LIBRO. Uskali Mäki (2009), The Methodology of Positive Economics

This page intentionally left blank
The Methodology of Positive Economics
Milton Friedmans 1953 essay The methodology of positive economics remains the most cited, influential, and controversial piece of methodological writing in twentieth-century economics. Since its appearance,
the essay has shaped the image of economics as a scientific discipline,
both within and outside academia. At the same time, there has been an
ongoing controversy over the proper interpretation and normative evaluation of the essay. Perceptions have been sharply divided, with some
viewing economics as a scientific success thanks to its adherence to
Friedmans principles, others taking it as a failure for the same reason.
In this book, a team of world-renowned experts in the methodology of
economics cast new light on Friedmans methodological arguments and
practices from a variety of perspectives. It provides the twenty-firstcentury reader with an invaluable assessment of the impact and contemporary significance of Friedmans seminal work.
Uskali Mki is Academy Professor at the Academy of Finland. He has
written extensively on the philosophy and methodology of economics
and is editor of a number of books, including The Economic World View
(Cambridge University Press, 2001), Fact and Fiction in Economics
(Cambridge University Press, 2002), and Handbook of the Philosophy of
Economics (2009).
The Methodology of Positive

Economics
Reflections on the Milton Friedman legacy
Edited by
Uskali Mki
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, So Paulo

Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521867016
Cambridge University Press 2009
This publication is in copyright. Subject to statutory exception and to the
provision of relevant collective licensing agreements, no reproduction of any part
may take place without the written permission of Cambridge University Press.
First published in print format 2009
ISBN-13
978-0-511-53988-6
eBook (EBL)
ISBN-13
978-0-521-86701-6
hardback
ISBN-13
978-0-521-68686-0
paperback
Cambridge University Press has no responsibility for the persistence or accuracy

of urls for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
For INEM
Contents
List of figures
List of tables
Contributors
Preface
page ix
x
xi
xvii
Part 1 The classical essay in twentieth-century economic

methodology
The methodology of positive economics (1953)

milton friedman
Part 2 Reading and writing a classic
45
1 Reading the methodological essay in twentieth-century

economics: map of multiple perspectives
uskali mki
47
2 Early drafts of Friedmans methodology essay

j. daniel hammond
68
Unrealistic assumptions and unnecessary confusions:

rereading and rewriting F53 as a realist statement
uskali mki
Part 3 Models, assumptions, predictions, evidence

4
5
90
117
The influence of Friedmans methodological essay

th oma s ma y er
119
Did Milton Friedmans positive methodology license

the formalist revolution?
d. wade hands
143
vii
viii
Contents
6
7
Part 4
8
9
10
11
12
13
Part 5
14
15
Appraisal of evidence in economic methodology

melvin w. reder
165
The politics of positivism: disinterested predictions from

interested agents
d a v id t e ir a s e r r an o a n d j e s s Z A M O R A B O N I L L A
189
Theoretical context: firm, money, expected utility,

Walras and Marshall
215
Friedmans 1953 essay and the marginalist controversy

r o g er e . b a c k h o u s e
217
Friedman (1953) and the theory of the firm

o l iv e r e . w i ll ia m s o n
241
Friedmans selection argument revisited

jack vromen
257
Expected utility and Friedmans risky methodology

c h r i s s ta r m e r
285
Milton Friedmans stance: the methodology

of causal realism
kevin d. hoover
303
On the right side for the wrong reason: Friedman on the

MarshallWalras divide
michel de vroey
321
Concluding perspectives
347
The debate over F53 after fifty years

mark blaug
349
Final word
milton friedman
355
Index
356
Figures
9.1 The sciences of choice and contract

page 251
11.1 Action/state payoff matrix
293
12.1 The reception of Friedmans positivism and the fate of causal
language in econometrics
318
ix
Tables
4.1 Effect of changes in economics on the acceptance

of F53
page 125
4.2 The Nobel Prize lectures and Friedmans criteria
133
12.1 Uses of causal language in Friedman and Schwartzs
Monetary History of the United States, 18671960
307
12.2 Examples of synonyms for causal language from Friedman
and Schwartzs Monetary History of the United States,
18671960
308
Contributors
r o g e r e . b a c k h o us e is Professor of the History and Philosophy of

Economics at the University of Birmingham. He is author of The
Penguin History of Economics / The Ordinary Business of Life (2002) and
many books and articles on the history of economics in the twentieth
century and on economic methodology. He has edited several volumes
on economic methodology and is a former editor of the Journal of
Economic Methodology. He is past Chair of the International Network
for Economic Method.
m a r k b l a u g is Professor Emeritus of the University of London and
Buckingham University, UK, and Visiting Professor of Economics at
the University of Amsterdam and Erasmus University of Rotterdam,
the Netherlands. He is a Fellow of the British Academy and a Foreign
Honorary Member of the Royal Netherlands Academy of Arts and
Sciences. His principal fields of interest are economic methodology
and the history of economic thought. His publications include
Introduction to the Economics of Education (1970), The Methodology of
Economics (1980), Great Economists since Keynes (1985), Economic
History and the History of Economics (1987), The Economics of Education
and the Education of an Economist (1987), Economic Theories. True or
False? (1990), Not Only an Economist. Recent Essays (1997), Whos Who
in Economics, ed. (1999).
k ev i n d . h o o v e r is Professor of Economics and Philosophy at Duke
University. He received his D.Phil from Oxford University. He was
previously a Research Fellow at Nuffield College, Oxford, and a
lecturer at Balliol College and Lady Margaret Hall, Oxford, as well as
a research associate and visiting economist at the Federal Reserve Bank
of San Francisco. He is the author of Causality in Macroeconomics and The
Methodology of Empirical Macroeconomics (both Cambridge University
Press, 2001), and of numerous articles in macroeconomics, monetary
economics, economic methodology, and the philosophy of science. He is
xi
xii
List of contributors
also the editor of Macroeconometrics: Developments, Tensions, and

Prospects (1995) and Real Business Cycle: A Reader (1998). Hoover is
past President of the History of Economics Society, past Chairman of
the International Network for Economic Methodology and a former
editor of the Journal of Economic Methodology. Hoover is best known
for his work on causality and on the new classical macroeconomics.
The latter work began with Two types of monetarism (Journal of
Economic Literature, 1984), which contained a careful examination of
Milton Friedmans economics and methodology, trying to locate why
Friedman disassociated himself from his apparent followers such as
Lucas and Sargent. These themes were further developed in The New
Classical Macroeconomics (1988).
j. da ni e l ha mmo nd is Professor of Economics at Wake Forest
University, Winston-Salem, NC. He received his PhD from the
University of Virginia, and has published in monetary economics,
economic methodology, and history of economics. He has written
extensively on the history of University of Chicago economics,
including Theory and Measurement: Causality Issues in Milton
Friedmans Monetary Economics (Cambridge University Press, 1996),
Making Chicago Price Theory: Friedman-Stigler Correspondence 1945
1957 (editor with Claire H. Hammond, Routledge, 2006), and An
interview with Milton Friedman on methodology, Research in the
History of Economic Thought and Methodology (1992). Hammond has
served as an elected member of the executive committee of the History
of Economics Society, and in 20012 as the Societys president.
d . w a d e h a n d s is Professor of Economics at the University of Puget
Sound in Tacoma, WA. He has published widely on various topics in
the history of economic thought and economic methodology. Some of
his published work is about the history and methodology of the Chicago
school, of which Milton Friedman is a member. His most recent book is
Reflection Without Rules (Cambridge University Press, 2001). He is also
one of the editors (along with John Davis and Uskali Mki) of The
Handbook of Economic Methodology (1998). His textbook Introductory
Mathematical Economics is in its second edition (2003). He is an editor
of the Journal of Economic Methodology.
u s ka l i m ki is Academy Professor at the Academy of Finland
and former Professor at the Erasmus Institute for Philosophy and
Economics. He has published on the assumptions issue, realism and
realisticness, idealizations and models, causation, explanation, the
rhetoric, sociology, and economics of economics, and the foundations
xiii
of New Institutional and Austrian economics; in journals such as the

Journal of Economic Literature, Philosophy of the Social Sciences, Studies in
the History and Philosophy of Science, Perspectives on Science, Erkenntnis,
Kyklos, Journal of Economic Methodology, History of Political Economy,
Economics and Philosophy, Cambridge Journal of Economics. He is editor
of The Economic World View: Studies in the Ontology of Economics
(Cambridge University Press, 2001); Fact and Fiction in Economics:
Models, Realism, and Social Construction (Cambridge University Press,
2002); and a co-editor of The Handbook of Economic Methodology
(1998); Economics and Methodology. Crossing Boundaries (1998);
Rationality, Institutions and Economic Methodology (1993). He is Chair
of the International Network for Economic Method and a former editor
of the Journal of Economic Methodology.
t h o m a s m a ye r was born in Vienna in 1927, emigrated to England in
1938 and to the United States in 1944. He received his BA from
Queens College, CUNY, in 1948 and his PhD from Columbia
University in 1953. He taught at West Virginia University, University
of Notre Dame, Michigan State University, University of California
at Berkeley, and for most of his professional life as professor at the
University of California at Davis from which he retired in 1992. His
initial interests were macroeconomics, monetary theory, and policy,
where he could best be classified either as an eclectic or a moderate
monetarist. Much of his work has addressed Friedmans monetarist
thinking. Shortly before retirement he shifted mainly to methodology
where he advocated a de-emphasis on formalist economics. During
this period, he has published on Friedmans methodology. He is the
author of Permanent Income, Wealth and Consumption; The Structure of
Monetarism (with others); Money, Banking and the Economy (with
J. Duesenberry and R. Aliber); Monetarism and Monetary Policy; Truth
vs. Precision in Economics; Doing Economic Research; Monetary Policy; and
The Great Inflation in the United States. He lives in Berkeley, California,
and is a citizen of the USA.
m e l v i n w . r ed e r is Gladys G. and Isidore Brown Professor of Urban
and Labor Economics Emeritus at the Graduate School of Business at
the University of Chicago; previously Professor of Economics at
Stanford University. Book publications include Economics: the Culture
of a Controversial Science (1999), Rationality in Psychology and Economics
(coeditor) (1986), Labor in a Growing Economy (1957), Studies in the
Theory of Welfare Economics (1947).
xiv
c h r is s t ar m e r is Professor of Experimental Economics at the

University of Nottingham, UK. Starmer was awarded a PhD for an
experimental investigation of decision under risk in 1992 from the
University of East Anglia (UEA). He worked as Lecturer then Senior
Lecturer at UEA and was visiting Associate Professor at Caltech before
moving to Nottingham in 2000. His main research interests have
focused on experimental economics, theories of decision making and
economic methodology. This combination makes him the best expert
on Friedmans work on expected utility and on methodology. Starmer
has published articles on these topics in American Economic Review,
Econometrica, Economic Journal, Economica, Journal of Economic
Literature, Journal of Economic Methodology, Quarterly Journal of
Economics, and Review of Economic Studies. Starmer is currently
Director of the Centre for Decision Research and Experimental
Economics (CeDEx) at the University of Nottingham.
d a v i d t e i r a s e r r a n o is associate professor in the Department of
Logic, History and Philosophy of Science (UNED, Madrid) and
research associate of the Urrutia Elejalde Foundation. His PhD
thesis, Azar, economa y poltica en Milton Friedman, won the best
dissertation award of the Spanish Society for Logic and Philosophy of
Science for the period 20013. He has published papers in the Journal of
Economic Methodology, History of Political Economy and Studies in History
and Philosophy of Science, among other journals.
j a c k v r o m e n is professor of the philosophy of science at the Faculty of
Philosophy of the Erasmus University Rotterdam, where he is also part
of EIPE (Erasmus Institute for Philosophy and Economics). His main
research interest is in philosophy of economics, with a special focus on
conceptual and meta-theoretical aspects of the relation between
evolutionary biology and evolutionary theorizing in economics. He
has published on Friedmans and Alchians evolutionary argument in
the theory of the firm. Recent publications include Ontological
commitments of evolutionary economics, in The Economic World
View (ed. Uskali Mki, Cambridge University Press, 2001), and
Stone Age minds and group selection what difference do they
make? in Constitutional Political Economy (2002) and Cognitive
theory meets evolutionary theory what promise does evolutionary
psychology hold out for economics? in Cognitive Developments in
Economics (ed. S. Rizzello, 2003). He is a past member of the
Executive Board of INEM (The International Network for Economic
Methodology) and Book Review Editor of the Journal of Economic
Methodology.
xv
o l i v e r p . w il l i a m s o n is Professor of the Graduate School and Edgar

F. Kaiser Professor Emeritus of Business, Economics, and Law at the
University of California, Berkeley. Williamson is a member of the
National Academy of Sciences (USA) and a Fellow of the American
Academy of Arts and Sciences and the Econometric Society. He is a
past president of the American Law and Economics Association, the
Western Economics Association, and the International Society for New
Institutional Economics. His books include Markets and Hierarchies
(1975), The Economic Institutions of Capitalism (1985), and The
Mechanisms of Governance (1996).
j e s s p. z a m o r a b o n i l l a is PhD in Philosophy and in Economics,
professor of Philosophy of Science at Universidad Nacional de
Educacion a Distancia (Madrid, Spain), as well as Academic
Coordinator of the Urrutia Elejalde Foundation (devoted to the study
of the connections between Economics and Philosophy). His main
research has been in the fields of scientific realism, verisimilitude,
economic methodology, and the economics of scientific knowledge.
Some recent papers by him are Verisimilitude and the scientific
strategy of economic theory (Journal of Economic Methodology, 1999),
Scientific inference and the pursuit of fame (Philosophy of Science,
2002), and Meaning and testability in the structuralist theory of
science (Erkenntnis, 2003), as well as a special issue (as invited
editor) on Darwinism in Economics in the journal Theoria. He is a
member of the Executive Board of the International Network for
Economic Method, of the Executive Board of the Spanish Society for
Logic and Philosophy of Science, and was a founder of the Iberian
American Society for Economic Methodology.
m i c h e l d e v r o e y is Professor of Economics at Universit catholique
de Louvain, Belgium. He has written extensively on the history of
economic thought. His particular research interest is the history of
macroeconomics. He is also a member of the editorial board for
European Journal of Economic Thought and a founding member of the
European Association for the History of Economic Thought. His latest
book is Involuntary Unemployment: The Elusive Quest for a Theory (2004).
Preface
Milton Friedman is known for a variety of contributions. Professional

economists acknowledge his work in monetary economics and some other
fields of economic inquiry. Many others recognize his name in connection
with the late twentieth-century free market ideology an ideology advising
how to conduct economic policy and how to organize social life more
generally. His writings in these areas are numerous, influential, and
controversial.
Friedman has also contributed to what may be called the ideology of
science an ideology advising how to conduct economic inquiry and how
to organize academic life in general. On this topic, his recognized writings
are less numerous, often considered reducible to one piece, The methodology of positive economics, that was published as the lead essay in his
Essays in Positive Economics in 1953 (dubbed throughout this volume as
F53). The essay was to become the most cited, the most influential, the
most controversial piece of methodological writing in twentieth-century
economics. It is also poorly understood and indeed hard to understand,
given its richness and obscurities. And it remains highly topical for the
foundational debates in and around economics in the early twenty-first
century. These are among the ingredients of a classic.
These features of F53 gave sufficient grounds for organizing a conference around it. The conference, organized by the Erasmus Institute for
Philosophy and Economics, took place in Rotterdam, the Netherlands, on
1213 December 2003. The occasion of the conference was the fiftieth
anniversary of the publication of F53. Prior to the conference, in January
2003, there was a panel session on F53 at the ASSA meetings in
Washington, DC, organized by Thomas Mayer. Milton Friedman was
connected to the panel via a telephone connection he had an opportunity
to comment on our ideas and to reply to our questions. Papers originating
from that panel were subsequently published in the Journal of Economic
Methodology later that year. The Erasmus conference was an ambitious
expansion of that panel.
xvii
xviii
Preface
Papers for the conference were commissioned from what was close to
the best possible team of scholars on the theme. They were asked to highlight a number of well-circumscribed issues that are contained in F53 or
that F53 has given rise to. The assignment given to the contributors was
broad and simple. They were asked to put themselves in the shoes of a
twenty-first-century reader wishing to check by himself or herself what to
make of F53: in the light of (1) the text of the essay, (2) the historical
and intellectual context of the essay, and (3) the later developments of
economics and its methodology what is the twenty-first-century reader
advised to think of the contents of F53 and their descriptive and prescriptive
adequacy? The assignment was indeed broad, but each contributor was
invited to approach it from a separate limited angle. Taken together, the
responses, reported in the chapters of the present book, provide a rather
comprehensive account of F53 for the twenty-first-century reader.
The harvest is no doubt thus far the most ambitious collective attempt
to examine and assess F53 in its various contexts. The flow of commentaries, criticisms, utilizations, and debates over the credentials of F53 has
continued for half a century now. I dare claim that the volume youre
holding in your hands makes significant progress towards a deeper understanding of F53 and the difficult issues that it set out to address in the mid
twentieth century and that economists and their critics keep arguing about
half a century later.
I invited Milton Friedman to join us in Rotterdam, but at the age of 91, he
was not eager to travel. Revealing his enduring interest in methodological
issues, he did read the contributions and wrote a brief note that is included
at the end of this volume. To my knowledge, this is the first time that he has
publicly spelled out his views about what others have written about his
essay, but unsurprisingly perhaps, he keeps his statement very general and
polite (while in private correspondence and conversations, he has been
active in reacting to various criticisms and suggestions in more substantive
ways). He had decided to stick to his old private rule according to which
he will let the essay live its own life. It remains a challenge to the rest of us to
live our academic lives together with the methodological essay that he left
behind. This is the challenge the current volume sets out to meet.
Special thanks go to Frank Hindriks, Caterina Marchionni, and
Giorgos Papadopoulos for great help in preparing and running the conference, and to Samuli Pyhnen for splendid assistance in preparing the
volume.
We would also like to thank the University of Chicago Press for allowing
us to reproduce F53 in its full and original form.
u s ka l i m k i
Part 1
The classical essay in twentieth-century

economic methodology
The Methodology of Positive Economics*
N HIS admirable book on The Scope and Method of Political

Economy John Neville Keynes distinguishes among "a positive
science . . . [ , ] a body of systematized knowledge concerning
what is; a normative or regulative science . . . [ , ] a body of systematized knowledge discussing criteria of what ought to be
. . . ; an art . . . [,] a system of rules for the attainment of a
given end"; comments that "confusion between them is common and has been the source of many mischievous errors"; and
urges the importance of "recognizing a distinct positive science
of political economy."1
This paper is concerned primarily with certain methodological
problems that arise in constructing the "distinct positive science"
Keynes called forin particular, the problem how to decide
whether a suggested hypothesis or theory should be tentatively
accepted as part of the "body of systematized knowledge concerning what is." But the confusion Keynes laments is still so
rife and so much of a hindrance to the recognition that economics can be, and in part is, a positive science that it seems well
to preface the main body of the paper with a few remarks about
the relation between positive and normative economics.
I. T H E RELATION BETWEEN POSITIVE AND NORMATIVE
ECONOMICS
Confusion between positive and normative economics is to

some extent inevitable. The subject matter of economics is regarded by almost everyone as vitally important to himself and
within the range of his own experience and competence; it is
* I have incorporated bodily in this article without special reference most of my
brief "Comment" in A Survey of Contemporary Economics, Vol. II (B. F. Haley,
ed.) (Chicago: Richard D. Irwin, Inc., 1952), pp. 455-57.
I am indebted to Dorothy S. Brady, Arthur F. Burns, and George J. Stigler for
helpful comments and criticism.
1. (London: Macmillan & Co., 1891), pp. 34-35 and 46.
4
Essays in Positive Economics
the source of continuous and extensive controversy and the occasion for frequent legislation. Self-proclaimed "experts" speak
with many voices and can hardly all be regarded as disinterested;
in any event, on questions that matter so much, "expert" opinion
could hardly be accepted solely on faith even if the "experts"
were nearly unanimous and clearly disinterested.2 The conclusions of positive economics seem to be, and are, immediately
relevant to important normative problems, to questions of what
ought to be done and how any given goal can be attained. Laymen and experts alike are inevitably tempted to shape positive
conclusions to fit strongly held normative preconceptions and to
reject positive conclusions if their normative implicationsor
what are said to be their normative implicationsare unpalatable.
Positive economics is in principle independent of any particular
ethical position or normative judgments. As Keynes says, it
deals with "what is," not with "what ought to be." Its task is to
provide a system of generalizations that can be used to make
correct predictions about the consequences of any change in
circumstances. Its performance is to be judged by the precision,
scope, and conformity with experience of the predictions it
yields. In short, positive economics is, or can be, an "objective"
science, in precisely the same sense as any of the physical sciences. Of course, the fact that economics deals with the interrelations of human beings, and that the investigator is himself
part of the subject matter being investigated in a more intimate
sense than in the physical sciences, raises special difficulties in
achieving objectivity at the same time that it provides the social
scientist with a class of data not available to the physical sci2. Social science or economics is by no means peculiar in this respectwitness
the importance of personal beliefs and of "home" remedies in medicine wherever
obviously convincing evidence for "expert" opinion is lacking. The current prestige and acceptance of the views of physical scientists in their fields of specializationand, all too often, in other fields as wellderives, not from faith alone,
but from the evidence of their works, the success of their predictions, and the dramatic achievements from applying their results. When economics seemed to provide
such evidence of its worth, in Great Britain in the first half of the nineteenth century, the prestige and acceptance of "scientific economics" rivaled the current prestige of the physical sciences.
entist. But neither the one nor the other is, in my view, a fundamental distinction between the two groups of sciences.3
Normative economics and the art of economics, on the other
hand, cannot be independent of positive economics. Any policy
conclusion necessarily rests on a prediction about the consequences of doing one thing rather than another, a prediction
that must be basedimplicitly or explicitlyon positive economics. There is not, of course, a one-to-one relation between policy
conclusions and the conclusions of positive economics; if there
were, there would be no separate normative science. Two individuals may agree on the consequences of a particular piece
of legislation. One may regard them as desirable on balance and
so favor the legislation; the other, as undesirable and so oppose
the legislation.
I venture the judgment, however, that currently in the Western world, and especially in the United States, differences about
economic policy among disinterested citizens derive predominantly from different predictions about the economic consequences of
taking actiondifferences that in principle can be eliminated by
the progress of positive economicsrather than from fundamental differences in basic values, differences about which men
can ultimately only fight. An obvious and not unimportant example is minimum-wage legislation. Underneath the welter of
arguments offered for and against such legislation there is an
underlying consensus on the objective of achieving a "living
wage" for all, to use the ambiguous phrase so common in such
discussions. The difference of opinion is largely grounded on an
implicit or explicit difference in predictions about the efficacy
of this particular means in furthering the agreed-on end. Proponents believe (predict) that legal minimum wages diminish
poverty by raising the wages of those receiving less than the
minimum wage as well as of some receiving more than the
3. The interaction between the observer and the process observed that is so prominent a feature of the social sciences, besides its more obvious parallel in the physical
sciences, has a more subtle counterpart in the indeterminacy principle arising out of
the interaction between the process of measurement and the phenomena being measured. And both have a counterpart in pure logic in Godel's theorem, asserting the
impossibility of a comprehensive self-contained logic. It is an open question whether
all three can be regarded as different formulations of an even more general principle.
minimum wage without any counterbalancing increase in the

number of people entirely unemployed or employed less advantageously than they otherwise would be. Opponents believe
(predict) that legal minimum wages increase poverty by increasing the number of people who are unemployed or employed
less advantageously and that this more than offsets any favorable effect on the wages of those who remain employed. Agreement about the economic consequences of the legislation might
not produce complete agreement about its desirability, for differences might still remain about its political or social consequences;
but, given agreement on objectives, it would certainly go a long
way toward producing consensus.
Closely related differences in positive analysis underlie divergent views about the appropriate role and place of trade-unions
and the desirability of direct price and wage controls and of
tariffs. Different predictions about the importance of so-called
"economies of scale" account very largely for divergent views
about the desirability or necessity of detailed government regulation of industry and even of socialism rather than private
enterprise. And this list could be extended indefinitely.4 Of
course, my judgment that the major differences about economic
policy in the Western world are of this kind is itself a "positive"
statement to be accepted or rejected on the basis of empirical
evidence.
If this judgment is valid, it means that a consensus on "correct" economic policy depends much less on the progress of
normative economics proper than on the progress of a positive
economics yielding conclusions that are, and deserve to be,
widely accepted. It means also that a major reason for dis4. One rather more complex example is stabilization policy. Superficially, divergent views on this question seem to reflect differences in objectives; but I believe
that this impression is misleading and that at bottom the different views reflect primarily different judgments about the source of fluctuations in economic activity
and the effect of alternative countercyclical action. For one major positive consideration that accounts for much of the divergence see "The Effects of a Full-Employment Policy on Economic Stability: A Formal Analysis," infra, pp. 117-32. For a
summary of the present state of professional views on this question see "The Problem of Economic Instability," a report of a subcommittee of the Committee on
Public Issues of the American Economic Association, American Economic Review,
XL (September, 1950), 501-38.
tinguishing positive economics sharply from normative economics is precisely the contribution that can thereby be made to
agreement about policy.
II. POSITIVE ECONOMICS
The ultimate goal of a positive science is the development of

a "theory" or "hypothesis" that yields valid and meaningful
(i.e., not truistic) predictions about phenomena not yet observed. Such a theory is, in general, a complex intermixture of
two elements. In part, it is a "language" designed to promote
"systematic and organized methods of reasoning."5 In part, it
is a body of substantive hypotheses designed to abstract essential features of complex reality.
Viewed as a language, theory has no substantive content; it
is a set of tautologies. Its function is to serve as a filing system
for organizing empirical material and facilitating our understanding of it; and the criteria by which it is to be judged are
those appropriate to a filing system. Are the categories clearly
and precisely defined? Are they exhaustive? Do we know where
to file each individual item, or is there considerable ambiguity?
Is the system of headings and subheadings so designed that we
can quickly find an item we want, or must we hunt from place
to place? Are the items we shall want to consider jointly filed
together? Does the filing system avoid elaborate cross-references?
The answers to these questions depend partly on logical, partly
on factual, considerations. The canons of formal logic alone can
show whether a particular language is complete and consistent,
that is, whether propositions in the language are "right" or
"wrong." Factual evidence alone can show whether the categories of the "analytical filing system" have a meaningful empirical counterpart, that is, whether they are useful in analyzing
a particular class of concrete problems.6 The simple example of
"supply" and "demand" illustrates both this point and the pre5. Final quoted phrase from Alfred Marshall, "The Present Position of Economics" (1885), reprinted in Memorials of Alfred Marshall, ed. A. C. Pigou (London: Macmillan & Co., 1925), p. 164. See also "The Marshallian Demand Curve,"
infra, pp. 56-57, 90-91.
6. See "Lange on Price Flexibility and Employment: A Methodological Criticism," infra, pp. 282-89.
ceding list of analogical questions. Viewed as elements of the

language of economic theory, these are the two major categories
into which factors affecting the relative prices of products or
factors of production are classified. The usefulness of the dichotomy depends on the "empirical generalization that an enumeration of the forces affecting demand in any problem and of the
forces affecting supply will yield two lists that contain few items
in common."7 Now this generalization is valid for markets like
the final market for a consumer good. In such a market there
is a clear and sharp distinction between the economic units that
can be regarded as demanding the product and those that can be
regarded as supplying it. There is seldom much doubt whether
a particular factor should be classified as affecting supply, on
the one hand, or demand, on the other; and there is seldom much
necessity for considering cross-effects (cross-references) between the two categories. In these cases the simple and even
obvious step of filing the relevant factors under the headings
of "supply" and "demand" effects a great simplification of the
problem and is an effective safeguard against fallacies that
otherwise tend to occur. But the generalization is not always
valid. For example, it is not valid for the day-to-day fluctuations
of prices in a primarily speculative market. Is a rumor of an
increased excess-profits tax, for example, to be regarded as a
factor operating primarily on today's supply of corporate equities in the stock market or on today's demand for them? In
similar fashion, almost every factor can with about as much
justification be classified under the heading "supply" as under
the heading "demand." These concepts can still be used and
may not be entirely pointless; they are still "right" but clearly
less useful than in the first example because they have no meaningful empirical counterpart.
Viewed as a body of substantive hypotheses, theory is to be
judged by its predictive power for the class of phenomena which
it is intended to "explain." Only factual evidence can show
whether it is "right" or "wrong" or, better, tentatively "accepted" as valid or "rejected." As I shall argue at greater length
below, the only relevant test of the validity of a hypothesis is
7. "The Marshallian Demand Curve," infra, p. 57.
comparison of its predictions with experience. The hypothesis

is rejected if its predictions are contradicted ("frequently" or
more often than predictions from an alternative hypothesis);
it is accepted if its predictions are not contradicted; great confidence is attached to it if it has survived many opportunities
for contradiction. Factual evidence can never "prove" a hypothesis; it can only fail to disprove it, which is what we generally
mean when we say, somewhat inexactly, that the hypothesis has
been "confirmed" by experience.
To avoid confusion, it should perhaps be noted explicitly that
the "predictions" by which the validity of a hypothesis is tested
need not be about phenomena that have not yet occurred, that
is, need not be forecasts of future events; they may be about
phenomena that have occurred but observations on which have
not yet been made or are not known to the person making the
prediction. For example, a hypothesis may imply that such and
such must have happened in 1906, given some other known
circumstances. If a search of the records reveals that such and
such did happen, the prediction is confirmed; if it reveals that
such and such did not happen, the prediction is contradicted.
The validity of a hypothesis in this sense is not by itself a
sufficient criterion for choosing among alternative hypotheses.
Observed facts are necessarily finite in number; possible hypotheses, infinite. If there is one hypothesis that is consistant with
the available evidence, there are always an infinite number that
are.8 For example, suppose a specific excise tax on a particular
commodity produces a rise in price equal to the amount of the
tax. This is consistent with competitive conditions, a stable demand curve, and a horizontal and stable supply curve. But it
is also consistent with competitive conditions and a positively
or negatively sloping supply curve with the required compensating shift in the demand curve or the supply curve; with monopolistic conditions, constant marginal costs, and stable demand
curve, of the particular shape required to produce this result;
and so on indefinitely. Additional evidence with which the
8. The qualification is necessary because the "evidence" may be internally contradictory, so there may be no hypothesis consistent with it. See also "Lange
on Price Flexibility and Employment," infra, pp. 282-83.
io
hypothesis is to be consistent may rule out some of these possibilities; it can never reduce them to a single possibility alone
capable of being consistent with the finite evidence. The choice
among alternative hypotheses equally consistent with the available evidence must to some extent be arbitrary, though there
is general agreement that relevant considerations are suggested
by the criteria "simplicity" and "fruitfulness," themselves notions that defy completely objective specification. A theory is
"simpler" the less the initial knowledge needed to make a prediction within a given field of phenomena; it is more "fruitful"
the more precise the resulting prediction, the wider the area within which the theory yields predictions, and the more additional
lines for further research it suggests. Logical completeness and
consistency are relevant but play a subsidiary role; their function is to assure that the hypothesis says what it is intended
to say and does so alike for all usersthey play the same role
here as checks for arithmetical accuracy do in statistical computations.
Unfortunately, we can seldom test particular predictions in the
social sciences by experiments explicitly designed to eliminate
what are judged to be the most important disturbing influences.
Generally, we must rely on evidence cast up by the "experiments" that happen to occur. The inability to conduct so-called
"controlled experiments" does not, in my view, reflect a basic
difference between the social and physical sciences both because
it is not peculiar to the social scienceswitness astronomy
and because the distinction between a controlled experiment and
uncontrolled experience is at best one of degree. No experiment
can be completely controlled, and every experience is partly controlled, in the sense that some disturbing influences are relatively constant in the course of it.
Evidence cast up by experience is abundant and frequently
as conclusive as that from contrived experiments; thus the
inability to conduct experiments is not a fundamental obstacle
to testing hypotheses by the success of their predictions. But
such evidence is far more difficult to interpret. It is frequently
complex and always indirect and incomplete. Its collection is
often arduous, and its interpretation generally requires subtle
II
analysis and involved chains of reasoning, which seldom carry

real conviction. The denial to economics of the dramatic and
direct evidence of the "crucial" experiment does hinder the
adequate testing of hypotheses; but this is much less significant
than the difficulty it places in the way of achieving a reasonably
prompt and wide consensus on the conclusions justified by the
available evidence. It renders the weeding-out of unsuccessful
hypotheses slow and difficult. They are seldom downed for good
and are always cropping up again.
There is, of course, considerable variation in these respects.
Occasionally, experience casts up evidence that is about as direct,
dramatic, and convincing as any that could be provided by controlled experiments. Perhaps the most obviously important example is the evidence from inflations on the hypothesis that a
substantial increase in the quantity of money within a relatively
short period is accompanied by a substantial increase in prices.
Here the evidence is dramatic, and the chain of reasoning required to interpret it is relatively short. Yet, despite numerous
instances of substantial rises in prices, their essentially one-toone correspondence with substantial rises in the stock of money,
and the wide variation in other circumstances that might appear
to be relevant, each new experience of inflation brings forth
vigorous contentions, and not only by the lay public, that the
rise in the stock of money is either an incidental effect of a rise
in prices produced by other factors or a purely fortuitous and
unnecessary concomitant of the price rise.
One effect of the difficulty of testing substantive economic
hypotheses has been to foster a retreat into purely formal or
tautological analysis.9 As already noted, tautologies have an
extremely important place in economics and other sciences as a
specialized language or "analytical filing system." Beyond this,
formal logic and mathematics, which are both tautologies, are
essential aids in checking the correctness of reasoning, discovering the implications of hypotheses, and determining whether supposedly different hypotheses may not really be equivalent or
wherein the differences lie.
But economic theory must be more than a structure of tautol9. See "Lange on Price Flexibility and Employment," infra, passim.
12
ogies if it is to be able to predict and not merely describe the

