Professional Documents
Culture Documents
The Mathematization of Economic Theory
The Mathematization of Economic Theory
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review
By GERARD DEBREU*
Sciences of the United States, 34 are ES been an inaccessible ideal toward which
Fellows. From 1969 to 1990, 30 economics economic theory sometimes strove. During
Nobel awards were made, and 25 of the that period, this striving became a powerful
laureates are, or were, ES Fellows. Since it stimulus in the mathematization of eco-
was first presented to Paul Samuelson in nomic theory.
1947, the John Bates Clark medal of the The great theories of physics cover an
American Economic Association has been immense range of phenomena with a
given to 21 economists, of whom 20 are ES supreme economy of expression. Of this,
Fellows; and of the 26 living past presidents James Clerk Maxwell (1865) had given a
of our Association, 13 are ES Fellows. notable example, as he described the elec-
One may wish that those counts had not tromagnetic field by means of eight equa-
been made. One may argue about points of tions at the time when mathematical eco-
their interpretation. But they belong in our nomics was born and came of age in the
common knowledge, and their thrust is un- middle of the 19th century. This extreme
equivocal. They indicate how extensive the conciseness is made possible by the privi-
mathematization of economics and how leged relationship that developed over sev-
deep the accompanying change of our field eral centuries between physics and mathe-
were over the past five decades. matics. In turn, the former presented the
The perception of the depth of that latter with open problems, or found to ques-
change is reinforced by a comparison of the tions raised by physical theory ready-made
levels of mathematics required in 1940 and answers discovered by mathematicians in
in 1990 to follow the development of eco- their abstract universe. Sometimes the
nomic theory in every direction it was tak- causal linkage of research done in each one
ing. Fifty years ago, basic undergraduate of the two fields could not easily be unrav-
preparation in mathematics was almost al- eled; and, on occasion, the same scientist
ways sufficient. Today, graduate training in made inextricably intertwined contributions
mathematics is necessary. If, instead of be- to both disciplines.
ing a follower, one wishes to be an active The benefits of that special relationship
participant in that development along its were large for both fields; but physics did
most technical avenues, a high degree of not completely surrender to the embrace of
mathematical professionalism is called for. mathematics and to its inherent compulsion
Several faculty members of the 13 depart- toward logical rigor. The experimental re-
ments of economics mentioned previously sults and the factual observations that are at
were actually identified as mathematicians the basis of physics, and which provide a
by their doctorates; four of them served as constant check on its theoretical construc-
chairmen of those departments during the tions, occasionally led its bold reasonings to
past 25 years. If still sharper focus brings violate knowingly the canons of mathemati-
out the intellectual leaders of that develop- cal deduction.
ment, prominent among them is John von In these directions, economic theory could
Neumann, one of the foremost mathemati- not follow the role model offered by physi-
cians of his generation. cal theory. Next to the most sumptuous
In that development process, mathemati- scientific tool of physics, the Superconduct-
cal economics was continuously redefined as ing Super Collider whose construction cost
new territories were included within its out- is estimated to be on the order of $1010
ward-moving frontier and as topics that were (David P. Hamilton, 1990; see also Science,
once at that frontier became standard parts 5 October 1990), the experiments of eco-
of the graduate, if not of the undergradu- nomics look excessively frugal. Being denied
ate, economic-theory curriculum. a sufficiently secure experimental base, eco-
nomic theory has to adhere to the rules of
II logical discourse and must renounce the
facility of internal inconsistency. A deduc-
Before the contemporary period of the tive structure that tolerates a contradiction
past five decades, theoretical physics had does so under the penalty of being useless,
of an interval of real numbers has long been bibliographical key (John Eatwell et al.,
familiar in descriptions of economic data. It 1987-1989). Walras's consumers and pro-
became familiar in economic theory as well ducers have been freed from many of their
after Robert J. Aumann (1964) showed that, constraining characteristics; Edgeworth's
in a pure exchange economy composed of universe of two consumers and two com-
insignificant agents, the formation of desta- modities has been vastly expanded.
