14039_49603_08_cha06.

qxp

2/7/2007

5:43 PM

Page 126

6
Is IS-LM a static or dynamic
Keynesian model?
Warren Young

Introduction
In my book (Young, 1987), I outlined the history of IS-LM, and based my
approach upon recollections and documents, as Hicks put it (1973). About
a decade later, this time with William Darity (Darity and Young, 1995), we
surveyed the early mathematical models of Keynes General Theory. In both
cases, the focus was on the original IS-LM approach and subsequent models based upon it. In this paper, I focus upon some overlooked aspects of the
interrelationships between these various interpretations and representations,
that is, between the early models themselves, over the decade after the original IS-LM approach of Roy Harrod, James Meade and John Hicks, and deal
with the impact of what Paul Samuelson called the KeynesLange system.
I then turn to the somewhat overlooked development of the dynamic
approach to IS-LM, as manifest in the work of Samuelson (1941) and
Franco Modigliani (1944), and the relationship between them and the
approach of Lawrence Klein (1947). Suggestions as to the possibility of utilizing archival resources to investigate these issues will be put forward.

From KeynesLange to ClowerLeijonhufvud

Oskar Langes 1938 paper The Rate of Interest and the Optimum Propensity
to Consume attempted to solve what can be termed the MalthusRamsey
problem. Lange noted that Malthus posed the problem as early as 1820 in
his Principles of Political Economy with unsurpassed clarity when he wrote
(Malthus, 1820: 89, 36970)
If consumption exceeds production, the capital of the country must
be diminished, and its wealth must be gradually destroyed from its
126

14039_49603_08_cha06.qxp

2/7/2007

5:43 PM

Page 127

Is IS-LM a static or dynamic Keynesian model?

127

want of power to produce; if production be in great excess above consumption, the motive to accumulate must cease from the want of will to
consume. The two extremes are obvious; and it follows that there must
be some intermediate point, though the resources of political economy
may not be able to ascertain it, where taking into consideration both
the power to produce and the will to consume, the encouragement to
the increase of wealth is greatest.
In his 1938 paper, Lange termed this point the optimum propensity to
save which maximizes investment (1938: 24, italics in original), and
went on to say that the general theory of interest outlined in his paper
enables us to solve this problem and to determine the optimum propensity to save which maximizes investment accordingly. He continued
(1938: 24)
Since investment per unit of time is a function of both the rate of
interest and expenditure on consumption a decrease of the propensity
to consume (increase in the propensity to save) has a twofold effect. On
the one hand the decrease of expenditure on consumption discourages investment, but the decrease in the propensity to consume also
causes . . . a fall of the rate of interest which encourages investment on
the other hand. The optimum propensity to consume is that at which
the encouraging and the discouraging effect of a change are in balance.
A decade earlier, in his classic EJ paper A Mathematical Theory of Savings
(1928), Frank Ramsey had re-stated Malthuss problem as How much of
its income should a nation save? (1928: 543). Ramsey went on to define
the maximum obtainable rate of enjoyment or utility as Bliss (1928: 545).
The notion of the optimum propensity to save and balance is implicit in
Ramseys assertion that (1928: 545)
in all cases we can see that the community must save enough [our
emphasis] either to reach bliss after a finite time, or at least to approximate to it indefinitely . . . Enough must therefore be saved to reach or
approach bliss some time, but this does not mean that our whole
income should be saved. The more we save the sooner we shall reach
bliss, but the less enjoyment we shall have now, and we have to set the
one against the other [my emphasis].
While Langes 1938 paper was reprinted in Gottfried Haberlers volume
Readings in Business Cycle Theory (1944), it received detailed attention only

14039_49603_08_cha06.qxp

128

2/7/2007

5:43 PM

Page 128

The Keynesian Tradition

a decade after its publication, in Kleins book The Keynesian Revolution (1947)
and especially in Samuelsons Foundations of Economic Analysis (1947).
Klein asserted that Langes notion of the optimum propensity to consume
was similar to that of the earlier underconsumptionist [Malthusian] idea
of the proper balance of consumption out of income (1947: 135).
Klein, however, took the view that (1947: 135)
As far as we are concerned, the optimum propensity to consume (in the
schedule sense) [italics in original] is that propensity which interacts
with the investment schedule to give a full-employment level of
national income, and there are an infinite number of consumption
functions which will do this.