consequences of action; if it is to be something different from
disguised mathematics.10 And the usefulness of the tautologies
themselves ultimately depends, as noted above, on the acceptability of the substantive hypotheses that suggest the particular
categories into which they organize the refractory empirical
phenomena.
A more serious effect of the difficulty of testing economic
hypotheses by their predictions is to foster misunderstanding of
the role of empirical evidence in theoretical work. Empirical
evidence is vital at two different, though closely related, stages:
in constructing hypotheses and in testing their validity. Full and
comprehensive evidence on the phenomena to be generalized or
"explained" by a hypothesis, besides its obvious value in suggesting new hypotheses, is needed to assure that a hypothesis
explains what it sets out to explainthat its implications for
such phenomena are not contradicted in advance by experience
that has already been observed.11 Given that the hypothesis is
10. See also Milton Friedman and L. J. Savage, "The Expected-Utility Hypothesis and the Measurability of Utility," Journal of Political Economy, LX
(December, 1952), 463-74, esp. pp. 465-67.
11. In recent years some economists, particularly a group connected with the
Cowles Commission for Research in Economics at the University of Chicago,
have placed great emphasis on a division of this step of selecting a hypothesis
consistent with known evidence into two substeps: first, the selection of a class
of admissible hypotheses from all possible hypotheses (the choice of a "model"
in their terminology); second, the selection of one hypothesis from this class
(the choice of a "structure"). This subdivision may be heuristically valuable in
some kinds of work, particularly in promoting a systematic use of available
statistical evidence and theory. From a methodological point of view, however,
it is an entirely arbitrary subdivision of the process of deciding on a particular
hypothesis that is on a par with many other subdivisions that may be convenient
for one purpose or another or that may suit the psychological needs of particular
investigators.
One consequence of this particular subdivision has been to give rise to the socalled "identification" problem. As noted above, if one hypothesis is consistent
with available evidence, an infinite number are. But, while this is true for the
class of hypotheses as a whole, it may not be true of the subclass obtained in
the first of the above two stepsthe "model." It may be that the evidence to
be used to select the final hypothesis from the subclass can be consistent with
at most one hypothesis in it, in which case the "model" is said to be "identified";
otherwise it is said to be "unidentified." As is clear from this way of describing
the concept of "identification," it is essentially a special case of the more general
13
consistent with the evidence at hand, its further testing involves

deducing from it new facts capable of being observed but not
previously known and checking these deduced facts against additional empirical evidence. For this test to be relevant, the deduced
facts must be about the class of phenomena the hypothesis is
designed to explain; and they must be well enough denned so
that observation can show them to be wrong.
The two stages of constructing hypotheses and testing their
validity are related in two different respects. In the first place,
the particular facts that enter at each stage are partly an accident of the collection of data and the knowledge of the particular
investigator. The facts that serve as a test of the implications
of a hypothesis might equally well have been among the raw
material used to construct it, and conversely. In the second place,
the process never begins from scratch; the so-called "initial
stage" itself always involves comparison of the implications of
an earlier set of hypotheses with observation; the contradiction
of these implications is the stimulus to the construction of new
problem of selecting among the alternative hypotheses equally consistent with
the evidencea problem that must be decided by some such arbitrary principle
as Occam's razor. The introduction of two substeps in selecting a hypothesis
makes this problem arise at the two corresponding stages and gives it a special
cast. While the class of all hypotheses is always unidentified, the subclass in a
"model" need not be, so the problem arises of conditions that a "model" must
satisfy to be identified. However useful the two substeps may be in some contexts, their introduction raises the danger that different criteria will unwittingly
be used in making the same kind of choice among alternative hypotheses at two
different stages.
On the general methodological approach discussed in this footnote see Tryvge
Haavelmo, "The Probability Approach in Econometrics," Econometrica, Vol. XII
(1944), Supplement; Jacob Marschak, "Economic Structure, Path, Policy, and
Prediction," American Economic Review, XXXVII (May, 1947), 81-84, and
"Statistical Inference in Economics: An Introduction," in T. C. Koopmans (ed.),
Statistical Inference in Dynamic Economic Models (New York: John Wiley & Sons,
1950); T. C. Koopmans, "Statistical Estimation of Simultaneous Economic Relations," Journal of the American Statistical Association, XL (December, 1945),
448-66; Gershon Cooper, "The Role of Economic Theory in Econometric Models,"
Journal of Farm Economics, XXX (February, 1948), 101-16. On the identification problem see Koopmans, "Identification Problems in Econometric Model
Construction," Econometrica, XVII (April, 1949), 125-44; Leonid Hurwicz, "Generalization of the Concept of Identification," in Koopmans (ed.), Statistical Inference in Dynamic Economic Models.
14
hypotheses or revision of old ones. So the two methodologically

distinct stages are always proceeding jointly.
Misunderstanding about this apparently straightforward process centers on the phrase "the class of phenomena the hypothesis
is designed to explain." The difficulty in the social sciences of
getting new evidence for this class of phenomena and of judging its conformity with the implications of the hypothesis makes
it tempting to suppose that other, more readily available, evidence is equally relevant to the validity of the hypothesisto
suppose that hypotheses have not only "implications" but also
"assumptions" and that the conformity of these "assumptions"
to "reality" is a test of the validity of the hypothesis different
from or additional to the test by implications. This widely held
view is fundamentally wrong and productive of much mischief.
Far from providing an easier means for sifting valid from invalid
hypotheses, it only confuses the issue, promotes misunderstanding about the significance of empirical evidence for economic
theory, produces a misdirection of much intellectual effort devoted to the development of positive economics, and impedes
the attainment of consensus on tentative hypotheses in positive
economics.
In so far as a theory can be said to have "assumptions" at
all, and in so far as their "realism" can be judged independently
of the validity of predictions, the relation between the significance of a theory and the "realism" of its "assumptions" is almost the opposite of that suggested by the view under criticism.
Truly important and significant hypotheses will be found to have
"assumptions" that are wildly inaccurate descriptive representations of reality, and, in general, the more significant the theory,
the more unrealistic the assumptions (in this sense).12 The reason is simple. A hypothesis is important if it "explains" much
by little, that is, if it abstracts the common and crucial elements
from the mass of complex and detailed circumstances surrounding the phenomena to be explained and permits valid predictions on the basis of them alone. To be important, therefore,
a hypothesis must be descriptively false in its assumptions; it
12. The converse of the proposition does not of course hold: assumptions that
are unrealistic (in this sense) do not guarantee a significant theory.
15
takes account of, and accounts for, none of the many other
attendant circumstances, since its very success shows them to
be irrelevant for the phenomena to be explained.
To put this point less paradoxically, the relevant question to
ask about the "assumptions" of a theory is not whether they
are descriptively "realistic," for they never are, but whether they
are sufficiently good approximations for the purpose in hand.
And this question can be answered only by seeing whether the
theory works, which means whether it yields sufficiently accurate
predictions. The two supposedly independent tests thus reduce
to one test.
The theory of monopolistic and imperfect competition is one
example of the neglect in economic theory of these propositions.
The development of this analysis was explicitly motivated, and
its wide acceptance and approval largely explained, by the belief that the assumptions of "perfect competition" or "perfect
monopoly" said to underlie neoclassical economic theory are a
false image of reality. And this belief was itself based almost
entirely on the directly perceived descriptive inaccuracy of the
assumptions rather than on any recognized contradiction of
predictions derived from neoclassical economic theory. The
lengthy discussion on marginal analysis in the American Economic Review some years ago is an even clearer, though much less
important, example. The articles on both sides of the controversy
largely neglect what seems to me clearly the main issue-the
conformity to experience of the implications of the marginal
analysisand concentrate on the largely irrelevant question
whether businessmen do or do not in fact reach their decisions
by consulting schedules, or curves, or multivariable functions
showing marginal cost and marginal revenue.13 Perhaps these
13. See R. A. Lester, "Shortcomings of Marginal Analysis for Wage-Employment Problems," American Economic Review, XXXVI (March, 1946), 62-82;
Fritz Machlup, "Marginal Analysis and Empirical Research," American Economic
Review, XXXVI (September, 1946), 519-54; R. A. Lester, "Marginalism, Minimum Wages, and Labor Markets," American Economic Review, XXXVII (March,
1947), 135-48; Fritz Machlup, "Rejoinder to an Antimarginalist," American
Economic Review, XXXVII (March, 1947), 148-54; G. J. Stigler, "Professor
Lester and the Marginalists," American Economic Review, XXXVII (March,
1947), 154-57; H. M. Oliver, Jr., "Marginal Theory and Business Behavior,"
American Economic Review, XXXVII (June, 1947), 375-83; R. A. Gordon,
16
two examples, and the many others they readily suggest, will serve
to justify a more extensive discussion of the methodological
principles involved than might otherwise seem appropriate.
III. CAN A HYPOTHESIS B E TESTED BY THE
REALISM OF ITS ASSUMPTIONS?
We may start with a simple physical example, the law of falling bodies. It is an accepted hypothesis that the acceleration of
a body dropped in a vacuum is a constantg, or approximately
32 feet per second per second on the earthand is independent
of the shape of the body, the manner of dropping it, etc. This
implies that the distance traveled by a falling body in any specified time is given by the formula s = % gt2, where s is the distance traveled in feet and t is time in seconds. The application
of this formula to a compact ball dropped from the roof of a
building is equivalent to saying that a ball so dropped behaves
as if it were falling in a vacuum. Testing this hypothesis by its
assumptions presumably means measuring the actual air pressure and deciding whether it is close enough to zero. At sea level
the air pressure is about 15 pounds per square inch. Is 15 sufficiently close to zero for the difference to be judged insignificant?
Apparently it is, since the actual time taken by a compact ball to
fall from the roof of a building to the ground is very close to the
time given by the formula. Suppose, however, that a feather is
"Short-Period Price Determination in Theory and Practice," American Economic
Review, XXXVIII (June, 1948), 265-88.
It should be noted that, along with much material purportedly bearing on the
validity of the "assumptions" of marginal theory, Lester does refer to evidence on the conformity of experience with the implications of the theory, citing
the reactions of employment in Germany to the Papen plan and in the United
States to changes in minimum-wage legislation as examples of lack of conformity.
However, Stigler's brief comment is the only one of the other papers that refers
to this evidence. It should also be noted that Machlup's thorough and careful
exposition of the logical structure and meaning of marginal analysis is called
for by the misunderstandings on this score that mar Lester's paper and almost
conceal the evidence he presents that is relevant to the key issue he raises. But,
in Machlup's emphasis on the logical structure, he comes perilously close to presenting the theory as a pure tautology, though it is evident at a number of
points that he is aware of this danger and anxious to avoid it. The papers by
Oliver and Gordon are the most extreme in the exclusive concentration on the
conformity of the behavior of businessmen with the "assumptions" of the theory.
17
dropped instead of a compact ball. The formula then gives wildly

inaccurate results. Apparently, 15 pounds per square inch is
significantly different from zero for a feather but not for a ball.
Or, again, suppose the formula is applied to a ball dropped from
an airplane at an altitude of 30,000 feet. The air pressure at
this altitude is decidedly less than 15 pounds per square inch.
Yet, the actual time of fall from 30,000 feet to 20,000 feet, at
which point the air pressure is still much less than at sea level,
will differ noticeably from the time predicted by the formula
much more noticeably than the time taken by a compact ball to
fall from the roof of a building to the ground. According to the
fOFHllila, the velocity of the ball should be gt and should therefore increase steadily. In fact, a ball dropped at 30,000 feet will
reach its top velocity well before it hits the ground. And similarly with other implications of the formula.
The initial question whether 15 is sufficiently close to zero for
the difference to be judged insignificant is clearly a foolish question by itself. Fifteen pounds per square inch is 2,160 pounds
per square foot, or 0.0075 ton per square inch. There is no possible basis for calling these numbers "small" or "large" without
some external standard of comparison. And the only relevant
standard of comparison is the air pressure for which the formula
does or does not work under a given set of circumstances. But
this raises the same problem at a second level. What is the meaning of "does or does not work"? Even if we could eliminate errors
of measurement, the measured time of fall would seldom if ever
be precisely equal to the computed time of fall. How large must
the difference between the two be to justify saying that the theory
"does not work"? Here there are two important external standards of comparison. One is the accuracy achievable by an alternative theory with which this theory is being compared and which
is equally acceptable on all other grounds. The other arises when
there exists a theory that is known to yield better predictions but
only at a greater cost. The gains from greater accuracy, which
depend on the purpose in mind, must then be balanced against
the costs of achieving it.
This example illustrates both the impossibility of testing a
18
theory by its assumptions and also the ambiguity of the concept

"the assumptions of a theory." The formula 5 = \ gt2 is valid
for bodies falling in a vacuum and can be derived by analyzing
the behavior of such bodies. It can therefore be stated: under a
wide range of circumstances, bodies that fall in the actual atmosphere behave as if they were falling in a vacuum. In the language
so common in economics this would be rapidly translated into:
the formula assumes a vacuum. Yet it clearly does no such thing.
What it does say is that in many cases the existence of air pressure, the shape of the body, the name of the person dropping the
body, the kind of mechanism used to drop the body, and a host
of other attendant circumstances have no appreciable effect on
the distance the body falls in a specified time. The hypothesis
can readily be rephrased to omit all mention of a vacuum: under
a wide range of circumstances, the distance a body falls in a
specified time is given by the formula s = i gt2. The history
of this formula and its associated physical theory aside, is it
meaningful to say that it assumes a vacuum? For all I know
there may be other sets of assumptions that would yield the same
formula. The formula is accepted because it works, not because
we live in an approximate vacuumwhatever that means.
The important problem in connection with the hypothesis is
to specify the circumstances under which the formula works or,
more precisely, the general magnitude of the error in its predictions under various circumstances. Indeed, as is implicit in the
above rephrasing of the hypothesis, such a specification is not
one thing and the hypothesis another. The specification is itself
an essential part of the hypothesis, and it is a part that is peculiarly likely to be revised and extended as experience accumulates.
In the particular case of falling bodies a more general, though
still incomplete, theory is available, largely as a result of attempts to explain the errors of the simple theory, from which the
influence of some of the possible disturbing factors can be calculated and of which the simple theory is a special case. However, it does not always pay to use the more general theory because the extra accuracy it yields may not justify the extra cost
of using it, so the question under what circumstances the simpler
theory works "well enough" remains important. Air pressure
19
is one, but only one, of the variables that define these circumstances; the shape of the body, the velocity attained, and still
other variables are relevant as well. One way of interpreting the
variables other than air pressure is to regard them as determining
whether a particular departure from the "assumption" of a
vacuum is or is not significant. For example, the difference in
shape of the body can be said to make 15 pounds per square inch
significantly different from zero for a feather but not for a
compact ball dropped a moderate distance. Such a statement
must, however, be sharply distinguished from the very different
statement that the theory does not work for a feather because
its assumptions are false. The relevant relation runs the other
way: the assumptions are false for a feather because the theory
does not work. This point needs emphasis, because the entirely
valid use of "assumptions" in specifying the circumstances for
which a theory holds is frequently, and erroneously, interpreted
to mean that the assumptions can be used to determine the circumstances for which a theory holds, and has, in this way, been
an important source of the belief that a theory can be tested
by its assumptions.
Let us turn now to another example, this time a constructed one
designed to be an analogue of many hypotheses in the social
sciences. Consider the density of leaves around a tree. I suggest
the hypothesis that the leaves are positioned as if each leaf deliberately sought to maximize the amount of sunlight it receives,
given the position of its neighbors, as if it knew the physical laws
determining the amount of sunlight that would be received in
various positions and could move rapidly or instantaneously from
any one position to any other desired and unoccupied position.14
Now some of the more obvious implications of this hypothesis
are clearly consistent with experience: for example, leaves are in
general denser on the south than on the north side of trees but,
as the hypothesis implies, less so or not at all on the northern
14. This example, and some of the subsequent discussion, though independent
in origin, is similar to and in much the same spirit as an example and the approach in an important paper by Armen A. Alchian, "Uncertainty, Evolution,
and Economic Theory," Journal of Political Economy, LVIII (June, 1950),
211-21.
20
slope of a hill or when the south side of the trees is shaded in

some other way. Is the hypothesis rendered unacceptable or invalid because, so far as we know, leaves do not "deliberate" or
consciously "seek," have not been to school and learned the
relevant laws of science or the mathematics required to calculate
the "optimum" position, and cannot move from position to position? Clearly, none of these contradictions of the hypothesis is
vitally relevant; the phenomena involved are not within the
"class of phenomena the hypothesis is designed to explain"; the
hypothesis does not assert that leaves do these things but only
that their density is the same as if they did. Despite the apparent
falsity of the "assumptions" of the hypothesis, it has great plausi^
bility because of the conformity of its implications with observation. We are inclined to "explain" its validity on the ground
that sunlight contributes to the growth of leaves and that hence
leaves will grow denser or more putative leaves survive where
there is more sun, so the result achieved by purely passive adaptation to external circumstances is the same as the result that would
be achieved by deliberate accommodation to them. This alternative hypothesis is more attractive than the constructed hypothesis
not because its "assumptions" are more "realistic" but rather
because it is part of a more general theory that applies to a
wider variety of phenomena, of which the position of leaves
around a tree is a special case, has more implications capable
of being contradicted, and has failed to be contradicted under
a wider variety of circumstances. The direct evidence for the
growth of leaves is in this way strengthened by the indirect evidence from the other phenomena to which the more general
theory applies.
The constructed hypothesis is presumably valid, that is, yields
"sufficiently" accurate predictions about the density of leaves,
only for a particular class of circumstances. I do not know what
these circumstances are or how to define them. It seems obvious,
however, that in this example the "assumptions" of the theory
will play no part in specifying them: the kind of tree, the character of the soil, etc., are the types of variables that are likely to
define its range of validity, not the ability of the leaves to do
complicated mathematics or to move from place to place.
21
A largely parallel example involving human behavior has been

used elsewhere by Savage and me.15 Consider the problem of predicting the shots made by an expert billiard player. It seems not
at all unreasonable that excellent predictions would be yielded by
the hypothesis that the billiard player made his shots as if he
knew the complicated mathematical formulas that would give the
optimum directions of travel, could estimate accurately by eye
the angles, etc., describing the location of the balls, could make
lightning calculations from the formulas, and could then make
the balls travel in the direction indicated by the formulas. Our
confidence in this hypothesis is not based on the belief that
billiard players, even expert ones, can or do go through the
process described; it derives rather from the belief that, unless
in some way or other they were capable of reaching essentially
the same result, they would not in fact be expert billiard players.
It is only a short step from these examples to the economic
hypothesis that under a wide range of circumstances individual
firms behave as if they were seeking rationally to maximize their
expected returns (generally if misleadingly called "profits") 16 and
had full knowledge of the data needed to succeed in this attempt;
as if, that is, they knew the relevant cost and demand functions,
15. Milton Friedman and L. J . Savage, "The Utility Analysis of Choices Involving Risk," Journal of Political Economy, LVI (August, 1948), 298. Reprinted
in American Economic Association, Readings in Price Theory (Chicago: Richard
D. Irwin, Inc., 1952), pp. 57-96.
16. It seems better to use the term "profits" to refer to the difference between
actual and "expected" results, between ex post and ex ante receipts. "Profits" are
then a result of uncertainty and, as Alchian {op. cit., p. 212), following Tintner,
points out, cannot be deliberately maximized in advance. Given uncertainty, individuals or firms choose among alternative anticipated probability distributions
of receipts or incomes. The specific content of a theory of choice among such
distributions depends on the criteria by which they are supposed to be ranked.
One hypothesis supposes them to be ranked by the mathematical expectation
of utility corresponding to them (see Friedman and Savage, "The ExpectedUtility Hypothesis and the Measurability of Utility," op. cit.). A special case
of this hypothesis or an alternative to it ranks probability distributions by
the mathematical expectation of the money receipts corresponding to them. The
latter is perhaps more applicable, and more frequently applied, to firms than
to individuals. The term "expected returns" is intended to be sufficiently broad
to apply to any of these alternatives.
The issues alluded to in this note are not basic to the methodological issues
being discussed, and so are largely by-passed in the discussion that follows.
22
calculated marginal cost and marginal revenue from all actions

open to them, and pushed each line of action to the point at which
the relevant marginal cost and marginal revenue were equal.
Now, of course, businessmen do not actually and literally solve
the system of simultaneous equations in terms of which the mathematical economist finds it convenient to express this hypothesis,
any more than leaves or billiard players explicitly go through
complicated mathematical calculations or falling bodies decide to
create a vacuum. The billiard player, if asked how he decides
where to hit the ball, may say that he "just figures it out" but
then also rubs a rabbit's foot just to make sure; and the businessman may well say that he prices at average cost, with of course
some minor deviations when the market makes it necessary. The
one statement is about as helpful as the other, and neither is a
relevant test of the associated hypothesis.
Confidence in the maximization-of-returns hypothesis is justified by evidence of a very different character. This evidence is
in part similar to that adduced on behalf of the billiard-player
hypothesisunless the behavior of businessmen in some way or
other approximated behavior consistent with the maximization
of returns, it seems unlikely that they would remain in business
for long. Let the apparent immediate determinant of business
behavior be anything at allhabitual reaction, random chance,
or whatnot. Whenever this determinant happens to lead to behavior consistent with rational and informed maximization of
returns, the business will prosper and acquire resources with
which to expand; whenever it does not, the business will tend
to lose resources and can be kept in existence only by the addition of resources from outside. The process of "natural selection"
thus helps to validate the hypothesisor, rather, given natural
selection, acceptance of the hypothesis can be based largely on
the judgment that it summarizes appropriately the conditions
for survival.
An even more important body of evidence for the maximization-of-returns hypothesis is experience from countless applications of the hypothesis to specific problems and the repeated
failure of its implications to be contradicted. This evidence is
extremely hard to document; it is scattered in numerous memo-
23
randums, articles, and monographs concerned primarily with

specific concrete problems rather than with submitting the
hypothesis to test. Yet the continued use and acceptance of the
hypothesis over a long period, and the failure of any coherent,
self-consistent alternative to be developed and be widely accepted,
is strong indirect testimony to its worth. The evidence for a
hypothesis always consists of its repeated failure to be contradicted, continues to accumulate so long as the hypothesis is used,
and by its very nature is difficult to document at all comprehensively. It tends to become part of the tradition and folklore
of a science revealed in the tenacity with which hypotheses are
held rather than in any textbook list of instances in which the
hypothesis has failed to be contradicted.
IV. T H E SIGNIFICANCE AND ROLE OF THE
"ASSUMPTIONS" OF A THEORY
Up to this point our conclusions about the significance of the

"assumptions" of a theory have been almost entirely negative:
we have seen that a theory cannot be tested by the "realism"
of its "assumptions" and that the very concept of the "assumptions" of a theory is surrounded with ambiguity. But, if this
were all there is to it, it would be hard to explain the extensive
use of the concept and the strong tendency that we all have to
speak of the assumptions of a theory and to compare the assumptions of alternative theories. There is too much smoke for there
to be no fire.
In methodology, as in positive science, negative statements
can generally be made with greater confidence than positive statements, so I have less confidence in the following remarks on
the significance and role of "assumptions" than in the preceding
remarks. So far as I can see, the "assumptions of a theory" play
three different, though related, positive roles: (a) they are often
an economical mode of describing or presenting a theory; (b)
they sometimes facilitate an indirect test of the hypothesis by
its implications; and (c), as already noted, they are sometimes
a convenient means of specifying the conditions under which the
theory is expected to be valid. The first two require more extensive discussion.
24

A. THE USE OF "ASSUMPTIONS" IN STATING A THEORY
The example of the leaves illustrates the first role of assumptions. Instead of saying that leaves seek to maximize the sunlight they receive, we could state the equivalent hypothesis, without any apparent assumptions, in the form of a list of rules for
predicting the density of leaves: if a tree stands in a level field
with no other trees or other bodies obstructing the rays of the
sun, then the density of leaves will tend to be such and such; if
a tree is on the northern slope of a hill in the midst of a forest
of similar trees, then . . . ; etc. This is clearly a far less economical
presentation of the hypothesis than the statement that leaves
seek to maximize the sunlight each receives. The latter statement
is, in effect, a simple summary of the rules in the above list, even
if the list were indefinitely extended, since it indicates both how
to determine the features of the environment that are important
for the particular problem and how to evaluate their effects. It
is more compact and at the same time no less comprehensive.
More generally, a hypothesis or theory consists of an assertion
that certain forces are, and by implication others are not, important for a particular class of phenomena and a specification
of the manner of action of the forces it asserts to be important.
We can regard the hypothesis as consisting of two parts: first,
a conceptual world or abstract model simpler than the "real
world" and containing only the forces that the hypothesis asserts
to be important; second, a set of rules defining the class of
phenomena for which the "model" can be taken to be an adequate
representation of the "real world" and specifying the correspondence between the variables or entities in the model and observable phenomena.
These two parts are very different in character. The model is
abstract and complete; it is an "algebra" or "logic." Mathematics
and formal logic come into their own in checking its consistency
and completeness and exploring its implications. There is no place
in the model for, and no function to be served by, vagueness,
maybe's, or approximations. The air pressure is zero, not "small,"
for a vacuum; the demand curve for the product of a competitive

25
producer is horizontal (has a slope of zero), not "almost horizontal."
The rules for using the model, on the other hand, cannot possibly be abstract and complete. They must be concrete and in
consequence incompletecompleteness is possible only in a conceptual world, not in the "real world," however that may be interpreted. The model is the logical embodiment of the half-truth,
"There is nothing new under the sun"; the rules for applying
it cannot neglect the equally significant half-truth, "History never
repeats itself." To a considerable extent the rules can be formulated explicitlymost easily, though even then not completely,
when the theory is part of an explicit more general theory as in
the example of the vacuum theory for falling bodies. In seeking
to make a science as "objective" as possible, our aim should be
to formulate the rules explicitly in so far as possible and continually to widen the range of phenomena for which it is possible
to do so. But, no matter how successful we may be in this attempt, there inevitably will remain room for judgment in applying the rules. Each occurrence has some features peculiarly its
own, not covered by the explicit rules. The capacity to judge
that these are or are not to be disregarded, that they should or
should not affect what observable phenomena are to be identified with what entities in the model, is something that cannot be
taught; it can be learned but only by experience and exposure
in the "right" scientific atmosphere, not by rote. It is at this
point that the "amateur" is separated from the "professional"
in all sciences and that the thin line is drawn which distinguishes
the "crackpot" from the scientist.
A simple example may perhaps clarify this point. Euclidean
geometry is an abstract model, logically complete and consistent.
Its entities are precisely denneda line is not a geometrical
figure "much" longer than it is wide or deep; it is a figure whose
width and depth are zero. It is also obviously "unrealistic." There
are no such things in "reality" as Euclidean points or lines or
surfaces. Let us apply this abstract model to a mark made on a
blackboard by a piece of chalk. Is the mark to be identified
with a Euclidean line, a Euclidean surface, or a Euclidean solid?
26
Clearly, it can appropriately be identified with a line if it is being

used to represent, say, a demand curve. But it cannot be so
identified if it is being used to color, say, countries on a map,
for that would imply that the map would never be colored; for
this purpose, the same mark must be identified with a surface.
But it cannot be so identified by a manufacturer of chalk, for
that would imply that no chalk would ever be used up; for his
purposes, the same mark must be identified with a volume. In
this simple example these judgments will command general
agreement. Yet it seems obvious that, while general considerations can be formulated to guide such judgments, they can never
be comprehensive and cover every possible instance; they cannot have the self-contained coherent character of Euclidean
geometry itself.
In speaking of the "crucial assumptions" of a theory, we are,
I believe, trying to state the key elements of the abstract model.
There are generally many different ways of describing the model
completelymany different sets of "postulates" which both
imply and are implied by the model as a whole. These are all
logically equivalent: what are regarded as axioms or postulates
of a model from one point of view can be regarded as theorems
from another, and conversely. The particular "assumptions"
termed "crucial" are selected on grounds of their convenience
in some such respects as simplicity or economy in describing the
model, intuitive plausibility, or capacity to suggest, if only by
implication, some of the considerations that are relevant in judging or applying the model.
B. THE USE OF "ASSUMPTIONS" AS AN
INDIRECT TEST OF A THEORY
In presenting any hypothesis, it generally seems obvious which

of the series of statements used to expound it refer to assumptions and which to implications; yet this distinction is not easy
to define rigorously. It is not, I believe, a characteristic of the
hypothesis as such but rather of the use to which the hypothesis
is to be put. If this is so, the ease of classifying statements must
reflect unambiguousness in the purpose the hypothesis is designed
to serve. The possibility of interchanging theorems and axioms in
27
an abstract model implies the possibility of interchanging "implications" and "assumptions" in the substantive hypothesis
corresponding to the abstract model, which is not to say that
any implication can be interchanged with any assumption but
only that there may be more than one set of statements that
imply the rest.
For example, consider a particular proposition in the theory
of oligopolistic behavior. If we assume (a) that entrepreneurs
seek to maximize their returns by any means including acquiring
or extending monopoly power, this will imply (b) that, when demand for a "product" is geographically unstable, transportation
costs are significant, explicit price agreements illegal, and the
number of producers of the product relatively small, they will
tend to establish basing-point pricing systems.17 The assertion
(a) is regarded as an assumption and (b) as an implication because we accept the prediction of market behavior as the purpose
of the analysis. We shall regard the assumption as acceptable if
we find that the conditions specified in (b) are generally associated with basing-point pricing, and conversely. Let us now
change our purpose to deciding what cases to prosecute under
the Sherman Antitrust Law's prohibition of a "conspiracy in restraint of trade." If we now assume (c) that basing-point pricing
is a deliberate construction to facilitate collusion under the conditions specified in (b), this will imply (d) that entrepreneurs
who participate in basing-point pricing are engaged in a "conspiracy in restraint of trade." What was formerly an assumption
now becomes an implication, and conversely. We shall now regard the assumption (c) as valid if we find that, when entrepreneurs participate in basing-point pricing, there generally
tends to be other evidence, in the form of letters, memorandums,
or the like, of what courts regard as a "conspiracy in restraint
of trade."
Suppose the hypothesis works for the first purpose, namely,
the prediction of market behavior. It clearly does not follow that
it will work for the second purpose, namely, predicting whether
there is enough evidence of a "conspiracy in restraint of trade"
17. See George J. Stigler, "A Theory of Delivered Price Systems," American
Economic Review, XXXIX (December, 1949), 1143-57.
28
to justify court action. And, conversely, if it works for the

second purpose, it does not follow that it will work for the first.
Yet, in the absence of other evidence, the success of the hypothesis for one purposein explaining one class of phenomena
will give us greater confidence than we would otherwise have
that it may succeed for another purposein explaining another
class of phenomena. It is much harder to say how much greater
confidence it justifies. For this depends on how closely related
we judge the two classes of phenomena to be, which itself depends in a complex way on similar kinds of indirect evidence,
that is, on our experience in other connections in explaining by
single theories phenomena that are in some sense similarly
diverse.
To state the point more generally, what are called the assumptions of a hypothesis can be used to get some indirect evidence on the acceptability of the hypothesis in so far as the
assumptions can themselves be regarded as implications of the
hypothesis, and hence their conformity with reality as a failure
of some implications to be contradicted, or in so far as the assumptions may call to mind other implications of the hypothesis
susceptible to casual empirical observation.18 The reason this
evidence is indirect is that the assumptions or associated implications generally refer to a class of phenomena different from the
class which the hypothesis is designed to explain; indeed, as is
implied above, this seems to be the chief criterion we use in deciding which statements to term "assumptions" and which to
term "implications." The weight attached to this indirect evidence depends on how closely related we judge the two classes
of phenomena to be.
Another way in which the "assumptions" of a hypothesis can
facilitate its indirect testing is by bringing out its kinship with
other hypotheses and thereby making the evidence on their
validity relevant to the validity of the hypothesis in question.
For example, a hypothesis is formulated for a particular class
18. See Friedman and Savage, "The Expected-Utility Hypothesis and the
Measurability of Utility," op. cit., pp. 466-67, for another specific example of
this kind of indirect test.
29
of behavior. This hypothesis can, as usual, be stated without

specifying any "assumptions." But suppose it can be shown that
it is equivalent to a set of assumptions including the assumption
that man seeks his own interest. The hypothesis then gains indirect plausibility from the success for other classes of phenomena
of hypotheses that can also be said to make this assumption; at
least, what is being done here is not completely unprecedented
or unsuccessful in all other uses. In effect, the statement of assumptions so as to bring out a relationship between superficially
different hypotheses is a step in the direction of a more general
hypothesis.
This kind of indirect evidence from related hypotheses explains in large measure the difference in the confidence attached
to a particular hypothesis by people with different backgrounds.
Consider, for example, the hypothesis that the extent of racial
or religious discrimination in employment in a particular area or
industry is closely related to the degree of monopoly in the
industry or area in question; that, if the industry is competitive,
discrimination will be significant only if the race or religion of
employees affects either the willingness of other employees to
work with them or the acceptability of the product to customers
and will be uncorrelated with the prejudices of employers.19 This
hypothesis is far more likely to appeal to an economist than to
a sociologist. It can be said to "assume" single-minded pursuit
of pecuniary self-interest by employers in competitive industries;
and this "assumption" works well in a wide variety of hypotheses
in economics bearing on many of the mass phenomena with which
economics deals. It is therefore likely to seem reasonable to the
economist that it may work in this case as well. On the other
hand, the hypotheses to which the sociologist is accustomed have
a very different kind of model or ideal world, in which singleminded pursuit of pecuniary self-interest plays a much less important role. The indirect evidence available to the sociologist on
19. A rigorous statement of this hypothesis would of course have to specify
how "extent of racial or religious discrimination" and "degree of monopoly"
are to be judged. The loose statement in the text is sufficient, however, for
present purposes.
30
this hypothesis is much less favorable to it than the indirect

evidence available to the economist; he is therefore likely to view
it with greater suspicion.
Of course, neither the evidence of the economist nor that of
the sociologist is conclusive. The decisive test is whether the
hypothesis works for the phenomena it purports to explain. But
a judgment may be required before any satisfactory test of
this kind has been made, and, perhaps, when it cannot be made
in the near future, in which case, the judgment will have to be
based on the inadequate evidence available. In addition, even
when such a test can be made, the background of the scientists
is not irrelevant to the judgments they reach. There is never
certainty in science, and the weight of evidence for or against
a hypothesis can never be assessed completely "objectively."
The economist will be more tolerant than the sociologist in judging conformity of the implications of the hypothesis with experience, and he will be persuaded to accept the hypothesis tentatively by fewer instances of "conformity."
V. SOME IMPLICATIONS FOR ECONOMIC ISSUES
The abstract methodological issues we have been discussing

have a direct bearing on the perennial criticism of "orthodox"
economic theory as "unrealistic" as well as on the attempts that
have been made to reformulate theory to meet this charge.
Economics is a "dismal" science because it assumes man to
be selfish and money-grubbing, "a lightning calculator of pleasures and pains, who oscillates like a homogeneous globule of
desire of happiness under the impulse of stimuli that shift him
about the area, but leave him intact"; 20 it rests on outmoded
psychology and must be reconstructed in line with each new development in psychology; it assumes men, or at least businessmen, to be "in a continuous state of 'alert,' ready to change
prices and/or pricing rules whenever their sensitive intuitions
. . . detect a change in demand and supply conditions"; 21 it
20. Thorstein Veblen, "Why Is Economics Not an Evolutionary Science?"
(1898), reprinted in The Place of Science in Modern Civilization (New York,
1919), p. 73.
21. Oliver, op. cit., p. 381.
31
assumes markets to be perfect, competition to be pure, and

commodities, labor, and capital to be homogeneous.
As we have seen, criticism of this type is largely beside the
point unless supplemented by evidence that a hypothesis differing in one or another of these respects from the theory being
criticized yields better predictions for as wide a range of phenomena. Yet most such criticism is not so supplemented; it is
based almost entirely on supposedly directly perceived discrepancies between the "assumptions" and the "real world." A particularly clear example is furnished by the recent criticisms of
the maximization-of-returns hypothesis on the grounds that
businessmen do not and indeed cannot behave as the theory "assumes" they do. The evidence cited to support this assertion is
generally taken either from the answers given by businessmen
to questions about the factors affecting their decisionsa procedure for testing economic theories that is about on a par with
testing theories of longevity by asking octogenarians how they
account for their long lifeor from descriptive studies of the
decision-making activities of individual firms.22 Little if any evidence is ever cited on the conformity of businessmen's actual
market behaviorwhat they do rather than what they say they
dowith the implications of the hypothesis being criticized, on
the one hand, and of an alternative hypothesis, on the other.
22. See H. D. Henderson, "The Significance of the Rate of Interest," Oxford
Economic Papers, No. 1 (October, 1938), pp. 1-13; J. E. Meade and P. W. S.
Andrews, "Summary of Replies to Questions on Effects of Interest Rates," Oxford
Economic Papers, No. 1 (October, 1938), pp. 14-31; R. F. Harrod, "Price and Cost
in Entrepreneurs' Policy," Oxford Economic Papers, No. 2 (May, 1939), pp. 1-11;
and R. J. Hall and C. J. Hitch, "Price Theory and Business Behavior," Oxford
Economic Papers, No. 2 (May, 1939), pp. 12-45; Lester, "Shortcomings of Marginal Analysis for Wage-Employment Problems," op. cit.', Gordon, op. cit. See Fritz
Machlup, "Marginal Analysis and Empirical Research," op. cit., esp. Sec. II, for
detailed criticisms of questionnaire methods.
I do not mean to imply that questionnaire studies of businessmen's or others'
motives or beliefs about the forces affecting their behavior are useless for all
purposes in economics. They may be extremely valuable in suggesting leads to
follow in accounting for divergencies between predicted and observed results;
that is, in constructing new hypotheses or revising old ones. Whatever their
suggestive value in this respect, they seem to me almost entirely useless as a
means of testing the validity of economic hypotheses. See my comment on
Albert G. Hart's paper, "Liquidity and Uncertainty," American Economic Review, XXXIX (May, 1949), 198-99.
32
A theory or its "assumptions" cannot possibly be thoroughly

