Working Paper No.

803

What Do We Know About the Labor Share and the Profit Share? Part I: Theories
by
Olivier Giovannoni*
Levy Economics Institute of Bard College

May 2014

* Correspondence: ogiovann@bard.edu. Translation assistance was provided by Dam Linh Nguyen, while Lei
Lu provided research assistance for the technology section. Both Linh and Lei acknowledge financial assistance
from the Bard Summer Research Institute. All remaining errors remain the author’s sole responsibility.

The Levy Economics Institute Working Paper Collection presents research in progress by
Levy Institute scholars and conference participants. The purpose of the series is to
disseminate ideas to and elicit comments from academics and professionals.

Levy Economics Institute of Bard College, founded in 1986, is a nonprofit,

nonpartisan, independently funded research organization devoted to public service.
Through scholarship and economic research it generates viable, effective public policy
responses to important economic problems that profoundly affect the quality of life in
the United States and abroad.

Levy Economics Institute

P.O. Box 5000
Annandale-on-Hudson, NY 12504-5000
http://www.levyinstitute.org

Copyright © Levy Economics Institute 2014 All rights reserved

ISSN 1547-366X
ABSTRACT
This series of working papers explores a theme enjoying a tremendous resurgence: the
functional distribution of income—the division of aggregate income by factor share. This first
installment surveys some landmark theories of income distribution. Some provide a
technology-based account of the relative shares while others provide a demand-driven
explanation (Keynes, Kalecki, Kaldor, Goodwin). Two questions lead to a better understanding
of the literature: is income distribution assumed constant?, and is income distribution
endogenous or exogenous? However, and despite their insights, these theories alone fail to fully
explain the current deterioration of income distribution.

Subsequent installments are dedicated to analyzing the empirical literature (part II), to the
measurement and composition of the relative shares (part III), and to a study of the role of
economic policy (part IV).

Keywords: Wage Share; Labor Share; Profit Share; Ergodicity; Technology

JEL Classifications: D33, E24, E25

1
CONTENTS

1 INTRODUCTION: HOW BIG OF A SLICE DOES THE PIE-MAKER GET? .......... 3

2 KEYNES: HINTS, CONCERNS, BUT NO THEORY .................................................... 7

3 THE INCOME DISTRIBUTION POLITICAL ECONOMY OF MICHAL KALECKI

10
3.1 Assumptions and Framework ........................................................................................... 10
3.2 The Desired Share of Wages ............................................................................................ 12
3.3 Power Relations and Class Struggle................................................................................ 14
3.4 Policy Implications .......................................................................................................... 15

4 KALDOR AND PASINETTI: INCOME DISTRIBUTION FOR FULL

EMPLOYMENT ........................................................................................................................ 17
4.1 Stylized Facts ................................................................................................................... 18
4.2 The Model ........................................................................................................................ 18
4.3 The Pasinetti Critique ...................................................................................................... 19
4.4 Full Employment vs. Harrodian Dynamics ..................................................................... 21
4.5 A Full Employment Keynesian Model.............................................................................. 22

5 THE INCOME DISTRIBUTION BUSINESS CYCLE MODEL OF RICHARD

GOODWIN ................................................................................................................................. 25
5.1 The Model ........................................................................................................................ 25
5.2 Solution and Interpretation .............................................................................................. 28
5.3 Analysis and Contribution ............................................................................................... 30

6 TECHNOLOGY AND THE SHAPE OF THE PRODUCTION FUNCTION ............ 32

6.1 Production Functions: Cobb-Douglas vs. CES ............................................................... 33
6.2 Technical Change in the CES Production Function ........................................................ 37
6.3 Criticisms of the Cobb-Douglas Form............................................................................. 40

7 WHAT HAVE WE LEARNED? (NON)ERGODICITY AND THE ROLE OF

ECONOMIC POLICY .............................................................................................................. 44

8 REFERENCES ................................................................................................................... 47

2
1 INTRODUCTION: HOW BIG OF A SLICE DOES THE PIE-MAKER GET?

Of the tendencies that are harmful to sound economics, the most seductive, and in my opinion the
most poisonous, is to focus on questions of distribution. [...] But of the vast increase in the
well-being of hundreds of millions of people that has occurred in the 200-year course of the
industrial revolution to date, virtually none of it can be attributed to the direct redistribution of
resources from rich to poor. The potential for improving the lives of poor people by finding
different ways of distributing current production is nothing compared to the apparently limitless
potential of increasing production.
– Robert E. Lucas Jr. (2003)

This paper addresses the functional distribution of income, i.e., it discusses the size of the
slice of the economic pie going to each factor of production, as a reward. As the aggregate shares
in output are considered, the functional distribution of income is deeply rooted in macroeconomic
analysis. The functional distribution of income is not to be confused with the personal distribution
of income, which studies the distribution of income across individuals, or households, and which
has traditionally received a microeconomic treatment.2 However, it would be a mistake to pitch
both distributions against each other or to believe that they have nothing in common. As will be
shown, the same underlying forces may be shaping both the functional and the personal income
distributions, in the same way that there can be two related symptoms of the same disease.
The issue of income distribution is both old and new. On one hand, income distribution is
the oldest question in economics: How much of the economic pie does each factor, and therefore
social class, receive? Ricardo (1817) famously opens his magnum opus by elevating the question
to the principal problem of political economy:

2
The work of James K. Galbraith on the macroeconomic dimension of inequality is a rare and welcome exception.
3
 The produce of the earth – all that is derived from its surface by the united application of
labour, machinery and capital, is divided among three classes of the community, namely, the
proprietor of the land, the owner of the stock or capital necessary for its cultivation, and the
labourers by whose industry it is cultivated. But in different stages of society, the proportions of
the whole produce of the earth which will be allotted to each of these classes, under the names of
rent, profit, and wages, will be essentially different [...] To determine the laws which regulate this
distribution is the principal problem in Political Economy [...]
– David Ricardo (1817), p. 1

Besides Ricardo, economists have treated the topic in a way that can only be described as
bipolar (Solow, 1958). Going down history lane, the times of manic interest were under the
Physiocrats and classical economists (including, of course, Marx), the early 20th century and its
first statistical inquiries, as well as the 1950s and 60s. The depressive phases fill the gaps. The
topic fell notably into oblivion during the marginalists’ times as well as in the 30s, 40s, 70s and
80s. In the 1990s interest in income distribution grew, albeit for its most visible manifestation at
the time, the personal type (inequality). The 2000s saw a marked deterioration of the functional
distribution, in the U.S. and worldwide, and with it, an increase in the research devoted to
understanding the underlying causes. The topic has gained enormous traction since the
mid-2000s, an interest reinforced by the global crisis of 2008 and the greater availability of
distribution statistics (Giovannoni, 2013b).
A reason for the distribution of income to fall into oblivion may be that relative shares
have been fairly constant over long periods of time (Giovannoni, 2013c), so much so that the
relative constancy of the factor shares came to be considered alternatively a “bit of a miracle”
(Keynes 1939), a “stylized fact” (Kaldor 1961) or even a law (“Bowley’s law”).3 Long ago,
Keynes noticed that the stability of the labor share, over a fifty-five year period, is

A well-known statistical phenomenon [which] confirms the probability of constant or

diminishing, rather than increasing, profit per unit of output when output increases. I mean the
stability of the proportion of the national dividend accruing to labour, irrespective apparently of
the level of output as a whole and of the phase of the trade cycle. This is one of the most

3
A term coined by Samuelson (1964a) referring to the works on aggregate wages and national income of the British
statistician Arthur Bowley (1969-1957). See Bowley (1900, 1920, 1937)
4
surprising, yet best-established, facts in the whole range of economic statistics both for Great
Britain and for the United States [...] It is the stability of the ratio for each country which is chiefly
remarkable, and this appears to be a long-run, and not merely a short-period, phenomenon. [...]
– John Maynard Keynes (1939), pp.48-9

But the labor share cannot reasonably be considered constant anymore. The actual
stability of the distribution of income came to be put into question starting in the 1970s. Real
wages and labor productivity became disconnected, leading the US labor share to track downward
around that time. The trend became clearly visible in the 1980s (see Fig. 1). The good economy
and low unemployment of the late 1990s did produce an increase as expected,4 but it was unable
to invert the trend. The major recession of 2008 barely produced an uptick. In the 2000s, the trend
won, and the labor share has been decidedly falling, posing a serious puzzle to economists since
then. It is this puzzle, the reasons why the labor share has been falling while it used to be constant,
that this paper aims to elucidate. Just what is known about the labor and profit shares?
In this first of a four-part series, some theories devised during the “years of high theory”5
are investigated. The present paper highlights a few contributions characteristic of those years.
This paper does not claim any exhaustive account. Rather, it is better understood as consisting of
a series of lightposts, casting light at distant intervals, in such a way that a general direction
emerges but many areas remain in the darkness. To cast light onto those areas, the reader is
directed to the subsequent installments of this first theoretical part addressing, in turn, the
empirical evidence of the determinants of the labor share (part II), the labor share measures and
structural features (part III) and economic policies (part IV).

4
The labor share is procyclical; see Giovannoni (2013c) for more details.
5
The sentence harks back to the special issue of the Cahiers d’Économie Politique (Papers in Political Economy)
entitled “What have we learned on income distribution since the ‘years of high theory’?”, number 61. We here extend
the “years of high theory” from the 1920s to the 1970s.
5
 Figure 1 Labor Share of GDP in the U.S. Nonfarm Business Sector
74%

72%

70%

68%

66%

64%

62%

60%
1953
1947
1950

1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
2013
Source: BLS, productivity and costs tables

The present paper is roughly organized in a chronological way in order to highlight the
progression (or, the differences, departures) in economic thought. In turn, it will show the hints
left by Keynes (section 2) as well as the contributions of Kalecki, Kaldor and Pasinetti and
Goodwin (sections 3, 4 and 5, respectively). Section 6 describes the role of technology and the
related assumptions regarding production functions, while section 7 presents some concluding
remarks in the form of finding a common thread among those theories.

