Agenda, Volume 9, Number 2, 2002, pages 121-134

Mathematicians as Great Economists: John

Forbes Nash, Jr.
Matthew J. Ryan

S
ylvia Nasars recent book, A Beautiful Mind, and a film of the same title
about the life of John Nash have piqued public curiosity about this most
interesting of characters. However, an important aspect to Nashs story is
neglected by these sources. This is the tale of the rise and fall, and rising again, of
his ideas, and their ultimately enormous influence on social science. Economic
and political theory are now heavily based upon the rational strategic models of
human behaviour that Nash pioneered. The impact of this new theory on policy is
also apparent. It has altered the conduct of monetary policy and the relationship
between central bankers and elected public representatives; utterly transformed the
work of the ACCC and other regulatory bodies; informed decisions about whether
and how to auction public assets; and revised thinking on the negotiation of
international trade and other agreements.
What follows is a selective and, more importantly, an interpreted history of
why and when the influence of Nashs ideas came to be so strongly felt,
particularly within economics. I hope that the history is accurate; but the
interpretation is of course subjective, and proffered as a stimulus to further debate.
The nub of the debate is, I believe what makes a mathematician a great
economist? As is well known, Nash is a mathematician, and attended but a single
course in economics undergraduate International Trade at Carnegie Tech (now
Carnegie Mellon University). The work by which he is best known to economists
is contained in three papers a mere 30 pages published between 1950 and
1953: Nash (1950b; 1951; 1953). Indeed, his Nobel citation refers primarily to the
second of these, which contains the substance of Nashs PhD thesis. After 1953,
Nashs attention turned to mathematical problems of no obvious interest to
economists.
Why should a mathematician, with such a paucity of direct involvement in
the enterprise of economics research, have been bestowed (albeit amid
unprecedented controversy) with two of economics highest honours: the Nobel
Prize (1994) and fellowship of the Econometric Society (1990)? The case of
Nashs PhD supervisor, Albert Tucker, provides a useful comparison. In 1951, the
very same year in which Nashs Nobel Prize-winning paper appeared, Tucker and
his co-author, Harold Kuhn, published a famous paper on the mathematics of
optimisation. The Kuhn-Tucker Theorem now appears routinely in economics
textbooks, and is known to virtually every graduate student of the subject. But
Kuhn and Tucker are no more regarded as part of the economics Pantheon than are

Matthew Ryan is a Lecturer and ARC Senior Research Associate in the School of
Economics at the Australian National University.
122 Matthew Ryan

the inventers of calculus. Just because economists use mathematics, this does not
make mathematicians famous contributors to economics. Why has John Nash
been singled out?
To understand the answer, we must relate the tale of the turbulent relationship
between mathematics and economics. Let us begin by describing the intellectual
landscape into which Nash first makes his appearance.

Setting the Scene

The year 1948, in which John Nash entered Princeton as a new PhD student, saw
the confluence of two powerful forces that would have a lasting impact on
economic research. The first was the increasing acceptance of mathematical
formalism in economic analysis, culminating in the publication of Paul
Samuelsons Foundations of Economic Analysis in 1947.
The second was the emergence of a new field of mathematics, called game
theory. John von Neumanns and Oskar Morgensterns Theory of Games and
Economic Behavior was the first substantive treatise in this new field. Its first
edition appeared in 1944, and an important second edition the first to include
the famous expected utility model in 1947.
As the title of von Neumanns and Morgensterns book suggests, and its first
chapter puts beyond doubt, the authors were convinced that game theory held
significant promise as a tool for mathematical economic research. The rapid
convergence of these two streams of intellectual development therefore seemed
imminent.

