Until recently Chicago economists, especially Milton Friedman, saw themselves as working in a Marshallian tradition.

TO BE NOTED: Via The Economist's View:

"Marshallian General Equilibrium Analysis
David Colander (Middlebury College)
In an assessment of Alfred Marshall, Paul Samuelson (1967) writes
that “The ambiguities of Alfred Marshall paralyzed the best brains in the
Anglo-Saxon branch of our profession for three decades.“ In making this
assessment he carried on a tradition of Marshall-bashing that has a long
history in economics, dating back to Stanley Jevons and F. Y. Edgeworth,
who accused Marshallian economists of being seduced by “zig zag windings of
the flowery path of literature.” (Edgeworth, 1925)
These harsh assessments of Marshall and his approach to economics
have had their influence on the modern profession and, other than historians
of economic thought, few young economists know much about him. Fewer still
would see themselves as Marshallians.1
Today, Marshall is best remembered for his contribution to partial
equilibrium supply and demand analysis.2 For the true economic theorists of
the 1990s, however, this contribution is de minimus; the partial equilibrium
approach is for novice economists with no stomach for real economic theory—
general equilibrium. The profession’s collective view of Marshall in the 1990s
1Until recently Chicago economists, especially Milton Friedman, saw themselves as working in a Marshallian
tradition. More recently, however, younger Chicago economists know little of Marshall, and work in the same
Walrasian general equilibrium framework as does the majority of the profession.
2 Even here, Marshall’s contribution is questioned. As Humpries and(to come) (1994) argue, Marshall was neither
first, nor clearest, in his presentation of partial equilibrium supply and demand.
Marshallian General Equilibrium Analysis
2
is that Marshall is passé--at most a pedagogical stepping stone for
undergraduate students, but otherwise quite irrelevant to modern economics.
The motto of recent 20th century economics has been:
Marshall is for kids and liberal arts professors; real economists
(professors at universities) do Walras.
Since Marshall’s name is synonymous with partial equilibrium
analysis, the title of this paper will seem strange to many. (One well known
economist, upon hearing it, labeledthe title an oxymoron.) Most economists
think of general equilibrium analysis as synonymous with Walrasian general
equilibrium analysis. In this paper I argue that this is not true. Marshall
was centrally concerned with general equilibrium analysis; he was, after all, a
classical economist, and he drew on, and saw his work as extending, the work
of Adam Smith, David Ricardo, and John Stuart Mill, all of whom were
concerned with general equilibrium, not partial equilibrium, issues.
I shall also argue that the profession’s negative assessment of
Marshall is wrong. Specifically, I shall argue that, conceptually, Marshallian
general equilibrium analysis is at a much higher level than Walrasian
general equilibrium analysis, and, becasue it is, is far more compatible with
modern developments in economics than is Walrasian general equilibrium.
Thus, Marshall's work is not a stepping stone to Walras, but is instead a
stepping stone beyond Walras. It is consistent with a fundamentally different
conception of general equilibrium, one which recognizes that the
mathematical formulation of a meaningful general equilibrium model is
much more intractable than those with which Walras and later Walrasians
dealt.
Marshallian General Equilibrium Analysis
3
Marshall's Interest in General Equilibrium Analysis
Marshall’s interest in general equilibrium is more than simply a
conjecture of mine. In Note 20 (Note 21 of 2-9th edition) of Principles of
Economics, Alfred Marshall discusses the issues of general equilibrium in his
“bird’s eye view of joint demand, composite demand, joint and composite
supply when all arise together." In discussing this note in a letter to J.B.
Clarke, Marshall (1908) comments that “my whole life has been and will be
given to presenting in realistic form as much as I can of my Note 21. If I live
to complete my scheme fairly well, people will, I think, realize that it has
unity and individuality."
Consistent with this view we can find discussions of interrelationships
among markets in his Principles. (See, for example, p. 711.) But what those
discussions present are observations of realities, not analytics. As I will argue
below, Marshall used the real world observations as a guide to the
interrelationship among markets because he believed that an analytic
understanding of these interrelationships was beyond the mathematical
specifications of the time. Given that belief, it is not surprising that
Marshall’s discussions were not about abstract mathematical
interrelationships, but were about observed interdependencies that
acknowledged institutional realities. In a sense Marshall used the actual
economy as an analog for the analytic model. If, in the short run, observed
prices were relatively fixed and quantities were variable, then one solution to
the complicated general equilibrium model underlying the economy must be
relatively fixed nominal prices and fluctuating quantities. Observations
Marshallian General Equilibrium Analysis
4
served as the basis for his discussions of the interrelationships among
markets.