"realistic" in the immediate descriptive sense so often assigned
to this term. A completely "realistic" theory of the wheat market
would have to include not only the conditions directly underlying the supply and demand for wheat but also the kind of
coins or credit instruments used to make exchanges; the personal
characteristics of wheat-traders such as the color of each trader's
hair and eyes, his antecedents and education, the number of
members of his family, their characteristics, antecedents, and
education, etc.; the kind of soil on which the wheat was grown,
its physical and chemical characteristics, the weather prevailing
during the growing season; the personal characteristics of the
farmers growing the wheat and of the consumers who will ultimately use it; and so on indefinitely. Any attempt to move very
far in achieving this kind of "realism" is certain to render a
theory utterly useless.
Of course, the notion of a completely realistic theory is in
part a straw man. No critic of a theory would accept this logical
extreme as his objective; he would say that the "assumptions"
of the theory being criticized were "too" unrealistic and that
his objective was a set of assumptions that were "more" realistic
though still not completely and slavishly so. But so long as the
test of "realism" is the directly perceived descriptive accuracy
of the "assumptions"for example, the observation that "businessmen do not appear to be either as avaricious or as dynamic
or as logical as marginal theory portrays them"23 or that "it
would be utterly impractical under present conditions for the
manager of a multi-process plant to attempt . . . to work out
and equate marginal costs and marginal revenues for each productive factor"24there is no basis for making such a distinction, that is, for stopping short of the straw man depicted in the
preceding paragraph. What is the criterion by which to judge
whether a particular departure from realism is or is not acceptable? Why is it more "unrealistic" in analyzing business behavior to neglect the magnitude of businessmen's costs than the
23. Oliver, op. cit., p. 382.
24. Lester, "Shortcomings of Marginal Analysis for Wage-Employment Problems," op. dt., p. 75.
33
color of their eyes? The obvious answer is because the first

makes more difference to business behavior than the second;
but there is no way of knowing that this is so simply by observing
that businessmen do have costs of different magnitudes and eyes
of different color. Clearly it can only be known by comparing
the effect on the discrepancy between actual and predicted behavior of taking the one factor or the other into account. Even
the most extreme proponents of realistic assumptions are thus
necessarily driven to reject their own criterion and to accept the
test by prediction when they classify alternative assumptions
as more or less realistic.25
The basic confusion between descriptive accuracy and analytical relevance that underlies most criticisms of economic theory
on the grounds that its assumptions are unrealistic as well as
the plausibility of the views that lead to this confusion are both
strikingly illustrated by a seemingly innocuous remark in an
article on business-cycle theory that "economic phenomena are
varied and complex, so any comprehensive theory of the business
cycle that can apply closely to reality must be very complicated."26 A fundamental hypothesis of science is that appearances
are deceptive and that there is a way of looking at or interpreting
or organizing the evidence that will reveal superficially disconnected and diverse phenomena to be manifestations of a more
fundamental and relatively simple structure. And the test of this
hypothesis, as of any other, is its fruitsa test that science has
25. E.g., Gordon's direct examination of the "assumptions" leads him to
formulate the alternative hypothesis generally favored by the critics of the
maximization-of-returns hypothesis as follows: "There is an irresistible tendency to price on the basis of average total costs for some 'normal' level of
output. This is the yardstick, the short-cut, that businessmen and accountants
use, and their aim is more to earn satisfactory profits and play safe than to
maximize profits" (op. tit., p. 275). Yet he essentially abandons this hypothesis,
or converts it into a tautology, and in the process implicitly accepts the test
by prediction when he later remarks: "Full cost and satisfactory profits may
continue to be the objectives even when total costs are shaded to meet competition or exceeded to take advantage of a sellers' market" (ibid., p. 284).
Where here is the "irresistible tendency"? What kind of evidence could contradict this assertion?
26. Sidney S. Alexander, "Issues of Business Cycle Theory Raised by Mr.
Hicks," American Economic Review, XLI (December, 1951), 872.
34
so far met with dramatic success. If a class of "economic phenomena" appears varied and complex, it is, we must suppose, because we have no adequate theory to explain them. Known facts
cannot be set on one side; a theory to apply "closely to reality,"
on the other. A theory is the way we perceive "facts," and we
cannot perceive "facts" without a theory. Any assertion that
economic phenomena are varied and complex denies the tentative
state of knowledge that alone makes scientific activity meaningful; it is in a class with John Stuart Mill's justly ridiculed statement that "happily, there is nothing in the laws of value which
remains [1848] for the present or any future writer to clear up;
the theory of the subject is complete."27
The confusion between descriptive accuracy and analytical
relevance has led not only to criticisms of economic theory on
largely irrelevant grounds but also to misunderstanding of
economic theory and misdirection of efforts to repair supposed defects. "Ideal types" in the abstract model developed by economic
theorists have been regarded as strictly descriptive categories
intended to correspond directly and fully to entities in the real
world independently of the purpose for which the model is being
used. The obvious discrepancies have led to necessarily unsuccessful attempts to construct theories on the basis of categories
intended to be fully descriptive.
This tendency is perhaps most clearly illustrated by the interpretation given to the concepts of "perfect competition" and
"monopoly" and the development of the theory of "monopolistic"
or "imperfect competition." Marshall, it is said, assumed "perfect competition"; perhaps there once was such a thing. But
clearly there is no longer, and we must therefore discard his
theories. The reader will search long and hardand I predict
unsuccessfullyto find in Marshall any explicit assumption about
perfect competition or any assertion that in a descriptive sense
the world is composed of atomistic firms engaged in perfect
competition. Rather, he will find Marshall saying: "At one
extreme are world markets in which competition acts directly
from all parts of the globe; and at the other those secluded
27. Principles of Political Economy
p. 436.
(Ashley ed.; Longmans, Green & Co., 1929),
35
markets in which all direct competition from afar is shut out,

though indirect and transmitted competition may make itself felt
even in these; and about midway between these extremes lie the
great majority of the markets which the economist and the business man have to study.'728 Marshall took the world as it is;
he sought to construct an "engine" to analyze it, not a photographic reproduction of it.
In analyzing the world as it is, Marshall constructed the
hypothesis that, for many problems, firms could be grouped into
"industries" such that the similarities among the firms in each
group were more important than the differences among them.
These are problems in which the important element is that a
group of firms is affected alike by some stimulusa common
change in the demand for their products, say, or in the supply
of factors. But this will not do for all problems: the important
element for these may be the differential effect on particular firms.
The abstract model corresponding to this hypothesis contains
two "ideal" types of firms: atomistically competitive firms,
grouped into industries, and monopolistic firms. A firm is competitive if the demand curve for its output is infinitely elastic
with respect to its own price for some price and all outputs, given
the prices charged by all other firms; it belongs to an "industry"
defined as a group of firms producing a single "product." A
"product" is defined as a collection ef units that are perfect
substitutes to purchasers so the elasticity of demand for the
output of one firm with respect to the price of another firm in
the same industry is infinite for some price and some outputs. A
firm is monopolistic if the demand curve for its output is not
infinitely elastic at some price for all outputs.29 If it is a monopolist, the firm is the industry.30
As always, the hypothesis as a whole consists not only of this
abstract model and its ideal types but also of a set of rules, mostly
28. Principles, p. 329; see also pp. 35, 100, 341, 347, 375, 546.
29. This ideal type can be divided into two types: the oligopolistic firm, if
the demand curve for its output is infinitely elastic at some price for some
but not all outputs; the monopolistic firm proper, if the demand curve is
nowhere infinitely elastic (except possibly at an output of zero).
30. For the oligopolist of the preceding note an industry can be defined as a
group of firms producing the same product.
36
implicit and suggested by example, for identifying actual firms

with one or the other ideal type and for classifying firms into
industries. The ideal types are not intended to be descriptive;
they are designed to isolate the features that are crucial for
a particular problem. Even if we could estimate directly and
accurately the demand curve for a firm's product, we could not
proceed immediately to classify the firm as perfectly competitive
or monopolistic according as the elasticity of the demand curve
is or is not infinite. No observed demand curve will ever be
precisely horizontal, so the estimated elasticity will always be
finite. The relevant question always is whether the elasticity is
"sufficiently" large to be regarded as infinite, but this is a question that cannot be answered, once for all, simply in terms of
the numerical value of the elasticity itself, any more than we
can say, once for all, whether an air pressure of 15 pounds per
square inch is "sufficiently" close to zero to use the formula
s = igt2. Similarly, we cannot compute cross-elasticities of demand and then classify firms into industries according as there
is a "substantial gap in the cross-elasticities of demand." As
Marshall says, "The question where the lines of division between different commodities [i.e., industries] should be drawn
must be settled by convenience of the particular discussion."31
Everything depends on the problem; there is no inconsistency in
regarding the same firm as if it were a perfect competitor for
one problem, and a monopolist for another, just as there is none
in regarding the same chalk mark as a Euclidean line for one
problem, a Euclidean surface for a second, and a Euclidean
solid for a third. The size of the elasticity and cross-elasticity of
demand, the number of firms producing physically similar products, etc., are all relevant because they are or may be among the
variables used to define the correspondence between the ideal and
real entities in a particular problem and to specify the circumstances under which the theory holds sufficiently well; but they
do not provide, once for all, a classification of firms as competitive or monopolistic.
An example may help to clarify this point. Suppose the problem is to determine the effect on retail prices of cigarettes of an
31. Principles, p. 100.
37
increase, expected to be permanent, in the federal cigarette tax.

I venture to predict that broadly correct results will be obtained
by treating cigarette firms as if they were producing an identical
product and were in perfect competition. Of course, in such a
case, "some convention must be made as to the" number of
Chesterfield cigarettes "which are taken as equivalent" to a
Marlborough.32
On the other hand, the hypothesis that cigarette firms would
behave as if they were perfectly competitive would have been
a false guide to their reactions to price control in World War II,
and this would doubtless have been recognized before the event.
Costs of the cigarette firms must have risen during the war.
Under such circumstances perfect competitors would have reduced the quantity offered for sale at the previously existing
price. But, at that price, the wartime rise in the income of the
public presumably increased the quantity demanded. Under conditions of perfect competition strict adherence to the legal price
would therefore imply not only a "shortage" in the sense that
quantity demanded exceeded quantity supplied but also an absolute decline in the number of cigarettes produced. The facts contradict this particular implication: there was reasonably good adherence to maximum cigarette prices, yet the quantities produced
increased substantially. The common force of increased costs presumably operated less strongly than the disruptive force of the
desire by each firm to keep its share of the market, to maintain
the value and prestige of its brand name, especially when the
excess-profits tax shifted a large share of the costs of this kind
of advertising to the government. For this problem the cigarette
firms cannot be treated as if they were perfect competitors.
Wheat farming is frequently taken to exemplify perfect competition. Yet, while for some problems it is appropriate to treat
cigarette producers as if they comprised a perfectly competitive
industry, for some it is not appropriate to treat wheat producers
as if they did. For example, it may not be if the problem is the
differential in prices paid by local elevator operators for wheat.
Marshall's apparatus turned out to be most useful for problems in which a group of firms is affected by common stimuli,
32. Quoted parts from ibid.
38
and in which the firms can be treated as if they were perfect

competitors. This is the source of the misconception that Marshall "assumed" perfect competition in some descriptive sense.
It would be highly desirable to have a more general theory than
Marshall's, one that would cover at the same time both those
cases in which differentiation of product or fewness of numbers
makes an essential difference and those in which it does not. Such
a theory would enable us to handle problems we now cannot and,
in addition, facilitate determination of the range of circumstances
under which the simpler theory can be regarded as a good enough
approximation. To perform this function, the more general theory
must have content and substance; it must have implications
susceptible to empirical contradiction and of substantive interest
and importance.
The theory of imperfect or monopolistic competition developed
by Chamberlin and Robinson is an attempt to construct such
a more general theory.33 Unfortunately, it possesses none of the
attributes that would make it a truly useful general theory. Its
contribution has been limited largely to improving the exposition
of the economics of the individual firm and thereby the derivation
of implications of the Marshallian model, refining Marshall's
monopoly analysis, and enriching the vocabulary available for
describing industrial experience.
The deficiencies of the theory are revealed most clearly in
its treatment of, or inability to treat, problems involving groups
of firmsMarshallian "industries." So long as it is insisted that
differentiation of product is essentialand it is the distinguishing
feature of the theory that it does insist on this pointthe definition of an industry in terms of firms producing an identical product cannot be used. By that definition each firm is a separate
industry. Definition in terms of "close" substitutes or a "substantial" gap in cross-elasticities evades the issue, introduces fuzziness and undefinable terms into the abstract model where they
have no place, and serves only to make the theory analytically
meaningless"close" and "substantial" are in the same category
33. E. H. Chamberlin, The Theory of Monopolistic Competition (6th ed.;
Cambridge: Harvard University Press, 1950); Joan Robinson, The Economics
of Imperfect Competition (London: Macmillan & Co., 1933).
39
34
as a "small" air pressure. In one connection Chamberlin implicitly defines an industry as a group of firms having identical cost
and demand curves.35 But this, too, is logically meaningless so
long as differentiation of product is, as claimed, essential and not
to be put aside. What does it mean to say that the cost and demand curves of a firm producing bulldozers are identical with
those of a firm producing hairpins?36 And if it is meaningless for
bulldozers and hairpins, it is meaningless also for two brands
of toothpasteso long as it is insisted that the difference between
the two brands is fundamentally important.
The theory of monopolistic competition offers no tools for the
analysis of an industry and so no stopping place between the
firm at one extreme and general equilibrium at the other.37 It is
therefore incompetent to contribute to the analysis of a host of
important problems: the one extreme is too narrow to be of
great interest; the other, too broad to permit meaningful generalizations.38
VI. CONCLUSION
Economics as a positive science is a body of tentatively accepted generalizations about economic phenomena that can be
used to predict the consequences of changes in circumstances.
34. See R. L. Bishop, "Elasticities, Cross-elasticities, and Market Relationships," American Economic Review, XLII (December, 1952), 779-803, for a
recent attempt to construct a rigorous classification of market relationships
along these lines. Despite its ingenuity and sophistication, the result seems to
me thoroughly unsatisfactory. It rests basically on certain numbers being
classified as "large" or "small," yet there is no discussion at all of how to
decide whether a particular number is "large" or "small," as of course there
cannot be on a purely abstract level.
35. Op. cit., p. 82.
36. There always exists a transformation of quantities that will make either
the cost curves or the demand curves identical; this transformation need not,
however, be linear, in which case it will involve different-sized units of one
product at different levels of output. There does not necessarily exist a transformation that will make both pairs of curves identical.
37. See Robert Triffin, Monopolistic Competition and General Equilibrium Theory
(Cambridge: Harvard University Press, 1940), esp. pp. 188-89.
38. For a detailed critique see George J. Stigler, "Monopolistic Competition
in Retrospect," in Five Lectures on Economic Problems (London: Macmillan & Co.,
1949), pp. 12-24.
40
Progress in expanding this body of generalizations, strengthening

our confidence in their validity, and improving the accuracy of
the predictions they yield is hindered not only by the limitations
of human ability that impede all search for knowledge but also
by obstacles that are especially important for the social sciences
in general and economics in particular, though by no means peculiar to them. Familiarity with the subject matter of economics
breeds contempt for special knowledge about it. The importance
of its subject matter to everyday life and to major issues of
public policy impedes objectivity and promotes confusion between scientific analysis and normative judgment. The necessity
of relying on uncontrolled experience rather than on controlled
experiment makes it difficult to produce dramatic and clear-cut
evidence to justify the acceptance of tentative hypotheses. Reliance on uncontrolled experience does not affect the fundamental methodological principle that a hypothesis can be tested
only by the conformity of its implications or predictions with
observable phenomena; but it does render the task of testing
hypotheses more difficult and gives greater scope for confusion
about the methodological principles involved. More than other
scientists, social scientists need to be self-conscious about their
methodology.
One confusion that has been particularly rife and has done
much damage is confusion about the role of "assumptions" in
economic analysis. A meaningful scientific hypothesis or theory
typically asserts that certain forces are, and other forces are not,
important in understanding a particular class of phenomena.
It is frequently convenient to present such a hypothesis by
stating that the phenomena it is desired to predict behave in
the world of observation as if they occurred in a hypothetical and
highly simplified world containing only the forces that the
hypothesis asserts to be important. In general, there is more
than one way to formulate such a descriptionmore than one
set of "assumptions" in terms of which the theory can be presented. The choice among such alternative assumptions is made
on the grounds of the resulting economy, clarity, and precision
in presenting the hypothesis; their capacity to bring indirect
evidence to bear on the validity of the hypothesis by suggesting
41
some of its implications that can be readily checked with observation or by bringing out its connection with other hypotheses
dealing with related phenomena; and similar considerations.
Such a theory cannot be tested by comparing its "assumptions"
directly with "reality." Indeed, there is no meaningful way in
which this can be done. Complete "realism" is clearly unattainable, and the question whether a theory is realistic "enough"
can be settled only by seeing whether it yields predictions that
are good enough for the purpose in hand or that are better than
predictions from alternative theories. Yet the belief that a theory
can be tested by the realism of its assumptions independently of
the accuracy of its predictions is widespread and the source of
much of the perennial criticism of economic theory as unrealistic.
Such criticism is largely irrelevant, and, in consequence, most
attempts to reform economic theory that it has stimulated have
been unsuccessful.
The irrelevance of so much criticism of economic theory does
not of course imply that existing economic theory deserves any
high degree of confidence. These criticisms may miss the target,
yet there may be a target for criticism. In a trivial sense, of
course, there obviously is. Any theory is necessarily provisional
and subject to change with the advance of knowledge. To go
beyond this platitude, it is necessary to be more specific about
the content of "existing economic theory" and to distinguish
among its different branches; some parts of economic theory
clearly deserve more confidence than others. A comprehensive
evaluation of the present state of positive economics, summary
of the evidence bearing on its validity, and assessment of the
relative confidence that each part deserves is clearly a task for
a treatise or a set of treatises, if it be possible at all, not for a
brief paper on methodology.
About all that is possible here is the cursory expression of a
personal view. Existing relative price theory, which is designed
to explain the allocation of resources among alternative ends
and the division of the product among the co-operating resources
and which reached almost its present form in Marshall's Principles of Economics, seems to me both extremely fruitful and
deserving of much confidence for the kind of economic system
42
that characterizes Western nations. Despite the appearance of

considerable controversy, this is true equally of existing static
monetary theory, which is designed to explain the structural or
secular level of absolute prices, aggregate output, and other
variables for the economy as a whole and which has had a form
of the quantity theory of money as its basic core in all of its
major variants from David Hume to the Cambridge School to
Irving Fisher to John Maynard Keynes. The weakest and least
satisfactory part of current economic theory seems to me to be in
the field of monetary dynamics, which is concerned with the process of adaptation of the economy as a whole to changes in conditions and so with short-period fluctuations in aggregate activity.
In this field we do not even have a theory that can appropriately
be called "the" existing theory of monetary dynamics.
Of course, even in relative price and static monetary theory
there is enormous room for extending the scope and improving
the accuracy of existing theory. In particular, undue emphasis
on the descriptive realism of "assumptions" has contributed to
neglect of the critical problem of determining the limits of
validity of the various hypotheses that together constitute the
existing economic theory in these areas. The abstract models
corresponding to these hypotheses have been elaborated in considerable detail and greatly improved in rigor and precision.
Descriptive material on the characteristics of our economic system and its operations have been amassed on an unprecedented
scale. This is all to the good. But, if we are to use effectively
these abstract models and this descriptive material, we must
have a comparable exploration of the criteria for determining
what abstract model it is best to use for particular kinds of
problems, what entities in the abstract model are to be identified
with what observable entities, and what features of the problem or of the circumstances have the greatest effect on the
accuracy of the predictions yielded by a particular model or
theory.
Progress in positive economics will require not only the testing
and elaboration of existing hypotheses but also the construction
of new hypotheses. On this problem there is little to say on a

43
formal level. The construction of hypotheses is a creative act
of inspiration, intuition, invention; its essence is the vision of
something new in familiar material. The process must be discussed in psychological, not logical, categories; studied in autobiographies and biographies, not treatises on scientific method;
and promoted by maxim and example, not syllogism or theorem.
Part 2
Reading and writing a classic
Reading the methodological essay in

twentieth-century economics:
map of multiple perspectives
Uskali Mki
Even outrageously unrealistic assumptions are just fine insofar as the theory
or model involving them performs well in predicting phenomena of interest. Most economists and many non-economists will attribute this principle
to Milton Friedman. Many will consider the principle itself outrageous,
while others praise Friedman for having formulated it so persuasively.
Friedmans The methodology of positive economics was published
in 1953 as the opening chapter of his Essays in Positive Economics. There is
no doubt that this short essay of forty pages became the most cited, the most
influential, and the most controversial piece of methodological writing in
twentieth-century economics. The power and diversity of its impact and
reception are evident. The essay (henceforth F53) has helped shape
many economists conceptions of how to do good economics and how to
defend certain ways of doing economics. It has also provoked many others
to voice their disagreement with the claims of the essay, to argue that
the view of science suggested by F53 is deeply flawed, even dangerous
for the cognitive aspirations and social responsibilities of economics.
These disagreements have not diminished in the course of the years
since the publication of F53. Whenever someone makes reference to
Friedmans methodological view (perhaps using other expressions
such as Friedmans as-if methodology or Friedmans instrumentalism), the context typically reveals that the expression is used either
approvingly or disapprovingly, and often uncompromisingly so.
The legacy of F53 has led a double life. In the last half a century, F53
has prompted numerous scholarly commentaries and criticisms in an
attempt to understand and (mostly) to resist its message. But the influence
of F53 evidently reaches far beyond such explicit commentaries. In fact its
effective impact may have flown along its own separate ways, uninfluenced by the various criticisms that have been leveled against the
essay. There has been a popular legacy of F53, based on a vague understanding of its key ideas, sustained and transmitted without any detailed
analysis or reflection on the text of the essay, often without citation or
47
48
Uskali Mki
reference. The popular legacy manifests itself in introductory chapters of

economics textbooks in which it is explained to students why the assumptions of economic models do not need to be realistic, and in oral communication at seminars and conference sessions when reference is made to
as-if methodology or the importance of predictive implications in testing models. Parallel to this popular legacy, and surely motivated by it,
there has been a tradition of more or less systematic F53 studies aiming
at detailed critical scrutiny of the claims and arguments of the essay. This
has constituted what may be called the reflective legacy of F53. Numerous
studies have been published in journals and books, critically analyzing and
interpreting and explaining F53 from a variety of angles, focusing on its
different aspects, and drawing conclusions that rival or complement one
another. The present volume adds many important and novel insights and
analyses to this body of reflective literature. It is hoped that they will have
consequences for the contents of the popular legacy.
In both of these incarnations of its legacy, the popular and the reflective,
F53 has been highly influential. It has shaped the perceptions of the
economist-of-the-street, and it has provoked specialists in economic
methodology to engage in serious scholarly studies in hermeneutic interpretation and analytical criticism. I should add the important observation
that its influence is not restricted to the economics profession: F53 has
shaped the image of economics also amongst other social scientists. One
often comes across statements about the nature of economic reasoning
that make reference to F53, thereby suggesting that F53 gives a relatively
accurate picture of how economics is being done by economists. In most
such situations, the image is not intended as a flattering one. Indeed,
many social scientists other than economists take it as evident that a
discipline dependent on working with highly unrealistic assumptions is
of dubious scientific value, thus a methodology seeking to justify such a
practice must be at least equally suspect as a theory of science. To the
extent that the broader public image of economics derives from F53, it is
usually not one of praise and celebration.
Yet another very important motivating observation is that F53 seems to
have been poorly understood. One might be surprised by this claim: given its
strong impact, and given the large number of reflective commentaries that
have appeared, one would expect F53 to be well understood by now. But
this is surely not the case. On the one hand, strong impact, as in the
popular legacy, is perfectly compatible with a poor understanding of what
exactly makes that impact. Even more embarrassingly, poor understanding sometimes helps a piece of writing to make a popular impact. On the
other hand, even though one might have higher expectations in regard to
the explicitly reflective commentaries, they too may have failed to generate
Reading the methodological essay in twentieth-century economics
49
sufficiently solid understanding of F53 for understandable reasons. Some

of these reasons derive from the fact that F53 is inherently very hard to
understand. This difficulty is not only due to its richness, but also to its
obscurities, ambiguities, and inconsistencies. One may even take these
characteristics to support the radical conclusion that there is no single
correct way of interpreting F53, rather there are multiple justifiable readings. Another possible reason for failure, or imperfect success, is the
familiar mismatch between ones interpretive concepts (employed by
readers) and ones target (the text). Any reflective commentary of a text
approaches its target equipped with conceptual tools and philosophical
frames, and these may not be appropriate for the task of getting the
meaning of the text right. No doubt F53 has been approached from angles
that are too limited or even plainly misguided to result in a solid understanding given the richness and complexity of F53 itself.
F53 was published more than half a century ago, but the interest in
understanding and assessing the essay is not at all just historical. Its key
claims remain as topical as ever: they are repeatedly being invoked by
economists, generation after generation. Nowadays, they have special
contemporary relevance in the context of the new debates around economic theory and method. In its own time, F53 was motivated by empirically based attacks against marginalist profit maximization. As Roger
Backhouse explains in his chapter below, F53 can only be understood
against the background of the so-called marginalist controversy in the
1940s. Together with Fritz Machlups methodological polemics
(Machlup 1946, 1955), F53 became a major defensive statement that
sought to undermine the empirical criticisms that were leveled against
marginalist maximization assumptions. Today, we are again witnessing
massive attacks against the standard assumptions of rationality, and there
are new attempts to replace them by more realistic assumptions. The
difference is that in the late 1930s and early 1940s, the critical strategy
was one based on field survey, simply asking business managers what they
were doing, while nowadays the critical conclusions derive from experiment and brain scan, with psychologists and neuroscientists contributing
to the collection of critical evidence. Even though F53 itself badly underestimated the role of experiment in economics, the topicality of its methodological arguments is evident.
F53 also set out to defend perfect competition against developments in
the theory of monopolistic competition that it described as an entirely
misguided pursuit of greater realisticness. As Oliver Williamsons chapter
below shows, post-F53 developments in theories of the firm and market
structure have not been obedient to these proscriptions; they have rather
taken the theories toward opening the black box of the business firm and
50
Uskali Mki
thereby toward more realistic images of motivation and procedure within

firms. Monopolistic competition has also recently made a comeback
packaged with increasing returns in explaining phenomena of growth
and agglomeration. The challenge is to assess these developments in
terms of F53 and to assess F53 in terms of these developments.
The topicality of F53 is also fortified by the chronic difficulties in
running reliable predictive tests in economics. The core principle
endorsed by F53 appears to require that there be a set of well-defined
and agreed-upon objective rules for measuring predictive success given
that predictive success is supposed to be all that matters in testing a theory.
F53 itself does not formulate such rules, and as we know, economists
constantly struggle with the consequences of not having such rules.
Naturally, there are established practices of prediction and testing, but
they are not and cannot be completely governed by explicit rules.
Testing by prediction is far from anything straightforward and uncontestable. Paradoxically perhaps, F53 itself supports this claim, as I point
out in chapter 3 below. Indeed, there is a line of thought in F53 that seems
to have been missed by most readers: predictive tests are not fully objective and reliable, they unavoidably involve subjective judgment and social
convention. Whether or not this makes F53 a self-undermining piece of
writing, it is obvious that it does imply a call for greater caution in
appealing to predictive tests, and for greater attention to the details of
the structure of actual predictive tests.
The above are among the features of F53 that taken together give
sufficient grounds for organizing a collective reconsideration of the essay
half a century after its publication. Contributions were commissioned
from scholars with complementary interests and competences to offer
insight into the conundrums of F53 from a variety of perspectives. The
assignments were designed with the goal of having a number of analyses
complementing one another, while avoiding excessive overlap between
the contributions. Each contribution would have a separate focus, and
together they would provide a comprehensive and rich but still incomplete picture of F53 in context. The contributors were asked to help a
twenty-first-century reader to have a reasoned opinion of F53, its contents, and their descriptive and prescriptive adequacy. Meeting this challenge would be based on examining the text of the essay, its historical and
intellectual context, and the developments of economics and its methodology after 1953.
Just like the various commentaries of F53 that have been published
during the last half a century, the contributions to this volume approach
F53 from different angles, addressing different questions and pursuing
the answers by following different strategies. In order to assess the various
51
contributions one has to understand what they seek to accomplish, what