Mathematical economics and the neoclassical agenda

The first major economic treatise to make substantial use of mathematics that is
to say, elementary calculus and algebra was the Recherches sur les Principes
Mathematiques de la Theorie des Richesses (1838) by Antoine Augustin Cournot.
This was, at the time, a rather radical departure from the norm, and the book was
not well received, prompting Cournot to publish a non-mathematical version, the
Principes de la Theorie des Richesses, in 1863. Far from being lauded for
introducing new discipline and rigour into academic discourse on political
economy, the mathematical method was scorned as irrelevant window-dressing, or
worse, as a deluded subversion of social science.
However, by the late 19th century, the tide was beginning to turn. The
impressive work of Jevons, Walras, Edgeworth, Pareto, Marshall and others
slowly won over the nay-sayers. Acceptance of the utility of mathematical
analysis spread amongst the agenda-setters of the discipline. As Irving Fisher
(1898:136) observed in reviewing the 1897 English translation of Cournots
Principes Mathematiques:

Marshalls diagrams and formulae were called dangerous, falsely

accurate, academic playthings. ... To-day few economists can be found
 Mathematicians as Great Economists: John Nash 123

who regard diagrams as useless curiosities or as waste of valuable page

space.

It should be noted that diagrams of this sort are now used to teach first-year
undergraduates! The cogniscenti of 1898 may indeed have embraced the
mathematical method, but the level of technical analysis was rather low.
The next milestone in the progress of mathematical economics was
Samuelsons Foundations, which actively contributed to raising the level of
analysis in economic science. So reads Samuelsons 1970 Nobel Prize citation.
Samuelson (http://www.nobel.se/economics/laureates/1970/samuelson-bio.html)
himself put it more pithily, claiming his mathematics saved economists from
practising mental gymnastics of a peculiarly depraved type. His Foundations
also brought the new mathematical approach to the rank-and-file of graduate
students, academics and other practising economists.
However, it is inevitable that the desire to impose the discipline of
mathematical formalisation on economics should entail a narrowing of the
research agenda. Some topics were better suited than others to expression in
precise mathematical terms.
As an example, suppose that one is interested as economists are in
understanding the functioning of competitive markets. A natural starting place for
the analysis is a bargain struck between two traders. However, pre-War
economists had virtually given up hope of formalising the process by which
bargaining determines the division of surplus; that is, price. This is vividly
expressed in the exasperated words of Edgeworth (1881:46): the division, he
declared, must necessarily be determined by higgling dodges and designing
obstinacy, and other incalculable and often disreputable accidents.
Sadly, the analysis of higgling dodges continues to elude adequate
formalisation. What are we to do? If one cannot predict the outcome of
individual bargains, how are we to understand complex interacting markets with
thousands of traders? Fortunately, the study of markets in the large turns out to
be easier than that of bargaining in the small. By studying dense markets, with
many buyers and many sellers, individual negotiation skills become irrelevant. A
market price asserts itself inexorably upon the process of trading.
In these densely populated markets, traders cease to eyeball each other across
the negotiating table, and instead operate anonymously, consulting a price posted
by the invisible hand of market forces, and submitting their trades to a central
clearing house. This is the paradigm of the so-called Walrasian market, which
quickly came to dominate economic theorising.
All direct human interaction is mired in the vagaries of human psychology
and emotion. Resorting to the idealisation of homo rationalis does little to
improve matters. In an interactive social environment, ones own best course of
action can be determined only once one has formed a belief about how others will
act. Since these others face a symmetric decision problem, an impasse is
reached. Like the bargaining problem, many avenues of obvious economic
enquiry were blocked in this manner. With face-to-face human interaction
124 Matthew Ryan

effectively off the agenda, economics had thus become a rather asocial branch of
social science.