Figure

Samuelson, in Foundations of Economic Analysis (1947), went much further in his treatment of what he termed the KeynesLange system
(1947: 354) not only providing it with a specific mathematical form
but describing it as an equilibrium system (1947: 27677). Samuelson
went on to say, however, that If we are to derive meaningful theorems,
we must clearly proceed to a consideration of a more general dynamic
system which includes the stationary Keynesian analysis as a special case.
He then proceeded to develop a differential system of dynamical equations leading to an equilibrium which is stable under conditions which
hold unambiguously. On this basis Samuelson establishes . . . theorems
which are not only useful but important that are derived directly from
the equilibrium conditions of the dynamized Keynes-Lange system, that
is if the equilibrium is stable (1947: 27879). We will discuss this in detail
below, in the context of our discussion of Samuelson (1941), which is
the source of the presentation in his 1947 volume. In his conclusion to
Foundations (1947), Samuelson further developed the question in the
comparative dynamics of the KeynesLange system he earlier analysed
in terms of comparative statical analysis regarding the influence on
investment of a change in thriftiness (1947: 353). As he put it (1947:
353), the wider problem was what happens to capital in the long run
[italics in original] as a result of a change in thriftiness. In Samuelsons
view (1947: 35354),
Thus, if instead of simply asking what level of consumption maximizes
current investment, we widen Professor Langes question and seek the
levels of consumption leading to the most capital at each instant of time
[italics in original], we shall find that capital formation in a run of any
length is only maximized if at each instant the Lange criteria are met.

14039_49603_08_cha06.qxp

2/7/2007

5:43 PM

Page 129

Is IS-LM a static or dynamic Keynesian model?

129

Now, in their 1947 books, neither Klein nor Samuelson took notice of the
extension of the KeynesLange system by Timlin (1942) whose work was,
however noted by Modigliani in his seminal 1944 Econometrica paper
(1944: 45). Her extension of the KeynesLange system proceeded in a
number of ways. First, she reemphasized the simultaneous and interdependent nature of the system. Secondly, she stressed the interest rate as
the key to the solution of the simultaneous and interdependent system;
this emanating from its dual role as both independent and dependent
variable. Thirdly, and most importantly, she developed a diagrammatic
representation of Langes Walasian system which did not reflect the static
equations of the IS-LM interpretations of the Keynesian system la Hicks,
Harrod and Meade (Young, 1987; Darity and Young, 1995). Rather, it
emphasized the system of the shifting equilibrium [my emphasis] which
lies at the heart of the KeynesLange General Equilibrium system according
to Timlin (1942: 7). This being said, and in light of the original aspects of
Mabel Timlins extension of the KeynesLange system, we called it the
LangeTimlin system accordingly (Young, 1987).
At this point, however, we must briefly recount the reinvention, some
two decades later, of the forgotten LangeTimlin system by Robert Clower
and Axel Leijonhufvud in their works in the 1960s (Clower, 1965;
Leijonhufvud, 1967, 1968, 1969; also see Clower and Leijonhufvud, 1975;
Leijonhufvud, 1983; Young, 1987). In his well-known 1965 article The
Keynesian Counter-Revolution, for example, Clower made no mention
of either Langes 1938 paper or Timlins 1942 book, the importance of
which we have seen above. This is difficult to understand because Clowers
general approach is foreshadowed in Langes 1938 paper, and in the 1965
paper, he presents Keyness system in a very similar way to Langes and
Timlins further development of it.
In outlining his interpretation of what he considered the cardinal element underlying Keynes General Theory, which he called the dualdecision hypothesis of household (consumption-saving) behaviour, Clower
attempted to reconcile Keynes with Walras Law, which, as he said, was
formally equivalent to Says Law (Clower, 1965: 275, 279, 289). Clowers
justification for his approach will not be discussed here. Suffice to say,
however, that according to Clower, Keynes either had a dual-decision
hypothesis [of household behaviour] at the back of his mind, or most of
the General Theory is theoretical nonsense (Clower, 1965: 28790).
Interestingly, Lange, in his 1938 paper, introduced the income identity
Y C I, as the sum of the budget equations of the individuals, and
noted investment or saving decisions can be different. He then distinguished between Walrass approach the equality of the value of the