6
2 KEYNES: HINTS, CONCERNS, BUT NO THEORY

[...] there is evidence that in its early stages, Keynes’ own thinking tended to develop in this
direction [to study income distribution] -only to be diverted from it with the discovery (made some
time between the publication of the Treatise on Money and the General Theory) that inflationary
and deflationary tendencies could best be analysed in terms of the resulting changes in output and
employment, rather than in their effects on prices.
– Nicolas Kaldor (1956), p. 83

Keynes’ theory of income distribution can be assessed through his two most important
works, the Treatise on Money and The General Theory of Employment, Interest and Money. The
term “distribution” is cited thirty-two times in the General Theory, but no book, chapter, or
section title includes the word “distribution” – there is not even a “distribution” entry in the index.
This count indicates a certain interest but is scarcely an indication of a main theme. The reason for
this relative absence is that the purpose of these two books is not the study of income distribution;
in those works Keynes talks respectively about money and aggregate demand, that is, what it
takes to generate employment and income, not how the income is distributed once created (see
Kaldor’s quote).
For Davidson (1960), Keynes discusses income distribution as soon as 1930 in the famous
parable of the widow’s cruse:
If entrepreneurs choose to spend a portion of their profits on consumption [...] the effect
is to increase the profit on the sale of liquid consumption goods by an amount exactly equal to the
amount of profits which have been thus expended [...] Thus, however, much of profits
entrepreneurs spend on consumption, the increment of wealth belonging to the entrepreneurs
remains the same as before. Thus, profits, as a source of capital increment for entrepreneurs, are
a widow’s cruse which remains undepleted, however much of them may be devoted to riotous
living. When, on the other hand, entrepreneurs are making losses, and seek to recoup these losses
by curtailing their normal expenditures on consumption i.e., by saving more, the cruse becomes
the Danaid jar which can never be filled-up; the effect of this reduced expenditure is to inflict on
the producers of consumption-goods a loss of an equal amount. Thus the diminution of their
wealth as a class, is as great, in spite of their saving, as it was before.
– John Maynard Keynes (1930), p. 139
7
 Through this metaphor, Keynes specifies that the more capitalists spend, the greater the
total amount of profits they receive. Conversely, entrepreneurs’ spending cuts necessarily mean a
lower overall profit level. Profits thus appear as a special category of income in the sense that they
must be spent to generate more income (and new profits). If there is no profit and/or if capitalists
do not spend, production and employment will stagnate. This reasoning is also attributed to
Kalecki (by Joan Robinson), as the adage “capitalists earn what they spend and employees spend
what they earn.” Keynes thus isolates a very special variable in the functioning of the economy:
profits. The widow’s cruse parable tells us about the thinking of Keynes c.1930: the question of
the functional distribution of income between wages and profits is not far—but a question Keynes
discusses only “in passing.” In 1930, Keynes preferred to treat the determinants and the level of
income (both sector consumption / production), but does not compare to the aggregate level of
income.
These observations give Keynes a special place among the theories of income distribution
—for in Keynes there is no distribution theory per se, just hints at one. The father of
macroeconomics and master of economic aggregates knows that income distribution matters but
his purpose is elsewhere.6 Keynes does not develop a theory of income distribution, and the
marginalists’ approach appears unchallenged, simple, and intellectually appealing (see section 6
esp. the Euler Theorem).
Keynes came really close to an analysis of income distribution in the General Theory.
Indeed his major work is centered on three specific real variables: wages, employment, and
output. Those alone define the share of labor:
W wN
= (1)
Y Y

where W is the wage bill, w is the nominal wage, the level of employment N , and Y
the level of production (or value added). After painfully precise definitions and analysis of each of
those three variables we expect Keynes to add a “synthesis” section addressing income
6
Keynes had plans to develop his “real economy” analysis, following developments the monetary Treatise on
Money. It is sometimes said that Keynes hesitated to embark on a theory of income distribution in the early thirties,
but after the article by Richard Kahn (1931), Keynes preferred the revision and integration of his earlier ideas on
uncertainty (Treatise on Probability), on currency (the Treatise on Money) and a demand analysis. The result is the
General Theory. See
Kahn, R. (1931) The Relation of Home Investment to Unemployment, The Economic Journal, 41, 162, 173-198.
Kahn, R. (1933) The Elasticity of Substitution and the Relative Share of a Factor, The Review of Economic Studies,
1, 1, 72-78.
8
distribution.7 But Keynes only shares, as a conclusion, a concern with income distribution: that
the “inequitable distribution of wealth and incomes” was an outstanding failure of the capitalist
economy. The book ends there. Overall, the introductory paragraphs to chapter 18 might best
expose Keynes’ thought:
We take as given [...] the degree of competition [...] as well as the social structure
including the forces [...] which determine the distribution of the national income. This does not
mean that we assume these factors to be constant; but merely that, in this place and context, we
are not considering or taking into account the effects and consequences of changes in them.
– John Maynard Keynes (1936), p.245

7
Keynes treated equally brief personal distribution of wealth and income (i.e, of the income inequality) in the
conclusion in terms of social philosophy of the General Theory.
9
3 THE INCOME DISTRIBUTION POLITICAL ECONOMY OF MICHAL KALECKI

At the same time that Keynes wrote his magnum opus, Michal Kalecki started a series of working
papers specifically on income distribution. Kalecki refined his views in a series of articles and
books (Kalecki 1935, 1938, 1942, 1954, 1962). For Kalecki, income distribution is inherently
related to the ability of “capitalists” to pass wage increases on to prices. Thus income distribution
is primarily a matter of degree of imperfect competition and, within it, the balance of power
between actors. Income distribution, market structure and pricing are interrelated and explain
economic growth and the business cycle. Among the different vintages of Kalecki’s theory we
will present the latest (1954) and most preferred by Kalecki himself.

3.1 Assumptions and Framework

Kalecki’s model is best understood when exposed “from the ground up,” i.e., starting from its
assumptions; there are five primary ones. Note that not all assumptions are relevant to Kalecki’s
theory of income distribution; for that purpose, only 1, 2 and 4 are necessary. Assumptions 3 and
5 set the broader framework.
1. There are two antagonistic classes: “capitalists” and “workers.”
2. Imperfect (oligopolistic) competition is the norm. Each firm is facing a
downward-sloping demand curve, meaning that each firm is demand-constrained. This is
in contrast to perfect competition where each firm faces an infinitely elastic demand
curve, so that demand changes do not affect prices -only costs and prices do (Lopez and
Assous, 2010).
3. Information is imperfect. Information (or lack thereof) plays a critical role in
investment decisions.
4. Less-than-full employment. As a result of 2 and 3, the economy is considered in a
situation of chronic underemployment of the factors of production, i.e. excess capacity
prevails.
5. The economy is a monetary economy in the sense that credit plays an important role.
“Capitalists” and the State can finance their spending by borrowing from banks.

10
 A note on the method seems necessary as well, for two things single out Kalecki’s
theory among heterodox economists. First, Kalecki does not assume price rigidity. A separate and
related theory of prices is developed where prices are endogenous and a function of the degree of
monopoly, the same parameter which, we will see, matters for the distribution of income. Second,
Kalecki is interested in the macroeconomic picture and especially in the distribution of income in
the aggregate, but he takes pains to derive a model that is micro-founded.
Now that the framework is set we can derive Kalecki’s profit equation. The latter stems
from the two national accounting identities defining output, one from the spending and one from
the income side. First, assuming away the government and foreign sectors for simplification,

Y  Cw  Cc  I (2)

where C w , Cc represents workers’ and capitalists’ consumption, respectively. Second,

production is the sum of total wages and total profits ( ∏ ), so that:

Y W  (3)

Assuming that all wages are consumed, i.e., W = Cw , and combining equation (2) and
equation (3) we arrive at the definition of profits as

 = Cc  I (4)

Equation (4) is an identity so that the direction of causality is unknown. However,

Kalecki remarks that if there is a profit, it must be the case that some output was sold in the first
place. Thus, there must have been some demand first and causality in (4) must run from
expenditure (conveniently put on the RHS) toward profits (voluntarily placed on the LHS).
Kalecki finds a result also envisioned in Keynes and confirmed in Kaldor: capitalists earn (  )
what they spend ( I  Cc ), while workers spend what they earn ( W = Cw ).

11
3.2 The Desired Share of Wages
There are several vintages of Kalecki’s theory of income distribution, the most famous of which
involves the share of profits as fixed and given (Kalecki 1954). The exogeneity of the profit share
is justified by the market structure, and the existence of the capitalist class which has the
decision-making power. In this formulation, the share of profits ( ß ) that is desired by capitalists
is defined as  * :

 = * (5)

Therefore, the wage share is defined as  * = 1   * , i.e., it is derived as a residual when

the balance of power leans towards capitalists.
The share of profits  =  / Y and equation (5) can be used to extract the level of activity
such as:
1
Y* =  (6)
*

Equation (6) states that the production level depends both on the share and the level of
profits. Coefficient 1/  * can be called an “income distribution multiplier” (my words, not
Kalecki’s) and is equal to the inverse of the share of profits desired by capitalists. The level of
expenditure that capitalists contribute to profits is given, so that profits are also given. This means
that production varies inversely with the share of profits. Capitalists must curtail production in
order to increase their profit share. Thus, with constant spending on the capitalists’ part, a higher
profit share can only be achieved if output and employment are limited.
Profits are defined as the sales proceeds minus total costs. For simplification we define
total costs as comprising only of wages and raw materials (or intermediate products). Sales
proceeds in turn are defined as prime costs (wages and raw materials) marked up by a factor k ,
that Kalecki calls the degree of monopoly. The degree of monopoly is defined as the price-to-cost
ratio in a particular industry. By definition then (see Lopez and Assous, 2010)

12
  = (k  1)(W  Pmat ) (7)

where Pmat represents the price of raw materials, i.e. costs other than wages. Using
equation (3) and (7) we get an expression for output

Y = W  (k  1)(W  Pmat ) (8)

Dividing through by the wages bill W and taking the inverse we arrive at the wage share
.