Game theory

Game theory is the mathematical analysis of precisely these sorts of human

interactions. It is the science of strategic decision-making. In the abstract, a
strategic decision may be defined as one whose consequences are not uniquely
determined by the decision-makers own actions plus exogenous random forces
(chance), but are also affected by the deliberate actions of other people. Such
problems are fundamental to any social science.
As its name suggests, the origins of game theory may be found in attempts to
devise winning strategies in popular parlour games, such as chess, poker and
paper, rock, scissors. Important contributions to statistics and probability arose
from the study of games of chance; and so too did these apparently frivolous
musings upon games of strategy lead ultimately to powerful insights into such
weighty matters as military tactics, oligopolistic behaviour, evolutionary biology,
and politics.
Game theory existed as a branch of mathematics before Nash began writing
on the subject, but only just. The French mathematician Emile Borel and the
Hungarian wunderkind John von Neumann had written a few papers on the subject
in the 1920s, and there were some other early contributors of lesser significance.
The key result to emerge from this early literature was von Neumanns famous
Minimax Theorem (1928).
This theorem provided a solution to so-called two-player, zero-sum games.
In such games, player payoffs always sum to zero: the winnings of one player
must be exactly matched by the losses of the other. The Minimax Theorem
suggested an optimal way to play such games.
As a simple example, consider the game of Matching Pennies. The players
are designated One and Two, and each player places a penny on the ground
simultaneously. If both match (that is, both pennies are heads up, or both are
tails up), Player One keeps both pennies; otherwise both pennies go to Player
Two. This is a zero-sum contest, since one player must gain a penny and the other
must lose one. The game is illustrated schematically below. The payoffs to any
given strategy pair is indicated in the corresponding cell of the matrix, with Player
Ones payoff listed first.

PLAYER TWO
Heads Tails
Heads +1,-1 -1,+1
PLAYER ONE
Tails -1,+1 +1,-1

Because each players attempt to maximise her own position in a zero-sum

contest is simultaneously an attempt to inflict maximal damage upon the other, it
 Mathematicians as Great Economists: John Nash 125

is natural to consider the worst-case scenario associated with each strategy (that is,
the scenario in which the rivals strategy inflicts the minimal payoff). For any
given strategy, its worst-case scenario payoff is called its security level. A
maximin strategy is one that maximises the players security level. This is an
obvious defensive strategy in such games.
Interestingly, as von Neumann and others had recognised, a maximin strategy
may require the player to randomise. The maximin strategy for a player in
Matching Pennies, for example, is to toss the penny, rather than place it. This
gives a 0.5 probability of winning no matter what the other player does. Hence,
the tossing strategy guarantees a payoff of zero (in expected value). Any other
strategy, randomised or otherwise, has a security level that is strictly less than
zero. (When players toss their coins, Australian readers may recognise Matching
Pennies as a simple version of Two Up.)
The Minimax Theorem demonstrates that if one player adopts a maximin
strategy, then the other can do no better than to do likewise. Therefore, maximin
strategies are mutual best responses. They are in a relationship of equilibrium
with respect to each other.
It therefore seems compelling that the best advice to give a player in a two-
person, zero-sum contest is to play a maximin strategy. Importantly, while players
in some games may have many different maximin strategies, the Minimax
Theorem implies that it doesnt matter which is chosen. Each can be expected to
perform as well as the other(s). This is sometimes expressed by saying that the
strategies are interchangeable.
In 1944, with the publication of von Neumanns and Oskar Morgensterns
Theory of Games and Economic Behavior (henceforth TGEB), the scattered
contributions to game theory were synthesised and extended. It also set an agenda
that took considerable hold of the discipline for years to come.
For our purposes, two features of this book are noteworthy. First, it contains
essentially no new important game-theoretic results. Its theoretical centrepiece
remains the Minimax Theorem. In particular, the authors analysis of non-zero-
sum games, and games with more than two players, produced a lot of interesting
ideas, but no firm conclusions. A solution for such games remained elusive.
Second, Morgenstern, an economist at Princeton, had perceived the potential
value of game theory to economics, and convinced von Neumann to pitch the
book in this direction. Indeed, TGEB was very favourably reviewed by leading
economic theorists such as Leonid Hurwicz (1945), Jacob Marschak (1946) and
Richard Stone (1948).
Unfortunately, most economic problems are not zero-sum in nature. For
example, the bargaining problem involves the potential for mutual advantage
through trade, and with respect to bargaining, TGEB was unable to improve on the
dispiriting conclusion of Edgeworth. Morgenstern was aware of this imperfect
match between the tools that TGEB was able to provide and the needs of
economists. Nevertheless, in the section on Necessary Limitations of the
Objectives he recalls from the history of the physical sciences that in the study of
problems which were modest as compared with the ultimate aims, methods were
126 Matthew Ryan

developed which could be extended further and further (von Neumann and
Morgenstern (1947:6)).

And so, the stage is set ...