their goals and constraints and other presuppositions are. It may therefore
be useful to outline a map of approaches to F53, providing the beginnings
of a typology of readings or interpretations. Such a map should help assess
the readings in their own terms and to compare their merits and
ambitions.
In the remainder of this chapter, I propose to organize the various ways
of reading and examining F53, each of them giving rise to different types
of project. I am thinking of a map that is on five dimensions. The first
dimension is related to the functions of reading F53, from taking it as a
target of critical examination to using it without examining it, for example
as a source of information or inspiration. The second dimension consists
of a range of internal characteristics of F53, from those of its elements that
focus on the logical aspects of economic theorizing to those that highlight
its ontological, semantic, and social aspects. The third dimension is constituted by a variety of external contexts of F53, from the context of its
writing to the context of its reading and reception. The fourth dimension
consists of perspectives that range from normatively assessing various
features of F53 to analyzing and explaining those features. The fifth
dimension is concerned with the sources of various labels attached to
F53, from extra-economics to intra-economics, and more.
I will expand on this map and will illustrate by locating various contributions to F53 studies that have appeared in the course of the last half
a century, including those published in the present volume.
1
Examining and using
The first dimension has to do with the foci and functions of reading F53,
ranging from examining it to using it. At one end, one may approach F53
as itself the primary target of study, examining and explaining it, analyzing
and scrutinizing it, contextualizing and criticizing it, and so on. At the
other end of the spectrum, the reader of F53 may take it as a source or
means of information and inspiration, of persuasive argument and authoritative judgment, of reproach and ridicule. Here, without examining F53
itself, one may quote or cite it, and then proceed to focus on other things,
such as the nature of economic models, contemporary debates in economic theory and method, issues around the public image of economics,
or challenges of economics education.
There is no dichotomy here, but rather a continuum of foci along this
dimension. For example, many reflective readers of F53 nowadays examine the text of the essay in order to use the findings as evidence for an
account of the methodological outlook of the author of F53 more broadly.
52
Uskali Mki
In this case, one examines F53 without drawing conclusions about F53
only, but may use the outcome of such investigations together with other
sources of evidence for portraying Milton Friedmans methodology of
doing economic research. This strategy appears in several studies, including the book on Friedman by Abraham Hirsch and Neil De Marchi
(1990), Daniel Hammonds and Thomas Mayers various works, as well
as the chapter below by Kevin Hoover.
This observation gives us an important distinction between examining
F53 in order to understand it, and examining F53 in order to understand
Friedmans economics and its methodology. It is important to be explicit
about the difference between the two goals of reading, given that they are
not always kept distinct. Both are legitimate tasks, but they require different standards to be appropriately assessed. For example, in considering an
interpretive attempt to understand F53 alone, one should not judge the
attempt straightforwardly in terms of its success or failure to provide
understanding of Friedmans economics. A good account of F53 may
not be a good account of Friedmans methodology, in some cases not even
part of it. Indeed, Friedmans methodology and the methodology of F53 are
not one and the same thing, even though the two are likely to be connected. That they are not the same thing is exemplified, among other
things, by incongruences between the dictates of F53 and Friedmans
statements and practices elsewhere, such as unrealisticness of assumptions considered irrelevant here, but relevant there. It is notable that prior
to the 1980s, virtually all commentary on F53 focused on F53 without any
interest in Friedmans methodology. The situation has since then
changed, perhaps due to the changes in Milton Friedmans status within
economics: next to straight F53 studies, we now also have studies of
Friedmans methodology.
Given that all the other dimensions on my map lean towards the
examination side of the first dimension, I will not say more about this
here, but will instead focus on the use side. The popular legacy of F53 is a
matter of use without examination. As an example in this category, textbook authors sometimes make reference to F53 when explaining to the
students why they should not worry about unrealistic assumptions.
In the present volume, all contributions are within the reflective legacy,
but in some of them the use component is more pronounced than the
examination component.
Melvin Reders chapter deals with the role of visions or general
frames in economic theorizing, offering an extensive discussion of
contemporary debates over involuntary unemployment and sticky
wages and other methodological issues concerned with what qualifies as
evidence in economics. In these parts of Reders chapter, F53 is not so
53
much examined itself as used for launching the broader discussion. Also
dealing with issues of evidence, the focus of Chris Starmers chapter is on
the articles on expected utility jointly authored by Friedman and Savage,
published in the neighborhood of F53, namely in 1948 and 1952.
Pointing out parallels and incongruences between the methodological
positions revealed in these articles and F53, Starmer raises critical questions about using normative appeal as indirect evidence in support of the
expected-utility hypothesis. Starmer puts his observations also in the
context of current debates over the assumptions of rationality.
The mixture of examination and use is similar in Oliver Williamsons
chapter, which discusses some key ideas of F53 in the context of the post1953 developments away from the neoclassical theory of the firm
defended by F53. These developments lead away from just one theory
toward many theories, and from the unrealistic image of theories as
production functions with profit maximization toward increasing realisticness in motivation, then in process. Jack Vromens chapter can be seen
as a combination of examination and use: tracing various pre-F53 and
post-F53 versions of the evolutionary argument as designed and used by
various economists, he manages to help us see the distinct characteristics
of the argument as presented in F53 itself while at the same time
informing us about the variety of such arguments available in economics.
2
Internal characteristics of F53
The second dimension on my map consists of a range of internal characteristics of F53 noticed by the reader, namely what F53 itself is taken to say
about economic theories and scientific reasoning, and how it says it. The
accounts of F53 here focus on interpreting and assessing its stated or
implied view of economics as a scientific discipline. The text of F53 serves
as the main source of evidence for these accounts, while the goal is to
identify and examine the claims and arguments of the essay. Given the
richness of the internal characteristics of F53, any reading is bound to be
somewhat selective among them some more, some less and there are
many ways of making ones selections.
At one end, readers may focus narrowly on just how F53 deals with
assumptions, predictive implications and data as well as their logical
relationships in economic theorizing. Most commentaries from the 1950s to
the 1970s were largely constrained by this perspective, from Rotwein
(1959) through Samuelson (1963) and Simon (1963) to Melitz (1965),
Bear and Orr (1967), Brunner (1969), Wong (1973), and Boland (1979).
F53 was read as stating that the truth of the assumptions of economic
theories is irrelevant, and their falsehood is just fine, thus it makes no sense
54
Uskali Mki
to try to test them against empirical data, while the assumptions entail
predictive implications that are to be checked against the data, and this
constitutes an appropriate test of the theory itself. Commentators then
critically examined the logic of this argument about the logic of testing
economic theory. Some of them recognized the ambiguity of assumption in F53 and elsewhere, and checked the consequences with alternative specifications of the meaning of the term. Among the premises
shared by many such narrowly logical accounts of F53 were the ideas that
unrealisticness (of an assumption, or model) amounts to falsehood; that
the falsehood of its assumptions amounts to the falsehood of a theory or
model; and that assumptions are considered mainly from the point of view
of their role in yielding predictions. Obviously, these premises enable
creating a very limited image of F53.
The portrayal of F53 can then be enriched by adding a mixed variety of
further characteristics to its description, including how F53 depicts the
ontological, semantic, and social aspects of scientific reasoning. As for the
ontological aspects, what does F53 suggest regarding what is presupposed
about the constitution of economic reality in economic theorizing? For
example, does the profit maximization assumption presuppose something
about the deliberations and purposeful activities of business managers or
owners, or rather about the real pressures of the selection mechanisms of
competitive markets? How is one to interpret Friedmans suggestion that
the assumption of profit maximization summarizes the conditions of firm
survival through natural selection in the competitive market? These questions have been discussed since Koopmans (1957) and Winter (1964).
Jack Vromens chapter below joins this tradition in his extended analysis
of Friedmans selection argument in support of the maximization assumption. Chris Starmer devotes a section to this sort of argument in his
chapter on expected utility. Kevin Hoovers discussion below on
Friedmans poorly articulated attitudes towards real-world causation is
another example of addressing ontological issues. Mere logical relations
between sentences in a model are insufficient for capturing real causal
connections.
Moreover, rather than just taking the truth-values of assumptions and
predictions as given, and then examining their logical relationships, we
may ask: what does F53 suggest regarding what it is for a theory and its
assumptions to be true, or fail to be so? What functions do true and false
assumptions serve in a model? How should we interpret the as-if formulation from the point of view of truth and falsehood, and what epistemic
purpose is being served? What other ways are there for a theory or model
to be realistic or unrealistic? These questions require attention to the
precise claims about the world that theories and their components are
55
used for making. For example, an assumption may appear false when
formulated as a straightforward assertion about the world (the air pressure
is nil, the market is perfect), but may have a chance of being true when
rephrased as a claim about the negligibly small effect of some factor (the
air pressure is negligibly small, the deviation from market perfection is
negligible). This is a theme that was not perfectly lucid in F53, but was
later introduced by Alan Musgrave (1981) and followed up in Mki
(2000). There is another important idea that was not explicitly developed
by F53, but on generous interpretation may be viewed as compatible with
it. This is the idea that a model may appear false due to its narrowness and
its containment of false assumptions, yet it may be used for making true
claims about some significant facts about the world (Mki 1992). This
idea is further developed in my second chapter to this volume, as are the
semantics and ontology of the as-if locution. Furthermore, next to truth
and falsehood, there are other kinds of realisticness and unrealisticness
that play important implicit roles in F53s reasoning, such as those related
to observationality and partiality.
As to the social aspect of economic theorizing, what does F53 say or
imply about the ways in which theorizing is shaped by various social
factors within academia and outside it? These factors range from the
enabling and constraining roles of academic communities and conventions to the pressures from economic policy and ideology. This has been a
strikingly neglected theme in the legacy of F53, both popular and reflective. The chapter by Teira and Zamora in this volume looks at one aspect
of F53 from the point of view of a contract between the agents (economists) and principals (consumers) of economic knowledge. My own
work has stressed the central role of social shaping of economics in the
reasoning of F53 and combines this with the rest of the spectrum, namely
the logical, ontological, epistemological, and semantic perspectives (e.g.
Mki 1986, 1992, 2000).
3
External contexts of F53
My map has another important dimension along which readings and

commentaries of F53 are located. This third dimension consists of the
range of external contexts of F53. External here means external to F53,
ranging from academic or intellectual contexts (such as the situation
within economics and the philosophy of science) to political and broader
cultural contexts in society at large and both of these at the more
particular level of Friedmans individual biography and at the more general level of academic and other social institutions. One may discuss F53
as a response to developments in economics and the philosophy of
56
Uskali Mki
science, and one may examine its later impact on these developments.
Most of the systematic commentary on F53, when paying attention to
external contexts at all, has thus far focused on such more narrowly
academic contexts. This is also the case in the present volume, even
though observations about the economist expert (by Teira and Zamora)
and free market ideology (by Reder) reach beyond the narrow academic
horizons.
Let us make the discussion on external context more focused by drawing a distinction between the context of production and the context of
consumption. At one end, the focus is put on the context of writing: the
intellectual challenges and inspirations that provoked and shaped the production of F53 such as the marginalist controversy triggering the textual
production of F53, and the Marshallian tradition guiding it, as well as
Friedmans own work as an economist and as a policy adviser. A prime
example of these accounts is the book on Friedmans economics by
Abraham Hirsch and Neil De Marchi (1990). Among other things, they
suggest ways in which Wesley Mitchells methodological views may have
influenced those of Friedman while at the NBER, and they point out what
they claim to be close affinities between Friedmans methodology and
John Deweys pragmatist account of inquiry. Daniel Hammonds extensive work on Friedmans methodology and its origins is another major
example in this category (e.g. Hammond 1996; on the idea of contexts,
see Hammond 1992b).
These authors have been driven by the ambition of identifying
Friedmans authentic convictions in economic methodology. They
believe the internal characteristics of F53 alone do not suffice as evidence
for those convictions, and that only when combined with lots of other
evidence related to Friedmans work in economics and his broader intellectual background can reliable conclusions be drawn about Friedmans
real methodological beliefs. Hammonds contribution to the present volume, tracing the development of the early drafts of F53, examines the
context of writing in almost as fine detail as it can get. Roger Backhouses
chapter examines the well-known role of the marginalist controversy in
the context of writing F53. Complementing Hammonds story,
Backhouse offers new information on the peculiar characteristics of F53
as a statement on theories of the firm and market structure by consulting
the correspondence between Friedman and Don Patinkin in the late
1940s. Michel De Vroeys chapter deals with Friedmans convictions in
relation to the Marshallian and Walrasian traditions. Chris Starmer casts
new illumination on the methodological issues in Friedman and Savages
joint work on expected utility and its role in inspiring some of the arguments in F53 such as the use of as if in formulating assumptions and
57
the use of predictive tests in assessing theories. Teira and Zamora argue
that the first section of F53 on the positivenormative distinction was
motivated by Friedmans experience with policy advice. Painting with a
broader brush, Melvin Reders chapter situates F53 in the long tradition
of the frame of the Invisible Hand of the free market.
At the other end, the focus is on the context of reading: of consumption
rather than production, of ex post influence and reception rather than ex
ante inspiration and creation. The context is that of the career of F53 once
it was born. There are weak and strong versions of these consumptionist
accounts. The weak versions deal with the (actual, possible, desirable,
inadvisable) reception and influence while at the same time keeping an
eye on the authors beliefs and intentions. The strong versions bracket the
authors attitudes entirely. Thomas Mayers and Wade Handss chapters
below are largely in the weak consumptionist category. Their main interest lies in the reception and influence of F53, but they contrast this with
what they view as Friedmans true beliefs and intentions. They raise
questions about whether F53 has been read appropriately, whether its
reception and influence are justified from the point of view of Friedmans
authentic methodological convictions. For example, one may ask whether
F53 licensed the formalist revolution as Wade Hands does and then
answers in the negative (whereas a section in my second chapter suggests
that F53 itself encouraged a torso version of its overall message, and
that the torso version can easily be used for licensing formalism).
On the other hand, the strong consumptionist is only interested in the
readers reception rather than the authors intention. The goal is to read a
methodological text from the point of view of its (actual or possible)
reception by the relevant audiences, including its (actual or possible)
interpretations and influences while completely ignoring questions
about the authors beliefs and goals. This is what I call reception methodology in analogy with reception aesthetics. My own second chapter
purports to be a more or less pure exercise in reception methodology.
The important difference between my chapter and those of Mayer and
Hands lies in the fact that Mayer and Hands provide a role for the
authentic beliefs and intentions of the author of F53, while I am bracketing them off altogether and discussing the text of F53 only. A further
difference is that Mayer and Hands write about the actual reception by
others in historical terms, while I am offering my own novel reading of the
essay. Theirs is a project of contrasting some other peoples readings of
F53 with their own reading of it, where their own reading is supposed to
yield access to the views that the author of F53 really held. My project
cannot be contrastive in that way since here I pay no attention to
Friedmans true beliefs and intentions.
58
Uskali Mki
Powerful support seems to be available to such a strong receptionist

approach: Milton Friedman himself has refrained from commenting on
the commentators of F53 in public, and he has done so deliberately. He
made an early decision to let F53 live its own life, leaving it at the mercy of
its readers as it were. This is confirmed in his final word at the end of the
present volume. He has enjoyed watching the readings of, and debates
over, F53 and has only privately provided feedback to its various interpretations. All this seems to offer strong support to the receptionist
approach from the most authoritative source. There is a tension here,
though. Reception methodology was supposed to ignore the authors
beliefs and intentions, so it might seem inappropriate to appeal to this
same authors stated intentions in defense of reception methodology. The
obvious way of relieving the tension simply points out that one should not
conflate the two levels of appeal to the authors attitudes: the levels of the
justification of a reading and of the justification of a strategy of reading.
The receptionist reading of F53 makes no appeal to the authors intentions, while it is at most the justification of the receptionist strategy that
makes higher-order appeal to the authors attitudes. Further relief is due
to the fact that, in the end, the former is not dependent on the latter: we
may adopt a receptionist strategy without Friedmans permission.
Another way of justifying a receptionist reading of F53 is based on the
recognition that this is how most readers view F53 anyway. This is very
much part of the popular legacy of F53, and this is part of much of the
reflective legacy as well. Most readers are not in the least interested in
Milton Friedmans peculiar intellectual trajectories or his authentic methodological convictions. They rather read F53 as a source of insights and
arguments that they can put into use when themselves dealing with troubling methodological issues in economics. When connecting the methodological arguments of F53 to the practice of economics, readers do not
usually connect them to Friedmans practice but rather to that of their
own and of others, whenever it seems useful as issues of justification and
critique of that practice arise.
The above remarks have dealt with the intra-academic context of F53.
It should also be possible to put F53 in a broader extra-academic context.
One could read it as being embedded in the cultural and political currents
of its time, both reflecting and shaping them. Not only would this mean
examining the image that F53 conveys of economics and of the economist expert as a politically powerful role amongst the sciences, but also in
society more broadly. The contribution by Teira and Zamora to this
volume takes some steps toward this direction.
Further steps could be taken by locating F53 in a historically and
politically unique external context. I have in mind a way of reading that
59
has not been practiced yet (not at least systematically), but that is not only
a possible approach but also one that I anticipate will be attempted in the
near future. This is based on recognizing a larger societal context in which
the production and consumption of F53 have taken place. This larger
context has to do with the overall transformations in the social sciences
(and the philosophy of science) that took place in the 1950s and that have
been linked with the broader sociopolitical conditions of the time, including the Cold War and the role of powerful funding agencies and research
institutes (for examples, see Amadae 2003; Mirowski 2002; Reisch 2005).
Cold War history of science is already a major stream of scholarly work,
and it will not be surprising if there is another small current of it within the
flow of F53 studies.
Evaluating and explaining
The fourth dimension of my map has to do with whether F53 is being read
from a normative or from an explanatory perspective. At one end, one sets
out to evaluate the soundness of the claims and arguments put forth in the
essay. Most evaluations in the reflective legacy have been critical of the
contents of F53. These include commentaries such as those by Koopmans
(1957), Rotwein (1959), Samuelson (1963), Melitz (1965), Caldwell
(1980), Blaug (1980), and Hausman (1992a).
Many of these criticize F53 for failing to see that good science not only
pursues predictive success but also asks and answers explanatory whyquestions about the phenomena investigated and in answering such
questions, the assumptions cannot be too much off the mark. Others
point out that even an economist interested in reliable predictions should
prefer realistic assumptions. Some question the objectivity of the reliance
on the predictive performance of economic theories without spelling out
very precise standards in terms of which to measure such performance.
Many critics also complain about the apologetic and complacent attitude
endorsed by F53: its arguments seek to justify a dogmatic commitment to
perfect competition and profit maximization against empirical challenges.
Another class of criticisms claims to identify mistaken statements in F53
(such as claiming that the Walrasian rendering of monopolistic competition theory is motivated by photographic accuracy). Yet another set of
complaints (my favorite) points out the strategic function of ambiguities
and inconsistencies in the reasoning of F53, as well as incompatibilities
between F53 and Friedmans practice as an economist (such as defending
unrealistic assumptions in some situations and criticizing them in other
contexts; suggesting reliance on the objectivity of predictive tests while
60
Uskali Mki
denying their objectivity; holding dogmatic attitudes while advocating

strong fallibilism).
So the mainstream of evaluative readings has been critical, and indeed
only a minority of reflective readings, such as Karl Brunners (1969), has
been supportive, usually with qualifications. Many evaluations are mixtures of approval and disapproval. In his chapter below, Williamson
criticizes F53 for an anti-pluralist endorsement of a single all-purpose
theory of the firm while admitting that the central idea of F53, namely
testing theories by predictions, is basically right. My own reading of F53 is
a mixture of praise and blame. The popular legacy is a different story
again: the proportion of favorable evaluations of F53 has been larger than
within the reflective legacy. Evaluations in this class typically do not come
with qualifications; they express simple attitudes in terms of yesyes, or
nono.1
In all evaluative judgments of F53, one invokes, explicitly or implicitly,
a set of standards. Such standards are concerned with the desirability and
possibility of things such as truth, reliability, explanation, causal knowledge, predictive success, epistemic progress, consistency, consensus, and
value-freedom. Further investigations are needed to turn such standards
explicit whenever they are implicit, to specify the various meanings of such
key terms whenever there is vagueness and ambiguity, and to identify and
critically examine further presuppositions whenever there is conflict
between, or disagreement over, those standards. For example, one
needs to subscribe to the standard of consistency between scientific disciplines regarding their theories and empirical results if one sets out to
blame an economist or an economic theory for ignoring evidence produced by experimental psychology.
While evaluation is one project, another task is to explain F53. We can
exploit the ambiguity of explain here and divide this into two types of
exercise. The first kind of exercise aims at explaining what the various
phrases and arguments in F53 mean, how they are best understood (this
can also be called an analytical activity). Naturally, this is a task that must
be taken before evaluation. One must understand what one wants to
evaluate before actually evaluating it. But this principle is not always
respected. The obvious suspicion is that within the popular legacy,
1
In his final word at the end of this volume, Milton Friedman exercises some playful selfcriticism, saying that had F53 been more lucid, commentators would have silenced by
now. This contrasts with his spoken statement, conveyed via a telephone connection at the
panel discussion on the occasion of the fiftieth anniversary of F53 at the ASSA Meetings (in
Washington, DC, January 2003). He said he recently reread F53 and found its claims
basically right. Had he been a perfect receptionist about F53 himself, he would have said
neither of these things.
61
normative judgment is easily passed without any solid understanding of

the arguments and claims of F53. This is understandable as it is almost
definitive of the popular legacy that no deep and detailed reflective understanding is sought to support the assessment. But it is clear that the
reflective legacy has also failed to produce consistently sound understandings of the message of F53. This is evident, among other things, in the
multiplicity of incompatible philosophical labels attached to F53. It is also
notable that while Boland (1979) claimed that all previous critics have
failed to understand F53, it can be argued that Boland himself is in that
same boat.
The second use of explain is in terms of answering various why- and
how-questions about F53. Why and how was the essay written the way it
actually came about? Why and how was it received, or could have been
received? For example, combining this with the focus on the context of
writing, one may argue: this is how Friedman came to hold these views,
and/or this is what he intended to mean by what he wrote. Dan
Hammonds work exemplifies this approach. The work of many others,
including my own, is a combination of the normative and explanatory
perspectives.
Using these categories, one may envision a distinction between reading
F53 philosophically and reading it historically. In reading F53 philosophically one adopts the first kind of explanatory perspective that aims at
interpreting and analyzing intended or possible meanings and arguments
in F53, and based on some interpretation, the reader may adopt an
evaluative perspective, assessing the soundness of its claims. In reading
F53 historically, one adopts the second kind of explanatory perspective,
describing and explaining the temporal trajectory of relevant ideas those
in F53 as well as those that have shaped it and those that it has shaped at
personal and social levels. These two ways of reading complement one
another usefully, and can also be combined with one another.
5
Extra-economic and intra-economic labels
It is only natural that the message of a complex text such as F53 will be
simplified in terms of supposedly informative labels that help classify and
characterize its message as distinctive. Since the 1970s, labeling the
message of F53 has been a major task undertaken by its reflective readers.
The labels have mostly come from extra-economic sources, namely philosophy: F53 has been classified as exemplifying this or that particular
position in the philosophy of science. A few others have characterized
Friedmans position as Marshallian, using an intra-economic label.
And there are other possibilities. This is the final dimension on my map.
62
Uskali Mki
A number of philosophical labels have been ascribed to Friedman and

F53 (mostly by readers who are professional economists). The view has
been labeled as positivist, falsificationist, conventionalist, pragmatist, and
instrumentalist and as realist, social constructivist, and radical fallibilist
in my own unconventional reading of F53. Of these, instrumentalism
has become the most popular label attached to F53. Very roughly, according to an instrumentalist conception of scientific theory, a theory is an
instrument or tool for some purpose, such as problem solving or organizing the empirical data. Such instrumental purposes must be other than
truthfully representing the way in which the phenomena of interest come
about the latter is how realism would see the tasks of theorizing. Given
that Friedman argues that it is all right for the assumptions of a theory to
be false as long as the theory predicts well, instrumentalism has appeared
to be his philosophy of economic theory.
Insofar as I have been able to determine, instrumentalism was first
explicitly used by Bear and Orr (1967) in characterizing Friedmans
views. Later on, it was used by commentators such as Wong (1971),
Boland (1979), Blaug (1980), and Caldwell (1980, 1999). However, the
priority on this matter should perhaps go to philosopher of science Ernest
Nagel. His authoritative treatise The Structure of Science appeared in 1961.
The book contained a substantial discussion on conceptions of scientific
theory, classifying them into what he called descriptivist, instrumentalist, and realist views (Nagel 1961). In December 1962, the
American Economic Association Meetings included a panel session
devoted to F53, featuring Paul Samuelson, Herbert Simon, Andreas
Papandreou, Sherman Krupp, Fritz Machlup, and Nagel. In that session,
Nagel (1963) suggested a possible interpretation of F53 as largely instrumentalist, but he did not use the term instrumentalism (even though he
used the expression theory as a useful instrument). I presume he could
easily have done so, but my guess is that because he was addressing an
audience mainly consisting of economists, he did not want to burden his
account with unnecessary philosophical labels.
But is the instrumentalist reading right? This still remains a legitimate
question. Contributing to the establishment of the instrumentalist interpretation of F53, Larry Bolands 1979 paper in the Journal of Economic
Literature put forth the claim that F53 is a coherent defense of an instrumentalist conception of economics. This claim has two components in it,
one about coherence and the other about instrumentalism. Since the
publication of Bolands paper, I have disputed both of these claims. I
have argued that what F53 offers is far from a coherent and unambiguous
argument for any well-defined position, and also that the most plausible
interpretation of F53 (and perhaps Friedmans methodology in general) is
63
based on the opposite of instrumentalism, namely realism (Mki 1986,

1992, 2000). My contribution to this volume provides a detailed reading
of F53 in realist terms. The key to this revisionist reading is to reconsider
the functions of falsehood in economic models. Also in this volume, Kevin
Hoover gives further support to the realist reading of Friedmans methodology by offering evidence from Friedmans work in monetary economics.
The instrumentalist interpretation of F53 used to be the dominant one,
but may have to give way to a diametrically opposing realist reading.
The broad variety of attributions is likely to be partly due to incompetent or relaxed use of the philosophical labels, but part of the reason may
be rooted just in the ambiguities and inconsistencies of F53 itself. For
example, to claim that F53 is a positivist statement simply because of its
adherence to the positive/normative distinction might be taken as an
example of philosophical incompetence if no other evidence is offered
(or else it might be just a matter of using the label in a very narrow sense).
There are a few other positivist elements in F53 (see Mki 1986), but
they are not dominant (at the same time we should grant that the label
positivist has itself become disturbingly vague and uninformative).
Similarly, there is a falsificationist element in F53 (the notion of failure
to be contradicted by evidence), but it appears just in passing. Social
constructivist features such as the emphasis on the role of disciplinary
culture in shaping scholarly judgment are far more central to the argument of F53, even though they have been mostly ignored by the readers.
It has been suggested that Friedman was not a professional philosopher
and therefore F53 should not be interpreted and judged in philosophical
terms. This suggestion has been made by commentators such as Daniel
Hammond, Thomas Mayer, and Kevin Hoover, also in this volume.
When viewed through philosophical lenses, F53 may emerge as uninformed and obscure. The suggestion is that Friedmans methodology
should instead be inferred from his work as an economist and as being
embedded in the tradition of economics. This approach then is argued to
suggest that the proper label for characterizing Friedmans methodology
is Marshallian a label that is intra-economic in that it derives from
within the tradition of economics itself rather than being imposed from
outside economics, using philosophical labels such as positivist and
instrumentalist.
I have three responses to this suggestion. First, I believe it will be fruitful
to consider the intra-economic and extra-economic perspectives as complements rather than substitutes. They usefully complement and crosscheck one another rather than exclude each other as incompatible rivals.
Second, while it is obvious that F53 is not to be celebrated for its elaborate
philosophical clarity and systematicity, I do think F53 is a delightfully
64
Uskali Mki
useful piece of writing from a philosophical point of view. It is a rich pool

of philosophically intriguing insights that pose difficult challenges to the
philosophical analysis of economic theorizing and reasoning. Moreover,
published in 1953, many of its ideas anticipated later post-positivist
ideas and arguments in the philosophy of science (such as the centrality of
disciplinary community and culture, the theory-ladenness of observations, and the underdetermination of theories by data). Pointing this out
has been a theme in my contributions to F53 studies and it is also one
theme in my second chapter to this volume. Third, while it is obvious that
due to the ambiguities of F53 the message of the essay is hard to characterize using a single philosophical label, it may be equally problematic to
identify Friedmans methodology as Marshallian due to the simple fact
that Marshallian is far from an unambiguous label itself, with fixed and
well-defined contents.
Friedman himself drew a distinction between Marshallian and
Walrasian ways of doing economics and elsewhere not in F53
identified his own approach with the Marshallian mode. He actively
portrayed his position in negative terms by leveling criticisms against
what he depicts as the unacceptable Walrasian approach. As pointed
out by Michel De Vroey in this volume, Friedmans understanding of
Walrasian was not fully coherent and historically accurate. It would not
be surprising if the opposing Marshallian label were to suffer from
similar unclarities.
Hence, whether using philosophical labels (such as instrumentalism)
or intra-economics labels (such as Marshallian), one cannot escape the
hard task of offering further detailed specifications and elaborations. One
has to look into the details of F53 and Friedmans economics (and whatever other things are relevant) to make progress in understanding
Friedmans methodology (or in just understanding F53, were this to be
the goal). Interaction and collaboration between philosophical concepts
and those derived from within economics (while granting that there is no
strict boundary line between the two) will be advisable in future
investigations.
An extreme conclusion would be to label Friedmans methodological
position simply as Friedmanian. This intra-Friedmanite strategy
would have the advantage of evading the problems of extra-economic
and intra-economic labels of the above kind. Yet the task would not be
easy. How would one go about selecting the relevant features of the
various methodological statements and practices that can be ascribed to
Friedman? Would one include a comprehensive and complex set of
details and peculiarities, throughout his intellectual career and across
various rhetorical contexts, not forgetting about the numerous
65
inconsistencies within and between those statements and practices? This

would single out the position as unique and unlike any other position
actually or potentially held by others. Or should one rather pursue a
generous interpretation that would ignore the ambiguities and inconsistencies and try to capture the simple and abstract essence of Friedmans
methodological position, perhaps by way of comparing and contrasting
his statements with those of relevant others? This, too, might end up with
describing a unique position. (Note that one may also choose to include
some of the inconsistencies in ones depiction of the simple essence of
Friedmans position or the position of F53.) On top of such strategic
choices, this line also involves the challenge of collecting and processing
masses of textual and other evidence that contributes to the contents of
the label Friedmanian methodology.2 Anyhow, when pursued on a
broad front in search for Friedmans methodology, this line has mainly
historical relevance. In contrast, a reception methodologist reading F53
only from the point of view of its later receptions and functions is relatively
uninterested in the contents of Friedmanian methodology, provided
they have no major consequences for the reception of F53.
Finally, it is also possible to use intra-F53 labels when characterizing the
message of F53. One such possibility would be to say F53 exemplifies a
predictivist position. Indeed, even though the very term predictivism
does not appear in the text of F53, it seems an appropriate derivative of the
repeated emphasis in F53 on the role of prediction as the goal and standard of economic theorizing. Another intra-F53 label is to say that it
endorses an as-if methodology. This might appear even more accurate
given that F53 formulates many hypotheses in terms of the as-if
locution.
Again, the intra-F53 strategy of labeling is no less problematic than the
other three strategies. This is because neither of the labels, predictivism
or as-if methodology, is able to fix a well-defined and distinctive methodological position. Indeed, many different and rival methodological
views put stress on prediction, and likewise the as-if locution can be
employed to express a number of different ideas with a number of different philosophical commitments. Just as with the other three labeling
strategies, this gives rise to the further need to provide more detailed
analyses, this time of the different meanings and uses of both prediction
and the as-if. Thus to say that the central message of F53 is the importance
2
That the task would not be easy is revealed by Hammonds (1992a) interview with
Friedman on the latters methodology: the interview is revealing in that it manages to
reveal so little, suggesting that perhaps there is little more there to be revealed concerning
the essence of Friedmans consciously held methodology.
66
Uskali Mki
of testing theories by their predictive implications, as Williamson does in

his chapter, is not yet detailed enough to distinguish Friedman or F53
from many other similar but rival views.
My response to all four strategies of labeling extra-economic, intraeconomic, intra-Friedmanite, intra-F53 has been the same. Labels are
useful and necessary, but if unsuccessful, they can be uninformative and
even misleading. To make them accurate and informative, more details
and qualifications are needed about the targets they are used for naming,
and about the purposes these labels are intended to serve. Providing such
details and qualifications is the task of careful empirical and conceptual
inquiry into the available textual and other evidence. The present volume
sets out to contribute to just such inquiry.
References
Amadae, S. M. (2003). Rationalizing Capitalist Democracy. The Cold War Origins of
Rational Choice Liberalism. Chicago: University of Chicago Press
Bear, D. V. T. and Daniel Orr (1967). Logic and expediency in economic
theorizing. Journal of Political Economy, 75, 18896
Blaug, Mark (1980). The Methodology of Economics. Cambridge: Cambridge
University Press
Boland, Larry (1979). A critique of Friedmans critics. Journal of Economic
Literature, 17, 50322
Brunner, Karl (1969). Assumptions and the cognitive quality of theories.
Synthese, 20, 50125
Caldwell, Bruce J. (1980). A critique of Friedmans methodological instrumentalism. Southern Economic Journal, 47, 36674
(1992). Friedmans predictivist methodological instrumentalism a modification. Research in the History of Economic Thought and Methodology, 10, 11928
Friedman, Milton (1953). The methodology of positive economics. In Essays in
Positive Economics. Chicago: Chicago University Press, pp. 343
Hammond, J. Daniel (1992a). An interview with Milton Friedman on methodology. Research in the History of Economic Thought and Methodology, 10,
91118
(1992b). The problem of context for Friedmans methodology. Research in the
History of Economic Thought and Methodology, 10, 12947
(1996). Theory and Measurement: Causality Issues in Milton Friedmans Monetary
Economics. Cambridge: Cambridge University Press
Hausman, Daniel (1992). Why look under the hood? In Essays in Philosophy and
Economic Methodology. Cambridge: Cambridge University Press
Hirsch, Abraham and De Marchi, Neil (1990). Milton Friedman: Economics in
Theory and Practice. Hertfordshire: Harvester Wheatsheaf
Koopmans, Tjalling C. (1957). Three Essays on the State of Economic Science. New
York: McGraw-Hill
67
Machlup, Fritz (1946). Marginal analysis and empirical research. American

Economic Review, 36, 51954
(1955). The problem of verification in economics. Southern Economic Journal,
22, 121
Mki, Uskali (1986). Rhetoric at the expense of coherence: a reinterpretation of
Milton Friedmans methodology. Research in the History of Economic Thought
and Methodology, 4, 12743
(1992). Friedman and realism. Research in the History of Economic Thought and
Methodology, 10, 17195
(2000). Kinds of assumptions and their truth: Shaking an untwisted F-twist.
Kyklos, 53, 30322
Melitz, Jack (1965). Friedman and Machlup on the significance of testing economic assumptions. Journal of Political Economy, 73, 3760
Mirowski, Philip (2002). Machine Dreams: Economics Becomes a Cyborg Science.
Cambridge: Cambridge University Press
Musgrave, Alan (1981). Unreal assumptions in economic theory. Kyklos, 34,
37787
Nagel, Ernest (1961). The Structure of Science. London: Routledge and Kegan Paul
(1963). Assumptions in economic theory. American Economic Review, 53,
21119
Reisch, George A. (2005). How the Cold War Transformed Philosophy of Science.
Rotwein, Eugene (1959). On The methodology of positive economics.
Quarterly Journal of Economics, 73, 55475
Samuelson, Paul (1963). Problems of methodology discussion. American
Economic Review, Papers and Proceedings, 53, 2316
Simon, Herbert (1963). Problems of methodology discussion. American
Winter, Sidney G. (1964). Economic natural selection and the theory of the
firm. Yale Economic Essays, 4, 22572
Wong, S. (1973). The F-twist and the methodology of Paul Samuelson.
American Economic Review, 63, 31225
Early drafts of Friedmans methodology

essay
J. Daniel Hammond
Introduction
Milton Friedmans essay, The methodology of positive economics