1
= (9)
1  (k  1)( j  1)

with j = Pmat /W . In the particular and simpler case of the absence of raw materials (or if
their price is constant) we can rewrite equation (9) as

 = 1/k (10)

This last expression states that the wage share is inversely related to the degree of
monopoly. A very competitive market with low degree of monopoly will see a very high wage
share whereas an oligopolistic market will see a much lower labor share.
Therefore, the wage share is given by four factors, all interrelated:
1. The intensity of the class struggle, through which capitalists and unions clash;
2. The importance in total value added of imperfectly competitive firms;
3. The degree of monopoly, which the markup “reflects” (Kalecki, 1954);
4. The ratio of aggregate prices to materials prices, j .

The first three factors are self-explanatory or are taken from equation (9). Recall also that
the wage share arrived at in equation (9) and (10) are wage shares desired by capitalists. Would
capitalists always achieve their desired share of profits? Again, this depends on the balance of
power in the class struggle.
13
 In order to see why, we define the wages bill as:

Ԝ=Y–  (11)

W =  / *   (12)

1  *
W=  (13)
*
Equation (13) states that the desired income shares are given by the decisions of capitalists’
expenditures (reflected in profits). We now have a whole system based on the decisions of
capitalists:
1. Capitalists spend;
2. the amount of capitalists’ spending implicitly determines
(a) their income, and,
(b) overall income, out of which wages and salaries are paid, once profits have
been distributed,
(c) capitalist spending determines (among other things, see above) the distribution of
income.

3.3 Power Relations and Class Struggle

The evolution of the wage rate is part of a more complex mechanism. Kalecki considers two
cases: either an increase in the wage rate leads to a constant share of profits, or capitalists accept a
decline in the share of profits.
In the first case, as we have seen, the balance of power lies in capitalists’ hands: the
increase in the wage rate causes an extra cost of production for capitalists, but capitalists enjoy
enough monopoly and bargaining power to pass on the full wage increases on to the price of
production. In this case the increase in the nominal wage rate corresponds to a real wage
stagnation and workers’ purchasing power is unchanged.
The second case is more complex. If capitalists are not powerful enough in relation to
unions, then capitalists are not able to pass all wage increases on to prices. This causes two things.
On one hand, capitalists’ profits remain unchanged since they depend on capitalists’ willingness
to spend. On the other hand, salary increases lead to a larger wages bill, which mechanically

14
increases consumption. As a result, companies are subject to higher demand, at least in the short
run, and so capacity utilization rises. The end result is that consumption and investment are at
higher levels. This corresponds to a situation with greater production, but with a smaller share of
profits. The result: growth comes with a declining share of profits—we have wage-led growth.
Kalecki is more interested in this latter case. Under this scenario, there are sufficiently strong
unions to counteract capitalists’ decisions by increasing consumer demand; hence, giving rise to
profits and employment.

3.4 Policy Implications

The capitalist class dominates the economic system proposed by Kalecki. This dominance is
characterized by a too-high profit share to maintain or generate full employment and economic
growth. In that case, the share of profits hinders economic growth and what is needed to restore
growth and employment is wage-led growth.
Kalecki discusses an exogenous intervention to avoid such undesirable outcome: the
intervention of the state for wage-led growth (Lopez and Assous, 2010). This requires us to
generalize the framework presented so far to include a foreign and public sector.
In an open economy with a government sector, the above remains true, with the provision
that each variable is defined net of taxes, and we have:

  Cc  I  (G  T )  ( X  M ) (14)

which means that capitalists are no longer entirely controlling the profit level / share. A
higher fiscal deficit and/or trade surplus increases profits, output and employment, but it may
come at the expense of capitalists’ profit share. Capitalists may want net exports ( X  M ) to
increase, because their production would go up and exports may be a way to beat a competitor.
However, positive net exports may lead to a lower profit share, depending on the fiscal balance.

But what fiscal stance would capitalists prefer? As equations (6) and (14) indicate,
• A high fiscal deficit (G  T ) > 0 may increase profits, but this would reduce the
profit share further.

15
 • Any fiscal surplus (G  T ) < 0 may decrease profits but it will increase the profit
share. If capitalists’ objective is to maintain or even increase their profit share, the fiscal policy of
choice is that of relatively low spending and high taxes.

More precisely, Kalecki examines three cases:

• The budget is balanced state expenditure and a tax levied on wages of an equivalent
amount. In this case there is a crowding-out effect of consumption by the public expenditure so
that the net effect is zero for the entire system. The government spending multiplier is one.
• The budget is balanced by state expenditure and a tax levied on profits of an
equivalent amount. If capitalists’ spending decisions remain unchanged, then profits before tax
are not changed. We are in the case discussed previously with the same conclusions. Capitalists
have the option to either pass the tax increase entirely on to wages or on to prices. Kalecki notes
that the repercussion is never completely in one direction; mainly because a tax levied on profits
results in a decrease in the share of profits. Consequently, the reasoning is the same as it was
previously: the decline in the share of profits leads to greater consumption, and thus boosts the
economy. This balanced budget multiplier is greater than one.
• Finally, if the budget is in deficit, the effect is twofold: the deficit benefits both
capitalists and workers, and both growth and employment go up. The converse is that a fiscal
surplus results in a tax levied on aggregate demand so that the surplus is levied on the whole
dynamic of the system.
Kalecki’s model combines different concepts we have discussed so far, especially some
of the ideas of Keynes, whose early ideas Kalecki was not necessarily aware of. The major
contribution of Kalecki’s model is to propose a tractable framework in the case of
under-employment. The immediate conclusion is that in an economy with excess capacity, State
intervention is justified. The State can alter power relations between workers and capitalists and it
can influence the distribution of income. Whatever the vintage, the distribution of income remains
central to the theory of Kalecki. This distribution is defined by exogenous conditions such as state
intervention, but this exogeneity does not result in a neutrality of distribution. In all, for Kalecki, a
distribution in favor of profits—e.g., due to heightened bargaining power of capitalists over
workers—is both the cause and the consequence of lower economic performance.

16
4 KALDOR AND PASINETTI: INCOME DISTRIBUTION FOR FULL
EMPLOYMENT

[...] no hypothesis as regards the forces determining distributive shares could be intellectually
satisfying unless it succeeds in accounting for the relative stability of these shares in the advanced
capitalist economies over the last 100 years or so, despite the phenomenal changes in the
techniques of production, in the accumulation of capital relative to labor and in real income per
head.
– Nicholas Kaldor (1956), p.84

The Post-WWII the economics profession became very interested in the issue of economic
growth. The Keynesian proposition that had been holding for two decades is that economic
growth depends on effective aggregate demand. Heterodox economists, following Keynes (1936)
and Harrod (1939), insisted on investment while neoclassical economists insisted on the role of
savings and technological progress (Solow, 1956). But how were saving and investment related to
economic growth?
Kaldor (1956, 1957) finds that income distribution matters to the correlation of savings
with economic growth. This comes in contrast to the Solow (1956) growth model, for instance,
where income distribution does not matter for economic growth (Bertoli and Farina, 2007).8
Further, Kaldor demonstrates that, assuming heterogeneous saving rates for workers and capital
owners, workers’ savings rate does not matter much at all for income distribution. Only
investment and capitalists’ saving propensity matter; the same result as in Kalecki and Keynes!
The model was then reprised by Pasinetti (1962), who had identified a “logical drift,” but whose
correction does not change Kaldor’s main conclusion. For that reason we refer to an integrated
“Kaldor-Pasinetti” model.

8
Solow (1956) uses a Cobb-Douglas production function which by nature (or technical feature) assumes constant
relative factor shares. This is not withstanding Solow’s own skepticism in the constancy of such shares (Solow 1958).
See section 6 for further details.
17
4.1 Stylized Facts
Kaldor’s (1956, 1957) model discusses the features of an economy in a steady state position. The
economy grows at a constant rate given by population, technological progress and investment.
Resources are fully utilized and full employment prevails “in general” (see caveat below). All this
is derived from Kaldor’s assumptions, which he only made explicit later (Kaldor, 1961):
1. Constant labor productivity (output per capita),
2. Consistent capital productivity, hence
3. Constant capital-labor ratio,
4. Constant distribution of income,
5. Relative stability of real interest rates, and finally
6. Existence of large disparities in the rate of productivity growth.

Following Harrod (1939) this implies that the economy’s growth path is stable if
warranted savings are actually achieved. But how?