The year 1948 dawns, and Mr Nash goes to Princeton, the birthplace of TGEB.
Economists are increasingly receptive to sophisticated mathematical ideas, and
Morgenstern has alerted them to the potential value of game theory.
By the time Nash has completed his PhD in 1950, he will have provided
precisely the extensions to TGEB that economists sought, including a solution to
the famous bargaining problem. All seems in readiness, then, for economics to
embrace game theory. Instead, the basic ideas of game theory and its powerful
applications to economics were largely ignored for the next twenty-five years.

Two Important Results of John Nash

Let us first examine Nashs two key contributions Nash equilibrium and the
Nash bargaining solution in more detail. The first of these is a contribution to
the theory of non-cooperative games: those in which the players are unable to
write enforceable contracts. The bargaining problem, by contrast, is a cooperative
game, since the bargain may be written down and enforced through an external
mechanism (that is, the courts). Indeed, this important distinction between
cooperative and non-cooperative games is another of Nashs contributions.

Nash equilibrium

Recall that the appeal of the maximin solution to zero-sum games arises from two
features: maximin strategies offer a natural defensive posture; and a situation of
equilibrium results when players adopt such strategies. Nash observed that the
first property does not survive outside the zero-sum context, but showed that the
second alone is sufficient to extend the solution beyond it. A Nash (1950a; 1951)
equilibrium is a collection of strategies one for each player that are mutual
best responses. For zero-sum games, only maximin strategies may be played in a
Nash equilibrium, but Nash proved that equilibrium strategies exist for any non-
cooperative game, including games that are not zero-sum, and those with more
than two players. This is Nashs famous equilibrium existence result.
Nash equilibrium also conveyed a powerful contrarian message to economists
enamoured of the efficiency of Walrasian markets. Consider the following
example, cited in Milnor (1995). Some readers may recognise it as a variation on
the classic Prisoners Dilemma game, which is attributed to Albert Tucker but first
appeared in print in Nash (1951).

A group of 20 people is going to dinner, and each must choose between

a $10 meal and a $20 meal. Each values the expensive meal only $5
more than the cheaper one, so if each pays individually, all diners choose
 Mathematicians as Great Economists: John Nash 127

the $10 meal. But if the bill is split, the meal choice becomes a strategic
problem, and the only Nash equilibrium is for each to choose the $20
meal: any individual who opted for the $10 meal would forego $5 worth
of additional value which they might have secured for only 50 cents.

When the bill is split, each diner obtains some market power over the
price paid by the other diners for their meals. The individually rational exercise
of this market power leads to a socially inefficient outcome. Everyone is made
worse off.
Of course, economists in 1950 were already aware of the efficiency problems
with imperfectly competitive markets, but Nashs equilibrium provided a powerful
unifying principle with which to understand and generalise these phenomena.

The Nash bargaining solution

In bargaining problems, it is assumed that the capacity to sign binding contracts

eliminates efficiency problems. If an enforcement mechanism were available,
Milnors diners would readily have signed contracts committing themselves to the
cheaper meal, whether or not the bill was to be split. In bargaining, the problem is
not to ensure that maximal surplus is achieved, but to determine how it will be
shared.
Abstractly conceived, a two-person bargaining problem consists of a set of
feasible contracts, with associated payoffs to each bargainer. As Edgeworth, von
Neumann and Morgenstern, and others had observed, it is reasonable to suppose
that the chosen contract will satisfy:

Efficiency: No other feasible contract offers both bargainers a higher payoff.

However, this requirement typically leaves a large number of possibilities.

Choice amongst them, it was felt, could only be determined by understanding the
complex psychology of bargaining. Nash (1950b; 1953) demurred. The process
by which the ultimate bargain is negotiated may be complex, he agreed, but
elementary principles of rationality dictate some additional properties of its
outcome.

Symmetry: If the bargaining problem is symmetric (that is, if for every

feasible contract, there exists another which swaps the bargainers payoffs),
then the chosen contract should award each bargainer the same payoff.
Independence of Irrelevant Alternatives: If bargaining reaches a particular
outcome, and we discard some of the unused contracts and ask the parties to
renegotiate over the smaller set of contracts, they should opt for the same
outcome as before.