(F53), has been many things to many people since it was published in
1953 as the Introduction to Essays in Positive Economics. Generations of
graduate students have learned a lesson (the lesson for most) in what it
means for economics to be a science. Philosophers and methodologists
have seen it as Friedmans apologetic for this or that philosophy of science.
Some have used it as a whipping boy to expose the philosophical naivet of
economists. Others have used it to frame the parameters of the Chicago
school.1
For close to four decades after its publication those who have used the
essay treated it as the beginning and end of Friedmans methodology.
More recently this has changed as interpreters and critics have looked to
other works by Friedman, on methodology and on economics, and into
the context from which he wrote it for insights to the essays substance.2
A 1988 interview with Milton Friedman (Hammond 1992a) began to
uncover the unpublished historical background for Friedmans methodology. However, while he talked at length in the interview about his
methodology and speculated about its intellectual roots, Friedman was
unable to shed any light on the history of the essay itself. He said, I dont
really remember why I wrote that article to tell you the truth What I
dont know is whether the idea of the collection of essays came first, or the
essay came first (Hammond 1992a, 106). This history, the history of the
1
A few examples from the voluminous literature will suffice. Wong (1973) was the first of a
number of readers including Boland (1979) and Caldwell (1980, 1982, 1992) who interpreted Friedmans methodology as instrumentalism. Alan Walters (1987) claims that
Friedman introduced Poppers philosophy of science to economists. McCloskey (1983,
1985) tags the essay as one of the main texts of modernism, for which he uses the
Chicago school as a primary exemplar. See Hammond (1992b) for more on the various
interpretations.
My contributions to this include Hammond (1992b) and (1996) See also Hirsch and De
Marchi (1990).
68
Early drafts of Friedmans methodology essay
69
making of the essay, is important for our understanding of the text in the
narrow sense and Friedmans methodology in the broader sense. In the
Friedman Papers at the Hoover Institution are two preliminary drafts of
the essay along with other unpublished documents that shed considerable
light on why Friedman wrote the essay and what he intended the message
to be. The remainder of this chapter sketches the evolution of the essay
through the two preliminary drafts into the final published version. This
history of the essays making helps us see not only from whence the essay
came but also what was central and what peripheral in Friedmans argument as presented in Essays in Positive Economics.
Draft one: Descriptive validity vs. analytical

relevance in economic theory
The earliest draft of Friedmans methodology essay that I have been able
to identify (which I will label draft one) has the title Descriptive validity
vs. analytical relevance in economic theory. To judge from the citations
and from Friedmans correspondence it would appear that he began
writing it in late 1947 or early 1948 and finished in the summer of
1948.3 The paper is twenty-four typescript pages. Friedman opens with
the remark that theories are often judged by two criteria: the validity of
their assumptions and the validity of their implications. Neoclassical theory
that assumes perfect competition is judged unrealistic by the first criterion, and business cycle theories are rejected by the second. Friedman says
he will demonstrate that this view is fundamentally wrong and is productive of much mischief (Descrip, 1). He explains what he means in a
manner that has become familiar to readers of the 1953 essay, that the
two tests, when examined critically, reduce to one, since the only relevant
criterion for judging whether the assumptions are sufficiently good
approximations for the purpose in hand is whether they lead to valid
conclusions (Descrip, 1). He distinguishes between mere validity of
a theory, and significance or importance, arguing that significant theories
have assumptions that are inaccurate representations of reality. This is
3
The paper contains a reference to R. A. Gordons June 1948 American Economic Review
article, Short-period price determination in theory and practice. There is, on the other
hand no reference to A. G. Harts paper for the 1948 American Economic Association
meeting. As a discussant, Friedman criticized Harts suggestion that economists should
use survey evidence. This dating of draft one is also suggested by the fact that Friedman
wrote to George Stigler in November 1947, I have gotten involved for various irrelevant
reasons in a number of discussions of scientific methodology related to the kind of thing
you are talking about. In the course of these I have been led to go further than I had before
in distinguishing between description and analysis (MF to GS, November 19, 1947).
70
J. Daniel Hammond
because a significant theory necessarily abstracts the crucial elements

from the complex mass of circumstances surrounding the phenomenon
under explanation.
Friedman suggests that this insight will likely appear trite and hardly
worth explicit formulation and discussion (Descrip, 2). Friedman may
well have thought that on abstract methodological grounds he would find
no opposition to his argument. Or this may be a rhetorical gambit used to
establish credibility before actually making the argument. But, whether the
methodology could be made self-evident or not, there was in Friedmans
mind a serious problem, for economics was proceeding down the wrong
methodological path. He draws on two examples to illustrate this: monopolistic (imperfect) competition, which was a reaction against the assumptions of pure competition and pure monopoly, and a series of
American Economic Review (AER) articles criticizing and defending marginal analysis. Friedman indicates that he considers the second example
the clearer but less important of the two. He postpones detailed treatment
of the two cases until after discussion of the methodological issues.
This discussion opens with the familiar example of the formula s =
2
gt , used to relate the distance a body falls in a vacuum (s) to the force of
gravity (g) and the time the body is falling (t).4 Friedman interprets
applications of the hypothesis represented by this equation as being
made under the as-if condition. A ball falls from the roof of a building
as if it were in a vacuum. He considers the outcome of different applications: a feather dropped from the buildings roof, and the ball dropped from
an airplane at 30,000 feet. He claims that the example illustrates the
impossibility of testing a hypothesis by its assumptions, and moreover
the inherent ambiguity in the very idea of the assumptions of a theory.
Air pressure is only one of several factors that determine how well the
formula works. What are commonly regarded as the assumptions of a
theory are in fact parts of the specification of conditions under which the
theory is expected to work. And the only way to see which conditions
are critical or at what point any of them become critical in the context of
the others is to test the theory by its implications. Thus Friedman concludes that the formula is accepted because it works, not because we live
in an approximate vacuum whatever that means (Descrip, 45).
4
In the November 19, 1947 letter to Stigler Friedman used the distance formula to
introduce Stigler to his idea that the only way to determine how closely assumptions
correspond to reality is by checking the theorys predictions. The main purpose of this
letter was to respond to a draft of Stiglers essay on Chamberlin (1949). Friedman says,
I thoroughly agree with you The main additional point I would like to make is that you
do not really go at all far enough (MF to GS, November 19, 1947).
71
Friedman next turns to an example taken from Friedman and Savage

(1948) the billiards player. One might conceivably construct a complicated mathematical formula that would determine the best shot for any
given arrangement of the balls on the table. This formula then could be
used to predict the shots made by the billiards player as if he made the
appropriate calculations with the formula. Our confidence in the formula
would be based on our ability to use it to predict his shots, not on our
belief that he knew the mathematics and was actually able to make the
calculations. In this case it seems highly unlikely that even the description of the area of validity of the hypothesis would employ directly any of
the assumptions of the hypothesis (Descrip, 7). Friedman argues that it
is but a short step from these two examples to the hypothesis that businessmen behave as if they calculate marginal cost and marginal revenue to
maximize profit.
Next, he introduces the natural selection argument, saying that the
businessmans situation is like the billiards players. The latter will not
be an expert at billiards unless he can make shots as if he did the
calculations. The businessman likewise will not long stay in business
unless he reaches the same result that marginal theory implies. Any alternative theory that does not imply the same behavior as the profit maximization hypothesis does imply abundant opportunities for easy profit,
and according to Friedman this is clearly not the kind of world in which we
live. Since this implication of the alternative theories is false they must be
rejected in favor of the profit maximization hypothesis for a wide range of
circumstances. Expressing the idea differently, Friedman suggests that
even if firms chose their output randomly, but once and for all, and if their
funding was from internal sources only, the outcome via natural selection
would be the same as that implied by profit maximization.
As he continues explaining his position Friedman insists that the as if
profit maximization construct does not imply that we actually view businessmen as selfish, coldly-calculating, and heartless any more than the
formula s = gt2 implies that we live in a vacuum. Rather, each pulls parts
of reality that are more essential away from those which are less essential in
the circumstances. So, Friedman points out, the profit maximization
hypothesis should not be expected to be equally useful in all aspects of
human behavior.
This brings Friedman to what he considers the real but largely
neglected problem of profit maximization determining the limits of the
hypothesis, the circumstances under which, and the problems for which,
it gives reasonably good predictions (Descrip, 11). He speculates that the
hypothesis will prove to be of no great use in explaining the behavior of
individual firms. Its strength is that it identifies a force that operates on
72
J. Daniel Hammond
firms from the outside. For any given firm there will likely be aberrations,
but across firms these should cancel out, so that the hypothesis will work
best for industries.5
Friedman then shifts to the other issue at stake in the debates over
monopolistic competition and Marshalls industry analysis. This is the
assumption of perfect competition. He claims that critics are wrong in
attributing to Marshall the assumption that real-world industries in fact
are perfectly competitive. With several page citations from Principles he
claims that Marshall took the world as it was, he sought to construct an
engine to analyze it, not a photographic reproduction of it (Descrip,
13). Marshall used perfect competition in the as if mode, and then only
for situations in which the common forces from outside such as changes in
demand for firms product are dominant. For this kind of analysis it is not
possible to once and for all divide firms into industries by measures such
as cross-elasticity. Quoting Marshall, The question where the lines of
division between different commodities (i.e. industries) should be drawn
must be settled by convenience of the particular discussion (Descrip, 14;
Marshall 1920, 100).
Friedman uses examples of a Federal cigarette tax (where perfect competition is appropriate) and war-time price controls on cigarettes (where
it is not) to illustrate how everything depends on the problem; and there
is no inconsistency in regarding firms as if they were perfect competitors
for one problem, and perfect monopolists for another (Descrip, 14). In
the first case Friedman speculates that the firms will respond as if they
produced identical products in perfect competition. For the second there
was evidence from World War II that firms increased rather than reduced
production, presumably in order to maintain market share.
Friedman acknowledges that it would be desirable to have a general
theory of which perfect competition is a special case. But the standards for
this theory are that it have content and substance; it must have implications; and it must have some capable of empirical contradiction and some
of substantive interest and importance (Descrip, 17). The theory of
monopolistic competition might be thought of as a general theory, but it
has none of the essential features. It is based on the presumption that good
theories are realistic representations of reality in their assumptions and
that the differences between firms and their products are of essential
importance. Thus any attempt to apply this theory in the domain of the
real problems of the world, where common forces influence groups of
firms, requires that the theory be scuttled. Disciples of Chamberlin and
5
The usefulness of Marshallian industries was the central issue in Friedmans 1941 review
of Triffins Monopolistic Competition and General Equilibrium Theory.
73
Robinson, notably Triffin (1940), have recognized this impasse and

chosen fealty to the assumptions over applicability.
Friedman reiterates that a more general theory is to be sought, and
characterizes the work of von Neumann and Morgenstern on game theory
as such an attempt. Their advantage is in not confusing descriptive validity
and analytical relevance. However Friedman has doubts that the experiment will prove successful.
He next turns to the controversy over marginal analysis, suggesting that
the best example is the most recent article in the series, by R. A. Gordon
(1948). He quotes from Gordon:
There is still need to examine critically the assumptions underlying the conventional theory of the firm and the extent to which these assumptions meet the
test of business practice. The following pages elaborate upon a long felt belief
that conventional price theory has held to unnecessarily unrealistic assumptions,
with the result that our analytical tool box is not as useful as it should be. (Descrip,
1819; Gordon 1948, 265).
Friedman says that this passage indicates that Gordon is on a dead-end trail,
for it presumes that we can directly examine the validity of a theorys
assumptions, and it takes descriptive accuracy and comprehensiveness as a
standard for theory choice. Friedman makes his point by invoking the notion
of a completely representative theory of the wheat market. This would have
to account for a multitude of non-essential features such as the types of
money and credit used in trading, the color of hair and eyes of traders, as
well as their education, the type of soil on which the wheat is grown, and so
on. So the converse of Gordons statement, that if the assumptions are
slavishly realistic the theory is not useful, is closer to the truth.
What Gordon has failed to do is to test for the importance of the unreality of the assumptions in terms of how the theorys implications accord
with reality, and in comparison with an alternative theory. That is, in terms
of the example given earlier, Gordon is checking with the billiards player to
see how much math he knows. He is not observing his game. Gordons view
that theory is a description of what businessmen actually do is thoroughly
false. A sophisticated view of theory is that however businessmen may
make their decisions for a wide class of problems, the results are what they
would be if businessmen engaged in rational and informed profit maximization (Descrip, 21; emphasis his). Friedman strings together a number of
quotations from Gordons paper to show the argument. One of these is,
There is an irresistible tendency to price on the basis of average total cost
for some normal level of output. This is the yardstick, the short-cut, that
businessmen and accountants use, and their aim is more to earn satisfactory
profits and play safe than to maximize profits (Descrip, 22; Gordon 1948,
74
J. Daniel Hammond
275). Friedman remarks in a footnote that he has never been able to understand this kind of argument, that behavior is the outcome of an irresistible
tendency, in the face of observable and clearly false implications. His interpretation of how economists such as Gordon deal with this situation is that
they either eliminate all content from their theory, making it tautological, or
interpret it in a way that makes it de facto equivalent to the usual theory.
Friedman concludes his paper with a response to Gordons closing section,
which has the title The need for greater realism. The main point Friedman
makes here is that he has no dispute with Gordons account of the lack of
descriptive realism in standard theory. His quarrel is over implications.
Gordon thinks greater realism would make theory more useful; Friedman
does not. But Gordon offers no evidence that a more realistic theory is more
useful. In the absence of such evidence the greater complexity is a burden
with no counterweight. Friedman tries to make clear that he is not arguing
that marginal theory is useful, though he thinks it is. He is arguing over the
grounds for determining this. In his final statement Friedman suggests an
appropriate role for descriptive studies of business behavior. This is in deriving new hypotheses, an activity which is distinct from testing existing hypotheses. Construction of new hypotheses depends on intuition, inspiration,
and invention, and is not subject to formal methodological analysis.6
6
This became a point of dispute between Friedman and George Stigler and Arthur Burns.
In September 1948 Stigler wrote to Friedman:
Personally I would like to see it published (in part because Ive paraphrased the argument
in two paragraphs of my Chamberlin essay, and would like to give a more specific
reference). But I keep feeling that you arouse skepticism and opposition by stopping
where you do. Because surely in some sense an assumption can be more promising than
another It is surely possible to say something about some assumptions being more
promising than others, and yet not to take back any of the things you are saying at present.
If you can pierce this muddy frontier of your article, it would be a great improvement. (GS
to MF, September 1948)
Friedman responded:
I think part of the difficulty you have on the methodology problem arises out of the fact that
the issues it deals with pertain only to one small part of all work in economics. One might, I
suppose, separate out four kinds of things that economists and other scientists do: first, the
collection of data to provide something to generalize from; second, the derivation of
hypotheses to generalize the empirical uniformities discovered in the data; third, the
testing of these hypotheses; and fourth, the utilization of them. My strictures apply only
to the third of these steps The real problem that you raise arises, I take it, when
somebody proposes a theory which we havent as yet been able to test, and the question
arises, shall we use it instead of some alternative. Its at this point that one is most likely to
say that he is judging the theory by its assumptions and to say that he will have some
confidence in it if the assumptions are reasonable, and he will not if they are not. This is the
kind of point Arthur was raising most strenuously this summer against it. (MF to GS,
October 4, 1948)
75
Draft two: The relevance of economic analysis to

prediction and policy
The next draft has a different title, The relevance of economic analysis to
prediction and policy, and considerable new content. At forty-seven
pages, it is almost exactly twice as long as the first. The manuscript is
undated, but we can place it in the fall of 1952. The copy that I am
working from has a holographic note on the first page from Friedman to
Arthur Burns identifying this as the piece I spoke of, and Burns wrote a
November 26, 1952 letter in reaction to the paper.7 Also, the paper opens
with a footnote explaining that I have incorporated bodily in this article
without special reference most of my brief Comment (1952a) in A Survey
of Contemporary Economics, which was published in 1952.
Friedman opens with what is now familiar to readers of the published
version of The methodology of positive economics, a reference to
J. N. Keyness admirable book on The Scope and Method of Political
Economy, which Friedman uses to distinguish positive from normative
science and to make the point that there is confusion between the two.
Friedman claims that this confusion will arise so long as there are selfproclaimed experts with their own agendas.8 He writes of the distinctions
between positive and normative economics and of the influence which
properly runs one way, from positive to normative economics. He identifies
the task of positive economics as to make correct predictions about the
consequences of any change in circumstances, and claims that its performance is to be judged by the precision, scope, and conformity with experience of the predictions it yields (Friedman 1952b, 2). He judges that most
disputes over economic policy are the result of differences on positive
7
Burns wrote, In accepting the heart of your argument, I do not, however, accept the
details of your exposition. I have made perhaps four or five comments on the margin,
whereas forty of four hundred are probably called for (AB to MF, November 26, 1952).
Burns thought that Friedman had not explored sufficiently the bases on which confidence
rests in not yet tested theories. Thus he criticized Friedman for not making, or not making
clearly enough, the very sort of distinction that Friedman began to make in his letter to
Stigler about draft one, and which he attempted in draft two.
Burns also thought the illustrations from economic theory were too sketchy, that none
were presented with the precision required for scientific testing. He suggested that the
proper time for stating precisely the circumstances under which the hypothesis is expected
to hold is before its testing.
I have come across several pages of text that seem to have been intended as part of a draft of
Friedmans essay, in which he reproduces Keyness quotation from F. A. Walker that gives
the ubiquity of quackery as reason for concern with economic methodology. Friedman
indicates more faith than Keynes in the publics ability to discriminate genuine science
from quackery. He contends that the more important reason methodology is valuable is the
special difficulty in economics of producing clear-cut evidence for or against theories.
76
J. Daniel Hammond
analysis, different predictions about the consequences of taking action,

rather than differences in values. The minimum wage, the role of trade
unions, price controls, tariffs, regulation of business, and even socialism are
given as examples. He concludes that making progress in positive economics is of more importance than making progress in normative economics
proper. This introduction in draft two is different only in minor details from
that of the published essay. It has the same title, The relation between
positive and normative economics. But the introduction, with its references to Keynes and positive and normative science, is not in draft one.
The second section of draft two likewise has a title that is carried over
into the published version, Positive economics. The section opens with
the distinction, now familiar to readers of Friedmans essay but new in
draft two, between theory as language and as substantive hypotheses.
Friedman gives citations from Marshalls The present position of economics ([1885] 1925), and from his own The Marshallian demand
curve (1949), and Lange on price flexibility and employment (1946).
As a language (a set of tautologies) theory is to be judged by criteria
appropriate to a filing system, which are partly logical and partly factual.
As a body of substantive hypotheses it is judged by its predictive power
for the class of phenomena which it is intended to explain (Friedman
1952b, 7). He addresses the question of choice among hypotheses that are
equally consistent with the evidence, suggesting that the appropriate
criteria are simplicity and fruitfulness. He discusses the restricted
latitude for controlled experiments in the social sciences, but suggests
that often history provides evidence as conclusive as one would get from a
controlled experiment.9 Inflation is suggested as one phenomenon for
which the evidence is very clear. Yet even here the inability to produce the
crucial experiment makes weeding out unsuccessful hypotheses very
difficult.
Friedman speculates that the difficulty of testing hypotheses has
encouraged a retreat into purely formal or tautological analysis. Here
economics becomes little more than disguised mathematics. Properly
seen, tautologies are to be judged ultimately by how well they facilitate
9
In 1946 the University of Chicago received a grant from the Volker Fund to initiate a
Program of Factual Research into Questions Basic to the Formulation of a Liberal
Economic Policy (Free Market Study), with Aaron Director as director of the study.
Friedman was a member of the Executive Committee. The prospectus for the program
indicates the goal of modeling the studies after the experimental natural sciences. Studies
such as these are of the nature of experiments in the physical sciences, most of which are
unsuccessful (University of Chicago, A Program, 5). It also is based on the same
presumption that Friedman stated in the methodology essay, that Disagreement about
the appropriate economic policy for the United States arises in considerable measure from
disagreement about facts rather than from differences in objectives (A Program, 1).
77
empirical analysis. However, the more serious outcome of the difficulty of

testing hypotheses is misunderstanding of the role of empirical evidence,
as seen in the idea that theories can be tested by their assumptions. Here
Friedman breaks the role of empirical evidence into two methodologically distinct but temporally overlapping roles construction of new
hypotheses and testing of existing hypotheses.
The difficulty in the social sciences of getting new evidence for this class of
phenomena (that which the hypothesis is designed to explain) and of judging its
conformity with the implications of the hypothesis makes it tempting to suppose
that other, more readily available evidence, is equally relevant to the validity of the
hypotheses; to suppose that hypotheses have not only implications but also
assumptions, and that the conformity of these assumptions to reality is a
test of the validity of the hypothesis different from or additional to the test by
implications. (Friedman 1952b, 12; emphasis his)
Friedman continues with the argument that because explanatory theory

necessarily abstracts from complex reality in general, the more significant the theory, the more unrealistic the assumptions (Friedman 1952b,
13). The key is not whether assumptions are descriptively realistic, but
whether they are sufficiently good approximations for the purpose at
hand. And this question can be answered only by seeing whether they
work, which means whether the theory yields sufficiently accurate predictions. The supposedly two independent tests thus reduce to one
(Friedman 1952b, 13).
Up to this point the material in the section headed Positive economics is new to draft two, and with two exceptions identical to the section
with the same title in the published version.10 The exceptions are two
passages added to the final version. Just after making the distinction
between theory as language and as substantive hypotheses Friedman
adds a paragraph to the published version including past events not yet
observed in what he means by predictions (F53, 9). He adds another
passage nearby that gives an example (excise tax) of a situation where
predictions do not allow the investigator to choose between theories,
making it necessary to fall back on simplicity and fruitfulness for
theory choice.
The final paragraph of the section on Positive economics in draft two
introduces the reader to the two outcomes of muddled methodology that
Friedman treats, monopolistic competition and the AER debate over
10
By new to draft two I mean the material is not in draft one. Friedman had made similar
arguments elsewhere, as indicated by his references to Lange on price flexibility and
employment (1946), The Marshallian demand curve (1949), and Friedman and
Savages The expected-utility hypothesis and the measurability of utility (1952).
78
J. Daniel Hammond
marginalism. This paragraph is identical to the concluding paragraph on

pp. 1516 in the published version. Aside from the lengthy footnote citing
the AER articles (footnote 13 in the published version), it is similar to
material on pp. 23 of draft one. So here, fourteen pages into draft two and
fifteen pages into the published version the discussion comes to the point
at which draft one begins.
Section III contains discussion of the distance formula and the billiards
player taken from pages 3 through 8 of draft one. It also matches up for
the most part with section III in the published draft; in both versions
the sections title is Can a(n) hypothesis be tested by the realism of its
assumptions?
The opening paragraph lays out the distance formula and the applications: a compact ball dropped from a roof; a feather dropped from the
same roof; and the ball dropped from an airplane. Each of these circumstances produces a different result. The next paragraph poses the question
of how much difference between the predicted and measured distance is
sufficient to conclude that the formula does not work. This depends,
according to Friedman, on the accuracy of the predictions from an alternative hypothesis and on the relative costs of using the two hypotheses.
This discussion is virtually word for word the same as the published
version. The first paragraph is rewritten from draft one and the second
paragraph is new. The copy of draft one that I am working from has
penciled changes which mostly match up with the text in this second
draft, but they do not match up wholly. This suggests that there was still
another draft with at least minor changes between what I am treating as
draft one and draft two.
Friedmans use of the distance formula is completed with his argument
that the real task for use of the formula is to determine the circumstances
under which it works, or how well it works under different circumstances.
Here he makes the contrast between more and less general theories. But,
he contends, specifying the conditions under which the less general and
simpler hypothesis works, or the errors obtained with certain values for the
disturbing factors, is quite different from testing the hypothesis by determining whether the assumptions are true or false. This discussion is the
same as the published version except for two brief passages added to the
latter.11 These are the statement on p. 18 of the published version which
claims that this specification cannot be separated from the hypothesis
itself, and the special emphasis drawn on p. 19 to the difference between
11
These appear to be in response to Burnss criticism in the November 26, 1952 letter.
79
the specification Friedman is concerned with here and the use of assumptions to determine the circumstances for which the hypothesis holds.
Friedman then moves to the example of the positioning of leaves
around a tree, which is not in draft one. The text is the same as that in
the published version. Explanation is given in a footnote that Friedmans
use of the example is though independent in origin, similar to and in
much the same spirit as that used by Armen Alchian in his June 1950
Journal of Political Economy article.12 The comparison is drawn between a
simple hypothesis that leaves are positioned as if they sought out and
maximized the amount of sunlight they receive, and the more general
theory that sunlight contributes to growth and survivability of leaves. The
choice between these theories rests, once again, not in a comparison of
assumptions with reality but in their ability to predict accurately over a
wide variety of circumstances.
The billiards player example, originally in Friedman and Savage
(1948), is rewritten from the first draft, and identical to the published
version through the examples end in draft two. The point is made that our
confidence in the hypothesis that the expert billiards player makes shots as
if he could calculate angles and solve the mathematical formulas that
describe the optimal shot comes from the fact that unless he can achieve
roughly the same result he will not in fact be an expert billiards player. We
judge the formula on the basis of how well it predicts his shots, not on how
he says he makes his shots.
Thereafter the published version has material that is pulled forward
from further into draft two (section 6, The present state of positive
economics) and rewritten. This is the final two paragraphs of section 3
in the published version. Here Friedman returns to the maximization of
returns hypothesis to draw the parallel between the examples from outside
economics and the economic issue; natural selection ensures that successful (surviving) businessmens behavior is as if they maximized returns.
The second paragraph that Friedman pulls forward points to the difficulty
of documenting the success of the maximization of returns hypothesis
since the experience from countless applications of the hypothesis to
specific problems is scattered in numerous memorandums, articles,
and monographs concerned primarily with specific concrete problems
rather than with submitting the hypothesis to test (F53, 223).
Section IV of draft two has the same title (The significance and role of
the assumptions of a theory) and subsections as the published version.
The essential content is the same, though there are a number of changes
12
Friedman read and criticized several drafts of Alchians paper in 1949 and 1950 before it
was accepted for the Journal of Political Economy.
80
J. Daniel Hammond
and additions to the published version. But in relation to draft one this
section is wholly new. The discussion is Friedmans attempt to identify
some constructive use for assumptions, given the strong tendency we
all have to speak of the assumptions of a theory and to compare the
assumptions of alternative theories (Friedman 1952b, 20), and in
response to the criticisms of Stigler and Burns. Friedman warns, however,
that he is less confident in the positive comments that follow than he is in
the negative comments that make up much of the rest of the essay.
He identifies three positive roles for assumptions: (1) economical
presentation of a theory; (2) facilitation of indirect tests of the hypothesis
by its implications; and (3) specification of conditions under which
the hypothesis is expected to hold. As the third has been covered in
earlier parts of the essay, Friedman devotes the balance of this section to
the first two.
The beginning of the discussion of the role of assumptions in stating a
theory is identical in draft two and the published version. The assumptions sort out in a short-hand manner factors that are expected to influence the phenomenon under explanation into those that are more and
those that are less important. A hypothesis has two parts, the abstract
model and the rules which define the class of phenomena for which the
model is taken to be an adequate representation and which specify the
correspondence between variables in the model and observable phenomena. The model is abstract and complete; the rules are concrete and
necessarily incomplete. In the two paragraphs where Friedman lays out
these differences between the parts of hypotheses (pp. 245 in F53) there
are several changes made between draft two and the final. A sentence, At
this stage (the abstract model), such qualifications (e.g. that air pressure is
small, not zero) are signs of sloppiness and fuzzy-mindedness, not of
practicality or realism, is dropped. This would appear to be the result of
Arthur Burnss marginal notation, to me! Friedman also adds a sentence stressing the importance of making the rules explicit in the interest
of making science as objective as possible.
The example of Euclidian geometry as an abstract model with pure but
unrealistic lines and the application of this model with a chalk mark on a
blackboard is the same in this draft as in the final. So is the final paragraph
in the subsection, which sums up the use of assumptions for stating key
elements of the abstract model.
The next subsection discusses the use of assumptions as an indirect test
of a theory.13 The first three paragraphs of the published version are not in
13
Friedman sketched the outlines of this argument in his October 4, 1948 letter to George
Stigler. Stigler responded, I like your general position but want you to enlarge it,
81
draft two, which starts off by directly pointing out that assignment of parts
of a theory into assumptions and implications is a matter of convention. Implications can be treated as assumptions and vice versa. The
evidential value of this reversibility is limited, however, when the assumptions refer to a class of phenomena different from those the theory is
intended to explain. Here a brief elaboration in draft two is cut for the
final.
A second and closely related way in which assumptions can facilitate
indirect testing is when the kinship of two hypotheses can be demonstrated by restating them in terms of the same assumption, and evidence for
the implications of one provides indirect support for the other. Friedman
gives as an example the use of the self-interest assumption along with
market power to explain different degrees of racial discrimination across
industries or regions. Here the economists expectation that this economic hypothesis will account for the discrimination is based on success
in using it to explain other behavior. Friedmans changes here between
draft two and the final again bear the mark of Arthur Burnss comments.
The section concludes with observations on the choice of hypotheses in
the absence of a crucial test of the implications, and the role of the
scientists background in influencing the amount of evidence he requires
before making a judgment for or against a hypothesis. Here two sentences
are added to the published version that roughly double the length of the
final paragraph. They point out the frequent need for a conclusion in
advance of a satisfactory test and the impossibility of completely objective choice among hypotheses.
In section V, Some implications for economic issues, Friedman turns
to the methodologically charged economic issues that dominate the first
draft but in draft two are encountered twenty-six pages into the manuscript (twenty-seven pages into the published version). Section V shows
considerable editing and rearranging between draft two and the published
version, including the folding in of material from section VI, a section
which is dropped in the final.
Friedman quotes Veblen and H. M. Oliver to put the unrealistic
assumptions before his readers. He immediately thereafter pulls material
forward from section VI of draft two for the published version. This is the
comparison of criticism of orthodox theory on the basis of survey
evidence (e.g. Hall and Hitch 1939) with asking octogenarians how they
precisely as you are enlarging it in your letter to me. While some elaboration along these
lines will take some of the paradox out of your thesis (and in a certain sense weaken its
message unless you write very carefully), it will create sympathy for and receptiveness to
your thesis and make the paper much more influential (GS to MF, undated [October
1948]).
82
J. Daniel Hammond
account for long life. The first two references in footnote 22 of the
published version are added to the corresponding footnote in draft two.
Friedman recalls from earlier discussion of the distance formula
(s = gt2) that complete realism as demanded by the critics is impossible,
a point he illustrates here with another example indicating the absurdity of
a realistic theory of the wheat market. This example is taken from p. 19
of draft one. Friedman admits that he is attacking a straw man, since no
reasonable critic would demand a thoroughly realistic theory of business
behavior, but he argues that the critics themselves offer no basis for
stopping short of the straw man. They must either go all the way or fall
back on predictive accuracy as the test of how much realism is enough.
An allusion to the purported complexity of business cycle phenomena
dictating a complex theory is found at this point in draft two, as it is in the
final version. This and a reference to Sydney Alexanders 1951 AER paper
are new in draft two, but the shadow of draft ones title is found in the
opening sentence of this (and the next) paragraph, where Friedman refers
to the confusion between descriptive accuracy and analytical relevance.
Friedman next argues that the misguided effort to make abstract models descriptively accurate is illustrated clearly in the charge that Marshall
assumed the world was made up of perfectly competitive industries.
Friedman challenges Marshalls critics to find in Principles an assumption
that industries are perfectly competitive. What they will find, according to
Friedman, is that Marshall took the world as it is. This discussion is
taken with minor changes from draft one, beginning on p. 12. Marshalls
view as interpreted by Friedman is that grouping firms into industries
cannot be done once and for all. It is contingent on their being subject to a
common external stimulus and on the analytical purpose at hand. The
most prominent change in Friedmans explanation from draft one to draft
two (and the published version) is his use of the distinction between the
abstract model, in which there are the ideal types of perfect competition,
infinitely elastic demand curves for identical products, etc., and the rules
for classification in real-world applications. Otherwise Friedmans interpretation of Marshall is the same as in draft one.
The critics mistake is to suppose that finding the abstract models ideal
types in the real world is necessary for applications. Friedman draws on
the distance formula and chalk mark examples to emphasize the error in
this supposition. He also appeals to the example used in draft one to
illustrate how the cigarette industry can be treated as competitive or
monopolistic, depending on the problem at hand.
Friedman leads into his criticism of the theory of monopolistic competition by granting that it would be desirable to have a theory more
general than Marshalls. But to be useful it must have implications
83
susceptible to empirical contradiction and of substantive interest and

importance (Friedman 1952b, 34). Monopolistic competition is, according to Friedman, a failed attempt to construct such a theory. The failure is
seen most clearly in this theorys inability to treat problems involving groups
of firms. Proponents of the theory find themselves on the horns of a
dilemma; they must either introduce fuzziness and undefinable terms
into the abstract model where they have no place (Friedman 1952b, 35),
or forego the opportunity of treating problems faced by groups of firms,
retreating into the economics of the individual firm. In draft two (and
draft one) there is a concluding paragraph for section 5 which is dropped
in the published version. It suggests that the von NeumannMorgenstern
theory of games is more likely than monopolistic competition to provide a
general theory within which Marshallian analysis will be a special case.
Friedman doubts that even this attempt to develop the more general theory
will succeed; yet it is not doomed from the start by a confusion between
descriptive realism and analytical relevance (Friedman 1952b, 36).14
The title of section VI in draft two is The present state of positive
economics. This section is dropped in the published version. As mentioned above, some of the material is folded into section V. Other parts are
rewritten and placed in the Conclusion of the published version. Some
of that discussion, in which Friedman attempts to be more specific about
the confidence warranted for different branches of contemporary economic
theory, is apparently taken from an early draft of Friedmans essay on
Wesley Mitchell.15 Friedman begins with the caveat, all that is possible
here is a cursory expression of a personal view, together with a somewhat
more extensive examination of a particular hypothesis to exemplify the kind
of evidence that exists (Friedman 1952b, 367). The corresponding
14
15
In draft one Friedman says, I have real doubts that it will succeed. In draft two his
prognosis is, It may not succeed in furnishing a fruitful construction within which the
Marshallian analysis can find its place as a special case
In the Friedman Papers at the Hoover Institution there are fragments of two early drafts of
Friedmans review of Mitchells contributions to economics (1950). In a section of one
draft entitled The present state of economic theory, the same title as section VI of draft
two of the methodology essay, Friedman writes:
Finally, Mitchell was not only a great scientific worker, but he also was a distinguished
essayist. His numerous essays are always well organized and beautifully written, but they are
on a different level from his fundamental scientific work. They are far less profound and
often implicitly at variance with his own work. In particular, I feel that the remarks on
methodology in his essays are, to say the least, unfortunate; and have been the source of
much misunderstanding his scientific instincts, incorporated in his own work, are sounder
than his rationalizations of them. (Friedman Papers, Box 53, folder unlabeled)
The paragraph concludes with, Section 5 discusses briefly this point. In the margin there
is the holographic notation, Have not written such a section. Should I?
84
J. Daniel Hammond
discussion begins three paragraphs into the Conclusion of the published

version, on p. 41.
Friedman divides economic theory into relative price theory, which is
designed to explain the allocation of resources among alternative ends and
the division of the product among the cooperating resources and monetary theory, which is designed to explain the absolute level of prices,
aggregate output, and other variables for the economy as a whole
(Friedman 1952b, 37). This division itself is based on an as if; it is
made as if money was a veil that can largely be ignored when explaining relative prices, and real factors can be ignored when explaining the
price level, etc. Friedman then divides these two types of theory into
statics and dynamics.
He argues that the neo-classical theory of relative prices, which reached
almost its present form in Alfred Marshalls Principles of Economics is, in my
opinion, both extremely fruitful and deserving of much confidence for the
kind of economic system that characterizes Western nations (Friedman
1952b, 37). In a statement that is dropped from the published version he
contends that
The essence of this theory is that the phenomena with which it deals can be
explained or predicted by supposing men to behave as if they sought singlemindedly and successfully to pursue their own interests and as if their interests
were predominantly to maximize the money income they could wring from a
hostile environment though so bald and simplified a statement does less than
justice to the subtlety of the theoretical structure and the range of phenomena it is
capable of illuminating. (Friedman 1952b, 378)
The weakness of relative price theory is in capital theory, according to