bilizing coalitions is prevented if and only if Mathematics also dictates the imperative
all those agents base their decisions on a of simplicity. It relentlessly searches for
price system. short transparent proofs and for the theo-
The concept of a convex set (i.e., a set retical frameworks in which they will be
containing the segment connecting any two inserted. Participating in that pursuit, eco-
of its points) had repeatedly been placed at nomic theory was sometimes drawn by drives
the center of economic theory before 1964. toward greater generality and toward greater
It appeared in a new light with the intro- simplicity in the same direction, rather than
duction of integration theory in the study of in opposite directions. Cohort after cohort,
economic competition: if one associates with students of consumer theory have learned
every agent of an economy an arbitrary set about the concept of decreasing marginal
in the commodity space and if one averages rate of substitution for two commodities on
those individual sets over a collection of an indifference curve and about its exten-
insignificant agents, then the resulting set is sion to the multicommodity case. Notably
necessarily convex.2 But explanations of the more general, and notably simpler, is the
three functions of prices taken as examples concept of convexity of the set of points
can be made to rest on the convexity of sets preferred to a given point in the commodity
derived by that averaging process. Convexity space. Welfare economics presents another
in the commodity space obtained by aggre- instance. One of its main theorems formu-
gation over a collection of insignificant lates precisely the principle enunciated by
agents is an insight that economic theory Adam Smith (1776). If all the agents of an
owes in its revealing clarity to integration economy are in equilibrium relative to a
theory. price system, then they utilize their collec-
An economist who experiences such an tive resources optimally. The proof of that
insight belongs to the group of applied theorem (Kenneth J. Arrow, 1951) has be-
mathematicians, whose values he espouses. come so simple that it can be given without
Mathematics provides him with a language mathematical symbols. It is, at the same
and a method that permit an effective study time, of utmost generality; in relating two
of economic systems of forbidding complex- basic concepts of economic theory to each
ity; but it is a demanding master. It cease- other, it uses no assumption.
lessly asks for weaker assumptions, for In its attempts to attain its many objec-
stronger conclusions, for greater generality. tives, economic theory was helped by greater
In taking a mathematical form, economic abstraction. Preference theory supplies an
theory is driven to submit to those demands. example again. Significant research efforts
The gains in generality that it has achieved were expended on solutions of the integra-
as a result, in little more than a century, bility problem. That problem can be by-
stand out when the first formulations of passed altogether, and greater simplicity can
the theories of general equilibrium (Leon be achieved by moving from the commodity
Walras, 1874-1877) and of the core of an space to the more abstract space of the
economy (Francis Y. Edgeworth, 1881 pp. pairs of its points. In this space, whose
34-8) are placed side by side with the re- dimension is twice the number of commodi-
cent treatments of those subjects to which ties, the pairs of commodity points indif-
The New Palgrave is an introduction and a ferent to each other are now assumed to
form a smooth (hyper)surface. As another
20n this direct consequence of a theorem of A. A. instance of the generality permitted by ab-
Lyapunov, see Karl Vind (1964). straction, consider the notion of a commod-
ity, which can be treated as a primitive were speaking, and in spite of the wide
concept, with an unspecified interpretation, diffusion of their critiques, neither Leontief
in an axiomatic economic theory. A newly nor Gordon altered the course of the devel-
discovered interpretation can then increase opment they were assessing. In the past two
considerably the range of applicability of decades, economic theory has been carried
the theory without requiring any change in away further by a seemingly irresistible cur-
its structure. Thus, by making the transfer rent that can be explained only partly by the
of a good or service between two agents intellectual successes of its mathematiza-
contingent on the state of the world that tion.
will obtain, Arrow (1953) made possible the Essential to an attempt at a fuller expla-
immediate extension of the economic theory nation are the values imprinted on an
of certainty to an economic theory of uncer- economist by his study of mathematics.
tainty by a simple reinterpretation of the When a theorist who has been so typed
concept of a commodity. The theory of fi- judges his scholarly work, those values do
nancial markets has been influenced by that not play a silent role; they may play a deci-
view of uncertainty, and their practice has sive role. The very choice of the questions
not been unaffected. Finally, take the prob- to which he tries to find answers is influ-
lem of existence of a general equilibrium, enced by his mathematical background.
once considered to be one of the most ab- Thus, the danger is ever present that the
stract questions of economic theory. The part of economics will become secondary, if
solutions that were proposed in the early not marginal, in that judgment.
1950's paved the way for the algorithms for The reward system of our profession rein-
the computation of equilibria of Herbert E. forces the effects of that autocriticism. De-
Scarf (1973) and for several of the develop- cisions that shape the career of an economic
ments of applied general equilibrium analy- theorist are made by his peers. Whether
sis (Scarf and John B. Shoven, 1984). In this they are referees of a journal or of a re-
case, abstraction in economic theory led to search organization, members of an ap-
the study of fundamental problems of great pointment or of a promotion committee,
generality, but also to a broad range of when they sit as judges in any capacity, their
applications. verdicts will not be independent of their
own values. An economist who appears in
III their court rarely ignores his perception of
those values. If he believes that they rate
The list of advances that the mathemati- mathematical sophistication highly, and if
zation of economic theory helped or permit- he can prove that he is one of the sophisti-
ted is already long; and in one aspect it may cates, the applause that he expects to re-
appear lengthy. Ceteris paribus, one cannot ceive will condition his performance.