4.2 The Model

Kaldor extends Harrod’s theory by using the fact that any savings rate provides a split of income
between consumption and savings which is different for each social group. Kaldor uses two
identities:

Y W  (15)

S  S w  Sc (16)

where, SW and S C are the savings amounts provided respectively by workers and
capital-owners. Assuming that the only source of income for these two classes are wages and
profits, the savings rate of the two social categories is defined as:

sw = Sw /W , and sc = Sc / (17)

18
 And in a closed economy without government savings equals investment (I), so that

I  S = Sw  Sc = swW  sc  (18)

Rewriting,

I = swY  (sc  sw ) (19)

Solving for the profit share we get Kaldor’s famous equation

 1 I sw
=  (20)
Y sc  sw Y sc  sw

Since the share of profits cannot be negative or zero, we must add the condition

I
0 < sw < < sc < 1 (21)
Y

Thus, provided that equations (20) and (21) are verified, which they necessarily are since
they come from an identity, the savings rate in the economy is going to the one matching the
natural rate of growth, so that the economy will be on a persistent, stable, full employment,
growth path.

4.3 The Pasinetti Critique

Pasinetti (1962) complements Kaldor’s model six years after its publication, having identified a
“logical drift” leading to an “absurdity”: if people save, they become owners of capital and they
have to be compensated accordingly. A portion of profits must be attributed to workers;
otherwise, workers have no motivation to save. This addition to Kaldor’s theory did not change
much of the key findings, hence the name Pasinetti paradox. The share of total profits in value
added remains the same, with the same determinants, method and with the same assumptions. To
see why, consider equation (18) which becomes

19
 S  S w  S K = sW (W  W )  sK  K (22)

S = I = sW Y  (sK  sW ) K (23)

where the subscripts W and K denote variables for workers and capital-owners,
respectively. Solving for the profit share we get

 1 I  I K 
=
Y s K  sw  Y  sW  r.sw  sK K  Y  (24)
  

where r is the interest rate, i.e. the rate at which workers extend loans to capital-owners.
Pasinetti remarks that in the long run the interest rate must equal the rate of profit, in which case,
after a long derivation, we get the same result that Kaldor arrived at:

 1 I
= (25)
Y sK Y

Pasinetti presents an approach that is more complete but arrives at the same result. The
most striking conclusion remains that workers can in no way influence the distribution between
wages and profits. And the savings rate of workers still cannot influence the macroeconomic
division between wages and profits, which is solely determined by the decisions of capitalists.
Kaldor’s rule (attributed to Kalecki, and which is also present in the parable of the widow’s cruse
of Keynes) remains true: workers spend what they earn and capitalists earn what they spend.
Table 1 summarizes the different values of the profit share according to three cases studied by
Kaldor and Pasinetti.
Kaldor and Pasinetti arrive at the conclusion that a balanced growth path and full
employment are consistent with a single rate of profit and a certain level of income distribution.
The sustainability of the balanced growth path is maintained by the realization of the profit rates
described in Table 1 (below).

20
 Table 1 Formula Summary for Profit Shares and Profit Rates in Kaldor and Pasinetti

Profit share Π 𝑌= Profit rate Π 𝐾=

1 𝐼 𝑠𝑤 1 𝐼 𝑠𝑤 𝑌
General case − −
𝑠𝑐 − 𝑠𝑤 𝑌 𝑠𝑐 − 𝑠𝑤 𝑠𝑐 − 𝑠𝑤 𝐾 𝑠𝑐 − 𝑠𝑤 𝐾

1 𝐼 1 𝐼
Special case 𝑠𝑤 = 0
𝑠𝑐 𝑌 𝑠𝑐 𝐾

𝐼 𝐼
Special case 𝑠𝑤 = 0 and 𝑠𝑐 = 1
𝑌 𝐾

4.4 Full Employment vs. Harrodian Dynamics

The dynamics along the natural growth path are better explained in reference to the work of
Harrod, who introduces three growth rates:
• The actual growth rate g ,

• The natural growth rate, g n , necessary to achieve or maintain full employment,

• And finally the warranted growth rate, g w , which is the rate of investment in the
total desired product. Harrod decomposes the investment rate as:

I I K S/Y
= = = s. 1 (26)
Y K Y 

1 K
with  := =
AK Y

The warranted growth rate is g w = s/ ; it is the output growth rate compatible with
entrepreneurs’ investment. To Harrod, most entrepreneurs are optimistic; they anticipate profit
from increased production prospects and the more they invest the higher the warranted growth
rate.

21
 To achieve balanced growth and full employment in Harrod’s model, we need the three
growth rates to be equal: g = g w = g n . This requires that the expectations of entrepreneurs
generate an actual growth rate that coincides perfectly with the natural growth rate. Harrod points
out that this case is unusual, historically, and there is no reason for this to happen automatically
and always. As a result, full employment is not the norm.
The theoretical framework of full employment advanced by Kaldor thus appears as the
particular, ideal case; a case rarely considered by Harrod. Kaldor only introduces the possibility of
a minor imbalance. Kaldor introduces the realization of S = I ex-post and introduces the
possibility of a slight deviation S  I ex-ante. What will happen if the system is close to, but not
in, equilibrium?
When investment is higher than savings, the excess demand generates inflation in the
consumer goods sector. With sticky nominal wages, inflation lowers real wages, with two major
consequences. First, a rising price of consumer goods leads to income consisting of a smaller
proportion of consumption and a greater share of savings. Second, a lower real wage rate leads to
a lower labor productivity in perfect competition, which restores the constancy of the share of
profits. Therefore, and whichever way, Kaldor’s model presents the peculiarity that a slight
imbalance is automatically compensated for by price adjustment.

4.5 A Full Employment Keynesian Model

Kaldor and Pasinetti’s approach to income distribution follows the Keynesian tradition in the
very particular context of the “stylized facts,” namely, assuming full employment. It is therefore
not true in the general case, as Keynes believed that full employment occurs only “by chance,”
and that it is rare that g = g n = g w . Despite this criticism, Kaldor (1956, 1957, 1961) remained
faithful to the assumption of full employment.
Kaldor has been strongly criticized for this assumption. Samuelson (1964b) refers to
Kaldor as “Jean-Baptiste Kaldor” (p.345). Weintraub (1981) quipped “Lord Kaldor, in his
cheering section, highlights a defined share of the profits of the original application [investment],
or savings, [theory] deficient for many because of its foundations [in terms] of full employment”.
Marglin (1984) points out that the assumptions Kaldor makes are “more neoclassical than
neo-Keynesian” (p.534). From the point of view of this assumption of full employment, Pasinetti
is more explicit than Kaldor.

22
 One can make the following comments, all being equivalent:
1. Consequences of Kaldor’s stylized facts:
(a) The assumption of balanced growth implies the realization of full
employment.
(b) Kaldor is not interested in the ups and downs of the economy as it is assumed
to be stable in a long run.
(c) The central question is whether the savings rate (thus, incidentally, the
distribution and rate of return) accompanies or helps maintain full employment growth, as
opposed to how to achieve full employment from a situation of unemployment.

2. The distribution theory presented by Kaldor is the result of the savings of workers
and capitalists which generate a certain level of fully invested savings. This investment comes
with a certain level of economic growth, which is fixed by assumption. The distribution of income
resulting from investment decisions must be such that the share of profits accommodates a
constant growth. Therefore, for Kaldor, the distribution of income does not allow for the
achievement of stable growth or full employment, since by assumption, growth is stable and full
employment is realized.

3. Income distribution has an endogenous role of proportionality that cannot explain

the fluctuations in the growth rate. Near full employment, there is a certain growth rate
accompanied by a certain distribution, the two being in a constant ratio.
4. Kaldor’s model is an ergodic model: the future is a continuation of the past,
through an automatic (price!) adjustment mechanism. There is no regime change in which
accumulation is endogenous, and the possibility of underemployment is not taken into
consideration.

Such specific assumptions limit the reach of Kaldor’s model. One may even wonder what
the real Keynesian content of the model is. Is the model in the form of a “synthesis”?
Nevertheless, it does not contradict the spirit of Keynes’s view of full employment, a situation in
which Keynes saw “no objection” to neoclassical economics. This suggests that Keynes and
Kaldor are in agreement with the neoclassical authors in regards to full employment; income
distribution plays an accompanying role—not a determining role.

23
 Kaldor’s model, confined as it is, introduces three major insights. First is the introduction
of distribution in the economic discourse, which is underlying but not explicit in the work of
Keynes. Second: the conclusion that workers spend what they get and capitalists get what they
spend, is maintained. Third, even in a situation of full employment, it is demand that is driving the
activity in Kaldor’s system—just a level of demand that is assumed to be always right, as implied
by the above equations.

24
5 THE INCOME DISTRIBUTION BUSINESS CYCLE MODEL OF RICHARD
GOODWIN

Goodwin’s (1967) model addresses some of the shortcomings of Kaldor’s model. Full
employment is no longer assumed; the framework is not long-term growth, and income
distribution is no longer passive. Instead, Goodwin develops a class struggle model in which
workers and capitalists clash over income distribution. This leads to an income distribution cycle
generated and endogenously maintained by employment and growth.

5.1 The Model

Goodwin’s model is based on the following causal chain. At high levels of employment,
upward-wage pressure starts to appear; the profit share is compressed which limits investment
and therefore employment. But at that point the scarcity of jobs creates de facto the conditions of
the restoration of the share of profits and the restoration of growth and employment.
Goodwin begins his presentation by specifying the limits and assumptions of his analysis:
“the model presented here is completely diagrammed and [is] as a rather unrealistic model of
growth cycles” (p.54). Similar to Kaldor (1956), Goodwin provides precise assumptions “made
for reasoning practices”:
1. Technical progress changes at a constant rate. Or;
2. Labor grows at a constant rate
3. Only two factors of production exist: labor and “capital”
4. All amounts are expressed in real and net terms
5. All wages are consumed
Q 1
6. Productivity of capital is constant: AK = =
K 
7. The real wage rate increases near full employment.