Friedman. The quantity theory of money is deserving of confidence as an
explanation of monetary statics. However in monetary dynamics it is
difficult or impossible to single out any theory that can be regarded as
deserving of much confidence (Friedman 1952b, 39). There have been
high hopes for Keynesian theories but their predictions have not been
confirmed.
Here Friedman returns to the maximization of returns hypothesis to
indicate the types of evidence in its favor. There are three. The first is that,
if there is natural selection among firms, the hypothesis isolates the condition for survival. This harks back to the illustrations of the billiards
player and leaves around a tree in section III. In the published version
this discussion is moved to section III. The discussion is also in draft one.
The second and more important type of evidence is the record of
successful applications of the hypothesis to specific problems. However,
this is hard to document as it is scattered through various publications and
85
because the evidence is generated in the course of applications rather than

by a crucial test. The third type of evidence is the knowledge we have
that indicates businessmens pursuit of profit. This is the indirect
evidence obtained by reversing the place of assumptions and implications,
and is of a lesser order of importance.
Friedman points out that most criticism directed at the maximization of
returns hypothesis concerns the third, and least important, class of evidence. He compares the use of surveys of businessmen to determine how
they set prices and quantities with the use of surveys of octogenarians for
testing theories of longevity. This discussion, as we have seen, is moved
forward to the beginning of section V in the published version. Friedman
allows a role for survey evidence, but this is in constructing new hypotheses and revising old ones when their predictive power is weak. He repeats
that this evidence is not a test of the hypothesis.
He then turns to Gordons AER paper with its consideration of an
alternative to the maximization of returns hypothesis. This is that
There is an irresistible tendency to price on the basis of average total
costs for some normal level of output (Friedman 1952b, 43; Gordon
1948, 275). Friedmans challenge is (as on pp. 223 of draft one) that this
hypothesis is contradicted by observable implications. The counterevidence includes bargain sales, reductions in price associated with rises in
average total costs for some normal level of output, increased or
reduced prices associated with no changes in costs (Friedman 1952b,
43). The contradictions can be avoided by defining average total costs for
some normal level of output in a way that makes it depend on demand.
But this is no resolution of the problem, it either reduces the hypothesis to
a tautology or makes it indistinguishable from the maximization of returns
hypothesis.
Section V of draft two concludes with discussion taken from pp. 1011
in draft one. Here Friedman returns to the more general point that the
maximization of returns hypothesis is an as if proposition that is supposed to isolate the crucial aspect of behavior for the problem at hand. Its
successful application does not suppose or depend on actual behavior that
is selfish, coldly and calculatingly heartless, or all-wise and all-knowing
in any descriptive sense, any more than using the formula s = gt2 means
assuming we live in a vacuum (Friedman 1952b, 44). Furthermore, the
hypothesis cannot be expected to explain equally well behavior in all
settings. Precisely because it selects one from many aspects of human
behavior as central, it is both useful and limited in its applicability
(Friedman 1952b, 44). The challenge that remains for economists regarding the hypothesis is to determine the limits of its applicability. Friedman
86
J. Daniel Hammond
is skeptical that it will prove applicable to the problems of the individual

firm, focusing as it does on forces that operate on firms from the outside.
Section VI is the Conclusion of draft two. This is different from the
conclusion in the published version, which is almost totally new. In draft
two Friedman begins the concluding section with the observation we find
at the end of the published conclusion. This is that progress in positive
economics depends on both testing existing hypotheses and construction
of new hypotheses. Of the latter task there is very little that can be said in
formal methodological terms. This is one reason Friedman focuses on
testing hypotheses. The other reason is that the obstacles to crucial tests
lead economists to wrongly reject the principle of testing by implications.
This methodological confusion has:
fostered mathematical economics, with its emphasis on Walrasian general equilibrium analysis as an escape from the ceteris paribus of partial equilibrium analysis; it
explicitly motivated monopolistic competition analysis and explains its popularity;
it encouraged emphasis on arithmetical rather than economic considerations in all
branches of economic analysis, it was the battle cry of institutionalism and the
closely related emphasis on extensive statistical studies of economic phenomena; it
is a major source of the nave yet recurring notion that the progress of positive
economics depends critically on the progress of psychology; it has been manifested
most recently in the belief that a theory can be tested by asking questions of
consumers, producers, and the like. (Friedman 1952b, 46)
These tendencies have not been wholly without beneficial results, but
have impeded the testing of hypotheses. Alfred Marshalls emphasis
on the construction of an engine for the discovery of concrete truth has
tended to be submerged under the urge for descriptive realism
(Friedman 1952b, 47).
Conclusion
Friedman worked on the methodology essay over a period of several years,

most likely from late 1947 into 1953 when Essays in Positive Economics was
published. During the period form the mid 1940s through the mid 1950s
Friedman was heavily engaged in economic methodology. Methodology
is prominent in much of what he wrote during this period. And up to at
least 1957 he gave a number of seminars on methodology.16 There is also
16
There are at least eleven sets of notes for lectures in the Milton Friedman Papers and in his
personal files. Some are dated and identified as to the place of the lecture and others are
not. The earliest that is dated is March 22, 1950 at Vanderbilt University and the latest is
April 23, 1957 at Purdue University.
87
evidence, in the form of an outline, that at some point Friedman planned

to write a book on methodology.
Taking the first draft as its heart, the central methodological message of
the essay is that theory necessarily leaves out elements of reality, and
consilience of a theorys implications with evidence of the phenomena it
is supposed to explain is the only test of how crucial omitted factors are.
Apart from this, the very meaning of the assumptions of a theory is
ambiguous. Friedman used the distance formula and billiards player
examples to illustrate these points with theory that abstracts from reality
in the as if mode in non-economic settings.
As for the connections between methodology and economic theory
and Friedmans concern was always methodology within the context of
economic theory issues the essay was a defense of Marshalls economics
from two attacks. One of these was the monopolistic competition of
Chamberlin and Triffin, where Marshalls analytical tools of industries,
perfect competition, and pure monopoly were attacked. The other was the
critique of profit maximization and marginalism by Gordon and others.
Friedman thought the limits of the applicability of Marshalls approach
had not been determined.
Friedman was imbued with Marshall. His label for his methodology,
a label he actually used in some of his methodological writings,
was Marshallian. But for Friedman Marshallian methodology and
Marshallian theory went hand in hand. He told students in the opening
lecture of Economics 300a at the University of Chicago at about the time
he was writing the methodology essay, Marshalls Principles viewed
contemporaneously, i.e., as if he were writing today instead of a century
ago, is still the best book available in economic theory. This is a sad
commentary on the economics of our time. Marshalls superiority is
explained primarily by his approach to economics as contrasted with the
modern approach (Lecture Notes, p. 1).
Draft two of the essay and the final published version bear the marks of
Burnss and Stiglers responses to draft one. They were particularly concerned that Friedman would lose his readers by appearing to take an
overly extreme position on the realism of assumptions. They urged and
he responded with two changes. One was to emphasize that his concern
was only with the stage of the research enterprise at which one is testing
theory. He thought that there was little to say about the derivation of
theory.17 The other was to consider positive roles for assumptions.
17
By derivation Friedman means something other than logical derivation from a set of
axioms. For him theory derivation is making use of existing theory and data to come up
with new hypotheses.
88
J. Daniel Hammond
The natural selection example of leaves position around a tree is new

in draft two, fitting in with the two examples carried over from draft one.
The idea of natural selection, however, is not new, having been illustrated
with the billiards player example. The positivenormative distinction
and reference to J. N. Keyness The Scope and Method of Political
Economy are new in draft two and largely peripheral to the heart of the
essay. Friedmans addition of this introduction to the essay has not been
without effects, for some readers of the essay have taken away the association of Friedman and the essay with philosophical positivism. A more apt
title in retrospect might have been The methodology of Marshallian
economics.
References
Alchian, Armen A. (1950). Uncertainty, evolution, and economic theory. Journal
of Political Economy. 58, 21121
Alexander, S. (1951). Issues in business cycle theory raised by Mr. Hicks.
Boland, Lawrence (1979). A critique of Friedmans critics. Journal of Economic
Burns, A. F. (1952). Letter to Milton Friedman, November 26, Box 43, Milton
Friedman Papers, Hoover Institution, Stanford University
Caldwell, Bruce J. (1980). A critique of Friedmans methodological instrumentalism. Southern Economic Journal, 47, 36674
(1982). Beyond Positivism. London: Allen and Unwin
(1992). Friedmans predictivist methodological instrumentalism a modification. In W. J. Samuels (ed.), Research in the History of Economic Thought and
Methodology, vol. 10. Greenwich, CT: JAI Press, 11928
Friedman, Milton (1941). Review of Monopolistic Competition and General
Equilibrium Theory by Robert Triffin. Journal of Farm Economics, 23, 38990
(1946). Lange on price flexibility and employment. American Economic Review.
36, 61331
(1947). Letter to G. J. Stigler, November 19. In Hammond and Hammond (2006)
(1948a). Descriptive validity vs. analytical relevance in economic theory, mimeo,
undated [1948], Box 43, Milton Friedman Papers, Hoover Institution,
Stanford University
(1948b). Letter to G. J. Stigler, October 4. In Hammond and Hammond (2006)
(1949). The Marshallian demand curve. Journal of Political Economy, 57, 46395
(1950). Wesley C. Mitchell as an economic theorist. Journal of Political Economy,
58, 46593
(1952a). Comment. In B. F Haley (ed.), A Survey of Contemporary Economics,
Chicago: Richard D. Irwin, 4557
(1952b). The relevance of economic analysis to prediction and policy, mimeo,
undated, Box 43, Milton Friedman Papers, Hoover Institution, Stanford
University
89
(1953). The methodology of positive economics. In his Essays in Positive

Economics. Chicago: University of Chicago Press, 343
(Undated). Lecture notes Price Theory. Undated, Box 76, Milton Friedman
Papers, Hoover Institution, Stanford University
Friedman, Milton and Leonard Savage (1948). The utility analysis of choices
involving risk. Journal of Political Economy, 56, 270304
(1952). The expected utility hypothesis and the measurability of utility. Journal
of Political Economy, 60, 46374
Gordon, R. A. (1948). Short-period price determination in theory and practice.
Hall, R. J. and C. J. Hitch (1939). Price theory and business behavior. Oxford
Economic Papers, 2, 1245
Hammond, J. Daniel (1992a). An interview with Milton Friedman on methodology. In W. J. Samuels (ed.), Research in the History of Economic Thought and
Methodology, vol. 10, Greenwich, CT: JAI Press, 91118
(1992b). The problem of context for Friedmans methodology. In W. J. Samuels
(ed.), Research in the History of Economic Thought and Methodology, vol. 10,
Greenwich, CT: JAI Press, 12947
Economics. Cambridge and New York: Cambridge University Press
Hammond, Claire H. and J. Daniel Hammond (2006). Making Chicago Price
Theory: Friedman-Stigler Correspondence, 19451958. London: Routledge
Hirsch, Abraham and Neil De Marchi (1990). Milton Friedman: Economics in
Theory and Practice. Ann Arbor: University of Michigan Press
Keynes, John Neville (1891). The Scope and Method of Political Economy. London:
Macmillan
Marshall, Alfred (1920). Principles of Economics, 8th edition London: Macmillan
([1885] 1925). The present position of economics. Reprinted in A. C. Pigou
(ed.), Memorials of Alfred Marshall. London: Macmillan
McCloskey, D. N. (1983). The rhetoric of economics. Journal of Economic
(1985).The Rhetoric of Economics. Madison: University of Wisconsin Press
Stigler, George J. (1948). Letter to Milton Friedman, undated (September 1948).
In Hammond and Hammond (2006)
(1948). Letter to Milton Friedman, undated (October 1948). In Hammond and
Hammond (2006)
(1949). Monopolistic competition in retrospect. In his Five Essays on Economic
Problems. New York: Longmans, Green Co.
Triffin, Robert (1940). Monopolistic Competition and General Equilibrium Theory.
Cambridge, MA: Harvard University Press
University of Chicago. A program of factual research into questions basic to the
formulation of a liberal economic policy. Box 79, Milton Friedman Papers,
Hoover Institution, Stanford University
Walters, Alan (1987). Friedman, Milton. In The New Palgrave Dictionary of
Economics, vol. 2. London: Macmillan
Wong, Stanley (1973). The F-twist and the methodology of Paul Samuelson.
Unrealistic assumptions and unnecessary

confusions: rereading and rewriting F53
as a realist statement
Uskali Mki
Introduction
Depending on the readers intellectual preferences and capacities, the

experience of reading Milton Friedmans 1953 essay on economic methodology (F53 for short) tends to be either one of relief and conviction
or one of irritation and provocation or perhaps some mixture of these
feelings. Many practicing economists have found the message of the essay
appealing and liberating, and its arguments convincing. Most specialists
in economic methodology, and some practicing economists, have found
F53 entirely defective, presenting a dangerously misguided doctrine of
scientific theory and method. Both of these responses may be based on an
incomplete understanding of the essay, or so I want to argue.
I will argue for two sets of general claims. First, there is no unambiguous
doctrine or argument presented in F53. The essay presents no unproblematically identifiable single methodology of economic theorizing. The
message of F53 cannot be captured by popular phrases like Friedmans
instrumentalism or Friedmans as-if methodology or Friedmans
argument in support of unrealistic assumptions or the F-twist. This is
because F53 also contains strong anti-instrumentalist elements; because
as if is itself highly flexible and compatible with a variety of philosophical
outlooks; and because F53 argues for unrealistic (and sometimes for realistic) assumptions from sets of mutually incompatible premises. What the
reader is served is an F-mix, a mixture of ingredients many of which are
ambiguous and some of which are hard to reconcile with one another. In
consequence, a variety of readers with different intellectual tastes will be
This is an expanded and revised version of a paper presented at the conference on F53 at
Erasmus University of Rotterdam in December 2003. And that version was a much
expanded version of the one presented at the ASSA meetings in Washington, DC, in
January 2003. After the Erasmus conference, it was presented at the universities of Aix-enProvence, Freiburg, Athens, and Lund. Useful comments from the various audiences are
acknowledged. In his characteristically generous fashion, Milton Friedman offered some
written comments that have been consequential.
90
Rereading and rewriting F53 as a realist statement
91
able to find in F53 their own selection of ideas that they will endorse or
oppose.
Second, the flexibility and malleability of the textual material creates room
for alternative readings of F53 that challenge its current reified textbook
image. In particular, it is possible to develop non-traditional interpretations
that might bring many former friends and foes of F53 or, more precisely, of
what has been believed to be the message of F53 closer to one another, if
not unite them entirely. In what follows, I will remove some unnecessary
confusions and ambiguities in F53, and by doing so, hope to pave the
way for a new sensible reading that will be found much less objectionable
by many more readers, practitioners and methodologists alike. To the
extent that my rereading fails to be a matter of unbiased discovery of what
is already there, hidden in the text of F53, it can also be taken as a project
of rewriting the essay. It is a matter of rewriting by selection and correction so as to eliminate its flaws and to make it more agreeable to a variety
of audiences. On this rereading (or rewriting) F53 emerges as a realist
(rather than instrumentalist) manifesto with strong fallibilist and social
constructivist sensitivities (in contrast to standard textbook positivism).1
One might object by saying that my rereading F53 as a realist and social
constructivist statement amounts to a reformulation beyond recognition,
or at least to a forced interpretation that does not come naturally. In
contrast, my conviction is that what I am about to offer is an obvious
interpretation that does come very naturally indeed. My interpretation is
more forced when ignoring parts of what F53 seems to be saying, while it
is less forced when based just on highlighting elements in F53 that have
been overlooked by other readers. But I must grant there is a possible
source of bias: the image of economics that emerges in F53 as it is being
reread and rewritten here is one that I myself endorse. I am rebuilding F53
as a statement that captures and conveys this rebuilders own methodological convictions.
A remark is needed on the strategy and motivation of this chapter by
way of locating it on the map of the various readings of F53 (the map was
outlined in chapter 1 above). First, mine is primarily an examination of
F53 itself rather than explicitly using F53 for some other purpose (though
I think it wonderfully serves other useful purposes). Second, my reading
goes beyond the narrow logical accounts (in terms of the relationship
between assumptions and predictions) and incorporates ontological,
semantic, and social ingredients in the image that F53 conveys of
1
This chapter follows the line of, and draws on, earlier work such as Mki 1986, 1989, 1992,
2000, 2003. These papers document my career in the F53 exegesis as one of swimming
against the stream.
92
Uskali Mki
economics. Third, the reading combines normative and descriptive perspectives in that it suggests remedying some flaws of F53 so as to identify
and reconstruct its sensible core ideas. Fourth, it attaches extra-economic
labels to F53, borrowed from philosophy and science studies. Finally, and
importantly, my reading is an exercise in what I called reception methodology: the focus lies on the context of consumption of F53 rather than its
production by its author. The context is that of the readers reception
rather than that of the authors intention. Reception methodology here
amounts to reading F53 from the point of view of its actual and possible
interpretations and influences among the relevant audiences.
It is part of my receptionist reading of the essay that I make no claims
about the authors intentions: I am not after Friedmans authentic convictions or anything related to what Friedman really meant (in fact I am
not convinced there are such things there to be discovered, at least in
F53 alone).2 The focus here is on the text of F53 while the authors
intentions and beliefs are bracketed. As a technical expression of this
I am not referring to the author but rather to the text: I am not saying
here Friedman argues but rather, F53 claims; not Friedmans
opinion seems to be but rather, F53 can be read as, and so on. My
attempt to rewrite F53 so as to make it more acceptable to various
audiences is also independent of anything that Milton Friedman the
author might have believed or intended, or of whether he would accept
the rewritten version of F53 as sound (even though I believe he would
have ample reason to accept it).
Let it be added that the receptionist reading of F53 is also consistent with,
and supported by, Milton Friedmans personal strategy in relation to F53, as
repeated in his statement at the end of this volume: he has chosen not to
respond to any commentaries on F53 by others, but has rather let the text
live its own life in the intellectual realm, at the mercy of its readers, as it were.
Assumptions and predictions, and the non-predictivist

torso
If there is one basic thesis conveyed by F53, it is this: economic theories

should not be judged by their assumptions but by their predictive implications
and in particular, the unrealisticness of the assumptions of a theory is no reason
2
The only way to determine what Friedmans possible authentic methodological convictions are is to read F53 together with his other methodological remarks conjoined
with a scrutiny of the implicit methodology of his work in economics. This is a very
legitimate historical and analytical project, but it does not lie at the heart of my reception
methodology.
93
for complaint or worry about the theory. The methodological advice given to
economists by F53 appears simple: when assessing a theory, dont pay
separate attention to its assumptions, instead focus on the predictions it
yields. As the much-cited statement puts it, the only relevant test of the
validity of a hypothesis is comparison of its predictions with experience.
The hypothesis is rejected if its predictions are contradicted (frequently
or more often than predictions from an alternative hypothesis); it is
accepted if its predictions are not contradicted (F53, 89). This passage
makes three important points about testing scientific theories: testing is by
predictive implications, not by assumptions; failed predictions play a key
role: acceptances are just failures to be rejected; and testing is comparative: what matters is the predictive performance of a theory relative to that
of alternative theories.3
It is clear that the main focus of attention of the text of F53 is on
assumptions, not on predictions. The historical context of F53 essentially
contained various attacks against some of the key assumptions of marginalist theory, blaming them for harmful unrealisticness. F53 sets out to
convince the skeptical or hesitant reader that unrealistic assumptions are
just fine, and the arguments of F53 revolve around this idea. Yet, in some
cases the reader would expect to be shown the full force of the basic thesis
in its entirety, including the role of comparative predictive performance.
Consider the treatment of Edward Chamberlins theory of monopolistic competition. F53 does not hide its hostility towards Chamberlins
theory an inferior theory that economics does not need at all in addition
to the superior simple models of perfect competition and perfect monopoly. The argument of F53 is straightforward: the creation of the theory of
monopolistic competition was explicitly motivated, and its wide acceptance and approval largely explained, by the belief that the assumptions of
perfect competition or perfect monopoly said to underlie neoclassical
economic theory are a false image of reality (15), and it is this misguided
motivation and flawed basis of acceptance that speaks against the theory.
What is noteworthy is that there is no appeal here to the superior predictive capacity of Friedmans favorite theories in contrast to the predictive
failures of Chamberlins theory. The realisticness or unrealisticness of
assumptions was not supposed to matter, but it seems they do, after all.
The key thesis of F53 is thereby turned into a modified non-predictivist
3
In these formulations, one may choose to hear echoes of Poppers doctrine of refutation
based on failed predictions as well as Lakatoss doctrine of comparative and dynamic
predictive performance. The English translation of Poppers Logik der Forschung appeared
in 1959 (Popper 1959), while Lakatoss relevant essays were published a decade or so later
(e.g. Lakatos 1970). Friedman had learnt about Poppers falsificationist views when
personally meeting him in the late 1940s.
94
Uskali Mki
torso: hail unrealistic assumptions, proscribe against the pursuit of realistic

assumptions.
This I believe is how many practicing economists have received the
message of F53. This explains its emancipatory effect on economists: it
helps liberate those employing models with unrealistic assumptions from
a sense of unease (see Mayer 1993). This also suggests why there is an easy
link between F53 and the defense of blackboard economics that is
mathematically highly refined and rigorous but accused of being unconnected to real-world facts and issues (cf. Hands 2003 and in this volume).
Whatever the methodological preferences of the author of F53 in fact we
know Friedman is an opponent of formalistic blackboard economics it is
not surprising to see appeals to F53 in justification of the assumptions that
help create the model worlds on the blackboard. From the point of view of
the context of consumption, it is the reception and use of F53 that matters.
A tension between the authors intended methodology and the use of the
torso version of the basic thesis of F53 does not lie within the purview of
reception methodology.
3
Unrealisticness as irrelevant and as a virtue
Friedmans claim about unrealistic assumptions appears in at least two

versions: one is in terms of irrelevance, the other is in terms of virtue. The
weaker version is the claim that unrealisticness is irrelevant for the goodness of theory: that it does not matter even if a theorys assumptions
are unrealistic. Now it is important to see that a consistent irrelevance
thesis would have to imply that the actual degree of both unrealisticness and
realisticness does not matter. In other words, no matter how unrealistic or
realistic the assumptions of a theory, this is irrelevant to the assessment of
the theory, since those properties are not among the ones we should
measure when evaluating a theory. But the bulk of F53 is not symmetric
in this way, thus not in line with the weak version: the essay keeps stressing
that it is high degrees of unrealisticness that do not matter. As the attack
in F53 against more realistic behavioral assumptions and the theory of
monopolistic competition indicates, the pursuit of high degrees of realisticness is regarded as a serious demerit of both a theorist who does so and
of a theory that is pursued. This contradicts the weak thesis in terms of
irrelevance.
The strong version acknowledges the asymmetry of F53 and does it in a
radicalized manner: it is the claim that unrealisticness is a virtue, that the
better the theory, the more unrealistic it is in its assumptions. Many
readers have found the strong version unacceptable, even outrageous.
Indeed, as a general rule, it must be mistaken. Even in some cases
95
where violating the truth may seem recommendable such as in assuming

a vacuum for falling cannon balls and assuming profit maximization for
certain purposes the strong version goes too far. F53s examples of
excellent scientific theories assume zero air pressure and profit maximization. The strong version suggests that there might be even better theories that assume that air pressure is infinitely large and that businessmen
aim at maximizing their losses these assumptions would be more unrealistic than the ordinary ones. But obviously, such unrealistic assumptions
would not be epistemically virtuous, thus the strong version is questioned.
There are obvious cases in which both versions appear questionable. In
the study of the used cars market, it is neither irrelevant nor virtuous for a
theory to falsely assume that information is symmetric, and in the study of
the computer software industry, one is not well advised to assume diminishing returns. Qualifications and further conditions are needed: neither
version of Friedmans thesis can be defended as a general principle. I will
have more to say about this in later sections.
Indirectly testing approximation to the truth
On closer scrutiny, it appears that F53 does not subscribe to either of the
above versions of the thesis consistently or without qualifications. Indeed,
the truth of assumptions appears as a relevant issue after all. F53 indicates
an interest in truth when suggesting that truth is something to be estimated: indeed, predictive tests serve as indirect tests of the approximate
truth of assumptions. The required degree of approximation is not simply
to be maximized (or minimized!) but is relative to the purposes that the
theory is supposed to serve: the relevant question to ask about the
assumptions of a theory is whether they are sufficiently good approximations for the purpose at hand (15). So there is a relevant question to
ask about assumptions, and the question is about their approximate truth.
The way to measure whether the required degree of approximation has
been achieved is to put the theory to the predictive test:
Complete realism [realisticness] is clearly unattainable, and the question
whether a theory is realistic enough can be settled only by seeing whether it
yields predictions that are good enough for the purpose in hand or that are better
than predictions from alternative theories. (41)
This implies that the unrealisticness of assumptions is not an irrelevant

issue at all, something to be ignored. On the contrary, one is advised to
pay attention to their actual degree of realisticness and to judge whether it
is sufficiently high for the purposes at hand. It appears that two kinds
of considerations shape these judgments: pragmatic and ontological.
96
Uskali Mki
Pragmatic considerations enter in the form of purposes: the appropriate

degree of (un)realisticness is relative to the purposes for which the theory
is put. Ontological considerations are concerned with the causal powerfulness of various factors: some factors are too weak to be included in the
model, while others play major causal roles and should be included. In
other words, it all depends on the difference those factors make:
Why is it more unrealistic in analyzing business behavior to neglect the magnitude of businessmens costs than the color of their eyes? The obvious answer is
because the first makes more difference to business behavior than the second; but
there is no way of knowing that this is so simply by observing that businessmen do
have costs of different magnitude and eyes of different color. Clearly it can only be
known by comparing the effect on the discrepancy between actual and predicted
behavior of taking the one factor or the other into account. (33)
Indeed, not only is it more unrealistic to ignore the magnitude of

businessmens costs than the color of their eyes, but such unrealisticness
also matters a great deal. It is advisable to consider costs but not the
eye color because they make a difference for the phenomena being
explained. And to determine whether or not they make a difference
relative to some purpose requires checking the predictive implications.
Another way of putting these ideas is to say that some of the assumptions
of a theory are to be paraphrased as statements about the negligibility of a
factor, and that predictive tests are a way of assessing such claims about
negligibility (Musgrave 1981; Mki 2000). I will return to this idea in a
later section.
Testing by assumptions and testing by predictions
F53 itself, and indeed most of the commentators on the issue, draw a
distinction between testing a theory by its assumptions and testing it by its
predictions. Accordingly, the participants in the debate are classified as
the assumptionists and the predictivists, the latter often also called
instrumentalists (Ill come back to the uses and abuses of this label later).
The assumptionists are supposed to test the assumptions directly and
thereby to test the theory, while the predictivists test the theory indirectly via its predictive implications.
Without qualifications, this is a misleading distinction. No direct
testing of assumptions is available. This should be easy to see: when one
seeks to test an assumption, one has to construe an argument in which that
assumption serves as one of the major premises and which entails a
predictive implication that one then compares with evidence. When testing the assumption of transitive preferences, one conjoins it with several
97
other assumptions and draws implications concerning choice behavior,

and then checks those implications against, say, experimental data.
I think the intended difference is between two or more domains of
testing by implications. Economists hold different views as to the relevant
domain of data that should be consulted as evidentially relevant. Thus,
some hold the view that the maximization assumption is to be tested
against the data concerning the individual behavior of business managers.
Others believe the assumption, and the theory of which it is a constitutive
element, should only be tested by drawing implications concerning market data. In both cases, one tests by predictive implications.
6
The whole truth and nothing but the truth
F53 uses a variety of expressions to refer to species of unrealisticness.

These include descriptively false, descriptively unrealistic, and
descriptively inaccurate (pp. 1415). None of them is being defined
in F53 other than through examples. But the examples reveal fatal
ambiguities. In particular, F53 conflates falsehood proper (violation of
nothing-but-the-truth) and incompleteness or narrowness (violation
of the whole truth). These are separate notions, but F53 fails to keep
them separate and so does much of other commentary on the realisticness issue. In particular, the violation of the whole truth does not imply
the violation of nothing but the truth about some part of the whole. Here
we come across with one major limitation of the logical accounts: they
miss the distinction between the whole truth and nothing but the truth,
focusing just on the latter and dealing with the former as if it were
reducible to the latter.
Consider an example. One of the targets that F53 seeks to rebut consists
of the criticisms of the maximization assumption in the marginalist
controversies launched by Hall and Hitch in the UK and Lester in the
USA. Friedman invites the reader to consider the recent criticisms of the
maximization-of-returns hypothesis on the grounds that businessmen
do not and indeed cannot behave as the theory assumes they do (31).
F53 here formulates the message of the criticism as the charge that the
maximization assumption violates nothing but the truth about business
behavior: businessmen do not, perhaps cannot, behave the way they are
assumed to behave.
Right after this, F53 sketches a completely realistic theory of the
wheat market that would have to mention the color of the traders
and farmers eyes and hair, antecedents and education, the physical and
chemical characteristics of the soil on which the wheat was grown, the
weather prevailing during the growing season, etc., etc. (32). Another
98
Uskali Mki
notion of realisticness is implied by such a completely realistic theory:

deviations from such a theory by leaving out various details would be
a matter of violating the whole truth about a certain subject. One can
easily imagine any number of further details that would have to be mentioned in such a theory of the wheat market, including the detailed career
of the causal and constitutive ancestors of each farmer, trader, parcel of
land, and seed of wheat during the process of cosmological, geological,
biological, and cultural evolution over the course of billions of years.
F53 implies a close connection or sameness between these two notions
of truth when suggesting that to criticize the maximization assumption is
as ridiculous as to insist on building such a completely realistic theory.
The reasoning is not sound: it is trivial that nobody will insist on having
such a theory but it also should be trivial that this has nothing to do
with the insistence that it is no recommendation for the maximization of
returns assumption as such that it violates nothing but the truth. The
violation of the whole truth (which is unavoidable) implies nothing about
the violation of nothing but the truth (which is avoidable).4 Only a small
fraction of all facts (even of the causally efficacious ones) are mentioned by
any theory, hence the violation of the whole truth; but statements about
those facts such as whether actors in the wheat market are maximizers
are not necessarily false. One cannot justify false assumptions by citing the
trivial fact that all theories are necessarily incomplete.
F53 thus conflates the whole truth with nothing but the truth. This
conflation may be seen to serve a rhetorical function: just like every
sensible person, my dear reader, you surely agree that the insistence on
a completely realistic theory is undesirable, therefore you are naturally
reasonable enough also to agree that to criticize the maximization assumption for falsehood is misguided.
Why is it important to resist this conflation? Keeping the two notions
distinct has implications for the assessment of the basic thesis of F53. One
may argue that good theories violate the whole (or comprehensive) truth
in that they isolate just narrow slices of the world and leave out most of it
such as most of the things listed in the completely realistic theory of the
wheat market. One may go further and argue that, subject to further
(perhaps ontological and pragmatic) constraints, the more a theory leaves
out the better it is. One may also point out that such theoretical isolations
are often accomplished by way of false idealizing assumptions (such as
those of a vacuum or homogeneous goods) that help exclude factors (such
as actual air pressure or product differentiation) that are viewed as
4
Many qualifications would have to be added to the two claims about avoidability.
99
irrelevant for a problem at hand. These are some of the ways in which
good theories may be unrealistic. More about this in a moment.
Assumptions and their roles: vacuum

and maximization
Here is another limitation of the logical accounts: they miss a relevant and
sufficiently rich idea of sentence role. In the logical accounts, the only
conceivable roles of sentences are those of premises and conclusions in
logical arguments. In order to understand economic theorizing, we need
to have a more qualified notion of role.
One of F53s famous examples is Galileos law of freely falling bodies.
In the formulation used, the law states that s = 1/2 gt2. One of the
assumptions behind the law is that of a vacuum:
[1] Air pressure = 0
F53 argues that for many purposes (such as predicting the fall of a cannon
ball) it does not matter even if the vacuum assumption is false. It then uses
this as an analogy in its defense of the maximization assumption:
[2] Producers and traders pursue maximum expected returns.
F53 argues that just as, in many cases, it does not matter if the vacuum
assumption is false, it does not matter even if the maximization assumption is false. This argument exploits a mistaken analogy, based on ignoring
the roles that these two assumptions [1] and [2] are supposed to play
within their home theories.
The purpose that assumption [1] serves is similar to that of [3]:
[3] All forces other than gravitation = 0
Assumptions [1] and [3] are idealizations. They idealize by assuming

that certain forces or causes are powerless, they have zero strength: no air
pressure, no magnetic forces, no winds, no other kinds of pushes and pulls
such as the pull of the Moon. The assumptions are unrealistic in that, in
the presently actual world, [3] is always false (for example, the pull of the
Moon, the Sun, and other planets are always there), while [1] is mostly
false but may sometimes be true if made true (vacuum conditions can be
artificially created). By means of such falsehoods, a model is constructed
in which all causes of the behavior of a falling body except for one are
neutralized. The one cause that remains non-neutralized is the gravity of
the Earth. The idealizing assumptions play the important role of helping
to isolate the impact of gravitation from other causal influences.
100
Uskali Mki
What about the profit maximization assumption? Is it also an idealization

that helps isolate one cause while neutralizing the influence of other causes?
One manifest difference between [2] and [1]&[3] is this: while [1]&[3]
explicitly refer to the factors that they serve to eliminate or neutralize, [2]
does not seek to neutralize what it is about, namely profit maximization.
Indeed, rather than removing the impact of profit maximization, it establishes profit maximization as a major motive of economically relevant
behavior. Instead of saying agents do not maximize, it says they do. The
analogy between maximization and vacuum seems to collapse.
If one were to formulate an analogue of assumptions [1] and [3] in
Galileos law, it would not be [2] as F53 suggests, but rather something like
[4] All other motives except the maximization motive have zero strength.
This is an idealization that helps isolate the motive of the maximization

of expected returns from all other possible motives (such as moral deliberations, sense of security, fairness, reputation, maximization of sales and/
or market share, etc.) in order to make it possible to examine the impact of
profit maximization on market phenomena in the absence of such other
motives. Maximization would be isolated as the motive rather than just a
motive among others. Now one may argue that this indeed is intended
already by assumption [2]: by assuming that firms are motivated by
expected maximum returns one thereby assumes that they are motivated
by nothing but expected maximum returns. In other words, assumption
[2] is intended to have assumption [4] as its implicit component. But this
does not save the suggested analogy from my rebuttal. In our more literal
moods, we might want to say that the profit maximization assumption is a
combination of [2] and [4]. This will help us retain the claim that there is
an analogy between [1] and [4] but not between [1] and [2].
The analogue of component [2] of the maximization assumption in
Galileos law is not the vacuum assumption but instead something like
[5] The gravitational field of the Earth pulls physical objects by force g.
It is the gravitation of the Earth that is being isolated by means of

