which was not industrial-
ized, and included Japan.
Kaldor specified the Verdoorn law using two alternative specifications, of
manufacturing productivity growth on manufacturing output growth and
manufacturing employment growth on manufacturing output growth. Later,
he and others insisted that only the latter specification serves as the true test
of the law (Kaldor 1975, pp. 891-92; Scott 1989, p. 340). It is worth
emphasizing that as long as one's data are consistent, only one of these
specifications needs to be estimated since, by definition,)** = x* +p*. Using
ordinary least-squares, the three measures £m = 1.0005*m + 1.0037pm,Jcm =
0.9985yM - 1.0017/?;n, and pm = 0.9960ym - 0.9959*n show'that Kaldor?s data
were consistent. In this case, just one of Kaldor's two equations—;tm on.ym
orpm onym — suffices to provide the relevant information. No wonder that
Kaldor arrived at identical standard errors of the slope coefficient v in both
of his regressions. In addition, the standard error of the regression and the
log likelihood of the estimates of the two regressions are identical because
the data are consistent.
In order to state the relevant statistical tests, I have reestimated the
Verdoorn "law" using Kaldor's data. The estimates, which came very close
to Kaldor's, are:

pm = 1.048 + 0.480y;n R2 = 0.823 (8)

standard error: (0.462) (0.070) ser/ym = 0.17

significance level: (0.047) (0.000) DW = 2.63 F-stat = 46.53

The value of the Durbin-Watson statistic borders on the indeterminate
range, but suggests negative serial correlation at the 5 percent significance
level. There is strong evidence that the residuals are normal (by the
Lillefors test), which validates the use of the statistical test of significance.
The likelihood ratio indicates that the probability that there is no Verdoorn
50 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

relation (that v = 0) is virtually zero.7

Several economists criticized Kaldor's estimation of the Verdoorn model
(Bairam 1987; Wulwick 1991). The sharpest critic was R. E. Rowthorn, a
junior colleague of Kaldor's in the Cambridge University economics de-
partment who was reputed to be a Marxist. Rowthorn argued that the
Verdoorn coefficient dp/dy measured returns to labor, rather than to scale,
which would follow if one assumed an underlying production function
(Rowthorn 1979, pp. 132). It is the case that Verdoorn derived the elasticity
of productivity in respect to output pjym from the production function,

ym = KK (a + b)>l (9)

Taking logs, differentiating with respect to time and dividing by x Verdoorn

arrived at

Next he defined the elasticity of productivity with respect to output (V),

and substituted that definition into (10) which resulted in the Verdoorn
elasticity

(12)

*• •(£)
Kdepends on the ratio of the growth of capital to the growth of employment
and both output elasticities, implying that the Verdoorn relation pertains to
the growth of total labor productivity (Thirlwall 1980, p. 387). The slope of
the Verdoorn law (in equations 7 and 9) then is

7. The log likelihood of equation 8 is -11.319 and the null hypothesis ofpm regressed on
the constant is -21.712. Given the chi-square of one degree of freedom, the probability in
favor of the null hypothesis in a large sample is less than 0.005.
 KALDOR'S GROWTH THEORY 51

%- = *-•*• (13)
K
dy y '
s
In contrast, Rowthorn substituted the identityxn ym -p into equation (10)
which yielded

=(f) *-
as the neoclassical model underlying the Verdoorn "law." The slope

d£_ _ b-\
(15)
dy b

which depended solely on labor's output elasticity could not measure (as
Kaldor interpreted it) returns to scale or the effects of technical progress.
The competing underlying models of the slope of the Verdoorn law in
equations (13) and (15) are observationally equivalent, so that there is no
logical basis for choosing between them. This problem vitiated Kaldor's
attack on the production function approach to modeling growth.