Why Marshall Shied Away from Developing a Formal General
Equilibrium Model
Why did Marshal focus his analysis on partial equilibrium and not
formally develop his conception of general equilibrium? One possible
explanation is that he was not the mathematician or conceptualizer that
Walras was, and that he knew he was incapable of formally specifying a
general equilibrium system. I think it is correct that he felt incapable of
specifying a meaningful formal general equilibrium system, but not because
he was unable to formulate a system such as Walras’s. One reason I believe
this is that Marshall was a trained mathematician, and by most accounts, a
good one. He understood simultaneous equations and had the ability to solve
systems of simultaneous equations. His Note 21 summarizes the essence of a
broad conception of general equilibrium better than any other one page
written on the subject.
I believe the reason Marshall didn’t formally analyze general
equilibrium issues is that he demanded intuitive correspondence between
math and his understanding of the economy. When that correspondence was
Marshallian General Equilibrium Analysis
5
not present, the math was irrelevant; such irrelevant math should be
discarded.3
Marshall’s recognition of the analytic intractability of the general
equilibrium problem, given the math available to him, and his desire for
concreteness in his economics, led him to shy away from abstract
specifications of general equilibrium. Leon Walras, meanwhile, had less
aversion to abstraction devoid of intuitive correspondence with reality, and
trod where others would not go. Unfortunately, it was a path that others
followed, and Walras’s version of the general equilibrium system has become
the foundation of modern 20th century economics, while Marshallian general
equilibrium economics never developed.4 Thus, when Paul Samuelson
3 In a well known letter to A. L. Bowley, Marshall wrote: "I had a growing feeling in the later years of my work at the
subject that a good mathematical theorem dealing with economic hypotheses was very unlikely to be good
economics; and I went more and more on the rules (1) Use mathematics as a shorthand language, rather than as an
engine of inquiry. (2) Keep to them until you have done. (3) Translate into English. (4) Then illustrate by examples
that are important in reallife. (5) Burn the mathematics. (6) If you can’t succeed in (4), burn (3). This last I did
often.” (Pigou, Memorials pg. 427.)
4I want to be careful to avoid the type of unfair criticisms of Walras that I believe earlier economists have made of
Marshall. I am not an expert on Walras, and what I am criticizing as Walrasian is what has been passed down as
Walrasian, not necessarily what a fair interpretation of Walras would include. I an sure that there are many subtleties
in Walras, which, if given a sympathetic reading, can lead one to conclude that Walras would have opposed what
came to be known as Walrasian general equilibrium--that is analysis of the aggregate economy that assumes a unique
equilibrium system in which an auctioneer sets price and no trading is done at disequilibrium prices even though the
system is always in disequilibrium.
Marshallian General Equilibrium Analysis
6
developed the mathmatical foundations of modern economics, (Samuelson, 1
), he developed them around Walrasian economics. Similarly when the
microfoundations to macro were developed, they were developed along
Walrasian general equilibrium lines.
To Marshall, once one mastered the intuition of the general
equilibrium reasoning, going through formal specification in the way Walras
did was laborious but trivial. Such an exercise was worth one page in an
appendix in the Principles. Anyone with reasonable training in math could
work out a system of general interrelated equations. Marshall did not do so
because it would not add much to our understanding, and would violate the
law of significant digits, since such a specification would be incomplete. The
problem was the interrelation between dynamic and static issues; such
interrelationships clearly existed and, in Marshall’s mind, invalidated any
static analytic conclusion at which one could arrive. Marshall followed the
maxim: better to be ambiguous and relevant than precise and irrelevant.
For example, Donald Walker (1994) argues that while this view of Walras follows from the fourth edition of
the Elements, the version most English speaking economists are familiar with (since that was the version translated),
in earlier versions there was a different, and what Walker believes is a more meaningful system--one that is closer to
the system I am attributing to Marshall, and that elsewhere (Colander 1995) I have called Post-Walrasian. Walker
calls this earlier version the “mature Walras” and attributes the later version to Walras’s intellectual decline that
began in the mid-1890s. Others Walrasian scholars I have talked to argue that the Walrasian system does not even
follow from the fourth edition.
I leave it for historians of thought to determine whether Walras and Marshall are closer in their view of
general equilibrium than my argument suggests, and whether the entire development of modern general equilibrium
is based on a wrong interpretation of Walras, or on the translation of the wrong edition of his book.