prefer less to more rigor, lesser to greater The same effects are also amplified by the
generality, or complexity to simplicity; but relentless pressure to publish exerted by his
other things are not equal, and in the esti- environment. There are indeed instances of
mate of many members of our Association extreme restraint in scientific publication,
the cost of that mathematization sometimes and some of them have become legend. The
outweighs its benefit. Two of its presidential mathematical papers of Bernhard Riemann
addresses notably confronted that difficult (1826-1866) take 506 pages in the volume
analysis and stressed the price that eco- that collected them (Riemann, 1876). The
nomics paid for its increased use of mathe- molecular structure of DNA was announced
matics. Wassily Leontief's (1971) observa- by James Watson and Francis Crick (1953)
tions were factual, and Robert A. Gordon's in a one-page article. But it is easier to
(1976) comments relevant when they were explain those examples away than to follow
made in 1970 and in 1975. They still are them. The environment of a scholar de-
today, for, in spite of their authorities, en- mands papers, and the temptation to supply
hanced by the platform from which they them without restraint may become over-
powering to an economic theorist who has ments of the phase that economic theory
developed proficiency in his research style. underwent and the effectiveness of attempts
The precocious development of that profi- to alter the course of its evolution will gain
ciency is a comparative advantage that a from a detailed analysis of the processes
mathematical approach bestows on him. that led to its present state.
The spread of mathematized economic
theory was helped even by its esoteric char- REFERENCES
acter. Since its messages cannot be deci-
phered by economists who do not have the Arrow, Kenneth J., "An Extension of the Ba
proper key, their evaluation is entrusted to sic Theorems of Classical Welfare Eco-
those who have access to the code. But nomics," in J. Neyman, ed., Proceedings
acceptance of their technical expertise also of the Second Berkeley Symposium on
implies acceptance of their values. Our pro- Mathematical Statistics and Probability,
fession may take pride in its exceptional Berkeley: University of California Press,
intellectual diversity, one of whose clearest 1951, 507-32.
symbols is an Ely lecture given by an eco- , "Le R6le des Valeurs Boursieres
nomic historian at a session chaired by a pour la Repartition la Meilleure des
mathematical economist. Yet that diversity Risques," in Econometrie, Paris: Centre
is strained by the increasing impenetrability National de la Recherche Scientifique,
to the overwhelming majority of our Associ- 1953, 41-8.
ation of the work done by its most mathe- Aumann, Robert J., "Markets with a Contin-
matical members. uum of Traders," Econometrica, Jan-
uary-April 1964, 32, 39-50.
IV Brown, Donald and Robinson, Abraham, "A
Limit Theorem on the Cores of Large
The bond that ties economists together in Standard Exchange Economies," Proceed-
their study of a common subject has not ings of the National Academy of Sciences
been tested only by differences in method- USA, May 1972, 69, 1258-60.
ologies. It has also been tried by differences Eatwell, John, Milgate, Murray and Newman,
in ideologies. In their endeavors to make Peter, eds., General Equilibrium, New
their field into a science, economists must York: Macmillan, 1987-1989.
renounce a favorite mode of thinking- Edgeworth, Francis Y., Mathematical Psychics,
wishful thinking; they must be impartial London: Paul Kegan, 1881.
spectators of a play in which they are the Gordon, Robert A., "Rigor and Relevance in
actors. While they attempt to keep that a Changing Institutional Setting," Ameri-
inhuman stance, they are pressed to give can Economic Review, March 1976, 66,
immediate answers to societal questions of 1-14.
immense complexity and thereby to aban- Hamilton, David P., "The SSC Takes On a
don the exacting slowness of the step-by-step Life of Its Own," Science, 17 August 1990,
scientific approach. Divisions according to 249, 731-2.
methodologies and ideologies, criticism from Jones, Lyle V., Lindzey, Gardner and Coggeshall,
outside and from inside, and intellectual Porter E., eds., An Assessment of
fashions that sweep our discipline make each Research-Doctorate Programs in the United
one of its steady developments remarkable. States: Social and Behavioral Sciences,
The mathematization of economic theory Washington DC: National Academy of
was one of them for a century and a half. Sciences Press, 1982.
During the past five decades it became one Leontief, Wassily, "Theoretical Assumptions
of the prime movers in the transformation and Non-observed Facts," American Eco-
of our field. The extent of that mathemati- nomic Review, March 1971, 61, 1-7.
zation has given rise to discordant assess- Maxwell, James C., "A Dynamical Theory of
ments of its effects and to attempts to the Electromagnetic Field," Philosophical
change its heading. The quality of assess- Transactions of the Royal Society of Lon-