The first equation in this context captures the increase in the capital stock of the
investment (which is the savings by identity). The author assumes that all profits are invested:

K  wg 
= I = S =  = 1  Q (27)
t  AL 

25
 where wg /AL is the wage share prevailing at full employment. Dividing through by K ,

we simplify the previous equation:

K
 wg Q
K = t = 1   (28)
K  AL K

where a dot above a variable indicates the rate of change of that variable from time t to
Q 1
t  1 . As before, we rename = , which is constant following assumption H1 , so that
K 

 wg 
K = 1   / (29)
 AL 

The rate of productivity growth is the gap between the growth rate of output and the growth rate of
employment. Since the growth rate of labor productivity is constant (as a result of H1 and H6).

Q  N = A L (30)

Since in the current framework any supplemental growth can only be achieved by extra
investment, Q = K , then (30) may be rewritten as

K  A L = N (31)

Combining (29) and (31), we find that

 wg 
N = 1   /  A L (32)
 AL 

26
Since both capital and labor feature constant productivity, equation (32) implies that the growth
wg
rate of employment is a decreasing linear function of the labor share in total income,  = .
AL

Denoting the labor force L , the rate of employment variable  is introduced as   N/L ,

  N  L . Combining this last identity with (32), it follows that

 =
1


 A L  L  (33)

Goodwin’s model assumes that the real wage rate increases in the neighborhood of full
employment. Hypothesis H 7 is interpreted with reference to Marx’s theory of a reserve army
and the empirical work of A.W. Phillips. Goodwin assumes an approximate growth rate of real
wages by a linear function of the rate of employment:

 =  a  b   A L (34)

Equations (33) and (34) are two differential equations which explain, respectively, the growth in
the employment rate and the growth of the wage share. We can rewrite those relationships in a
dynamic systems form:


 =

1


 A L  L  (35)
 =  a  b   AL
 

  1  


=  
 A L  L   
  t   
 
(36)

 =  a  b   A L 

 t

27
Goodwin notes that the last rewriting is of a most common type of differential equations called
Lotka-Volterra also known as predator-prey models. Such system is represented by a phase
diagram characterized by a cycle and an equilibrium point (see Figure 2).

Figure 2 The Goodwin Cycle

5.2 Solution and Interpretation

The equilibrium point E is the solution that simultaneously cancels the LHS of equations (36):

 
 t = 0 
 * = 1  A  L  
   *
 
L
(37)
 =0 
  = a  A L /b
 t

28
 The models’ features are that
• The solution, or equilibrium, is never reached but there is a constant movement
around it. The equilibrium point E = ( * ,  * ) is better seen as an “average”; the average of the
labor share and the employment rate over the business cycle. What is remarkable is that this
equilibrium is stable and independent of initial conditions, as pointed out by Goodwin (p.58). The
direction of rotation A  B  C  D  A on Figure 2is given by the sign of the model
parameters.
• Contrary to the equilibrium, the amplitude of the cycle does depend on the initial
conditions, so that starting from a very high wage share will make the cycle last longer. Whereas
there is only one equilibrium point, there may be several cycles for different initial conditions.
• The circular shape of the trajectories is due to the fact that we have assumed (or made
it in such a way) that the relationships between the ( ,  ) variables are linear. For nonlinear
relationships we would have led to concentric oval or ellipsoidal shapes. Finally, the cycle is a
closed cycle in Goodwin’s model: the model has the particular feature that, whatever the initial
conditions (  0 , 0 ), we return to the starting point.
The usual interpretation of the equations of Volterra type is that of predator-prey model,
which is in line with a Darwinian interpretation of the evolution and the preservation of species. A
classic illustration of this type of model is two populations of shark and fish living in balance in a
closed environment. Both populations have a common dynamic in the sense that if there is plenty
of prey (fish), then predators (sharks) are increasing in numbers, whereas scarcity of prey implies
a dwindling predator population. The converse is true for prey: the fish population increases when
there are few predators and decreases if they are abundant. Both populations tend to over-react
(overshoot) insufficient / excess of the other population.

29
5.3 Analysis and Contribution
We can describe the economics of Goodwin’s model using this analogy between fish and sharks.
The employment rate represents the fish population, or prey, and the wage share represents the
shark population, the predator. When the wage share is too high between A and C (i.e.,
superior to its equilibrium or average value), the employment rate decreases: the abundance of
sharks is reducing the number of fish. Similarly, when the wage share is lower than average, such
as between points B and D , the employment rate increases: due to the small number of fish, the
shark population increases).
Goodwin’s model may also be interpreted in reference to economic conditions. From
point A to point C , the employment rate goes down from its maximum to its minimum (point
A can be seen as a peak of the cycle and point C as the lowest point). This continued
deterioration in the employment rate occurs when the wage share is superior to its equilibrium
value  * ; that is to say when the profit share is below its average. According to Goodwin (p. 58),
downturns are due to the loss of profits. Conversely between point A and C , the employment
rate rises as the profit share rises.
In economic terms, the main features of the model can be summarized in five points:
1. By assumption, profits are assumed to be fully invested and wages entirely
consumed. Goodwin does not raise the question of market outlets and eventual leakage out of the
system, and their contractionary implications.
2. By assumption, labor productivity grows at a fixed rate. This assumption is similar
to the one in Kaldor’s model (1957), but since Goodwin does not assume a constant growth rate of
production, then we must have cyclical fluctuations —see equation (30).
3. Corollary: cyclical fluctuations are a source of under-employment (in labor),
because the growth rate of labor is constant.
4. Conclusion 1: The share of wages and the employment rate are interrelated: at
times in a positive way (quadrants II and IV, second equation in (37)); at times in a negative way
(quadrants I and III, first equation in (37)). According to Goodwin, the increase in employment
can only occur when businesses become more profitable, that is to say, have a lower payroll /
wages bill.
5. Conclusion 2: These two opposite effects compensate in time to form a dynamic,
closed system. Whatever the set of initial conditions, it always returns to the starting point through

30
rotation around the point of equilibrium.
In Goodwin’s model, the distribution of income is endogenous, deeply embedded, and
appears to function harmoniously with the economic system. It is not just accompanying the
business cycle: income distribution is (a part of) the business cycle, since it is able to restore or
deteriorate the level of production and employment.
Finally, we note that model is constructed so that there are closed trajectories around
equilibrium. Employment growth is automatically restored by income distribution (i.e. an
increase in the share of profits). This calls for three comments:
• If the equilibrium point E = ( * ,  * ) is unchanged, we should expect the
distribution of income to be oscillating around a long-run constant. Goodwin’s model cannot
account for a permanent economic situation of underemployment. It could nevertheless be
extended in that way by assuming that profits are not entirely re-invested, wages are not entirely
consumed or any other kind of leakage.
• If the equilibrium point E = ( * ,  * ) could also be assumed changing due to
changes in technology or other exogenously-defined variables.
• Finally, it is important to discuss the adequacy of applying the economic-biological
model. The theory of the evolution of species is a naturalist theory. In the real biological world,
there are also cases where species have disappeared even without human intervention. Not all
economic models need to have closed dynamics.

31
6 TECHNOLOGY AND THE SHAPE OF THE PRODUCTION FUNCTION

Until the laws of thermodynamics are repealed, I shall continue to relate outputs to inputs — i.e.
to believe in production functions.
– Paul A. Samuelson (1972), p.174.

Technological progress has often been viewed as a central element of economic growth.
In the most general sense, technological progress is based on the “yield” of the factors of
production, i.e., the marginal products of capital and labor. And since marginal products can
differ, technological progress can be biased towards either factor of production, and the relative
shares will change. To understand how and in what direction, we need to discuss different types of
technological progress and different types of production function.
Consider the most general aggregate production function:

Y = F ( L, Z , A) (38)

with two inputs: labor L , and a nondescript input Z which could be capital, skilled
labor or land. Parameter A is a technology index featuring F /A > 0 : a greater level of A
corresponds to “better technology” or “technological progress.” The following definitions apply:
1. Factor augmentation: Technical change is said to be L -augmenting if the
production function takes the more special form Y = F ( AL, Z ) , or Z -augmenting when
Y = F ( L, AZ ) . We have factor augmentation when only one factor is affected by technological
change.
2. Factor bias: L -biased technical change is different from being L -augmenting.
We say that technological progress is L -biased when

 FL '
>0 (39)
A FZ '

that is, if technical progress increases the marginal productivity of labor at a higher rate
than it increases the marginal productivity of Z (Acemoglu, 2002).

32
 Technological progress can be factor-biased:
• Technical progress is defined as Hicks-neutral if it does not affect the balance of
labor and capital in the production function (Hicks, 1932). Since Hicksian neutrality implies that
the marginal products of all factors increase at the same proportion, the production function can
be written by factoring out technical progress such as Y = A.F ( L, Z ) .
• An innovation is Solow-neutral (Solow, 1969) if it only affects the productivity of
capital: Y = F ( L, AZ ) .
• An innovation is Harrod-neutral (Harrod, 1942) if technology is labor augmenting:
Y = F ( AL, Z ) .