idealizing assumptions [1] and [3]. Likewise, it is the drive to maximize
that is being isolated by means of idealization [4]. Insofar as truth and
falsehood are concerned, it is an entirely different matter for assumptions [2] and [5] to be true or false, and for assumptions [4] and [1]&[3]
to be true or false. Moreover, the truth or falsehood of assumptions [1] and
[3] does not imply anything about the truth of [5]. Statement [5] may be
true even though [1] and [3] are false, and even if they are false to the extent
that Galileos law were to yield radically incorrect predictions, such as in the
case of a falling feather in the weekend market of my hometown plaza.
101
If the analogy between [2] and [5] were to be adequate in relevant respects,
one might argue that nothing follows from the acceptability, believability, or
truth of [4] for the respective attributes of [2] in analogy with the similar
connection between [1]&[3] and [5]. One might then say that agents may be
profit maximizers even though this is not the only thing that motivates them
in business transactions: [2] might be true even if [4] is false. Obviously, this
is not at all evident, and certainly would invite a lot of further qualifications.5
Anyway, the point relates to the important distinction to be made next.
8
Truth of assumptions and truth of theory
Most commentaries of F53 are based on the presumption that one can
determine the truth-value of a theory or model by way of checking the
truth-values of its assumptions. This is because the truth-value of a theory
is a function of the truth-value of its assumptions, such as the truth-value of a
theory being equal to the truth-value of the conjunction of its assumptions.
On this view, the falsehood of assumptions can be taken to imply the falsehood of the theory. F53 admits that many of the central assumptions of the
economic theories challenged by their critics are false. This has been interpreted by later commentators as implying a commitment to a special kind
of instrumentalist view of theory: economic theories and models are false
instruments of prediction (e.g. Wong 1973; Boland 1979; Caldwell 1992).
F53 itself performs better, and we can see this if we look a bit more
carefully at the relevant formulations. F53 uses expressions such as a theory
being descriptively false in its assumptions (14). We are not compelled to
read this as admitting that theories with false assumptions are false. Indeed, it
has been my contention that theories with false assumptions may be true,
and that realism (as a theory of theories) is perfectly comfortable with
unrealistic assumptions (e.g. Mki 1992). The truth-value of a theory
cannot be read off the truth-values of its assumptions. The key to understanding the gap between the two is to ask the question: what is the theory
about, what claim does it make about the world, if any? A theory may be
true about the functioning of some important causal factor while making
false assumptions about the existence and functioning of other factors.
Galileos law is true about, well, what it is about: namely the causal role of
gravity in shaping the behavior of falling bodies. The core assertions of the
5
The argument in the text about profit maximization and its analogues is in terms of
motives. Qualifications would be needed if other versions of the profit maximization
assumption were employed, such as one in terms of deliberate marginal calculations and
their implementations in maximizing behavior; and another in terms of behavior that ends
up with maximizing the profits of a firm regardless of the motives and mental operations of
its managers.
102
Uskali Mki
law are not about air pressure, magnetic forces, or the shape of the Earth,
but about the gravitational field of the Earth in relation to freely falling
bodies. False assumptions about factors other than gravity imply nothing
about the truth-value of the law statement itself.
Falsehood is a friend of truth. Not only does the concept of truth require
that of falsehood: what is not true is false. Not only does truth often
emerge as false beliefs are rejected and replaced by true ones. My somewhat radical statement is that one may claim to have identified significant
truths about the world by using a theory with false assumptions. In such
cases one characteristically uses those idealizing assumptions to eliminate
from consideration some less significant factors and thereby isolates, and
helps focus on, what are believed to be the most significant causes and
relationships. This is an important feature of the methodological image of
theory and method held by many practicing economists. This is part of the
realism I am reading in, or writing into, F53.
9
Manipulation of truth-values by paraphrasing

assumptions
The previous section suggested that a model with false assumptions may
be true. Let us next develop another way of reconciling truth and falsehood by reconsidering the falsehood of assumptions from a different
perspective. This is based on Alan Musgraves contribution (Musgrave
1981) and my suggested modifications and elaborations (Mki 2000).
The trick is to relocate truth claims by paraphrasing assumptions. The point
is this: one should not criticize what appears as a false assumption without
understanding what assertion is intended when using it. The previous
section focused on the truth claims made by means of models, while here
the focus lies on the truth claims intended in making assumptions.
On a correct understanding of what claims are intended when putting
forth this or that assumption, assumptions taken as such claims should be
true. If needed, one uses a paraphrase to determine the intended claim
when making an assumption. When suitably paraphrased, the assumptions of a theory had better be true. Such paraphrasings include turning
an assumption into an assertion of the negligibility of a factor or of the
applicability of a model to a domain of phenomena. The vacuum assumption in connection to predicting the behavior of a freely falling cannon ball
can be paraphrased as the statement that the impact of actual air pressure
is negligible, given ones purposes (while it is not negligible in the case of a
falling feather): the deviation from the truth is negligibly small, or the
degree of approximation to the truth is sufficiently close. Such statements
about negligibility (the causal significance of deviations from the vacuum
103
is negligible for the purpose at hand) should be required to be perfectly

true while the non-paraphrased assumptions (there is a vacuum)
provide more (even much more) or less distant approximations to the
truth. This difference is implicit in F53.
F53 says that for certain purposes (such as prediction with a certain
accuracy) and for certain falling bodies (such as cannon balls), the actual
impact of air resistance is negligibly small, therefore assuming that those
bodies fall in a vacuum is all right and Galileos law applies. F53 also
suggests that for some bodies, such as feathers, air resistance makes a nonnegligible difference, therefore the vacuum assumption is not justified.
This suggests using the vacuum assumption as an applicability assumption saying that the simplest formulation of the law of freely falling bodies
only applies in cases where actual air pressure has negligibly small consequences for the behavior of falling bodies. An applicability assumption
makes claims about the applicability of a model or law statement to some
domain. Claims about applicability had better be true, just as claims about
negligibility should be true.
These are all very reasonable claims and they indicate an appreciation
for the role that judgments about matters of fact play in theorizing. These
considerations do not fit smoothly with an instrumentalist view of theory.
They are rather realist considerations, but they do not require the nonparaphrased (yet paraphrasable) assumptions to be perfectly or even
moderately realistic. Yet they stress the principle that the appropriately
paraphrased assumptions had better be realistic.
In this connection, it is noteworthy that the specific Marshallian (or
anti-Walrasian) methodological preferences that lie behind F53 have little
appreciation for considerations of formal tractability as a constraint on
theorizing in contrast to more mathematically oriented styles of doing
economics (on tractability issues, see Hindriks 2005). The latter might
include paraphrasing an assumption that, taken as such, is false, as an
assertion that says: the use of the assumption enhances the formal tractability of the model. Such an assertion is about the intrinsic properties of a
model in relation to the available formal tools and skills, not about the
properties of the real world. Even if this strategy were to generate a
perfectly true assertion, it would not satisfy those Marshallian methodological preferences. This is also in line with a realist outlook: true
tractability assumptions do not describe facts about economic reality,
nor are they easily justifiable by reference to such facts.
The above reasoning was not tied to any particular assumptions. But
there is one particular assumption that F53 briefly discusses and suggests
justifying in a way that involves the paraphrase technique. Again, the idea
is that the assumption only appears to make false assertions about the
104
Uskali Mki
world, but when suitably paraphrased, falsehood is replaced by truth. The

assumption in question is profit maximization, understood literally as
dealing with goals and computations taking place in businessmens
extended minds: by going through a series of marginalist calculations,
they consciously infer what the maximizing behavior is, then take action
accordingly. The paraphrase of this assumption dispenses with the decision procedures behind overt behavior, and makes claims about that
behavior only. The assertion now made is about behavior that ends up
with maximum profits rather than about profit maximizing motives and
calculations. F53 indicates optimism that so paraphrased, the assumption
has a fair chance of being (at least approximately) true.
Moreover, F53 suggests justifying the assumption in terms of an evolutionary argument that offers indirect evidence in its support. The
mechanism of natural selection in competitive markets ensures that only
firms exhibiting profit maximizing behavior survive: unless the behavior
of businessmen in some way or other approximated behavior consistent
with the maximization of returns, it seems unlikely they would remain in
business for long (22). On this reasoning, it does not matter what precise
computational processes have preceded and generated overt business
behavior. Businessmen only behave as if they were deliberately maximizing. The realist response to this argument would be in two parts. First, just
as Koopmans (1957) insisted long ago, the evolutionary argument should
be taken to imply the obligation to develop an account of the evolutionary
mechanisms themselves as one core target of theory. Second, one should
have an account of the structure of the as-if locution that connects
profit maximization as a deliberately pursued goal and as manifest
behavior.
10
Two kinds of as if
It seems to be generally believed that the use of the as-if locution in

formulating theoretical claims commits one to an anti-realist (fictionalist
or instrumentalist) view about scientific theory. This is a mistake. The asif formulation is a flexible tool that can be used for expressing a number of
ideas about a theory and its relationship to the world. F53 itself appears to
be of two minds as to the import of as-if. It is one thing to say that
(a) phenomena behave as if certain ideal conditions were met, namely conditions
under which only the theoretically isolated real forces are active;
and it is quite another thing to say that

(b) phenomena behave as if those forces were real.
105
The difference between the two is striking and has important philosophical implications. Of the two, (a) is in the spirit of realism, while (b)
invites a fictionalist and instrumentalist reading. Formulation (a) says that
the behavior of certain phenomena is shaped by a real force isolated by the
theory, and that it is to be treated without considering the role of other
factors, suggesting that those phenomena behave as if the theoretical
isolation were materialized also in the real world. This is what F53 is
effectively saying in one of its general passages:
A meaningful scientific hypothesis or theory typically asserts that certain forces
are, and other forces are not, important in understanding a particular class of
phenomena. It is frequently convenient to present such a hypothesis by stating that
the phenomena it is desired to predict behave in the world of observation as if they
occurred in a hypothetical and highly simplified world containing only the forces
that the hypothesis asserts to be important. (40)
That this is in line with (a) should be evident. There are a number of
causally relevant forces in the world. But they are not equally relevant,
thus the economist proceeds by assuming that it is just the most powerful
forces that shape the explanandum phenomenon, that they function in
isolation from all other forces: the phenomenon occurs as if nothing but
those powerful forces were in operation. The reality of none of these forces
is denied. If one wants to identify a fiction here, it is the isolation that is a
fiction, not the isolated forces (Mki 1992, 2004).
In its concrete illustrations, on the other hand, F53 is closer to formulation (b). The imaginative hypothesis of the leaves of a tree is the most
striking of its examples. F53 suggests the hypothesis that the leaves
[around a tree] are positioned as if each leaf deliberately sought to maximize the amount of sunlight it receives, given the position of its neighbors,
as if it knew the physical laws determining the amount of sunlight that
would be received in various positions and could move rapidly or instantaneously from any one position to any other desired and unoccupied
position (19). However, so far as we know, leaves do not deliberate
or consciously seek, have not been at school and learned the relevant laws
of science or the mathematics required to calculate the optimum position,
and cannot move from position to position (20). In contrast to the general
statement cited above, this suggests that the forces cited after as if are not
real: we are fully aware of the fictionality of the forces postulated, and this
awareness is expressed in terms of the as if. This represents a fictionalist
use of as if. Our realist rereading of F53 will ignore this passage.
Note that the fictionalist use of the as if rules out yet another way of
using the as if, namely its epistemological use in expressing uncertainty.
In this case, what follows an as if is a hypothesis describing one possible
106
Uskali Mki
causal trajectory, and it is taken to be the task of further research to

establish whether what seems possible is also actually the case. This is
not an option for the fictionalist use of the as if but can (and should) be
accommodated in a realist account. (Mki 1998)
11
Theoretical isolation and explanatory unification
In line with the realist use of the as if, F53 contains passages suggesting
that theory construction is a matter of theoretical isolation whereby economists abstract essential features of complex reality (7). This is a widely
endorsed idea in economics and elsewhere: the real world is complex,
therefore we need to build simple models that theoretically isolate causally
significant aspects of the world. This general idea also provides justification for the employment of false assumptions in such models. The simplest version of Galileos law isolates the significant impact of gravity on
the falling body. It excludes the impact of other forces by false idealizations such as those of vacuum, absence of magnetic forces, flatness of
the earth. On this picture, it is the task of false assumptions to help isolate
major causal factors.
F53s remarks about these matters seem to be ontologically motivated.
The economist is supposed to theoretically isolate essential features of
complex reality which suggests that it is a (higher-order) feature of
reality itself that some of its features are essential. The same idea is
conveyed by another important passage in F53. This one introduces the
realist notions of deceptive appearances and ontologically grounded
explanatory unification:
A fundamental hypothesis of science is that appearances are deceptive and that
there is a way of looking at or interpreting or organizing the evidence that will
reveal superficially disconnected and diverse phenomena to be manifestations of a
more fundamental and relatively simple structure. (33)
One is invited to read this roughly as follows. The world is, really, not as it
appears to be: appearances are deceptive manifestations of more fundamental structures. Scientific theories are required to capture those fundamental
structures. This suggests a distinction between reality and appearance. To
this the passage adds the notion of explanatory unification, the idea of
theoretically describing what appears diverse as really constituting a unity.
A theory unifies apparently disconnected phenomena by showing them to be
manifestations of the same fundamental structure. Unification amounts to
showing that those phenomena are really connected and only apparently
disconnected, and this is accomplished by successfully representing how
things are related in the way the world works. This is a matter of ontological
107
unification (Mki 2001), and is only attainable by using a theory that

truthfully manages to isolate the key causes and relations underlying such
apparently disparate phenomena. This gives us a realist notion of explanatory unification.
Once again, the leaves-of-a-tree example offers a different story. It, too,
appeals to the virtue of unification. Indeed, in F53 the hypothesis that
the leaves of a tree actively and deliberately seek some goals emerges,
after all, as inferior precisely because it does not exhibit the virtue of
unifying power. The alternative hypothesis according to which the leaves
only passively adapt to external circumstances is presented as superior
because it unifies, because it is part of a more general theory that applies
to a wider variety of phenomena, of which the position of leaves around a
tree is a special case, has more implications capable of being contradicted,
and has failed to be contradicted under a wider variety of circumstances
(20). This passage, taken alone, would be compatible with the realist
notion of ontological unification. But the account given by F53 about
this example departs from the realist view after all: even though F53 is
explicit that the power to unify constitutes a significant criterion of theory
choice, realist hopes are undermined as no role is supposed to be played
by what is justifiably believed to be the case in the real-world causation of
leaves behavior. The argument here in F53 ignores the fact that the core
assumptions of the activist hypothesis are false and those of the passivist hypothesis are more realistic in isolating the key lines of that
causation (20). What is claimed to be decisive is not related to what causal
processes are actually believed to take place in nature, but rather to the
range of conclusions that can be derived from the two hypotheses and the
range of theoretic connections they exhibit. What the argument appeals to
is derivational unification (Mki 2001): the power of theories to generate
large sets of logical conclusions. The power of theories to describe what is
really going on and what real unity there might be in the world is
viewed as irrelevant, thus this argument is a non-realist one.6 In our realist
rereading of F53 we simply put its treatment of this example aside.
12
Underdetermination and theoretical virtues
One of the troubling logical features of theorizing is that theory choice is

underdetermined by empirical evidence. No body of evidence is sufficient
for determining a unique choice of theory. Yet another reason for reading
F53 as an up-to-date statement of methodology is that it indicates
6
In the philosophy of science, the current interest in explanatory unification started evolving
after Michael Friedmans 1974 article and Philip Kitchers articles in 1976 and 1981.
108
Uskali Mki
admirable awareness of the underdetermination issue:7 predictive success

of a theory is not by itself a sufficient criterion for choosing among
alternative hypotheses (9). F53 then explains why this is so:
Observed facts are necessarily finite in number; possible hypotheses infinite. If
there is one hypothesis that is consistent with the available evidence, there are
always an infinite number that are. For example, suppose a specific excise tax on a
particular commodity produces a rise in price equal to the amount of the tax. This
is consistent with competitive conditions, a stable demand curve, and a horizontal
and stable supply curve. But it is also consistent with competitive conditions and a
positively or negatively sloping supply curve with the required compensating shift
in the demand curve or the supply curve; with monopolistic conditions, constant
marginal costs, and stable demand curve, of the particular shape required to
produce this result; and so on indefinitely. (9)
This is to say that the observational implications of an infinite number of

alternative hypotheses are identical; in the example, they all imply a rise in
price equal to the tax. F53 thus subscribes to what is generally recognized
as the thesis of observational equivalence between theoretical hypotheses.
This thesis has opened up a strategic opportunity for anti-realists to
undermine realist hopes for rationally believing scientific theories to be
true. Such anti-realists have used the opportunity by inferring from the
thesis of observational equivalence to the thesis of epistemic equivalence:
all the observationally equivalent theoretical hypotheses are also equally
believable. Given any number of observationally equivalent hypotheses,
there is no rational way of believing one of them to be true or closer to the
truth than the others (or even the stronger thesis that none of them
actually is true or closer to the truth). The choice between epistemically
equivalent hypotheses is epistemically arbitrary and must be based on
non-epistemic factors. Theory choice is not only empirically undecidable
but also epistemically undecidable.
F53 goes some way toward conceding epistemic equivalence and arbitrariness as an implication of observational equivalence: The choice
among alternative hypotheses equally consistent with the available evidence must to some extent be arbitrary (10). Immediately after this
F53 invokes extra-empirical virtues as criteria of making non-arbitrary (or
at least less arbitrary) choices between observationally equivalent theories:
there is general agreement that relevant considerations are suggested
by the criteria simplicity and fruitfulness, themselves notions that defy
7
The two classical statements of the underdetermination issue were published at about the
same time as F53. Quines Two dogmas of empiricism had appeared in 1951, while the
English translation of Duhems book was to be published in 1954 (the French original
appeared in 1904).
109
completely objective specification Logical completeness and consistency are relevant but play a subsidiary role (10) Note that the list of
favorite theoretical virtues includes, foremost, simplicity and fruitfulness
(F53s name for unifying power). The Walrasian virtues of logical
completeness and consistency are ranked lower, while mathematical elegance, another popular theoretical virtue, is not mentioned at all.
F53 thus conforms to the popular solution to the underdetermination
conundrum: it cites theoretical virtues as the remedy. Because empirical
evidence is insufficient for deciding which theories to accept and reject, we
need to supplement empirical virtues with theoretical virtues. When choosing between empirically equivalent theories, we choose the one that is theoretically more virtuous. Arbitrariness becomes eliminated or at least reduced.
This invites two observations. First, the argument from theoretical
virtues is a standard move amongst realists against those anti-realists
who infer from observational equivalence to epistemic equivalence and
arbitrariness. F53 would hence seem to stand in the realist camp on this
issue, the point being that of all the observationally equivalent theories, we
are to believe the one that is simplest and has maximum unifying power.
On the other hand, an instrumentalist may argue that such theoretical
virtues are epistemically irrelevant and thus fail to be of help in guiding the
allocation of belief, in driving theory choice towards the most truthlike
theory. It might therefore seem that in matters of underdetermination,
F53 holds an indeterminate position between realism and instrumentalism. I can think of one insight in F53 that might be used for making the
position more determinate: we have seen that F53 contains the realist idea
of ontological unification (in contrast to the instrumentalist idea of mere
derivational unification). Since F53 cites unification (or fruitfulness) as a
major theoretical virtue, we may conclude that it might be, after all,
inclined toward a realist solution to the underdetermination challenge.
The second observation about the position taken by F53 is this: even
though the conundrum of underdetermination can be alleviated by means
of an appeal to theoretical virtues, theory appraisal will not thereby turn
into a fully objective rule-governed affair. As we saw, simplicity and
fruitfulness are notions that defy completely objective specification
and cannot therefore completely eliminate all arbitrariness involved in
theory choice. As we will see next, this is a prominent theme in F53 even
though it has been overlooked by most readers.
13
Subjective and social dimensions of theory appraisal
Those who have read F53 as a positivist or falsificationist statement of

economics as a strictly rule-governed and evidence-constrained endeavor
110
Uskali Mki
will be disappointed upon being pointed out a few passages, completely

ignored by most commentators. If one takes the methodology of positive
economics to amount to a set of fixed and explicit rules for reasoning from
given evidence (or evidence fixed by convention or agreement), one
should be frustrated by the statements in F53 that emphasize the role of
subjective judgment, of the background of economists, of tradition and
disciplinary institutions, and of consensus amongst them. These statements in F53 further fortify the concession that strict predictivist tests
are unavailable in economics. These statements do not appeal just to
theoretical virtues such as simplicity and unifying power but rather to
the social nature of scientific work, its embeddedness in academic and
other institutions.
Here is a representative passage, making reference to something like the
methodological culture of economics in comparison to that of sociology:
Of course, neither the evidence of the economist nor that of the sociologist is
conclusive. The decisive test is whether the hypothesis works for the phenomena it
purports to explain. But a judgment may be required before any satisfactory test of
this kind has been made, and, perhaps, when it cannot be made in the near future,
in which case, the judgment will have to be based on the inadequate evidence
available. In addition, even when a test can be made, the background of the scientists
is not irrelevant to the judgments they reach. There is never certainty in science,
and the weight of evidence for or against a hypothesis can never be assessed completely
objectively. The economist will be more tolerant than the sociologist in judging
conformity of the implications of the hypothesis with experience, and he will be
persuaded to accept the hypothesis tentatively by fewer instances of conformity.
(30; emphases added)
Complete objectivity in testing a theory is thus unattainable: judgment

and persuasion are involved, and they are directed by the background of
the scientist and the degree of tolerance characteristic of the disciplinary
culture. In passing, F53 uses falsificationist jargon by referring to the
repeated failure to be contradicted as evidence for a theory, but it simultaneously acknowledges the role of social factors in shaping the fate of a
hypothesis. It recognizes the tenacity with which hypotheses are held against
negative evidence and the powerful role of tradition and continued use in
creating the image of an acceptable hypothesis: the discipline has its own
epistemically forceful folklore that supports a theory but is irreducible to an
explicit list of successful empirical tests. Here goes another key passage:
[The evidence for the maximization-of-returns hypothesis] is extremely hard to
document: it is scattered in numerous memorandums, articles, and monographs
concerned primarily with specific concrete problems rather than with submitting the
hypothesis to test. Yet the continued use and acceptance of the hypothesis over a long
period, and the failure of any coherent, self-consistent alternative to be developed
111
and be widely accepted, is strong indirect testimony to its worth. The evidence for a
hypothesis always consists of its repeated failure to be contradicted, continues to
accumulate so long as the hypothesis is used, and by its very nature is difficult to
document at all comprehensively. It tends to become part of the tradition and
folklore of a science revealed in the tenacity with which hypotheses are held rather
than in any textbook list of instances in which the hypothesis has failed to be
contradicted. (223; emphases added)
What F53 suggests here is that the social performance (embeddedness in

disciplinary institutions and culture, continued use, collective acceptance, tenacious grip) of a theory provides indirect evidence in its support.
The core idea is that acceptances and rejections of theories are not strictly
rule-governed responses to direct empirical evidence, they rather depend
on the subjective judgments of economists whose behavior is shaped by
their background and social context. Part of the reason for this can be
traced to the character of the rules of application that are described as
concrete and incomplete: The rules of using the model cannot possibly be abstract and complete. They must be concrete and in consequence
incomplete (25). It is this incompleteness that creates room for judgment and social shaping: no matter how explicitly we succeed in formulating the rules there inevitably will remain room for judgment in
applying the rules It is something that cannot be taught; it can be
learned but only by experience and exposure in the right scientific
atmosphere, not by rote (25). The set of rules is embedded in concrete
examples and it escapes explicit formulation: it is mostly implicit and
suggested by example (356).8
It is then only natural to wonder whether the acknowledgment of the
social shaping of theory appraisal is in tension with the optimism about
the prospects of positive economics expressed in the opening pages of
F53. The suspicion is that once you let the social influences in, the values
characteristic of normative economics are included in the package and
will play a role in undermining hopes of empirically grounded consensus.
Whatever consensus there is among economists may emerge as a mixture
of empirical, institutional, and ideological influences. The empirical
8
These are ideas that would be made more widely known by Thomas Kuhn in his 1962
book, The Structure of Scientific Revolutions. On rules of application, F53 anticipates Kuhns
idea: the rules of application are socially shaped, linguistically inexplicable, and embedded
in previous archetypical concrete examples of scientific practice. The orthodox positivist
idea of correspondence rules as explicit components of a theory governing its use is
replaced by Kuhn by the notion of a concrete exemplar whose implicit guidance of theory
application is brought into scientists awareness only through the socialization process of
scientific education. I first pointed out the Kuhnian features of the account given by F53 in
Mki 1986 (of which an earlier version was published as a discussion paper in 1981).
112
Uskali Mki
control of belief turns out to be regrettably weak, because of which

theories exhibit remarkable capacities of survival.
The denial to economics of the dramatic and direct evidence of the crucial
experiment does hinder the adequate testing of hypotheses; but this is much less
significant than the difficulty it places in the way of achieving a reasonably prompt
and wide consensus on the conclusions justified by the available evidence. It
renders the weeding-out of unsuccessful hypotheses slow and difficult. They are
seldom downed for good and are always cropping up again. (11)
Whatever label one may wish to use for these ideas pragmatism, collective conventionalism, or social constructivism it is obvious that they
reflect, and resonate with, the practitioners actual experience that cannot
be easily corrupted by positivist or falsificationist textbook teachings about
the scientific method. These statements convey highly agreeable insights
into the imperfect yet perfectly human reality of scientific work. But
this has a price: tensions emerge with some other ingredients of the image
of economics argued by F53. In my rereading, the above valuable observations about the social and institutional nature of economic inquiry will
be given precedence.
14
How to rewrite F53
I keep admiring F53 for its insightfulness and its capacity to give rise to
intriguing questions and to suggest challenging answers. F53 is a rich
source of issues and observations that may help us understand important
aspects of the intellectual endeavor of economics in its institutional setting. As I have indicated in the footnotes, F53 was ahead of its time in
regard to many issues: falsification, comparative testing, unification,
underdetermination, pragmatics of theorizing, social shaping of scientific
reasoning.9 At the same time, it appears to share the old methodological
wisdom in economics: theories seek to isolate what are believed to be
causally significant aspects of social reality.
If we take the F-mix to be based on ambiguities, inconsistencies and
confusions in F53, and if we have a taste for a coherent and defensible
account of economic methodology, then the task will be to remove such
flaws in F53 and to develop its more promising ingredients into a coherent
methodology of economics. This is a project of elimination, reinterpretation,
9
Based on this observation, I would take partial departure from the conventional claim that
F53 is an example of amateur philosophy of science. My qualified judgment is that F53
contains a set of amazingly sophisticated philosophical insights about scientific theorizing,
while at the same time failing to put them together in the form of a coherent system.
Rewriting is required precisely in order to remedy this failure.
113
and amendment: one of rereading and rewriting F53. I believe a solid and
sensible methodology will emerge as a result of such an exercise. A rewritten
F53 will also create new common ground for many participants in the
controversies over F53 itself and over economics and its methodology
more generally. Within such a shared framework, the remaining disagreements can then be debated in a more focused manner.
I have reread F53 by focusing on a selected set of ambiguities that
open up opportunities for reinterpretation. I have exploited these opportunities by highlighting the partly hidden realism in F53s conception of
economic science. On this basis, F53 could be rewritten as an unambiguous
and consistent realist manifesto. It conveys a methodology of economics
that conforms to the tradition of viewing theories or models as partial
but potentially true descriptions of causally significant mechanisms. Their
primary service is to convey explanatory understanding (answers to whyand how-questions) and only secondarily to yield predictions (answers to
what-, when-, and where-questions). The tradition runs from Senior, Mill,
and Cairnes to Menger, Marshall, and Robbins up to present-day economics. In regard to this matter, the often-heard claim about a radical rupture
from the Robbinsian to the Friedmanian regime in mainline economic
methodology seems exaggerated. What does change is the sophistication
with which it is now possible to put some of the key ideas such as the
following.
For the purposes of pursuing significant truths by theory, many of its
assumptions are allowed to be false. More strongly, false assumptions are
required for true theories. The best theories are simple thus violate
comprehensive truth and unify large classes of apparently diverse phenomena. Such phenomena are ontologically unified by explaining them as
manifestations of the same simple causal structures. So much for the
ontological underpinnings and veristic ambitions of economic theorizing.
Another set of sophistications pertains to epistemological concerns, the
issues related to assessing theory candidates. One is supposed to test
theory candidates by their comparative predictive and explanatory performance, but this will never yield conclusive results. Theory choices are
underdetermined by evidence, and they are shaped but not fully determined by theoretical virtues, thus room is created for various subjective
judgments as well as social constraints that reflect the institutional structure of economic inquiry. The recognition of this requires a social epistemology involving a strong fallibilism. Of these, fallibilism recommends
modesty and awareness of the possibility of being mistaken, while social
epistemology recommends critically assessing and redesigning the institutions of economic inquiry so as to improve their capacity in facilitating
the acquisition of significant truths about economic reality.
114
Uskali Mki
The outcome of this exercise in brief is this: the rewritten F53 = realism +
fallibilism + social epistemology. According to this methodology, excellent economists are epistemically ambitious in seeking deep truths about
underlying causal structures, and at the same time epistemically modest in
acknowledging the possibility of error and the shaping of inquiry by
academic and other institutions and again epistemically ambitious in
demanding the design and implementation of institutions that are
adequate for enhancing the acquisition of significant truths. This combination is not easy to keep in balance. For perfectly sound reasons,
therefore, F53 claims: More than other scientists, social scientists need
to be self-conscious about their methodology (40).
15
From reception to the authors intentions?
Even though the foregoing has been an exercise in reception methodology, a final speculation may be permitted about how all this might relate
to the intentions and beliefs of the author of F53. I have argued that F53
fails to unambiguously express any coherent set of beliefs and intentions
about economic methodology. The unavailability of a single coherent
methodology in F53 is consistent with two possibilities concerning
Milton Friedmans beliefs and intentions. The first possibility is that
Friedman did not have a coherent methodology in mind and that F53
reflects this fact accurately (for a rhetorical account of this option, see
Mki 1986). The second possibility is that Friedman had a coherent
methodology in mind but just failed to convey it in F53 due to flaws in
textual production.
Whatever the case, the ambiguities and inconsistencies in F53 have
enabled multiple receptions and mutually incompatible interpretations in
the long series of commentaries. In a special sense, we might say that any
such interpretation is underdetermined by textual evidence: the textual
evidence does not fix any of the interpretations as the correct one.
However, this is not a situation of underdetermination in precisely the
same sense that was recognized by F53 itself as discussed above: namely
that an infinite number of theories are consistent with the same set of data.
Here, various interpretations of F53 rely on reading selected passages
rather than the textual data in its entirety: the selected sets of evidence
overlap but are not identical (for example, most of the instrumentalist
interpretations of F53 have neglected those passages that are crucial for
my realist and social constructivist readings). This is bound to be so as no
interpretation is possibly consistent with the textual data as a whole. What
I am envisaging here is not exactly the same idea as the standard argument
115
from underdetermination but it may have interesting affinities with

actual situations in economics and elsewhere in science!
In a sense, then, the economic methodology possibly intended by the
author of F53 is underdetermined by F53 itself and is therefore open for
multiple interpretations. There is just one way I can think of that might
help reduce this multiplicity: analyze Friedmans other statements on
methodology, and analyze Friedmans work in economics for its implicit
methodology. This is a task for historians of economic methodology
interested in one Nobel laureates views not a task for reception methodology in the sense pursued here. Yet, were these two perspectives combined, I would not be surprised if it turned out to be the case that F53 as
rewritten along the lines suggested above is the methodological statement
that the author of F53 intended to write.
References
Boland, Lawrence A. (1979). A critique of Friedmans critics. Journal of Economic
Caldwell, Bruce J. (1992). Friedmans methodological instrumentalism: a
modification. Research in the History of Economic Thought and Methodology
10, 11928
Duhem, Pierre (1954). The Structure of Physical Theory. Princeton: Princeton
University Press
Friedman, Michael (1974). Explanation and scientific understanding. Journal of
Philosophy, 71, 519
Positive Economics. Chicago: Chicago University Press, 343
Hands, D. Wade (2003). Did Milton Friedmans methodology license the formalist revolution? Journal of Economic Methodology, 10, 50720
Hindriks, Frank (2005). Unobservability, tractability, and the battle of assumptions. Journal of Economic Methodology, 12, 383406
Hirsch, Abraham. and Neil De Marchi (1990). Milton Friedman: Economics in
Theory and Practice. Hertfordshire: Harvester Wheatsheaf
Kitcher, Philip (1976). Explanation, conjunction, and unification. Journal of
Philosophy 73, 20712
(1981). Explanatory unification. Philosophy of Science, 48, 50731
Kuhn, Thomas S. (1970/1962). The Structure of Scientific Revolutions. Chicago:
University of Chicago Press
Lakatos, Imre (1970). Falsification and the methodology of scientific research
programmes. In I. Lakatos and A. Musgrave (eds.), Criticism and the Growth of
Knowledge. Cambridge: Cambridge University Press, 91196
Milton Friedmans methodology, Research in the History of Economic Thought
and Methodology, 4, 12743
116
Uskali Mki
Methodology 10, 17195
(1998).As if. In J. Davis, W. Hands and U. Mki, (eds.) The Handbook of
Economic Methodology. Edward Elgar 1998, 257
(2000). Kinds of assumptions and their truth: shaking an untwisted F-twist.
Kyklos, 53, 30322
(2001). Explanatory unification: double and doubtful. Philosophy of the Social
Sciences. 31, 488506
(2003). The methodology of positive economics (1953) does not give us the
methodology of positive economics. Journal of Economic Methodology, 10,
495505
(2004). Realism and the nature of theory: a lesson from J. H. von Thnen for
economists and geographers. Environment and Planning A, 36, 171936
Mayer, Thomas (1993). Friedmans methodology of positive economics: a soft
reading. Economic Inquiry 31, 21323
Musgrave, Alan (1981). Unreal assumptions in economic theory: the F-twist
untwisted. Kyklos 34, 37787
Popper, Karl (1959). The Logic of Scientific Discovery. London: Unwin Hyman
Rotwein, Eugene (1959). On the methodology of positive economics. Quarterly
Journal of Economics, 73, 55475
Samuelson, Paul (1963). Problems of methodology discussion. American
Wong, Stanley (1973). The F-twist and the methodology of Paul Samuelson.
Part 3
Models, assumptions, predictions, evidence
The influence of Friedmans