14039_49603_08_cha06.qxp

130

2/7/2007

5:43 PM

Page 130

The Keynesian Tradition

capitaux neufs and the excess of income over consumption, i.e. saving
and that of his own system which, in Langes view, is as in the theory of
Mr. Keynes because it is an identity. According to Lange, whatever the
investment and savings decisions are, the volume of total income always
adjusts itself so as to equalize saving and investment actually performed.
This is a simple budget relationship, for the individuals incomes are equal
to the sum of expenditure on consumption and investment. Lange claims
that his identity corresponds to the sum of the budget equations in the
Walrasian system and shows how expenditure on consumption and
investment determine the total income. When this budget relationship
is taken account of, there is no need any more for a separate equation
indicating the equilibrium of saving and investment decisions based on
some given income, however defined.
Clowers 1965 attempt to link Keynes system with that of Walras by the
dual-decision hypothesis would seem, then, to originate in Langes 1938
paper and Timlins subsequent development of the KeynesLange system
(Lange, 1938: 14, 2223; Timlin, 1942). In his works published between
1967 and 1969, Leijonhufvud was even more explicit than Clower in his
attempt to base his views of Keynes theory upon a Walras-type system
(Leijonhufvud, 1967, 1968, 1969). Again, however, there is no mention of
Langes 1938 paper or Timlins 1942 book. Taking his lead from Clower,
Leijonhufvud asserted in his 1967 paper, for example, that the Clower
approach was the most suitable interpretation of Keyness General Theory
(1967: 4024). In an article published a decade after Clowers rediscovery
albeit unattributed of Langes approach and Timlins extension of it,
Clower and Leijonhufvud took what they called a Keynesian perspective
on the coordination of economic activities (Clower and Leijonhufvud,
1975). To be brief, they simply restated the link that Lange had made
between Keynes and Walras some 30 years earlier, which was subsequently
developed by Timlin more than two decades prior to Clowers 1965 paper.
In essence, Clower and Leijonhufvud simply turned the KeynesLange
and LangeTimlin systems into one variant of the neo-Walrasian synthesis
(Clower and Leijonhufvud, 1975: 184).

Samuelson, Modigliani and Klein: from statics to

dynamics, 19417
In the April 1941 issue of Econometrica, Samuelson published a paper
entitled The Stability of Equilibrium: Comparative Statics and Dynamics,
which was to become Chapter IX of Foundations of Economic Analysis (1947).
In this paper, he developed techniques . . . of . . . fruitful applicability in

14039_49603_08_cha06.qxp

2/7/2007

5:43 PM

Page 131

Is IS-LM a static or dynamic Keynesian model?