While these principles are not without their critics, particularly the latter, the
remarkable thing is that they should be sufficient to determine a unique outcome
128 Matthew Ryan

to any bargaining problem. Using an argument of almost magical simplicity, Nash

showed that these three principles can only be satisfied if the bargainers choose
the contract that maximises the product of their payoffs. The bargaining problem
had been brilliantly solved.

Denouement delayed

Precisely why Nashs results failed to ignite the ardour of economists at the time
remains unclear. However, at least two factors appear to have been important.
First, mathematicians held the line of TGEB, and rejected Nashs
equilibrium as an inadequate solution for general non-cooperative games. This
bad press from the mathematics community, and the consequent absence of
mathematicians actively developing Nashs ideas, must certainly have been
discouraging to potential end-users.
Second, economists proved reluctant to expand their horizons beyond the
entrenched neoclassical agenda. Appropriate new questions simply did not get
asked. To the extent that game-theoretic ideas infiltrated economics, they were
the ideas that economists found useful in understanding Walrasian markets.

The normative agenda in game theory

When Nash described his equilibrium existence result to von Neumann, the latter
cut Nashs explanation short with the dismissive words (Nasar, 1998:94): Thats
trivial, you know. Thats just a fixed point theorem.
Other mathematicians working on game theory were similarly unimpressed
by Nashs ideas. This is evident in the influential Games and Decisions by R.
Duncan Luce and Howard Raiffa (1957) a highly successful attempt to make
game theory accessible to a wider audience of social scientists. After reviewing
some of its properties, Luce and Raiffa (p. 104) find Nash equilibrium wanting,
and conclude that a unified theory for all possible non-cooperative non-zero-sum
games does not seem possible.
This judgement sounds almost shocking to modern ears. Nash equilibrium is
now accepted without question as the way to solve games. Why did
mathematicians reject it?
Let us examine von Neumanns charge of triviality first. With minimal
reading between the lines of history, it is entirely plausible that von Neumann was
either aware of the possibility of proving Nashs result, or capable of reaching
such a conclusion within a few short logical steps. The logic of equilibrium was
familiar to him from his zero-sum analysis, as we have already observed.
Moreover, he was also familiar with the basic mathematical tool fixed point
theory that Nash used to extend this equilibrium logic to general non-
cooperative games. Indeed, for a paper on economic growth written in 1932
von Neumanns one other foray into economics von Neumann (1945-6) proved
a fixed point theorem essentially equivalent to the one applied in Nash (1950a), an
 Mathematicians as Great Economists: John Nash 129

early version of Nashs equilibrium existence result1. In his growth paper, von
Neumann (1945-46:5.) even observes that the Minimax Theorem follows as a
direct corollary of this fixed point theorem. Thus, von Neumann was aware of the
logic of equilibrium, acquainted with fixed point theory, and aware of the utility of
the latter in connection with the former. Nashs result was surely within his grasp.
It is not surprising, therefore, that von Neumann saw Nashs result before
Nash had finished his explanation, and saw little novelty in it. One is therefore
lead to the question: Why had von Neumann not already written down the
extension of equilibrium to general non-cooperative games? The answer is
evident in the lukewarm reception to the idea expressed by Luce, Raiffa and others
in the game theory community. Game theorists simply did not think Nash
equilibrium a viable solution to non-zero-sum games. They felt it to be
conceptually flawed.
The reason for this, it seems clear, was the normative orientation of
mathematical game theory. True to its origins, game theory in the 1950s still
sought to advise players on how best to make their strategic decisions. Nash
equilibrium failed to generate unambiguous advice, since many non-zero-sum
games possess a multiplicity of equilibria that are not interchangeable. The best
advice to give one player depends upon the advice being received by the other,
even if both are being advised according to Nashs theory, and so we arrive at yet
another strategic impasse.
Social scientists, of course, are not concerned to advise people how the ought
to behave, but rather to understand how they do behave. They required a positive
theory of games. Nash equilibrium served this descriptive purpose perfectly.
Indeed, even Luce and Raiffa (1957:105) tempered their criticism of Nash with the
observation that:

Even if we were to reject equilibrium as a normative theory for non-

cooperative games ... it may still be that the notion is relevant as a
description of behavior.