VI. CONCLUSION
Kaldor's writings mark him as one of the great literary economists of our
age, but his growth theory has received little recognition from mainstream
economists. The primary motivation of his growth theory was critical. He
considered that the constraints imposed on neoclassical growth theory as a
consequence of the use of calculus and the need for mathematical tractabil-
ity were unwarranted. Unfortunately, his own formal models could be
derived from —but did not necessarily imply—the neoclassical model that
he set out to criticize. The dismissal of Kaldor by economists like H. Varian
("Kaldor had no theory") and P. Romer (Kaldor lacked "a tractable
alternative" to the Solow growth model), however oblivious to Kaldor's
insights into the growth process, is not entirely unreasonable (Romer
1986a, p. 1005).8 "Mathematics," as Samuelson once warned, "can certainly
be a hindrance, since it is only too easy to convert a good literary economist
into a mediocre mathematical economist" (Samuelson 1952, p. 65). In this
context, the limitations of Kaldor as a growth theorist were a largely product
of his time, the transition of economics from a mainly literary to a mainly
mathematical discourse.

8. H. Varian in conversation with the author at the Conference on Studies in Alternative

Research Programmes in Capri, October 15-18,1989.
52 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

REFERENCES
Arrow, K. J. 1962. "The Economic Implications of Learning by Doing,"
Review of Economic Studies, 29, 155-73.
Bairam, E. I. 1987. "The Verdoorn Law, Returns to Scale and Industrial
Growth: A Review of the Literature,"/4utfra//a/z Economic Papers, 26,
20-42.
Baumol,W.J. and E.N.Wolff. 1988. "Productivity Growth, Convergence,
and Welfare: Reply," American Economic Review, 78,1155-59.
Boyer,R.andP.Petit. 1991. "Kaldor's Growth Theories: Past, Present and
Prospects for the Future," in E. J. Nell and W. Semmler, eds., Nicholas
Kaldor and Mainstream Economics, Macmillian, London.
Champernowne, D. G. 1954. "On the Use and Misuse of Mathematics in
Presenting Economic Theory," Review of Economics and Statistics, 36,
369-72.
Clark, Colin 1940. The Conditions of Economic Progress, Macmillan, Lon-
don.
Economist. 1966. "The Explanation of SET," 5 November, pp. 547-78.
Gilbert, C. L. 1991. "Richard Stone, Demand Theory and the Emergence
of Modern Econometrics," Economic Journal, 101, 288-302.
Haache, G. 1979. The Theory of Economic Growth: An Introduction,
Macmillan, London.
Hahn, F. 1989. "Kaldor on Growth," in T. Lawson, J. G. Palma and J.
Sender, eds., Kaldor's Political Economy, Academic Press, London.
. and R. C. O. Matthews. 1964. "The Theory of Economic Growth:
A Survey," Economic Journal, 74, 779-902.
Harrod, R. 1965. "Review: Essays on Economic Policy, by N. Kaldor,"
Economic Journal, 75, 794-802.
Hill, A. T. P. and J. McGibbon. 1966. "Growth of Sector Real Product,"
Review of Income and Wealth, ser. 12,35-55.
Johnston, J. 1967. "Econometrics: Achievements and Prospects," Three
Banks Review, no. 73, 3-22.
Jones, G. T. 1933. Increasing Returns, Cambridge University Press,
London.
Kaldor, N. 1957. "A Model of Economic Growth," Economic Journal, 67,
591-624.
. 1958. "Capital Accumulation and Economic Growth", in F. A. Lutz
and D. C. Hague, eds., Theory of Capital, Macmillan, London.
. 1960. "A Rejoinder to Mr Atsumi and Professor Tobin," Review of
Economic Studies, 28, 121-23.
. 1966a. Causes of the Slow Rate of Economic Growth of the United
Kingdom, University of Cambridge Press, Cambridge.
. 1966b. "Marginal Productivity and the Macro-economic Theories
of Distribution," Review of Economic Studies, 33, 309-19.
 KALDOR'S GROWTH THEORY 53