6.1 Production Functions: Cobb-Douglas vs. CES

Two types of production functions stand out in the present context: the classic Cobb-Douglas
(1928) and the Constant Elasticity of Substitution (CES). By modeling the growth of the
American economy from 1899 to 1922, Cobb and Douglas (1928) proposed a simplified view of
the economy where the level of output is determined by the amount of labor and capital involved
in the production process. In the case of two factors and technical progress (later added in the
function by Solow 1956), the standard Cobb-Douglas form is:

Y = F ( L, K ) = AL K  (40)

where Y represents total production, L is the labor input, measured by the total
number of person-hours worked in a year; K is the capital input, measured by the monetary
worth of all machinery, equipment and buildings; A represents total factor productivity, and
(which can be intangible as it can range from technology to human capital).
Finally,  and  are the output elasticities of labor and capital, respectively, i.e., they
measure the responsiveness of output to a change in the levels of either the labor or the capital
input, ceteris paribus. For instance, if  equals 0.64, a 1% increase in labor usage would lead to
a 0.64% increase in output. In the case of an economy in perfect competition, the factors are paid
at their marginal product, so that

33
 W w FL' AL 1 K 
 = L = L = L =
 Y Y Y AL K 
   1 (41)
  = r K = FK K = AL K K = 
'


Y Y Y AL K 

The Cobb-Douglas production function is particularly interesting in the case of perfect

competition, in which case the function has the following features:
• The factor shares are directly readable on the production function itself as the capital
and labor exponents,
•  and  must be constant, or else the derivation above does not hold,
• Since aggregate income is composed of wages and profits and nothing else, it must
be that    = 1 ,
• As a consequence, a perfectly competitive economy characterized by a
Cobb-Douglas production function features constant returns to scale.9
Another way of proving this result is to call upon the so-called Euler theorem (one of
many) stating that for a continuous function y = f ( x1 ,, xn ) homogenous of degree one, i.e.,
featuring constant returns to scale, we have

y
Y = xi (42)
xi

On one hand we can apply this theorem on a production function such as Y = F ( K , L) to

get

Y Y
Y= L K (43)
L K

9
A production function Y = F ( K , L) whose inputs are each multiplied by a scalar  implies, in the case of a
Cobnb-Douglas case, that
α α+ ß
F(λΚ , ΚL) = (λK_)ß(λL) = λ .K ßLα = λ α+ ß .F (K, L)
If    = 1 , doubling the amount of capital and labor used in the production process will result in a doubled output
and the production function displays constant returns to scale; if    < 1 , the function has decreasing returns to
sale; if    > 1 , increasing returns to scale takes place.
34
 On the other hand, and by definition, we have

Y = w.L  r.K (44)

Y Y
By identification, it must be that = w and that = r or, in other words, it must be that
L K
factors are paid at their marginal product.
Summarizing: the Cobb-Douglas production function has such a particular form that the
factor shares are constant in perfect competition, i.e., factors are paid at their marginal product
and we have constant returns to scale. If more capital is used in the production process, the rate of
profit /K falls just enough to maintain a constant capital share.

Cobb-Douglas + perfect competition = constant labor share

Another way to arrive at a constant distribution of income (see Gollin, 2007) is to use a
production function with Harrod-neutral technical progress Y = F ( AL, K ) in the Solow (1957)
growth model. The production function F () need not be Cobb-Douglas. It is well-known that
along the balanced growth path

s
k* = f (k * ) (45)
  gA

where there is population growth, s is the exogenous savings rate, capital depreciation
is  , technical progress grows at a rate of g A , f features constant returns, and k = K/AL is
the capital stock per effective worker. Multiplying through by r and rearranging we get the
capital, or profit, share as

rk * sr
= (46)
f (k )   g A
*

35
 which is necessarily constant on the balanced growth path because s and r are
assumed constant. Thus, two very commonly used economic models, the Cobb-Douglas
production function and the Solow growth model, feature a constant distribution of income.
Factor shares are exogenous and have no driving role.
Another often-used production function is the Constant Elasticity of Substitution (CES)
type. In the following pages we focus our attention on the classic CES production function; see
Klump et al. (2011) for a survey of variations on the CES theme.
Pioneered by Arrow, Chenery, Minhas and Solow (1961), the classic CES production
function is a “generalized Cobb-Douglas production function” as it encompasses the
Cobb-Douglas as a special case. Mukerji (1963) considered a CES function for constant ratios of
elasticity of substitution and Bruno (1962) suggested a generalization of CES production function
to permit the elasticity substitution to vary. Thus, one of the most important differences between
the Cobb-Douglas and CES production functions is that the former has a unit elasticity of
substitution between labor and capital while the latter allows for non-unity elasticity.
The general two-factor CES production function takes the form of


Y = A K   (1   ) L 
1/
(47)

Here 0 <  < 1 is the relative share of capital and  captures the degree of
substitutability of the inputs. Parameter A depends upon the units in which the output and inputs
are measured and is therefore not directly interpretable as technology. The value of  is equal to
or less than 1. Note that if unit elasticity of substitution prevails  = 1 , the CES function
collapses to the Cobb-Douglas form Y = AKβ L1-β .
“Neoclassical” growth theory and the aggregate CES production function have a long
common history, starting with Solow’s (1956) seminal contribution (Klump et al., 2007).
However, the workhorse of growth theory has tended to be the Cobb-Douglas. One reason for this
general interest may reside in the long-held belief in a “stylized fact” of long-term economic
growth: the approximate constancy of factor shares. We have already proved that an elasticity of
substitution equal to unity, as suggested in the Cobb-Douglas production function, implies a
constant factor share and a constant capital-to-labor ratio. Any changes in factor proportions will
be exactly offset by changes in the marginal product of the factor inputs (Miller, 2008). In the case
36
of a CES production function, since the elasticity of substitution need not be unity, a constant
factor income share can only be achieved if technological progress is purely Harrod-neutral
(Klump et al., 2007).

6.2 Technical Change in the CES Production Function

 1
Consider the previous CES production function with  =



 1  1  1
Y = A  AL L    (1   ) AZ Z    (48)
 

By construction, AL and AZ are two separate technology terms, and AL is L

-augmenting, while AZ is Z -augmenting.10 Parameter  captures the relative importance of the

two factors and  [0,[ is the elasticity of substitution between labor and variable Z . This
quantity measures the extent to which firms can substitute capital for labor as the relative
productivity or the relative cost of the two factors changes.
The elasticity of substitution is defined as

Z
ln
= L (49)
MPL
ln
MPZ

• When the two factors are perfect substitutes,  =  , the production function
becomes linear: Y = AL L  (1   ) AZ Z and the isoquants are straight (see Figure 3. The
marginal rate of substitution of labor for capital at any point on an isoquant is a constant. The
movement of the isoquants depends on the values of AL and AZ .

10
AL is also L -complementary and AL is Z -complementary.
37
 Figure 3 CES Production Function and Perfect Factor Substitution

• When  = 0 , there is no substitution between two factors.

• When  = 1 , the production function is Cobb-Douglas.
The direction of the bias of technical change depends on the elasticity of substitution. To
understand this, calculate the relative marginal product of the two factors (Acemoglu, 2002):

 1 1

MPZ 1    AZ  
Z 
=     (50)
MPL   AL  L

As the relative quantity of factor Z is increasing, Z/L increases and its relative
marginal product is decreasing (ceteris paribus). This is the usual substitution effect, leading to a
downward sloping relative demand curve. The effect of AZ on the relative marginal product
depends on  :
• When the two factors are gross substitutes (σ > 1), an increase in AZ relative to AL
increases the relative marginal product of Z , so that a Z -augmenting technical change is also
Z -biased.
• When the two factors are gross complements (σ < 1), which is usually the case (see
Giovannoni [2013b] for an overview), the reverse holds: a Z -augmenting technical change is
actually L -biased. Intuitively, in this case of complementarity, an increase in the productivity of
Z increases the demand for the other factor—labor—by more than the demand for Z , which
38
creates “excess demand” for labor. As a result, the marginal product of labor increases by more
the marginal product of Z (Acemoglu, 2002).
The value of the elasticity of substitution has been shown to play a critical role in
influencing economic growth and the movements of the labor share (Irmen, 2011; Choi and
Rios-Rull, 2009). Both non-competitive factor prices and a non-unit elasticity of substitution can
explain the dynamics of the labor share, and the latter seems more important. Raurich et al. (2012)
derive the equation of the labor share when there is imperfect competition, and labor-augmenting
technology, so that the production function is


  1  1   1
Y = AL   (1   ) AL Z    (51)
 

Following Galí (1996) we obtain the labor share  t as

1
w L 1    Yt  
t = t =   (52)
Yt mt  AL L 

This equation clearly shows that the labor share depends on:
• The evolution of the markup mt

• The average labor productivity in efficiency units, Yt /( AL L) , when   1

Further, the average productivity can be written as

1
 Yt   1
  =  1
(53)
 AL L   Zt  
    1   
 AL Lt 

so that the labor share can be rewritten as

39
 1 1
t =  1 (54)
mt
 Zt  
    1   
 AL Lt 

The last equation indicates a relationship between the labor share and capital deepening
that depends on the value of  . Capital deepening, or capital intensity, refers to the accumulation
of capital per effective worker. If  > 1 , capital deepening reduces the labor share, and if  < 1 ,
capital deepening increases the labor share.
Table 2 summarizes the movement of labor and the direction of technical bias given the
value of  for the case of a capital-augmenting technology.