What evidence is there that Nash interpreted his own work in a positive
spirit? On the one hand, his 1951 paper reserves the term solution for cases in
which a games Nash equilibria are interchangeable. This suggests a normative
inclination. However, one may also argue that Nash was merely adopted a usage
of the term solution that was consistent with its normative origins in the earlier
work on zero-sum games. In other words, he may have seen his equilibrium as a
satisfying descriptive analysis for all games, but an adequate normative solution
only for games in which the additional requirement of interchangeability is
satisfied.
Indeed, there is compelling evidence elsewhere that Nash did seek to
motivate his equilibrium as a positive concept. Much of this evidence was

1
von Neumann (1945-6) is an English translation of von Neumann's paper, originally
published in German in 1938.
130 Matthew Ryan

obscured from public view because of the omission of a section on Motivation

and Interpretation from the published version of Nashs PhD thesis. In this
section, Nash (cited in van Damme and Weibull (1995, p.19, emphasis added))
says that he wishes to answer the question: (W)hat would be a rational
prediction of the behavior to be expected of rational players in the game in
question?.
He also goes on to propose what he calls a mass action interpretation of
equilibrium. He suggests that equilibria may come to be played through an
evolutionary process by which a population of players are randomly matched and
re-matched in the same contest. Their strategy choices evolve, and will only stop
evolving once an equilibrium has been reached, so that no deviating mutant can
obtain a superior outcome. This interpretation was independently re-invented in
the 1970s, and evolutionary game theory is now a thriving branch of the
discipline.
In the mass action story, the strategies of all players evolve together. There
is no need of interchangeability, since strategies are jointly determined. If
multiple equilibria exist, this only reflects the inability of Nashs logic to generate
unique predictions. The players themselves know perfectly well which
equilibrium they are playing; only we, as outside observers, are in doubt.
One also sees a methodological unity between Nash equilibrium, under the
mass action interpretation, and his bargaining solution. In both cases, Nash uses
elementary principles to narrow down the plausible outcomes of an underlying
process of considerable complexity. This methodology is now unquestioningly
accepted and powerfully applied by economists and other social scientists.
However, it was certainly not embraced immediately.

The apotheosis of Walrasian market theory

Why it was not so embraced remains an intriguing unsolved mystery. Certainly,

leading economists had more than enough mathematical facility to read and
interpret Nash for themselves. Indeed, Nashs papers were read, and their ideas
syphoned off by economists. However, in one of the great ironies of our story, the
ideas were applied to understand the non-strategic behaviour of Walrasian
markets, not strategic interaction.
Kenneth Arrow and Gerard Debreu were directly inspired by Nashs fixed-
point arguments (Debreu, 19522; Arrow and Debreu, 1954). By adapting these
arguments, they were able to show that, under fairly weak assumptions, there will
always exist a set of prices that balance supply and demand simultaneously on a
system of Walrasian markets. This is known as a Walrasian equilibrium.
Even the Minimax Theorem exerted an indirect influence on economic theory
at this time. Tjalling Koopmans activity analysis, for which he won a Nobel
Prize in 1975, is an application of duality theory in linear programming, which is

2
Interestingly, Debreu's paper was communicated to the National Academy by John von
Neumann.
 Mathematicians as Great Economists: John Nash 131

itself a corollary of the Minimax Theorem. Economists were not only out of
practice at asking the sorts of questions that game-theoretic tools were useful in
answering, the usefulness of the mathematics of games to the neoclassical agenda
contributed to the temporary demise of game theory itself within economics.
Indeed, this traditional agenda acquired a new momentum that sustained it into the
1970s.

The Game-theoretic Revolution

As Aumann (1985:43) states:

The Nash equilibrium is the embodiment of the idea that economic

agents are rational; that they simultaneously act to maximise their utility.
If there is any idea that can be considered the driving force of economic
theory, that is it.