131

order to analyse what he called the Keynesian system (1941: 113). As

he put it (1941: 113) I shall analyse in some detail the simple Keynesian
model as outlined in the General Theory. Various writers, such as Meade,
Hicks, and Lange, have developed explicitly in mathematical form the
meaning of the Keynesian system. He went on to outline an equational
system of the form:
C(i, Y) Y I
F(i, Y) I
L(i, Y) M
(1941: 114, equations 5658)
Samuelson then provided three relations to determine the three unknowns
[i, Y, I] in terms of three parameters [, , M] (1941: 114, equation 59),
going on to call this the Keynesian equilibrium system (my emphasis).
He then proceeded to devise a more general dynamic system, encompassing the stationary Keynesian analysis as a special case, and considered
two distinct cases, the first of which was based upon a differential system
(1941: 115). In order to do this, he replaced the static equations (1941: 114,
equations 5658) with dynamical ones (1941: 116, equations 6365),
and provided a solution set for them (1941: 116, equation 66). He then
established the conditions for the unambiguous stability of the dynamic
equations (1941: 116, equations 68 and 69).
The second case Samuelson considered was that of a dynamic system
based upon a difference equation set which, if none of the variables are
taken as given (1941: 119) was specified by him as (1941: 120, equation 85)
C(it, Y t 1) Yt It 0,
F(it, Yt) It 0,
L(it, Yt) M 0
He then went on to present the conditions that assured stability of equilibrium (1941: 120).
There are a number of important things to notice in Samuelsons
representation of what he called the Keynesian system. First, he utilized a consumption function, rather than a savings function. Second,
he specified consumption as being a function of both income and the
interest rate, as did Lange (Young, 1987: 80). And this, in contrast to the
approaches of Hicks, Harrod and Meade before him (Young 1987; Darity
and Young 1995: 23) and Modigliani (1944: 46) and Klein (1947: 199ff)
subsequently, all of whom utilized savings rather than consumption

14039_49603_08_cha06.qxp

132

2/7/2007

5:43 PM

Page 132

The Keynesian Tradition

functions in their respective models of the Keynesian system, as they

interpreted it. However, Samuelsons innovative step, as we put it over
a decade ago (Darity and Young, 1995: 29) was to introduce an income lag
in the consumption function, and to push the analytics into the realm
of solutions of difference equations. Moreover, as we noted (1995: 29,
note 19) while the mathematics of difference equations . . . [was] alien
to Keynes mode of presentation, Samuelson (1988) contends they were
very much present in the conceptual structure of his argument, if not in
the General Theory, then in Keyness Galton Lecture, published in the
Eugenics Review in 1937.
Less than three years later, in the January 1944 issue of Econometrica,
Modigliani published a paper entitled Liquidity Preference and the Theory
of Interest and Money. This has been widely recognized as the apex
of the neoclassical synthesis (Young, 1987: 1215; Darity and Young,
1995: 2426). What has received less attention is the dynamic model in
Modiglianis paper (1944: 6264), and no comparison, as far as I know, has
ever been made between Modiglianis dynamic model of 1944 vintage and
Samuelsons dynamic model of 1941 vintage. Moreover, Modigliani did not
mention Samuelsons dynamic model in his 1944 paper, although he did cite
Lange and Timlin respectively (1944: 46, note 4, 50, note 9, 52, note 11).
In Section 10 of his 1944 paper, which he called A dynamic model of the
Keynesian theory and the stability of equilibrium, Modigliani developed
a system of difference equations which he considered as the simplest
dynamic model of the theory he presented (1944: 62). Modigliani specified a system of difference equations (1944: 63, equations 2.12.6), and
went on to develop stability conditions (1944: 6364), going on to say
that if these conditions are satisfied, the variables approach their equilibrium values, which are the same as those obtained by solving the static
system he presented in this section on Macrostatic Systems (1944,
Section 2: 4648). It is important to note here that Modiglianis stability
condition for the system of difference equations (1944: 64) and that of
Samuelsons Case 2 system of difference equations (1941: 119, equation 83) are parallel; and this, while Modigliani did not cite Samuelson
(1941) at all.
In his 1947 book The Keynesian Revolution, Klein mentioned Samuelsons
1941 paper (1947: 112, note 25), but Modiglianis 1944 paper is not cited
by Klein at all. In addition, Klein did not include Modigliani as a supporter of Keynesian economics amongst economists working in the US
in the list he provided (1947: 184), which included Hansen, Samuelson,
Smithies, Mosak, Metzler, Hagen, Lerner, and even Lange, although he
added a caveat to the effect that some individuals in this group would not

14039_49603_08_cha06.qxp

2/7/2007

5:43 PM

Page 133

Is IS-LM a static or dynamic Keynesian model?