Marshallian General Equilibrium Analysis
7
Marshall was not the only economist of the time who did not make the
jump to Walrasian-style general equilibrium. Auguste Cournot and F. W.
Edgeworth were also superb mathematicians, and they too shied away from
developing a formal general equilibrium system. Only Walras made the jump
to a formal specification of the general equilibrium system. One possible
explanation for why Walras trod where others would not is that Walras
simply wasn’t a top flight mathematician, as evidenced by his failure to gain
admittance to the Ecole Polytechnique. Moreover, as Landreth and I argue in
(Landreth and Colander 1994), Walras relied on others to clear up
mathematical problems. For example, his development of marginal
productivity followed Wicksteed’s superior treatment. Similarly, his
knowledge of multivariate calculus was limited, and his early editions
demonstrated confusion about interdependent derivatives where cross
partials were required. Thus my conclusion on this question of mathematical
ability is that Marshall was lost in the “zig zag windings of the flowery path
of literature” by choice, not by relative lack of understanding or mathematical
ability compared to Walras.
What I am arguing is that Marshall understood the intricacies of
general equilibrium far better than did Walras, and knew that a formal
mathematical specification of those intricacies that was necessary to meet his
demand for correspondence between the math and the intuition was beyond
him. Consider his description of the stability of a supply demand equilibrium.
He writes:
When demand and supply are in stable equilibrium, if any accident
should move the scale of production from its equilibrium position,
there will be instantly brought into play forces tending to push it
Marshallian General Equilibrium Analysis
8
back to that position; just as, if a stone hanging by a string is
displaced from its equilibrium position, the force of gravity will at
once tend to bring it back to its equilibrium position....
But in real life such oscillations are seldom as rhythmical as
those of a stone hanging freely from a string; the comparison would
be more exact if the string were supposed to hang in the troubled
waters of a mill-race, whose stream was at one time allowed to flow
freely, and at another partially cut off. Nor are these complexities
sufficient to illustrate all the disturbances with which the
economists and the merchants alike are forced to concern
themselves. If the person holding the string swings his hand with
movements partly rhythmical and partly arbitrary, the illustration
will not outrun the difficulties of some very real and practical
problems of value. For indeed the demand and supply schedules do
not in practice remain unchanged for a long time together, but are
constantly being changed; and every change in them alters the
equilibrium amount and the equilibrium price, and thus gives new
positions to the centres about which the amount and the price tend
to oscillate. (Marshall, Principles pp. 346-347.)
As Barkley Rosser (1991) points out, the metaphor in this passage is a system
that exhibits chaotic, or at least partially chaotic, dynamics. To meaningfully
analyze such a system requires an interdependent system of equations
involving, at a minimum, complex second and third order differential
equations. The solutions to such systems are anything but simple; they
exhibit path dependency, and sensitive dependence on initial conditions.
Marshallian General Equilibrium Analysis
9
Marshall recognized this complexity, and did not try to fly before the
airplane had been invented. He knew he could not deal with the issues
formally, so he did the best he could to deal with them informally. In Walras,
observed reality was forced to be consistent with available mathematical
techniques. In Marshall what is, is what we observe, and if what we observe
doesn't fit the available math, then we will simply have to write about the
ambiguities in words, and wait for the mathematical techniques to develop.
Marshall introduced his period analysis with a market period, short
period, and long period. As Axel Leijonhufvud points out, this period
approach to studying the adjustment of potentially complex non-linear
systems was the type of approach physicists were using in studying problems
involving non-linear dynamics. It was known as adiabatic transformations in
the older thermodynamics literature. (Leijonhufvud 1995.)
My point is not that Marshall’s treatment of such issues was
satisfactory; it had serious problems, and Marshall knew it. For example, he
wrote that his treatment of time and the various runs was the weakest
element of his analysis. (Marshall, 1908) My point is that Marshall
recognized that these issues were of fundamental importance, and that the
then available mathematics was insufficient even to begin to handle those
problems. Since such complicated issues were central to understanding the
workings of the aggregate economy, why formulate formal models that
deviated so much from observations? Only now, in the 1990s, are economists
becoming sufficiently familiar with the math relevant to such situations—
non-linear dynamics, chaotics and complexity—to start to apply them in their
models.