. 1968. "Productivity and Growth in Manufacturing Industry: A

Reply," Economica, 35, 385-91.
. 1972. "The Irrelevance of Equilibrium Economics," Economic
Journal, 82, 1237-55.
. 1975. "Economic Growth and the Verdoorn Law: A Comment on
Mr Rowthorn's Article," Economic Journal, 85, 891-95.
. 1980. "The Economics of the Selective Employment Tax," in N.
Kaldor, Reports on Taxation 1, Holmes and Meier, New York.
. 1989. "Recollections of an Economist," in J. Kregel, ed., Recollec-
tions of Eminent Economists, 1, New York University Press, New York.
and Mirrlees, K. 1962. "A New Model of Economic Growth,"/tev/ew>
of Economic Studies, 25, 174-92.
Lucas, R. E. 1988. "On the Mechanics of Economic Development/Vowr/ia/
of Monetary Economics, 22, 3-42.
. 1990. "Why Doesn't Capital Flow from Rich to Poor Countries,"
American Economic Review, 80, 92-96.
Novick, D. 1954. "Mathematics: Logic, Quantity, and Method," Review of
Economics and Statistics, 36, 357-58.
OECD. 1966. Productivity Measurement, 3, Paris.
Quandt, R. E. 1976. "Some Quantitative Aspects of the Economics Journal
Literature," Quarterly Journal of Economics, 85, 741-55.
Romer, M. 1983. "Dynamic Competitive Equilibria with Externalities,
Increasing Returns and Unbounded Growth," unpublished Ph.D. dis-
sertation, University of Chicago.
. 1986a. "Increasing Returns and Long-Run Growth," Journal of
Political Economy, 94, 1002-37.
. 1986b. "Increasing Returns, Specialization, and External Econo-
mies: Growth as Described by Allyn Young," University Center for
Economic Research Working Paper, no. 64.
. 1989. "Capital Accumulation in the Theory of Long-run Growth,"
in R. J. Barro, ed., Modem Business Cycle Theory, Harvard University
Press, Cambridge.
Rowthorn, R. E. 1979. "A Note on Verdoorn's Law," Economic Journal,
89, 131-33.
Samuelson, P. A. 1988. "The Passing of the Old Guard," Eastern Economic
Journal, 14, 319-30.
. 1989. "Remembering Joan," in G. Fe'vwe\,ed., Joan Robinson and
Modern Economic Theory, New York University Press, New York.
Scott, M. F. 1989. A New View of Economic Growth, Clarendon Press,
Oxford.
Solow, R. J. 1956. "A Contribution to the Theory of Economic Growth,"
in J. E. Stiglitz and H. Uzawa, eds., Readings in the Modern Theory of
Economic Growth, MIT Press, Boston, 1969.
54 JOURNAL OF THE HISTORY OF ECONOMIC THOUGHT

. 1957. "Technical Change and the Aggregate Production Function,"

Review of Economics and Statistics, 39, 312-20.
. 1959. "Investment and Technical Progress," in K. J. Arrow, S.
Karlin, P. Suppes, eds., Mathematical Methods in the Social Sciences,
Stanford University Press, Stanford.
. 1962. "Technical Progress, Capital Formation, and Economic
Growth," American Economic Review, 52, 76-92.
Svennilson, I. 1945. "Industriarbetets vaxande avkastning i belysning av
svenska erfarenheter," in E. F. Heckscher, ed., Studier I Ekonomi Och
Historia, Almqvist and Wicksells, Uppsala.
. 1954. Growth and Stagnation in the European Economy, United
Nations Economic Commission for Europe, Geneva.
Tinbergen, J. 1988. "The Role of Errors in Scientific Development,"
Review of Social Economy, 46, 225-30.
Thirlwall, A. P. 1980. "Rowthorn's Interpretation of Verdoorn's Law,"
Economic Journal, 90, 386-88.
. 1983. "Introduction: A Plain Man's Guide to Kaldor's Growth
Laws," Journal of Post Keynesian Economics, 5, 341-58.
. 1987. Nicholas Kaldor, Wheatsheaf Publishers, Brighton.
. and Dixon, R. 1975. "Regional Growth-Rate Differences on
Kaldorian Lines," Oxford Economic Papers, 26, 201-14.
Verdoorn, P. J. 1949. "Factors that Determine the Growth of Labour
Productivity," L'Industria, English translation available upon request
from A. P. Thirlwall, University of Kent at Canterbury.
Wulwick, N. J. 1983. "Can Keynesian Economics Be Scientific: An
Historical Reconstruction," Eastern Economic Journal, 9, 190-204.
. 1990. "The Mathematics of Economic Growth," The Jerome Levy
Institute at Bard College Working Paper, no. 38.
. 1991. "Did the Verdoorn Law Hang on Japan?" Eastern Economic
Journal, 17, 15-19.
Young, A. 1928. "Increasing Returns and Economic Progress," Economic
Journal, 38, 518-42.