Table 2 The Elasticity of Substitution and the Labor Share of Income

𝝈>𝟏 𝝈=𝟏 𝝈<𝟏

Factors are Factors are
substitutes complements
Capital-biased Neutral Labor-biased
Capital deepening Technology, Technology, Technology,
leads to …
lower labor share constant labor share higher labor share

6.3 Criticisms of the Cobb-Douglas Form

I personally have faith that there is a fundamental unity in economic as in physical life [...] There
is law and relative regularity everywhere else - why not in production and distribution?
– Paul Douglas (1967), p. 22

There are reasons to be skeptical that the Cobb-Douglas production function provides an entirely
satisfactory approximation to reality, however. First, most estimates suggest that the aggregate
elasticity of substitution is significantly less than 1. Second, a production function with an
elasticity of substitution of 1 does not provide a framework for analyzing fluctuations in factor
shares, such as those [observed in reality]
– Daron Acemoglu (2003), p. 3
40
 Even though the Cobb-Douglas form was supported by the data from 1899–1922, its
accuracy in different industries and time periods has been called into question. It is clear that the
model lacks micro-foundations. The function assumes that the labor and capital shares of total
output are constant over time, which is not always true and has not always been true in every
circumstance that the function was taken to the data (see Giovannoni 2014c). Neither Cobb nor
Douglas provided any theoretical reason why the coefficients  and  should be constant over
time or be the same among different sectors of the economy. Those are mathematical imperatives,
not economic imperatives.
The use of macroeconomic production functions spread following Solow’s (1957) classic
growth model was introduced. It is rarely noticed that shortly afterwards Solow (1958) qualified
the constancy of relative shares as “a mirage” and was implicitly skeptical about the use of such
aggregated production functions.
Numerous studies have tried to assess whether the Cobb-Douglas or the CES production
function was more appropriate to macroeconomic forecasting. Miller (2008) finds that the
strength of the Cobb-Douglas is its ease of use and its seemingly good empirical fit across many
data sets. Unfortunately, the fact that the Cobb-Douglas model also fits the data well in cases
where some of its fundamental assumptions are violated suggests that many empirical tests of the
Cobb-Douglas model are picking up a statistical artifact rather than an underlying production
function.
Similar results can be traced back to Shaikh (1974), who criticized the Cobb-Douglas
production function for having a weak theoretical basis—it is an identity, really. Shaikh shows
that the empirical results do not in fact have much to do with production conditions at all. Instead,
Shaikh shows that when factor shares are constant, there are broad classes of production data
(output, capital, and labor) which can always be related to each other through a functional form
mathematically identical to a Cobb-Douglas function with “constant returns to scale,” “neutral
technical change,” and “marginal products equal to factor rewards.”

41
 Fraser (2002) paid attention to the issue of whether the data provides deductive support for
the “laws of production” as claimed by Cobb and Douglas (1928). Only the New South Wales
data and, to a lesser extent, the New Zealand data produce supportive results. Moreover, Fraser
ran collinearity diagnostics to reexamine the original series studied, and the result shows all data
are subject to collinearity and that the time series properties raise questions as to the statistical
robustness of the estimates presented by Douglas.
Another criticism of the Cobb-Douglas production function rests in his possible
misinterpretation of technical progress (Miller, 2008). The majority of production functions
assume that technical progress is Hicks-neutral, which does not change the marginal products of
capital or labor given a certain ratio of inputs. Because of such a strong assumption, it can be
shown that Cobb-Douglas is the only functional form that is able to explain the U.S. experience of
constant factor shares and a rising capital-labor ratio (Antràs 2004). However, this is simply
because Cobb-Douglas is the only functional form where Hicks-neutrality can be equivalently
expressed as labor-augmenting technical change. Antràs (2004) also suggests that the finding of
the constant shares in many older econometric investigations may be due to an omitted-variable
bias caused by the assumption of Hicks neutral technical change.
Furthermore, Raval (2011) found that the Cobb-Douglas production function has two
empirical implications that do not hold in the data: a constant cost share of capital and a strong
co-movement in average revenue product of capital and average revenue product of labor. Raval
finds that the cost shares of capital are different within four-digit-SIC (standard industry code)
industries, so simply assuming that they are constant can lead to an estimation bias. Also, the
average revenue product of labor is found to increase much more with revenue than the revenue
product of capital.
Finally, there is the concern of short-run versus long-run breadth of analysis. Did the
empirical investigations into the Cobb-Douglas form and into factor share feature enough
datapoints? Surely the Cobb-Douglas results hold only in the long run, but how long is the long
run?

42
 Swimming against the current, Jones (2003) presents a defense of the Cobb-Douglas
production function by presenting readers with four stylized facts:
• The growth rate in U.S. GDP per capita has not shown a considerable trend for the
last 125 years.
• The capital share shows a significant trend in many countries and in many U.S
industries over time.
• The estimates of the elasticity of substitution between capital and labor are often
below unity.
• The price of capital goods in the “equipment” category, such as computers, machine
tools, has been falling relative to the price of nondurable consumption goods, where
the falling price is taken to indicate that technical progress is being embodies in
capital goods at a faster rate than in consumption goods.
Jones (2003) then attempts to reconcile the above facts with a Cobb-Douglas function
despite these facts not being compatible with it. However, this attempt is cut short by Chirinko
(2002) which finds the value of the elasticity is still significantly below unity in the short run and
in the long run. Thus, capital and labor are found to be complements, and capital deepening leads
to a higher labor share because technological progress is labor-biased.
Comparatively, the CES production function has less restrictive assumptions about the
interaction of capital and labor in production, but its data fit is inferior. One possible reason may
be that various studies are not all measuring the same thing. The CES production function
contains a number of variants that can be tested with either cross-sectional or time-series data.
Klump and de La Grandville (2000) suggest that cross-study results would be much more
meaningful when they are within the same CES family.
Moreover, the inefficiency of CES results can be attributed to the fact that time series
estimates of the elasticity of substitution are not well measured by least squares regression.
Klump and Preissler (2000) found that not all variants of CES functions commonly used are
consistently specified. Therefore, there is no compelling evidence suggesting one should prefer
CES to the Cobb-Douglas for forecasting GDP and income shares. However, since the seemingly
perfect data fit of Cobb-Douglas is likely due to an accounting identity and mathematical feature
rather than an underlying production function, the CES specification is probably getting better
because of its allowance for a changing labor share and non-unitary elasticity of substitution, in a
word, for being more general.
43
7 WHAT HAVE WE LEARNED? (NON)ERGODICITY AND THE ROLE OF
ECONOMIC POLICY

At the end of our inquiry it appears that there is no single model of income distribution that
has emerged as a mainstream model (Kregel, 1973). The expression “model of income
distribution” is an expression more true in plural form than in singular form. Consider the factors
that each author introduces to explain factor shares:
• Keynes: capitalist propensity to consume
• Kalecki: degree of monopoly, ratio of raw materials prices to aggregate prices,
capitalist consumption and investment
• Kaldor, Pasinetti: investment share of GDP, saving propensities
• Goodwin: employment ratio (itself a negative function of labor productivity, of
labor supply, and of the productivity of capital, all three assumed constant)
• Technology: degree of substitution between factors, type of technological change

In addition, recall that the models are cast in either full-employment or imperfectly
competitive frameworks, and that income distribution is alternatively constant, drifting or cyclical
around full employment. How can we reconcile such different frameworks and models?
The present survey and summary can leave the reader feeling one of several ways. An avid
reader is probably happy to learn about the many facets of the economics of factor shares.
However, it is easy to get lost, overwhelmed or exasperated by the diversity and cacophony of the
main theoretical models. A critical reader will note, despairingly, that the theoretical approaches
detailed above are just that: theoretical. In practice there is no clear-cut division between
aggregate labor income and aggregate profits. Economic theories have nothing to say about the
apportionment of proprietors’ income or the classification of interest income or the reason why
the labor share seems to have fallen precipitously since the early 2000s. There is no consensus for
this in the theoretical literature, when it even addresses such specific questions (Giovannoni
2014b).
Thus, it is hard to escape the conclusion that, in matters of factor shares, economic theories
are useful but are not enough. The theory of international trade is more precise, the theory of
economic growth is more developed, and so it goes with many other branches of economics. But
we must go beyond those limitations. Is there a common thread to income distribution theories?

44
 As stated in the introduction, the ambition of this paper is to shed light on various theories,
and then try to discern a pathway. Some decisive progress can be made by using a taxonomy
based on the answers to the following two questions:
1. “Is income distribution assumed to be constant?” and
2. “Is income distribution treated as an exogenous factor (in the sense of driving the
economy) or as an endogenous factor (adjusting to the rest of the economy)?”

Table 3 Suggested Taxonomy of Income Distribution Models

Is income distribution assumed constant?

no yes
Kalecki Goodwin
exogenous (driving)
The distribution Keynes (cycle around constant)
of income is… Cobb-Douglas
endogenous (driven) CES prod. function
Kaldor-Pasinetti

The justification of the placement of each model is as follows, going clockwise from the
top right cell. The Goodwin model presents a cyclical model where income distribution alternates
as two driving variables with the employment ratio. The wage share is assumed constant on
average. The labor share from either the Cobb-Douglas or CES production functions are driven by
technology, which is the real force underlying the changes in income distribution. What income
distribution drives, however, is not clear. Kaldor’s model and Pasinetti critique make income
distribution appear as a result of the investment and saving decisions of the economic agents.
Those adjust to the slack on the labor market and devise the distribution of income which is
compatible with full-employment. Thus income distribution in Kaldor-Pasinetti is “constant at
full-employment” and adjusts ever so slightly to correct for any departure from full employment,
as described in Kaldor (1956). Finally, for Kalecki, and seemingly for Keynes, the relative shares
need not be constant but they are drivers of the whole economic system through, respectively, the
degree of monopoly and the marginal propensity to consume. Note that there is no theory of
income distribution in Keynes, but there is enough evidence to place him in the top left cell (see
details in section 2).