Marshallian General Equilibrium Analysis
10
A second reason Marshall did not formally specify his general
equilibrium system was that he was a cautious man; for example, although
he had worked out the central elements of partial equilibrium supply and
demand analysis, and his foundations of neoclassical economics, in the 1870s
when Menger and Jevons were espousing their claims, he did not publish
them until the 1890s—twenty years later. Keynes, reflecting on Marshall’s
cautious nature writes: “Jevons saw the kettle boil and cried out with the
delighted voice of a child; Marshall too had seen the kettle boil and sat down
silently to build an engine.” (Keynes 1956.) Marshall recognized that the
jump to general equilibrium was, by contrast to the jump to partial
equilibrium, a gigantic leap worthy of at least a 100-year wait, if partial
equilibrium took a 20-year wait.
The Marshallian General Equilibrium System
I admire Marshall, but I do not share his cautiousness. I have more the
personality, and the mathematical ability, of Walras. Moreover, mathematics
has developed enormously since the late 1800s; work in complexity theory,
non-linear dynamics, chaos theory, and the developments in computers has
given us tools needed to gain more understanding of complex systems—tools
Marshall did not have. In short, our formal tools have begun to catch up with
Marshall’s intuition. Thus, there are many similarities between this new
work and Marshall’s approach to general equilibrium.
These developments in math, combined with an inherent
incautiousness, place me in an ideal position to do what Marshall would not
do—to spell out a possible vision of his conception of general equilibrium, and
Marshallian General Equilibrium Analysis
11
to show how it contrasts with Walras’s. 5 It will be a broad vision, one that
will likely raise as many questions as it answers. But, I believe, that while I
will not adequately specify a Marshallian general equilibrium system, I will
make clear why it is what we should be working on, rather than adding yet
another detail to almost vacuous Walrasian vision.
Introducing Stability Through Institutions
The central organizing theme of the Marshallian general equilibrium
system that I am proposing is the following observation: Our economy may
be messy and sometimes chaotic, but it is nowhere near as chaotic as one
would expect of the solution to a general equilibrium system of simultaneous
equations. Realistic assumptions about interactions would cause a system of
simultaneous equations of a Walrasian type to exhibit dynamic path
dependencies, non-linearities, and strategic interdependencies, which will
making it far more chaotic than the observed reality. This means that our
economy cannot be described by such a system of simultaneous equations. It
follows that it is senseless to accept what might be called the Walrasian
fudge—the assumption of a Walrasian auctioneer who eliminated all dynamic
disequilibrium adjustment problems and brought about equilibrium.
5 I should make it clear that in this endeavor while, I believe, the conception I put forward in a Marshallian tradition, I
make no claims of it being the only general equilibrium conception consistent with Marshall. It is what I elsewhere
call a Post-Walrasian conception. What Marshall would have put forward, or what can be teased out of Marshall, is
infinitely debatable. I do not want to be part of that debate; my interest in the past is in its ability to generate ideas
about the present and future, not in the past itself.
Marshallian General Equilibrium Analysis
12
Mathematically, this fudge created the possibility of a solution to the system,
but it was a conceptually uninteresting solution to anyone who agrees with
the Marshallian observation. 6
For Marshall the question was what to do in specifying an alternative
system, and given his cautiousness, he simply did nothing. What I am
proposing is that he should have posited a the existence of a set of equations
combined with restrictions on the aggregate combinations of individual
actions, and hence on individual actions themselves, that eliminated these
instabilities. These restrictions are what might be the Marshallian fudge.
This Marshallian fudge involves a substantive role of existing
institutions and non-market coordinating mechanisms in providing the
coordination that is assumed in Walras. These institutions provide a
framework of coordination, but they also provide systemic constraints on the
decision making of individuals. Any analysis of individual decision making
must take into account these systemic constraints. An institution-less
economy would, in this Marshallian sense, be unstable; it would be
6There are two reasons why I believe Marshall could not accept the Walrasian fudge. The first is that he did not believe
that general equilibrium issues could be reasonably dealt with using a set of timeless interrelated simultaneous
equations because individuals do not have the capabilities to process the information necessary to deal with such a
system. The second is that if people did have the capabilities to deal with general equilibrium analysis, the result
would have been chaos since there were too many options and strategic interdependencies.
Marshallian General Equilibrium Analysis
13
characterized by anarchy and chaos.7 Because these restrictions embodied in
institutions provide the stability necessary to prevent chaos, such restrictions
must be included in the analysis. Institutions provide stability, but they also
provide restrictions on individual actions. You cannot assume stability
without institutions.