By the mid-1970s economics was ready for a new paradigm. The heyday of
research into Walrasian markets was over, and graduate students were looking for
new topics on which to write their dissertations. A game-theoretic revolution in
economics was about to take place.
Of course, economists had never entirely abandoned game theory, or Nashs
ideas. Both had persisted at the fringes of the discipline, sustained by the work of
theorists such as Robert Aumann, John Harsanyi, Lloyd Shapley and Martin
Shubik. It is also true, as I have attempted to convey, that the ideas of Nash were
so compellingly useful to economists, that it was only ever a question of when
they would be adopted, not if. Nevertheless, an explanation is required for their
suddenly being thrust centre stage at this particular juncture.
Definitive answers are elusive, but some influential factors are evident. The
first is the aforementioned loss of steam from the research program into Walrasian
markets. Second, political scientists have suggested that game theory may have
re-entered economics via politics, where it had spawned the important new
positive political theory school, lead by William Riker, in the 1960s (Amadae
and Bueno de Mesquita, 1999). Indeed, the use of game theory in political science
has seen a growing together of economics and politics. A new political economy,
formalised along game-theoretic lines, has emerged. Theories of voting, lobbying
and other political phenomena are now routinely taught in economics
programmes.
Finally, there was the important catalyst provided by Nashs two Nobel co-
laureates: Reinhard Selten and John Harsanyi. Selten pioneered the literature on
so-called refinements of Nash equilibrium. By tightening the predictive power of
equilibrium, the refined solutions greatly enhanced the attractiveness of game
theoretic tools for applied work. Harsanyi enhanced Nash equilibrium in the
opposite direction, by expanding its scope of application. Nash, like the game
theorists before him, had assumed that all players knew the payoff consequences
for all rivals of any given outcome of the game. This complete information
assumption is frequently unrealistic. For example, in the context of an auction, it
132 Matthew Ryan

requires that each bidder know every other bidders valuation of the object for
sale. Harsanyi indicated how the complete information assumption might be
dispensed with using an elaboration of the Nash logic known as a Bayesian Nash
equilibrium. A vast new array of interesting problems could now be successfully
tackled.
With respect to the contributions of Selten and Harsanyi, however, let me not
give the false impression that Nashs ideas were inadequate in their own right. It
was not that Seltens and Harsanyis embellishments were necessary to make Nash
equilibrium serviceable. The new ideas re-ignited interest in game theory, but
they did not render Nash equilibrium obsolete. On the contrary, as Aumann
(1985:48) observes: Nash equilibrium is without doubt the most successful
that is, widely used and applied solution concept in game theory. Indeed, most
applied work on problems of complete information employ nothing more than
simple versions of the Selten refinement that combine Nash equilibrium with the
even older idea of backward induction. This combination is already evident in
Nashs own work: see Nash (1953).

Game Theory and the Modern Agenda in Economics

To conclude, let me return to the question: what distinguishes the contributions of

Nash from those of his supervisor, Tucker? Unlike any other single idea, Nash
equilibrium has transformed the agenda in economics, and in social science more
broadly. While the Kuhn-Tucker Theorem allowed economists to do what they
had already been doing more effectively, Nashs equilibrium opened entirely new
avenues of research, and substantially altered both the scope and the focus of
economics. In particular, the credit for this must go to Nash himself, rather than to
the collective enterprise of game theory. Nash was the first to see clearly how
game theory might be used in a positive mode, and is one of its most
accomplished exponents. Reading the game theory literature of the 1950s, it is
striking how Nash stands out as the most modern, the most in tune with
contemporary thinking about strategic analysis in social science. This is indeed no
accident.
It scarcely needs to be emphasised that game theory is now ubiquitous within
economic research, from industrial organisation, whose existence one could
scarcely imagine without it, to strategic trade theory, law and economics, and even
macroeconomics. Indeed, game theory itself now exists as a field within
economics, entirely separate from the kindred branch of mathematics.
The analysis of markets now focuses heavily on the visible hand of the
strangely mis-named imperfect competition, rather than the invisible hand of
the shadowy Walrasian mechanism. To the extent that Walrasian markets are the
object of continued theoretical interest, this is directed at providing strategic
foundations in terms of individual bargaining. Nashs bargaining solution is the
key ingredient in this endeavour.
Most importantly, perhaps, is the way in which game theory has transformed
the dismal science into a truly social science, and reconnected it with its
 Mathematicians as Great Economists: John Nash 133