133

call themselves Keynesians (1947: 184, note 6). On the other hand, he
cited Langes 1938 paper in his book (1947: 135, note 13). Moreover, in
the technical appendix to his book, Klein made a specific reference to
the stability conditions in Samuelsons 1941 dynamic model when he
wrote Professor Samuelson has shown that if a dynamic model for which
our static system is the stationary solution, is to be stable then the denominator must be negative (1941: 205). This directly follows from the equilibrium condition Samuelson set out for his dynamical system of equations
as described above (1941: 116, equations 6369).

Suggested questions for archival research

The following questions arise, that can only be answered by means of
studying the papers of Clower, Leijonhufvud, Samuelson, Modigliani
and Klein respectively:
(1) Why are Langes 1938 paper and Timlins 1942 book not mentioned
by Clower and Leijonhufvud in any of their works?
(2) Why is Samuelsons 1941 paper not mentioned by Modigliani in his
1944 paper?
(3) Why is Modiglianis 1944 paper not mentioned by Klein in his 1947
book?
The papers of Clower, Leijonhufvud and Modigliani are accessible at
Duke University. Samuelsons papers are at MIT Special Collections (MC
403), while only those of Klein remain to be deposited in archives; however, some of the individuals involved can be directly approached with the
questions listed above and others, relating to the relationship between
their respective works. Answers to these questions should throw new light
on the interrelationship between some of the second generation of interpreters of Keyness General Theory, and provide at least a partial answer to
the question: is IS-LM a static and dynamic Keynesian model?

References
Clower, R. (1965) The Keynesian Counter-Revolution: A Theoretical Appraisal,
in F. Hahn and F. Brechling (eds), The Theory of Interest Rates (London:
Macmillan); reprinted in R. Clower (ed.) (1969), Monetary Theory (Harmondsworth:
Penguin).
, and A. Leijonhufvud (1975) The Coordination of Economic Activities:
A Keynesian Perspective, American Economic Review: Papers and Proceedings
vol. 65: 1828.

14039_49603_08_cha06.qxp

134

2/7/2007

5:43 PM

Page 134

The Keynesian Tradition

Darity, W., and W. Young (1995) IS-LM: an Inquest, History of Political Economy,
vol. 27: 141.
Haberler, G. (ed.) (1944) Readings in Business Cycle Theory (Philadelphia:
Blakiston).
Hicks, J. (1973) Recollections and Documents, Economica (n.s.) vol. 40: 211.
Klein, L. (1947) The Keynesian Revolution (New York: Macmillan).
Lange, O. (1938) The Rate of Interest and the Optimum Propensity to Consume,
Economica (n.s.) vol. 5: 1232.
Leijonhufvud, A. (1967) Keynes and the Keynesians: A Suggested Interpretation,
American Economic Review Papers and Proceedings, vol. 57: 40110.
(1968) On Keynesian Economics and the Economics of Keynes (Oxford
University Press).
(1969) Keynes and the Classics (London: Institute of Economic Affairs).
(1983) What Was the Matter with IS-LM, in J. Fitoussi, (ed.), Modern
Macroeconomic Theory (Oxford: Blackwell).
Malthus, T. (1820) Principles of Political Economy (London: J. Murray).
Modigliani, F. (1944) Liquidity Preference and the Theory of Interest and
Money, Econometrica, vol. 12: 4588.
Ramsey, F. (1928) A Mathematical Theory of Saving, Economic Journal, vol. 38:
543559.
Samuelson, P. (1941) The Stability of Equilibrium: Comparative Statics and
Dynamics, Econometrica, vol. 9: 97120.
(1947) Foundations of Economic Analysis (Cambridge, MA: Harvard
University Press).
(1988) The Keynes-Hansen-Samuelson Multiplier-Accelerator Model,
Japan and the World Economy, vol. 1: 319.
Timlin, M. (1942) Keynesian Economics (University of Toronto Press).
Young, W. (1987) Interpreting Mr Keynes: The IS-LM Enigma (Oxford: PolityBlackwell).