The Marshallian fudge follows from insights we get from the analysis
of complex systems. Inevitably those complex systems are not organized with
a single system of simultaneous equations; instead they are organized with
hierarchical structures that take advantage of the computational abilities of
the various levels. A metaphor for this approach is the way a computer is
organized. It has an operating system, software, and nested software.
Individuals operating at lower levels do not understand the workings of the
entire computer; they accept the rationality of their subsystem.
The essence of my proposed Marshallian general equilibrium analysis
is that it sees the interaction among sectors as being solved in a sequential
manner in which nested institutions of various longevities are accepted by
some set of individual decision makers. These institutions limit instability
with a corridor around existing situations. In normal times, individual
optimization is conducted given the multiple leveled constraints, but every so
often, perhaps because of a large autonomous shock, or simply spontaneous
7The irony of Marshall's general equilibrium system is that if it is taken seriously, it undermines the one contribution
that he is known for--partial equilibrium, because what is now known as partial equilibrium does not take into
account the constraints imposed on individual decision makers by general equilibrium institutions.
Marshallian General Equilibrium Analysis
14
dissatisfaction, individuals challenge these constraints; aggregate stability is
lost, and new institutional structures, and new constraints, emerge.
Marshallian rationality is fundamentally different than Walrasian
rationality, and its role in the system is different. Decisions are made
sequentially, and certain decisions, once made, become operating data for
lower level systems. Marshallian rationality can mean many different things
depending on what level one is operating at. For most decisions the
institutions, and the constraints they impose on individuals' decisions, are
the central feature of fixity in the Marshallian general equilibrium system,
and the shorter the run, the more institutions are assumed fixed.
Marshallian rationality is defined locally, not globally.8 In fact, Marshallian
systemic stability depends on individuals not exhibiting global rationality.
People’s limitations make it possible for institutions to develop; their bounded
rationality creates a stability that could not exist if everyone pushed
economic maximization to the limit.
But this Marshallian systemic stability is fragile; the economy is
always bordering on chaos, and when a sufficient number of individuals try to
take advantage of the niches in the system left by institutions—i.e., follow
economic rather than social restrictions—the institutions fail, stability is lost,
and a new set of institutions must be found to provide the necessary stability.
In short, the system takes advantage of people’s cost of computing and,
whenever, possible, chooses an institution that provides stability.
8Herbert Simon, and his bounded rationality, is the logical follower of Marshall.
Marshallian General Equilibrium Analysis
15
Notice the difference between the Marshallian and Walrasian
conception of economically rational actor. In the Walrasian conception the
ultra rational economic actor drives the system to equilibrium and serves a
useful purpose. In the Marshallian system such ultra-rational economic
actors can destroy the system by destroying the institutions that give it
stability.
A Mathematical Specification of Marshall’s General Equilibrium
System
Mathematically, Marshall’s jump to general equilibrium would not be
a single jump, but rather a set of jumps; these intermediate jumps complicate
the mathematics of general equilibrium enormously. It involves specifying all
decisions as a system of multiple nested equations.
y=f(g(h(k(l(x))))))
One could argue that such a layered problem could be reduced to a Walrasian
system by simply reducing this equation into a composite function:
y=f’(x)
That could be done, but the functional form would have no relationship to our
intuition. It would be non-continuous, and there would be no presumption of
its having any of the nice properties that we need it to have if we are to
analyze it. formally. The reason is that the broader optimization involves
complex programming problems that cause strategies to shift substantially as
the situation changes slightly. Marshall would argue that we can reasonable
Marshallian General Equilibrium Analysis
16
hope to understand the system only when people are operating within the
system—when they are accepting the restrictions that are imposed at all but
the lowest level. Our intuition doesn’t go beyond that. Hence, Marshall’s
limited focus of analysis.
Marshall sees it as impossible to go from intuition to specification of
composite functional form as is done in Walrasian general equilibrium
analysis. There is no presumption of correspondence of intuition and
functional form. The characteristics of the composite function will likely be
significantly different than the characteristics one would intuitively identify.
If you use the composite function rule, the composite function should
have built into it all the constraints that would follow from the intuition of
combining the various functions. One cannot continue to use one's intuition
about functional relationships as if a composite function were not used. But
that is precisely what is done in Walrasian general equilibrium. It was
because Marshall recognized the limitations of the mathematics of this
multiple jump that he chose the zig zags of literary exposition rather than the
assured failure of mathematical specification.