brethren: political theory, sociology, law, anthropology, and even biology. Social
and economic institutions are now viewed as mechanisms that determine the rules
of the games that people play. Normative questions of institutional design (so-
called mechanism design or implementation theory) may be addressed in an
intellectually satisfying manner. Dialogue on these issues of common interest is
now possible between economists, political scientists and other social
philosophers. This, ultimately, may be the most significant legacy of Nash
(Myerson, 1996:273-4):

Methodological limitations no longer deter us from recognising the

essential interconnections between economic, social, and political
institutions in economic development. ... Today, the original scope of
political economy has been restored to economic theorists, because of
the general methodology that John Nash introduced.

References
Amadae, S. and B. Bueno de Mesquita (1999), The Rochester School: The Origins of
Positive Political Economy, Annual Review of Political Science 2:269-295.

Arrow, K. and G. Debreu (1954), Existence of Equilibrium for a Competitive Economy,

Econometrica 22:265-290.

Aumann, R. (1985), What is Game Theory Trying to Accomplish?, pp. 28-76 in K.

Arrow and S. Honkapohja (eds), Frontiers of Economics, Basil Blackwell, Oxford.

Debreu, G. (1952), A Social Equilibrium Existence Theorem, Proceedings of the

National Academy of Sciences of the U.S.A., 38:886-893.

Cournot, A. (1987), Researches into the Mathematical Principles of the Theory of Wealth,
MacMillan, New York. (English translation by N. Bacon of Cournots original French
publication of 1838).

van Damme, E. and J. Weibull (1995), Equilibrium in Strategic Interaction: The

Contributions of John C.Harsanyi, John F. Nash and Reinhard Selten, Scandinavian
Journal of Economics 97(1):15-40.

Edgeworth, F. (1881), Mathematical Psychics, C. Kegan Paul and Co. London.

Fisher, I. (1898), Cournot and Mathemtical Economics, Quarterly Journal of Economics

12(2):119-138.

Hurwicz, L. (1945), The Theory of Economic Behavior, American Economic Review

35(5):909-925.

Luce, R. and H. Raiffa (1957), Games and Decisions, John Wiley and Sons, New York.

Marschak, J. (1946), Neumanns and Morgensterns New Approach to Static Economics,

Journal of Political Economy 54(2):97-115.
134 Matthew Ryan

Milnor, J. (1995), A Nobel Prize for John Nash, The Mathematical Intelligencer
17(3):11-17.

Myerson, R. (1996), John Nashs Contributions to Economics, Games and Economic

Behavior 14:287-295.

Nasar, S. (1998), A Beautiful Mind, Faber and Faber, London

Nash, Jr., J. (1950a), Equilibrium Points in N-Person Games, Proceedings of the

National Academy of Sciences of the U.S.A. 36:48-49.

Nash, Jr., J. (1950b), The Bargaining Problem, Econometrica 18:155-162.

Nash, Jr., J. (1951), Non-Cooperative Games, Annals of Mathematics 54(2):286-295.

Nash, Jr., J. (1953), Two-Person Cooperative Games, Econometrica 21:128-140.

von Neumann, J. (1945-6), A Model of General Economic Equilibrium, Review of

Economic Studies 13:1-9.

von Neumann, J. and O. Morgenstern (1947), The Theory of Games and Economic
Behavior, 2nd edition, Princeton University Press, Princeton, NJ.

Samuelson, P. (1947), Foundations of Economic Analysis, Harvard University Press,

Cambridge, MA.

Stone, R. (1948), The Theory of Games, Economic Journal 58:185-201.

This paper is based on a Toyota/ANU Public Lecture, 21 March 2002. My thanks

to the School of Mathematical Sciences at the ANU for inviting me to write and
speak on this topic. Thanks also to Gerry Mackie, Guillaume Rocheteau, two
anonymous referees, and seminar participants in the School of Economics at the
ANU, for helpful comments on an earlier draft.