45
 This taxonomy can be further refined by introducing the concept of (non)ergodicity as
exposed, for instance, in Davidson (2003). An ergodic economic system is an economy whose
future position is knowable in a deterministic way, possibly allowing for a stochastic error. In
such a world, Davidson argues, economic policy and Keynes are irrelevant for there is a natural
tendency of economies to self-correct. The future is knowable. Keynesian economics, Davidson
continues, is inherently nonergodic—the future is unknowable and economic agents make
decisions in radical uncertainty following rules of thumb and crowd movements.
It is the top left cell containing Kalecki and Keynes, and only this cell, which is compatible
with the idea of non-ergodicity as in Davidson (2003). Other cells, particularly when income
distribution is constant, are not compatible with nonergodicity: if income distribution is constant,
or cyclical, or if it is assumed to adjust to maintain full employment, then income distribution in
the future is knowable, and equilibrium will prevail. There is no need for economic policies
except perhaps insofar as to expedite the process of convergence towards equilibrium. In those
cases, income distribution is never “wrong” or inadequate, for it is the correct one that assures full
employment of resources.
Income distribution can only be a problem, to the contrary, if one adopts a Keynesian or
Kaleckian view of the economy. In those models income distribution can be inappropriate for full
employment, and an income distribution policy could be desirable. For Keynes an inadequate
distribution of income (outside of moral judgments) is one in which much income is diverted to
individuals with a low marginal propensity to consume (MPC); for Kalecki, capitalists can
confiscate much income to the detriment of workers. Hence the need of third actor, the State,
which can institute redistribution policies, industrial organization policies such as introducing
more competition, or introduce and support collective bargaining -among other possibilities.

46
8 REFERENCES

Acemoglu, D. (2002) Directed Technical Change, The Review of Economic Studies, 69,4,
781-809

Antràs, P. (2004) Is the US Aggregate Production Function Cobb-Douglas? New Estimates of the
Elasticity of Substitution, Contributions in Macroeconomics, 4, 1

Bertola G., R. Foellmi and J. Tweimuller (2006) Income Distribution in Macroeconomic Models,
Princeton: Princeton University Press

Bertoli, S and F. Farina (2007) The functional Distribution of Income: a Review of the
Theoretical Literature and of the Empirical Evidence Around its Recent Pattern in
European Countries, Department of Economic Policy, Finance and Development,
University of Siena, working paper 5/2007.

Bowley, A. (1900) Wages in the United Kingdom in the Nineteenth Century, Cambridge:
Cambridge University Press

_____ (1920) The Change in the Distribution of the National Income, 1880-1913, Oxford:
Clarendon Press

_____ (1937) Wages and Income in the United Kingdom since 1860, Cambridge: Cambridge
University Press

Bruno, M. (1962) A Note on the Implications of an Empirical Relationship Between Output per
Unit of Labor, the Wage Rate and the Capital-Labour Ratio, mimeo, Stanford University

Cobb, C. and P. Douglas (1928) A Theory of Production, The American Economic Review, 18, 1,
139-165

Choi, S., and J.V. Rios-Rull (2009) Understanding the Dynamics of Labor Share: The Role of
Noncompetitive Factor Prices, Annals of Economics and Statistics/Annales d’Économie et
de Statistique, 95/96, 251-277

Chirinko, R. (2002) Corporate Taxation, Capital Formation, and the Substitution Elasticity
Between Labor and Capital, National Tax Journal, 60, 339-355

Davidson, P. (1960) Theories of Aggregate Income Distribution, New Brunswick: Rutgers

University Press

_____ (2003) Financial Markets, Money and the Real World, Cheltenham: Edward Elgar
Publishing

Douglas, P. (1967) The Theory of Wages, New York: Kelley

Fraser, I. (2002). The Cobb-Douglas Production Function: an Antipodean Defense, Economic

Issues, 7, 39-58

47
Giovannoni (2014b) What Do we Know About the Labor Share and the Profit Share? Part II:
Empirical Evidence, The Levy Economics Institute of Bard College, working paper No.
forthcoming

_____(2014c) What Do we Know About the Labor Share and the Profit Share? Part III: Measures
and Structural Factors, The Levy Economics Institute of Bard College, working paper No.
forthcoming

Galí, J. (1996) Multiple Equilibria in A Growth Model with Monopolistic Competition, Economic
Theory, 8, 251-266

Gollin, D. (2008) Labour’s Share of Income, New Palgrave Dictionary of Economics, MacMillan

Goodwin, R. (1967) A Growth Cycle, in Feinstein (ed.) Socialism, Capitalism and Economic
Growth, Cambridge University Press, Cambridge

Irmen, A. (2011) Steady-state Growth and the Elasticity of Substitution, Journal of Economic
Dynamics and Control, 35, 8,1215–1228

Jones, C. (2003) Growth, Capital Shares, and a New perspective on Production Functions,
mimeo, University of California, Berkeley

Harrod, R. (1939) An Essay in Dynamic Theory, The Economic Journal, vol. 49, 193, 14-33

_____ (1948) Towards a Dynamic Economics, London: MacMillan

Hicks, J. (1932) The Theory of Wages, London: Macmillan

Kahn, R. (1931) The Relation of Home Investment to Unemployment, The Economic Journal, 41,
162, 173-198

_____ (1933) The Elasticity of Substitution and the Relative Share of a Factor, The Review of
Economic Studies, 1, 1, 72-78

Kaldor, N. (1956) Alternative Theories of Distribution, The Review of Economic Studies, 23, 2,
83-100

_____ (1957) A Model of Economic Growth, The Economic Journal, 67, 268, 591-624

_____ (1961) Capital Accumulation and Economic Growth, in Lutz & Hague (ed.) The Theory of
Capital, McMillan, London

Kalecki, M. (1935) Essai d’une Théorie du Mouvement Cyclique des Affaires, Revue d’Economie
Politique, 49, 285-305

_____ (1938) The Determinants of Distribution of the National Income, Econometrica, 6, 2,

97-112

_____ (1942) A Theory of Profits, Economic Journal, 206/207, 258-267. Reproduced in

48
 Osiatynski J.(1990) Collected Works of Michal Kalecki, Clarendon Press, Oxford

_____ (1954) Theory of Economic Dynamics, London: Allen & Unwin

_____ (1962) Studies in the Theory of Business Cycles, 1933-1939, London: Basil Blackwell

Keynes, J. M. (1930) A Treatise on Money, reproduced in volume V of Moggridge, D. (ed.)

(1973) The Collected Writings of John Maynard Keynes, London: MacMillan

_____ (1936) The General Theory of Employment, Interest and Money, Harcourt, Brace &
World, New-York, 1997 Prometheus Books edition, New York: Armonk

_____ (1939) Relative Movements of Real Wages and Output, The Economic Journal, 49, 193,
34-51

Klump, R., P. McAdam and A. Willman (2007) Factor Substitution and Factor-Augmenting
Technical Progress in the United States: a Normalized Supply-side System Approach, The
Review of Economics and Statistics, 89, 1, 183-192

Klump, R. and H. Preissler (2000) CES Production Functions and Economic Growth, The
Scandinavian Journal of Economics, 102, 1, 41-56

Klump, R. and O. de La Grandville (2000) Economic Growth and the Elasticity of Substitution:
Two Theorems and Some Suggestions, American Economic Review, 90, 1, 282–291

Kregel J. (1971) Rate of Profit, Distribution and Growth : Two Views, London: MacMillan

_____ The Reconstruction of Political Economy, second edition (1975), London: MacMillan

Lopez, J. and M. Assous (2010) Michal Kalecki, Great Thinkers in Economics, New York:
Palgrave MacMillan

Lucas, R. (2003) The Industrial Revolution: Past and Future, 2003 Annual report Essay, printed in
the May 2004 issue of The Region, Federal Reserve Bank of Minneapolis, available
online at
http://www.minneapolisfed.org/publications_papers/pub_display.cfm?id=3333http://ww
w.minneapolisfed.org/publications_papers/pub_display.cfm?id=3333

Miller, E. (2008) An Assessment of CES and Cobb-Douglas Production Functions, working

paper 2008/5, Congressional Budget Office, available online at
http://www.cbo.gov/sites/default/files/cbofiles/ftpdocs/94xx/doc9497/2008-05.pdfhttp://
www.cbo.gov/sites/default/files/cbofiles/ftpdocs/94xx/doc9497/2008-05.pdf

Mukerji, V. (1963) A Generalized SMAC Function with Constant Ratios of Elasticities of

Substitution, The Review of Economic Studies, 30, 233-236

Pasinetti, L. (1962) Rate of Profit and Income Distribution in Relation to the Rate of Economic
Growth, The Review of Economic Studies, 29, 4, 267-79

49
Raurich, X., H. Sala and V. Sorolla (2012) Factor Shares, the Price Markup, and the

_____ (1964b) Elasticity of Substitution Between Capital and Labor, Journal of

Macroeconomics, 34, 1, 181-198

Raval, D. (2011) Beyond Cobb-Douglas: Estimation of a CES Production Function with Factor
Augmenting Technology, US Census Bureau Center for Economic Studies Paper No.
CES-WP-11-05

Ricardo, D. (1817) Principles of Political Economy and Taxation, 1911 edition, London: John
Murray

Samuelson, P. (1964a) Economics: An Introductory Analysis, 6th edition, New York:

McGraw-Hill

A Brief Survey of post-Keynesian Developments, in Lekachman, R. (ed.) Keynes’ General

Theory: Reports of Three Decades, London: MacMillan

_____ (1972) Collected Scientific Papers, Cambridge, Mass.: MIT Press

Solow R. (1955) The Production Function and the Theory of Capital, The Review of Economic
Studies, 23, 2, 101-108

_____ (1956) A Contribution to the Theory of Economic Growth, Quarterly Journal of

Economics, 70, 1, 65–94

_____ (1957) Technical Change and the Aggregate Production Function, Review of Economics
and Statistics, 39, 3, 312-20

_____ (1958) A Skeptical Note on the Constancy of Relative Shares, American Economic
Review, 48, 4, 618-631

Shaikh, A. (1974) Laws of Production and Laws of Algebra: the Humbug Production Function,
The Review of Economics and Statistics, 56, 1, 115-120

Volterra, V. (1931) Théorie Mathématique de Lutte pour la Vie, Paris: Jacques Gabay

50