The mathematical specification of such a layered equilibrium is
extraordinarily difficult, and each layer involves a slight deviation from
intuition. Thus, when Robert Solow writes that Alfred Marshall seems to
have felt that, at every level of mathematical deduction, a little truth leaked
out, Solow was right. But Solow was suggesting that Marshall was wrong in
believing that; I am suggesting Marshall was right. If you are trying to
relate intuition and observation with formal specification, the tolerances of
deviation increase with each functional form one specifies.
Marshallian General Equilibrium Analysis
17
How does one deal with such a problem? By relying on what one sees,
not on what one deduces. This accounts for the Marshallian dynamics that
assumes prices are fixed, and quantities variable in the short run
adjustment, as opposed to Walrasian dynamics which sees prices variable
and quantities fixed.
Some Implications of the Marshallian Approach
There are many implications of this Marshallian approach to general
equilibrium for the way we do economics in the 1990s. For example, consider
the justification for the dynamics. In Walrasian economics one must search
for a microfoundation for such dynamics. Why don’t individuals allow prices
to fluctuate, since that would be optimal? In Marshallian general equilibrium
there is no such presumption, and thus the search for a contextless micro
foundations, a search that characterizes much of modern macro, is
meaningless. If institutions exhibit relatively fixed nominal prices, such fixity
is a macro systemic constraint that is imposed by institutional requirements.
A second, related, issue concerns optimality of the market system.
Since multiple institutions can be chosen, there is no presumption of systemic
optimality of a market system. Any conclusion about systemic optimality
follows only from a consideration of comparative institutions. There is no
assurance that the market system coordinates better than other systems. If it
does, this is an observable phenomenon, not a deduced fact. In fact, in the
Marshallian system the concept “market” has no meaning without a
specification of the institutions that make that market feasible.
Marshallian General Equilibrium Analysis
18
In the long run all interactions are possible, but like a computer
without an operating system, the long run institutional structure is
extremely user unfriendly. Changes in that institutional structure are made
with great trouble. There is no omnipotent being choosing the best system,
but, instead, there are individuals working within the institutional structure
they have. Bill Gates said that God created the world in seven days, but He
didn’t have an installed user base. It is not simply a single operating system
that our economy has, but is instead a multiple layered system of nested
software, and that nested system makes change, and any analysis of low level
decisions, extraordinarily complicated.
As a final example of where Marshallian general equilibrium theory
gives one a fundamentally different view of economic reality than does
Walrasian general equilibrium, let me consider a specific aspect of economics,
one that has been central to distinguishing different schools of economics: the
theory of distribution. In the Walrasian approach income distribution is
determined by marginal productivity. Assuming a linear homogeneous
production function, one has a complete theory of distribution. This theory of
marginal productivity is so built into our way of thinking that it is often not
questioned. Marshall, however, had serious reservations about it, and
understanding Marshallian general equilibrium explains why. In the
Marshallian general equilibrium approach, marginal productivity theory
influences distribution, but it is in no way a theory of distribution. You can
see Marshall's view where he writes:
This doctrine (of marginal productivity) has sometimes been put
forward as a theory of wages. But there is no valid ground for any
such pretension. The doctrine that the earnings of a worker tend to be
Marshallian General Equilibrium Analysis
19
equal to the net product of his work has by itself no real meaning; since
in order to estimate net product, we have to take for granted all the
expenses of production of the commodity on which he works, other
than his own wages.
But though this objection is valid against a claim that it
contains a theory of wages; it is not valid against a claim that the
doctrine throws into clear sight the action of one of the causes that
govern wages. (Principles p. 519.)
The problem Marshall had with marginal productivity theory is that
institutions have significant effects on distribution, and thus it is simply
wrong to talk about marginal productivity independent of institutions' effect
on income distribution. In game theoretic terms the argument is that to get
an acceptance of institutions, side deals must be made among participants
which place constraints on individuals and change the nature of equilibrium.
Let me give an example. Say you have two types of individuals: big
heads and big arms. Say also that there are three possible production
techniques that are possible. Two of these production techniques require
acquiescence among individuals; these two techniques are equally efficient in
the sense that when all workers are used, 100 units of output is forthcoming
from both techniques. Technique A, however, gives a MP of 3/4 to big arms
and 1/4 to big heads, while Technique B gives a MP of 3/4 to big heads and
1/4 to big arms. Techniques A and B require acceptance from both groups; if
no agreement is reached, Technique C must be used, which gives a MP of 1/2
for both, but which has a total output of only 40.
Marshallian General Equilibrium Analysis
20
Clearly each group will be better off with choosing either Technique A
or Technique B, but neither technique dominates the other. How do they
decide which technique to use? The obvious answer is to make an inviolable
social compact, embodied in an institution, to use one of the two techniques,
but to have, say, big arms receive certain side payments, perhaps 25, from big
heads. (Of course, big arms would want more since no compact is inviolate,
but let me ignore that complication here.)
In Walrasian economics such side payments resulting from prior deals
cannot be considered; there is no history and no institutions. In Marshallian
economics, to have a theory of distribution one requires both a theory of
history and a theory of institutions. In Marshallian economics, to judge any
outcome, it is not enough to look at marginal productivities at a point in time;
production has a social and historical component, and a particular result can
only be interpreted in its historical and social context. Walrasians make the
implicit assumptions that all these complications do not matter—that the
time inconsistency problem is not dealt with by individuals, and that
,somehow, all institutions are simply plopped down upon individuals.
Walrasian marginal productivity distribution theory ignores all that;
Marshallian general equilibrium distribution theory could not, and therefore
is much more complicated.
Conclusion: A Reversal of Samuelson’s Dictum
There is much more to be said about Marshallian general equilibrium,
but I believe that I have said enough to make my point. There is some depth
in Marshall that belies the many negative assessments of his work. My own
Marshallian General Equilibrium Analysis
21
feeling is that there is sufficient depth to warrant a reversal of the negative
assessments discussed above. To make my point clear, let me continue the
Paul Samuelson attack on Marshall with which I started this talk.
Samuelson writes:
I have come to feel that Marshall’s dictum that “it seems
doubtful whether any one spends his time well in reading lengthy
translations of economic doctrines into mathematics, that has not been
done by himself” should be exactly reversed. The laborious literary
working over of essentially simple mathematical concepts such as is
characteristic of much modern economic theory is not only
unrewarding from the standpoint of advancing science, but involves as
well mental gymnastics of a peculiarly depraved type. (Samuelson,
1955. pg 6. )
In the 1940s and 1950s, in certain aspects of economics Samuelson
was, I have no doubt, right. At that time there were many issues to be cleared
up, and his Foundations did clear up numerous issues. But the fact that
there then existed some poor intuitive literary economic analysis should not
condemn all intuitive literary economics, just as the fact that today that there
is some poor mathematical economics should not condemn all mathematical
economics.
What I am arguing is that there is a symbiotic relationship between
intuitive literary economics and formal mathematical economics. Both are
necessary; both can advance our knowledge. Some aspects of good literary
economics of a period become the core of good formal economics of a later
period. But we will only know which aspects when the formal math catches
Marshallian General Equilibrium Analysis
22
up with the intuition. The ideal would be a peaceful coexistence of the two.
But peaceful coexistence does not seem to be a stable equilibrium and instead
the profession seems to experience these cycles when Marshal’s Dictum or
Samuelson’s Dictum predominates. (Consider, for example, the Ricardo-Mill
cycle.) Whether Samuelson’s Dictum or Marshall’s Dictum is relevant
depends on what part of the cycle we are in. The 1930s-50s was a time for
formal mathematical economics to export ideas to intuitive economics. In my
view, the 1990s is a time for the reverse. More and more top economists are
accepting that we have come as far as we can with static Walrasian general
equilibrium.
The new reality of the 1990s is an acceptance that the general
equilibrium system relevant to our economy is formally complex. Because
that is the case, in the 1990s, Samuelson’s condemnation of Marshall needs
to be reversed. Thus, for the 1990s I suggest that the pendulum has swung
and the following reworking of Samuelson's above quotation is relevant.
Specifically: “The laborious mathematical working over of essentially simple
intuitive concepts such as is characteristic of much modern economic theory is
not only unrewarding from the standpoint of advancing science, but involves
mental gymnastics of a peculiarly depraved type.” “The intuitive
ambiguities of Walras’s general equilibrium, and Samuelson’s expansion of it,
have paralyzed the best brains in economics for the last five decades.” It is
only now that the profession is returning to the understanding of economic
issues that Marshall had at the turn of the century
Marshallian General Equilibrium Analysis
23
Bibliography
Colander, David. (ed.) Beyond Micro Foundations: Post Walrasian
Macroeconomics Cambridge University Press, forthcoming.
Edgeworth, F. Y. Paper Relating to Political Economy, Burt Franklin, New
York, 1925.
Humphries and---(citation to come) 1994.
Keynes, John Maynard. Essays and Sketches in Biography, Meridian, New
York. 1956.
Landreth, Harry and David Colander, History of Economic Thought,
Houghton Mifflin. Boston. 1994.
Leijonhufvud, Hicks, Keynes and Marshall, in Hagemann and Hamouda,
(eds)The Legacy of Hicks: His Contributions to Economic Analysis,
Routledge, New York and London. 1995.
Marshall, Alfred. Personal letter to J.B. Clark. 1908. (found in Pigou.
Memorials of Alfred Marshall.)
Marshall, Alfred. Principles of Economics, Macmillan, London, 9th edition,
1920 (1961)
Pigou, A.C. (editor) Memorials to Alfred Marshall. Kelley and Millman, New
York, 1956.
Rosser, Barkley. From Catastrophe to Chaos: A General Theory of Economic
Discontinuities , Kluwer, Boston, 1991.
Marshallian General Equilibrium Analysis
24
Samuelson, Paul. "The Monopolistic Competition Revolution" in Competition
Theory. R. E. Kuenne, (ed.), John Wiley, New York. 1967.
Samuelson, Paul. Foundations of Economic Analysis. Cambridge, Mass.
Harvard University Press, 1955.
Walker, Donald. "The Structure of Walras's Consumer Commodities Model in
the Mature Phase of His Thought" Review Economique. March, 1994.
Walker, Donald. The Adjustment Processes in Walras's Consumer
Commodities Model in the Mature Phase of his Work" Review Economique.
November, 1994.
Walras, Leon. Elements of Pure Economics Irwin Publishers. Homewood,
Illinois. (fourth edition) 1954. (translated by William Jaffe from the 1926
edition.)
Marshallian General Equilibrium Analysis
25
Only now, that the problems with Walrasian general equilibrium are
beginning to be taken seriously, do we see economists working in that
Marshallian tradition even if they don’t know it--work of Brian Arthur,
Clower and Leijonhufvud in macro, and Nate Rosenberg on technology are
the modern inheritors of the Marshallian mantel.
The third point I will argue in this paper is that Marshall's general
equilibrium will replace Walrasian general equilibrium in the next decade.
Thus, the motto of the 21st century will be:
Marshall is for real mathematicians and liberal arts professors; Walras is
for pseudo mathematicians which includes many of the professors at
universities.
Marshallian General Equilibrium Analysis
26
Why a Marshallian General Equilibrium Revival is Inevitable
What makes a Marshallian revival almost inevitable is the
development of new mathematical and computer techniques to deal with
these. Chaos theory, the analysis of complexity, non-linear dynamics
___________ ______ and dynamics vector game theory, and frequency
dependent equilibria give a research program in analytics that can begin to
get at some of these complications. At the same time computers have
developed that make analysis of such problems through simulation possible.
The Walrasian fudge need not be made to mathematically analyze these
issues. Thus I see a new condition emerging in the 21st century between real
mathematicians--who develop the math necessary to deal with the problems
in the full complexity in an intuitively satisfying way (as compared to the
pseudo mathematical types who force their intuition to fit their available
mathematical techniques) and the literature types who can interpret, and
Marshallian general equilibrium policy analysis--the analysis of institutions
Marshallian General Equilibrium
Summary of Paper to be presented at HES meetings
David Colander
In his rather harsh assessment of Alfred Marshall Paul Samuelson
writes that “the ambiguities of Alfred Marshall paralyzed the best brains in
the Anglo-Saxon branch of our profession for three decades.” This paper
argues that Samuelson is right, but interprets the events differently than did
Samuelson. It argues that that paralyzing effect was a good thing—necessary
because the mathematics available at the time were not up to the task of
considering issues of general equilibrium in the complexity that they needed
to be considered. It argues that the misplaced precision of Leon Walras’s
general equilibrium analysis sent the best brains in the profession on a wild
goose chase that reduced, as opposed to increased, our knowledge of economic
phenomena. Only now in the 1990s, with new developments in mathematics
and computers, are we even beginning to get to a level of mathematics where
we might be able to shed some light on general equilibrium issues through
analytic means.
The paper argues that Marshall understood the limitations of the
current math. It considers his Note 21 of Principles where he summarized the
essence of Walrasian general equilibrium theory. It explains why he felt that
was inadequate, and summarizes what I see, given current mathematical
techniques available, to be Marshall’s general equilibrium vision and
contrasts that Marshallian vision with Walras’s.
Marshallian General Equilibrium Analysis
28"