methodological essay
Thomas Mayer
Many leading methodologists have described the central role that Milton
Friedmans 1953 essay (henceforth referred to as F53) has played in
methodological discussions. (See, for instance, Hammond 1998;
Hoover 2001; Backhouse 2002.) Yet Friedman himself did not intend
his essay to be a contribution to methodology about which he did not
claim any expertise but only a description of the approach he found
useful in his practice. However, it does not necessarily follow from
Friedmans intention that his essay has had a great influence on the
practice of economics. Practicing economists pay little attention to freestanding discussions of methodology. At best they learn their methodology by seeing it put to work on substantive problems.1 Arguably,
Friedman and Schwartzs (1963) A Monetary History of the United States
has had more influence on the methodology of practicing economists than
did F53. The most pervasive methodological influence in macroeconomics in the last thirty years has been the insistence of new classical economists on reducing macroeconomics to microeconomics, and in this they
paid no attention to the debate about reductionism among philosophers of
science. All the same, I will go along with Friedmans intention, and deal
only with F53s effect on its intended audience, practicing economists.
And here I will look almost only at mainstream economists. Heterodox
economists have been highly critical of F53, so that it is unlikely to have
had much influence on them. I also omit its influence on econometrics
(for that see Hoover, 2004) and on government and business economists,
since their work is almost invisible to other economists. Several other
essays in the book containing F53 also discuss methodology. If these
discussions relate to ideas discussed in F53 I treat them as part of it, but
refer only occasionally to methodological claims that Friedman made in
other publications.
1
Partha Dasgupta (2002, 57) remarked that he knows no contemporary practicing economist whose investigations have been aided by the writings of professional methodologists.
119
120
Thomas Mayer
Before seeing how F53 has influenced economics we need to look at its
message (section I), at some difficulties in evaluating its influence (section
II), and at the reviews it initially received (section III). Due to limitations
of space and knowledge I limit the discussion to a survey of (a) three
general trends in economics: formal modeling, game theory, and econometric testing (section IV); (b) Nobel lectures (section V); and (c) certain
topics in macroeconomics (section VI). This emphasis on macroeconomics is not based on any deep principle, but only on my greater familiarity
with it. I end the discussion of macroeconomics with the new classical
counter-revolution, because the further one goes beyond 1953, the
greater is the probability that a change advocated in F53 is due to some
other influence. But I discuss Nobel prizes up to 2002 because of the lag
until work is recognized by the Nobel committee.2
1
Friedmans themes
F53 has one general and six specific themes. The former is the desirability
of combining theoretical and empirical work, and thus to heal the split
between theorists and institutionalists, to whom Friedman is closer than is
often appreciated (see Friedman 1949, 1950). He was, of course, not the
only one who attempted this. It was a general trend that would eventually
have prevailed even in his absence. In any case, what has dominated the
discussion is F53s specific theme that hypotheses should be not be tested
by the realisticness of their assumptions. A second theme of F53 is the
superiority of Marshallian over what Friedman calls Walrasian economics. This creates a terminological problem. As Hutchison (1954)
pointed out, the latter does not represent Walrass own views on methodology. Nor does it necessarily represent the methodology of those using a
general equilibrium approach; we now have computable general equilibrium models. But Friedmans terminology has caught on, so rather than
2
I do not consider the number of citations to F53 in the Social Science Citation Index (SSCI).
On the one hand, a citation count may understate a papers influence, since its influence
may be indirect via another paper that is cited instead of the original one. Indeed, once an
idea becomes well known, it no longer requires citation. Moreover, a paper may exert a
strong negating influence that does not generate a single citation: no citation appears if F53
prevented an economist from writing a paper criticizing a theory for its unrealistic assumptions. It is even possible for a papers influence to be inversely related to the frequency of its
citations. If it criticizes a research program convincingly, it may terminate work on that
topic and therefore not be cited. Another problem is that the SSCI ignores citations in
books. On the other hand, a citation count may overstate a papers influence. Many
citations are just hat-tipping references, made to show that the author is familiar with the
paper, or they may be from a paper that criticizes the cited paper. Moreover, an economist
may cite F53 as a justification for an unrealistic assumption he or she would have made in
any case.
121
use the more appropriate term formalistic I will use Friedmans term,
but with a lower-case W to mark the misuse of Walrass name. The third
theme is the rejection of introspection and other forms of casual empiricism, and the fourth is the distinction between normative and positive
economics. A fifth theme, relating to the context of discovery, is the
benefit of a continual interplay of attention to data and to theory construction. A final theme is the need for modesty about what economic
theory can accomplish. This theme (which deserves a paper of its own,
since it underlies most of Friedmans other themes and some of his
substantive work) is not well articulated and usually ignored, so, I, too,
will ignore it.3
1.1
Realisticness of assumptions
Friedmans claim that hypotheses should not be judged by the realisticness of their assumptions but by the predictive success of their implications does not seem so startling once one recognizes that F53 still allowed
testing by the realisticness of critical assumptions, since these can be
rephrased as implications (see Baumol 1954; Mayer 1995, ch. 7).4 If
Friedmans objection to testing hypotheses by the realisticness of their
assumptions is so much less startling than it appears at first glance why did
it almost monopolize the discussion? Likely reasons include the fact that
he stated it with much elan and at much greater length than his other
themes, its wide scope, and its seeming conflict with intuition. Moreover,
it was seen as a massive salvo in the ongoing battle about neoclassical
theory. And, while practicing economists might applaud Friedmans
deep insight into how economists work, for methodologists Friedmans
3
Hirsch and De Marchi (1990, 1549) in discussing both Friedmans methodological

writings and his practices categorize his themes somewhat differently as: (1) adopt an
outside view, that is, reject arguments from introspection; (2) begin with observations;
(3) continually test the implications of hypotheses, though not in a falsificationist mode;
(4) base empirical tests on the framework of the most reliable available knowledge; and
(5) focus on concrete problems. Perhaps because I deal with F53 and not with Friedmans
own methodological practice, I am using a different framework.
Not all the stated assumptions of a theory are necessary assumptions, some are expository
assumptions, or domain-defining assumptions, etc. (see Musgrave 1981). Moreover, the
necessary assumptions may be stated more broadly than required. A way to determine
whether an assumption is a necessary one and is stated no more broadly than required, is to
see if it can also be stated as an implication. But what about a theorys power to explain?
Friedman surely considered explanation in the sense of subsuming an observation (or a
narrow theory) into a (broader) theory to be important his reference to generality, and his
substantive work both before and after 1953 bear witness to that as does the fact that he
taught price theory. To Friedman, generality, and thus explanation, is important, but the
acid test is prediction.
122
Thomas Mayer
philosophically amateurish exposition, combined with his high status in

the profession, made F53 a tempting target, and the assumptions issue
seemed a weak spot.
1.2
Marshallian intuition vs. walrasian rigor
Many economists have (wrongly) read Friedmans objection to testing by

the realisticness of assumption, not as an argument for testing by predictive success, but as enhancing the importance of three other criteria for
theory choice, elegance, generality, and rigor, and hence as supporting
their insistence on formal derivation of hypotheses from explicitly stated
assumptions. So, assumptions again play a key role. But what matters now
is not their realisticness, but their congruity with the canonical assumption
of rational profit maximization, as well as their parsimony. But that is not
Friedmans criterion for theory choice.
The only relevant test of the validity of a hypothesis is comparison of its predictions
with experience Logical completeness and consistency are relevant, but play a
subsidiary role; their function is to ensure that the hypothesis says what it is
intended to say One of the effects of the difficulty of testing substantive
economic hypotheses has been to foster a retreat into purely formal or tautological
analysis [T]autologies have an extremely important role in economic analysis
But economic theory must be more than a structure of tautologies if it is to be
something different from disguised mathematics. (Friedman 1953, 812; italics in
original)
Friedman paid much attention to the shortcomings of such disguised

mathematics masquerading as economics. In F53 he referred readers to
another chapter of Essays on Positive Economics, a review of Oscar Langes
Price Flexibility and Employment, where he wrote:
Here is a brilliant display of formal logic, abstract thinking, complicated chains of
deductions, yet the analysis seems more nearly a rationalization of policy conclusions previously reached than a basis for them. What is there about the type of
theorizing employed that makes it sterile ? The basic sources of the defects
are the emphasis on formal structure, the attempt to generalize without first
specifying in detail the facts to be generalized, and failure to recognize that the
ultimate test of the validity of a theory is not conformity to the canons of formal
logic but the ability to deduce facts that have not yet been observed, that are
capable of being contradicted by observation, and that subsequent observations
do not contradict. (Friedman [1946] 1953, 277, 300)
Elsewhere in Essays Friedman (1953, 11213) wrote that: formal analysis

can seldom if ever give easy answers to hard problems. Its role is quite
different: to suggest the considerations relevant to an answer and to
123
provide a useful means of organizing the analysis. And in his discussion

of the Marshallian demand curve Friedman (1953, 91) told us that:
The test of the theory is its value in explaining facts, in predicting the consequences of changes in the economic environment. Abstractness, generality, mathematical elegance these are all secondary, Doubtless, most modern economic
theorists would accept these general statements But our work belies our professions. Abstractness, generality and mathematic elegance have in some measure
become ends in themselves, criteria by which to judge economic theory.5
Hammond (1990) considers Friedmans rejection of formalism a more

important component of F53 than rejection of testing by the realisticness
of assumptions, and calls Marshallianism the hallmark of Friedmans
methodology (Hammond 1996, 43).
1.3
Casual empiricism and introspection
Friedmans belief that appearances may be misleading, and his resulting

criticism of casual empiricism and introspection, is set out in more detail
in his review of Lange than in F53.6 There he showed that Langes analysis
is a deductive system based on certain assumptions that are just unverified
conjectures. To Friedman such casual empiricism is insufficient. And
though he did not mention this, this theme supports his insistence on
testing implications rather than assumptions since many assumptions can
only be tested casually.7 Along with rejecting casual empiricism he also
rejected reliance on the responses of firms to questions about what they do
and why.
1.4
Normative and positive economics
F53 claimed that the normative/positive distinction has received inadequate attention, and conjectured that disagreements on economic policy
among disinterested people in the West are primarily disagreements about
positive issues. In stressing the positivenormative distinction Friedman
5
Friedman, like many economists of his generation, became an economist because of the
suffering he saw during the Great Depression (see Snowdon and Vane 1999, 1245). That
may explain his emphasis on practicality.
Hirsch and De Marchi (1990, 19) argue that it is Friedmans rejection of introspection that
distinguishes his methodology from the traditional one. They also call confidence in
introspection undoubtedly a major reason that methodologists argue on the basis of
realistic assumptions (1990, 56).
Some assumptions, such as rational behavior, are hard to formulate as testable hypotheses,
and many others, such as a closed economy, are obviously not intended to be realistic, and
hence not something anyone would want to test.
124
Thomas Mayer
furthered the unfortunate tendency to ignore a third category of economics,

J. N. Keyness (1891) art of economics, that is the knowledge required to
apply positive and normative economics in a way that results in successful
policies. (See Colander 1991.) This is surprising since in his chapter
reviewing Abba Lerners Economics of Control Friedman castigated Lerner
for ignoring problems pertaining to the art of economics. Also, Friedman
lacked the postmodernists nose for any whiff of normative elements.
1.5
Interplay of theory and observation
Friedmans fifth theme concerns the process of discovery. He advised

against developing theory from a merely casual acquaintance with the
data. He also advised against just concentrating on the data, and either
being satisfied with mere description, or else expecting theory to spring
from the data like Athena from the head of Zeus. This interplay theme
characterizes Friedmans own research. In writing A Theory of the
Consumption Function (1957) he started with a thorough knowledge of
the data. In his monetary theory he carefully combined theory with his
empirical evidence. The primary reason A Monetary History of the United
States (1963) has been so influential is the masterly way it combines
detailed knowledge of historical facts with an overarching theory.
2
Some difficulties in estimating the influence of F53
Several major difficulties arise in gauging the influence of F53. First,

suppose that in its absence some other methodology, such as operationalism, had taken over. If so, then F53 had a great, but hidden, influence.
Second, what economists write might not accurately reflect how they act
or what motivates them.8 Thus, F53 may prevent an economist from
writing a paper criticizing a model for its assumptions, and yet he or she
may still not accept it because of these assumptions. Third, in the absence
of F53 someone else might have written a similar and just as influential
paper. There is nothing that can be said about this, except that a similar
thing is true of Darwin. So, I will ignore it.
Fourth, the influence of a methodological paper is often subtle and
more difficult to detect than the influence of other papers. Suppose an
economist who in 1935 analyzed inflation in terms of the money supply
and its velocity, analyzed it in 1937 in terms of the marginal efficiency of
8
An example of this is the treatment of Gurley and Shaws Money in a Theory of Finance
(1960). It was savaged when it appeared, but over the years some of its ideas became
influential without acknowledgments.
125
Table 4.1 Effect of changes in economics on the acceptance of F53

Computers,
Mathemat- econometrics, Increased Greater
Prestige of Quality of
ization
data
resources policy role Chicago economists
Assumptions vs.
predictions
Marshallian vs.
walrasian
Casual
empiricism,
introspection
Theory/data
interaction
+a
Note: + denotes influence in the direction of F53, denotes influence in the opposite
direction, ? denotes direction unknown, and a superscripted ? some doubt about the
direction.
a
Likely but not certain direction.
capital. The influence of Keynes is obvious. And when economists use

unit root tests Grangers influence is clear. But the influence of F53 may
be less obvious; Marshallian or walrasian outlooks are a matter of emphasis, and since F53 allows testing by assumptions that are also implications,
not every test by the realisticness of assumption shows a lack of influence
of F53. Nor is every instance of unconcern about the realisticness of
assumptions due to F53.
Another difficulty is inferring causation from correlation. Economics
may have evolved in ways advocated in F53, but primarily for other
reasons. These include (a) economists increased mathematical skills;
(b) the revolution in data analysis due to computers, new econometric
techniques and greater availability of data; (c) an increase in the funding
and volume of academic research; and (d) the increased role of economists in policy-making; (e) the greater prestige of Chicago economics
(which is, in part, endogenous to F53), and (f) an apparent improvement
in the intellectual quality of economists, probably due largely to improved
academic salaries. Table 4.1 summarizes the likely effects of these factors
for all of Friedmans themes, except the confounding of positive and
normative economics, a point at which economics has not changed much.
The increased use of mathematics (apart from its effect through econometrics) has had conflicting effects for F53. It increased formal modeling
which often requires tolerance for many unrealistic assumptions. But it
also encouraged a shift away from the empiricism and from the emphasis
126
Thomas Mayer
on predictive success that underlies Marshallian economics. This

decreased emphasis on genuine empiricism may perhaps also have
resulted in an increase in casual empiricism, and by emphasizing theory
might have reduced the interplay of theory and data.
The revolution in data analysis obviously stimulated Marshallian
empirical work. It also made it much easier to evaluate predictive success,
but had less effect on the ability to test the realisticness of assumptions, so
that it shifted emphasis toward the former.9 It also reduced the reliance on
casual empiricism and introspection. Its effect on the interaction between
theorizing and data analysis is ambiguous. There is an obvious positive
effect, but the ease with which one can now test a hypothesis, and, if the
data reject it, quickly formulate and test another version, has reduced the
pressure to rethink ones theories when the data do not oblige, and this
reduces the effectiveness of theory and data interaction.
The resources devoted to economic research have increased dramatically. By 1953 in the USA the shift in the self-image of faculty members
from that of learned gentlemen enjoying considerable leisure to that of
harried researchers had not yet run its full course, resources, such as
research assistants, were scare and universities were fewer and smaller.
With increased resources and numbers came an increased professionalization, and thus a greater belief that economics is an autonomous subject
and not just one of the social sciences. Hence, economists no longer had
to be so concerned if their assumptions seemed bizarre to other social
scientists and to the public, or even to themselves there is safety in
numbers. While Irving Fisher may have worried about how the assumptions he made in his books would strike the general reader, this is of no
concern to someone publishing in Econometrica. Increased professionalization also favored walrasian economics as economists focused more on
acclaim from fellow economists rather than from the general public.
Walrasian economics with its claims of rigor and elegance has relatively
more appeal to a self-absorbed profession that claims to be scientific,
while Marshallian economics has more appeal to those who address the
concerns of a broader public. On the other hand, it is quite possible that
with the expansion of the profession the opportunity for any particular
economists to make an even arguably worthwhile contribution to abstract
theory has declined, and hence Marshallian empirical work has filled the
gap. A rise in the ratio of research assistants to faculty researchers could
also shift research in an empirical direction, but with falling teaching loads
9
It did, however, also facilitate the use of questionnaires, but probably not by as much as
time-series data, particularly since it did not reduce the cost of interviewing, or mailing,
and editing surveys.
127
and increased pressure to publish, it is not clear that this ratio has risen.
The net effect on Marshallian versus walrasian economics is therefore
unclear.
Increased resources have reduced the need for casual empiricism and
introspection. But their effect on the interplay of theorizing and data
analysis is unclear. On the one hand, they have facilitated the use of
statistical data, but on the other, increased professionalization has
reduced the use of observations in the form of the reality checks provided
by personal every-day experience, which could now be dismissed as mere
anecdotes.
The greater role that economists now play in policy-making may perhaps have led them to stress predictive success more than realisticness of
assumptions, and it surely led them to prefer Marshallian to walrasian
theory. But, given the lack of information required for many policy issues,
and the pressure to come up with answers (and sometimes with answers
pleasing to policy-makers), it may have increased reliance on casual
empiricism. Also, it has probably increased the interplay of data and
theory since that is helpful in dealing with applied problems. In the period
following Friedmans paper the prestige of the Chicago economics department rose as its members gathered a disproportionate share of Clark
medals and Nobel prizes (see Reder 1982), and as more of their substantive work appeared. That economists tended to view markets more favorably also helped. This probably enhanced the standing of all of Friedmans
work, including F53. This leaves the higher intellectual caliber of entrants
into the profession. Supporters of Friedmans themes surely believe that
this facilitated their acceptance, while their opponents believe the
opposite.
Table 4.1 summarizes these influences. Only on the assumptions/predictions issue (and perhaps on the casual empiricism issue) did all the
changes, with the possible exception of the increased quality of economists, favor F53. On the other issues it is unclear whether the prevailing
winds favored or hindered the acceptance of F53. Hence, it is possible that
its apparent success was due less to its persuasive power than to good luck.
But one also cannot just dismiss the opposite idea that, except on the issue
of realisticness of assumptions, its successes were won in the face of strong
headwinds. This is an unsatisfactory conclusion, but when dealing with
causality unsatisfactory conclusions are hardly rare.
A final issue on causality is overdetermination. Were Friedmans
themes in the air at the time? There is no evidence for this, neither in
the reviews of F53 nor in the initial responses of methodologists. While
Terence Hutchison and Fritz Machlup shared some of Friedmans ideas,
128
Thomas Mayer
that is not the same as these ideas being in the air, particularly since
Hutchison and Machlup lacked Friedmans persuasive power.
Reviews of F53
Essays in Positive Economics was reviewed in The American Economic

Review, The Economic Journal, The Economic Record, Economica, The
Review of Economics and Statistics, The Southern Economic Journal, and
the Weltwirtschaftliches Archiv. Surprisingly, the Journal of Political
Economy did not review it. Nor did The Canadian Journal of Economics
and Politics, The Indian Journal of Economics, The Journal of the American
Statistical Association, Kyklos, Revu Economique or the Zeitschrift fr
Nationalkonomie, even though Friedman was at that time already well
known, having won the Clark medal two years earlier.10
William Vickrey (1954) commented extensively on the books macroeconomic chapters, while not mentioning F53 at all, and dismissed the
chapter criticizing Lange in one sentence. But William Baumol (1954,
465) called the book: certainly one of the most engrossing volumes that
has appeared recently in economic theory, and devoted about one-fifth
of his review to F53. (It accounts for 13 percent of the book.) He called it
highly stimulating (1954, 464), even though it rides hard on a single
point, and fails to present Friedmans viewpoint systematically (1954,
463). In particular, while it seems to suggest that economists should seek
only correlations, even if they are spurious, he believed that Friedman said
something much more acceptable that badly needs saying (1954, 465),
because he permitted testing by assumptions if these assumptions can be
reformulated as implications. While agreeing with the essence of
Friedmans argument he nonetheless objected to accepting a hypothesis
that has ridiculous assumptions, because we should ask whether it
abstracts in a way that omits aspects important for the problem at hand.
Peter Newman (1954) called the book an enjoyable feast, though
perhaps some readers should be warned to take it cum grano salis,
stimulating, provocative, often infuriating, but well worth reading
(1954, 259 and 260; italics in original). Only two and a half sentences of
his one-and-a-quarter page review dealt with F53. He referred to
Friedmans principle of testing hypotheses by predictive success instead
of realisticness of assumptions as salutary and rewarding [if] temperately used (1954, 206), but warned about taking this too far.
10
At the time both the Journal of Political Economy and the American Economic Review
reviewed books, The Journal of Economic Literature not yet having appeared. The Journal
of the American Statistical Association reviewed some economics books, but not this one.
129
Hutchison (1954) described the book as containing so much that is

acute, independent and stimulating, that though there is much to disagree with this collection of essays should be widely and warmly
welcomed (1954, 799). Almost half his review dealt with Friedmans
methodology, both in F53 and in other chapters. He did not mention
Friedmans strictures against testing by the realisticness of assumptions,
though he did discuss his insistence on testing by predictive success.
Unlike most reviewers, he also dealt with other aspects of Friedmans
methodology, such as Friedmans emphasis on the distinction between
normative and positive economics and his criticism of the prevailing
overemphasis on elegance, generality, and taxonomy.
Allen Clark (1954) called it a collection of remarkable essays (1954,
394) that refutes claims that neoclassical theory is of no use to policymakers. And what is profoundly more significant (1954, 394), it shows
that in some areas where we thought we had unequivocal knowledge, we
do not. In discussing F53, Allen summarized Friedmans criticism of
testing by the realisticness of assumptions, but did not go beyond that.
C. S. Soper (1954) devoted about a third of his review to F53, summarizing Friedmans belief that most disputes among economists concern
positive issues, and to his rejection of testing by the realisticness of
assumptions. While he mostly just summarized F53, he also argued that
Friedman, by stressing that a theorys predictions can only be tested for
the phenomena it seeks to explain, fails to tell us how to enlarge its sphere
of validity. In his two-paragraph review Erich Schneider (1955) dispatched F53 in two sentences, saying that Max Weber and Joseph
Schumpeter will already have familiarized German readers with its contents, but even so, that the clarity of Friedmans presentation makes it
enjoyable reading.11
In summary then, four of the seven reviews of F53 were laudatory, one
can at best be described as mixed, another was dismissive, and one made
little attempt at evaluation. None predicted its future importance. Nor did
the decision of many journals not to review Friedmans book suggest a
brilliant future for it. And only one reviewer, Hutchison, was able to see
sufficiently beyond the realisticness-of-assumptions issue to discuss at
some length Friedmans criticisms of empirically inapplicable formalism,
perhaps because that resembled Hutchisons own position. But the
reviews did have a commendable characteristic; while casual observation
11
Such a short review was unusual for this journal, but in the same issue Schneider also
reviewed in just two paragraphs the second edition of Michael Kaleckis Theory of
Economic Dynamics.
130
Thomas Mayer
suggests that many economists treated F53 as an apologetic for neoclassical price theory, none of the reviewers took that line.
4
An overview
Friedmans underlying theme combine theoretical and empirical work

by focusing on predictive success carried the day and the relative
importance of empirical work has increased (see Figlio 1994; Hoover
2004). By no means all of the credit here belongs to Friedman. Thus,
James Duesenberry (1949) had previously published an outstanding
example of such work, which, I was told, had much influence on
Harvard students. From outside economics there was the influence of
Popper and of positivism in general. Other social sciences moved in the
same direction. But surely Friedman deserves some of the credit, if only
because he furnished neoclassical economists with a defense against their
institutionalist critics, thus giving economic theory some breathing room,
while commending the use of data to test theories rather than just to
compile facts.
But the subsequent decline of old-line institutionalism should not be
entirely (or even largely?) attributed to Friedman. The rise of econometrics
also played a part. One could now be an empiricist in a more glamorous
way than by just gathering data if one could manage the required math.
And since many institutionalists could not, they had to relinquish their
claim to be the shining example of the empirical and hence scientific
approach to economics. The increased professionalization of economics
with its tendency to draw sharp disciplinary lines also played a role since it
meant that the preoccupation of some institutionalists with sociological
and historical problems tended to place them outside the limits of
economics. All the same, it seems plausible that Friedmans essay significantly and perhaps even substantially contributed to the decline of institutionalism; perhaps given his tribute to Mitchell (Friedman 1950) by
more than he wished.
Contemporary academic economics consists primarily of modeling
exercises, particularly in game theory and in exploring the implications
of asymmetric information. Some of these models try to earn their keep by
eliminating assumptions made in previous models. But they do this to
generalize them, and not to criticize them for unrealisticness. That all
models are false, but some are useful, has become a familiar clich.
Granted that occasionally economists criticize a paper for its unrealistic
assumptions (see, for instance, Waldman 2003), this is consistent with
F53 if these assumptions are also relevant implications. But Friedmans
victory is incomplete. While most economists no longer explicitly reject
131
some model for its unrealistic assumptions, they may still (perhaps unconsciously) be unwilling to accept it for that reason. And much more
seriously, Friedmans stricture that the implications of models need
tough-minded testing is often ignored. Tests of implications are often
perfunctory, playing tennis with the net down (Blaug 1980, 256; Mayer
2001).
Friedman has been much less persuasive on the superiority of
Marshallian economics. While most empirical work is Marshallian, within
theory the walrasian approach has become more common since 1953. But
only a small proportion of papers are purely theoretical, so that the
vaunted victory of walrasian economics is more a matter of Sunday
dress than workday dress. But it is to the professions Sunday dress that
many (most?) economists point with pride.
Casual empiricism is a pejorative term and, in principle, few economists
would now explicitly defend it, but few would have done so in 1953 either.
Whether the indifference curve between using casual empiricism and
admitting ignorance has changed is hard to say. What has changed massively is the budget constraint due to increased research funding and the
computer revolution. Friedman has, however, been more successful in
combating the Robbinsian emphasis on introspection, though it probably
still plays a significant implicit role by excluding certain models that do
not make intuitive sense. It is hard to say whether the surveys of motives
and beliefs that Friedman criticized are less frequent now than before
1953.12
While postmodernism has had virtually no influence on mainstream
practicing economists, so that Friedmans dictum to separate normative
from positive statements still holds sway, there is little reason to think that
it is obeyed any better or worse now than before 1953. It is, however, likely
that in theory construction there is now more interplay of theorizing and
data analysis than before 1953.
5
Nobel Prize lectures
Two considerations justify using Nobel lectures to characterize the way

economics has changed. First, in economics Nobels are given, not for
establishing indubitable results, but for initiating, or carrying to fruition, a
12
F53s discussion of surveys is perfunctory, but in a paper cited in F53 Friedman (1949)
had already set out his criticisms of surveys that ask agents to reveal their motives or
thinking. In that paper he did not criticize surveys of factual information, indeed he
himself (Friedman and Kuznets 1945) had used such evidence. As Boulier and
Goldfarb (1998) show, survey information now pervades empirical economics, but
most of it comes from objective questions, not questions about opinions and motives.
132
Thomas Mayer
particular sub-literature. They therefore highlight the growth points of

economics. Second, the Nobel committee largely represents the judgment
of the profession. My sample consists of all Nobel lectures up to 2002 with
the following exceptions: when several recipients split the prize I used the
lecture of only one of them, chosen somewhat arbitrarily. I also excluded
Klein, Tobin, and Modigliani because I cover their work in the next
section. Laureates have considerable latitude in their lectures. Most
focus on their own work, or work that derives directly from it, but six
did not, and are therefore excluded from the sample.13 Table 4.2 categorizes the remaining 22 lectures by four characteristics of F53. A plus
denotes consistency with the characteristic, a minus inconsistency, and
a zero the irrelevance of this characteristic.14 Such a classification is
inevitably subjective and sometimes somewhat arbitrary but then tracing
the influence of a paper is more like doing literary criticism than like doing
physics. (Appendix B of Mayer 2004, provides summary information on
the included Nobel lectures.) While the minuses in table 4.2 have a
straightforward interpretation, the pluses do not. They show only that
the lecture conforms to a certain recommendation of F53, and not necessarily that F53 influenced it. For example, someone who tests a hypothesis
by its implications may do so only because in this case it is quicker than
testing it by the realisticness of assumptions.
What is remarkable in table 4.2 is the paucity of minuses. If the Nobel
committees judgment is correct, few of the outstanding breakthroughs
have come from those employing a methodology contrary to F53. On the
issue of assumptions versus predictions there are 11 pluses, and only 4
minuses, while on the Marshallian/walrasian dichotomy there are 16
pluses and 4 minuses. On the rejection of casual empiricism there are 15
pluses and 1 minus. And on the interplay of data and theories 12 pluses
confront 2 minuses.
Such a tabulation is subject to the criticism that I followed the naive
empiricists bad advice of: go measure something, never mind what it is.
In depicting the thrust of mainstream economics, some prizes (e.g.
Samuelsons) may deserve more weight than others. Indeed, it seems
obvious that the frequencies in table 4.2 shortchange the formalist trend
in economics. With only 4 of the 23 prizes going to formalists (Samuelson,
Arrow, Debreu, and Harsanyi) the influence of formalism is surely
understated.
13
14
These six are: Kantorovich, Kuznets, Meade, Miller, Stigler, and Stone.
I exclude the normative/positive distinction because economics has changed little in that
respect.
133
Table 4.2 The Nobel Prize lectures and Friedmans criteriaa

Author and date
Assumptions vs.
predictions
Marshallian/
walrasian
Casual
empiricism
Interplay of theory
and data
Tinbergen, 1969
Samuelson, 1970
Arrow, 1972
Leontief, 1973
Hayek, 1974
Simon, 1978
Shultz, 1979
Debreu, 1983
Buchanan, 1986
Solow, 1987
Allais, 1988
Haavelmo, 1989
Coase, 1991
Becker, 1992
North, 1993
Harsanyi, 1994
Mirrlees, 1996
Scholes, 1997
Sen, 1998
McFadden, 2000
Akerlof, 2001
Smith, 2002
+
b
c
+
?f
+g
0
0
+
+
+
j
+
+k
0
0
+
0
+
?
+m
+
0
+d
+
+
+
+
+
+
+
+
+
+
+
0
+
+
+
+
0
0
+
e
+
+
0
+h
+
+
+
+
+
+
0
0l
+
0
+
+
+
+
0
0
+
+
+
?i
+
0
0
+
+
+
0
0
+
0
+
+
+
Note: + denotes consistent with F53, denotes inconsistent with F53, 0 unrelated to F53
with respect to that criterion, and ? hard to say.
a
This table does not cover all Nobel lectures; for the reasons some were excluded see the text.
b
Samuelson does not discuss this issue in his lecture, but has elsewhere strongly rejected
Friedmans stricture against testing by the realism of assumption.
c
Arrows advocacy of minimizing assumptions is in principle consistent with F53, but points
in a different direction.
d
Leontief uses a general equilibrium model, but is Marshallian in paying attention to the
details of specific industries.
e
This is not explicit in Hayeks lecture, but is inherent in his general position.
f
In his lecture Simon strongly criticizes Friedmans essay, but the substance of his lecture is
consistent with it, at least if it is given a soft reading.
g
Based on the general tenor of Schultzs lecture rather than on an explicit statement.
h
There is little empiricism in Buchanans lecture, but it plays a large role in the work of his
school.
i
No data are discussed in Buchanans lecture, but in the work of his school there may well be
an interplay of theory and data.
j
Coase does not criticize Friedman on this score on this lecture, but does so elsewhere.
k
Applicability of assumptions determines domain of theory.
l
There is casual empiricism, but at this stage of the development of the model one cannot
expect anything more.
m
By implication.
134
Thomas Mayer
Macroeconomics
This sections deals with the three major macro research programs,
Keynesian, monetarist, and new classical. It omits the Austrian and postKeynesian programs because they are heterodox, and new Keynesian
theory, and endogenous growth theory, because they came too late.
6.1
The Keynesian program
The major components of the traditional Keynesian program are the

consumption function, the investment function, a general equilibrium
version of monetary theory, econometric models, and growth theory. All
are positive theories that do not introduce normative elements to a greater
extent than is usual in economics.
The consumption function Here the most important breakthrough
was the permanent income theory (PIT) of Friedman (1957) and the
closely related life-cycle hypothesis (LCH) of Modigliani and associates,
followed by Halls reformulation of these theories in a rationalexpectations framework. All three followed F53 in judging their hypotheses by their predictive successes and not by the realisticness of their
assumptions. Specifically, they assumed intertemporal utility maximization even over long horizons. Friedman and (though more tentatively)
also Modigliani assumed that the savings/income ratio is uncorrelated
with permanent income. They did not justify these assumptions as realistic, but instead tried to show that their theories predicted consumption
accurately and resolved some puzzling anomalies.
These theories are Marshallian in treating income as exogenous. A more
walrasian approach would use as the budget constraint, not permanent
income, but the households potential labor supply multiplied by the
wage rate, plus non-human wealth. They also avoid the casual empiricism
and introspection on which Keynes relied when he grounded his consumption function in: The fundamental psychological law upon which we are
entitled to depend with great confidence, both a priori from our knowledge
of human nature and from the detailed facts of experience (Keynes 1936,
96; italics in original). Friedman (1957, ix), who stated that he developed
his theory from a close study of the data, gives the strong impression of an
interplay of data and theory. This is less so for Modigliani.15
15
Modigliani and Ando developed the theory initially like a ray of light while on a car trip.
Although Modigliani in developing the relative income theory had previously analyzed
consumption data, he had then neglected the problem until his conversation with Ando
(Modigliani 2001, 59).
135
Investment In the two decades after 1953 the two most notable
developments here were Dale Jorgensons neoclassical theory and James
Tobins q theory. The former is based on two insights: that earlier work on
investment had confounded the capital stock and investment (its first derivative), and on the need to pay attention to the institutional details of the tax
treatment of investment. The latter is based on the gap between the market
price of new capital and its costs of production. Neither is easy to trace to
F53. Jorgensons theory appeared fourteen years after F53. Its attention to
the distinction between capital and investment is a clarification of the internal
logic of the theory, and hence unrelated to F53. The inclusion of tax variables
may seem to show concern about the realisticness of assumptions. But
Friedman did not object to adding realistic details if they improve predictive
success. Tobins q theory, too, appeared only long after F53. Since it does
not attempt to explain the firms equity price, it is more a tool for predicting
investment than an explanatory theory. An instrumentalist reading of F53
would therefore make it seem Friedmanian, but, as the chapters by Mki and
Hoover in this volume show, such a reading is mistaken. Both theories are
Marshallian in that they focus on a concrete problem rather than on elegant
formulations. Developing them may well have involved a close interplay of
theory and data; no information on this is available.
Monetary and portfolio theories The most influential Keynesian
monetary theory is that of Tobin and his Yale school. It takes account of the
complexity of the financial system, with money being just one asset in a
continuum of liquid assets. Such an approach rejects F53. It seeks to give a
realistic picture of the financial system, and does not stress prediction. It is also
walrasian. And Tobin gives the impression that he relied on broadly based
observations (akin to casual empiricism), such as Treasury bills having characteristics similar to money, rather than on a close interplay of theory and data.
Econometric models of the economy Although F53 was not published until three years after Lawrence Kleins (1950) model it could have
influenced subsequent models. Or, if these models follow the methodology recommended in F53, that could account for their popularity, and
hence document the influence of F53.
Predictive success, not the realisticness of their assumptions, is the
main criterion by which these models ask to be judged.16 But that does
16
Predictive success is, however, not the only important criterion. Even if it predicts
accurately, we would reject a theory that explains the US price level by the league standing
of Manchester United. Friedman himself rejects large econometric models because he
believes that they require much more knowledge than we have.
136
Thomas Mayer
not bear testimony to the influence of F53, because prediction is clearly

their main purpose, even if one believes that scientific hypotheses should
be judged by their assumptions. Although they are general equilibrium
models in Friedmans terminology, they are actually Marshallian, because
they seek to provide empirical and practical knowledge and make no claim
to elegance. Casual empiricism may play a role in choosing their individual equations, but primarily they are chosen to accord with substantive
empirical results. Their development relies on a constant interplay of
observation and theory. Indeed, one of the criticisms of these models is
that they are overfitted.
Growth models In the early 1960s the first wave of growth theory
accounted for a substantial part of the work in macroeconomics. Much of
it consisted of varying some previously made assumptions. It essentially
stopped after a critical survey by Frank Hahn and Robbin Matthews
(1964), who wrote:
The variety of growth models is very great and with ingenuity can evidently be
almost indefinitely enlarged. This is very largely due to the rather extreme level of
abstraction employed and the very artificial nature of the problems considered
We want theories that can be used as plumbers use a spanner not simply abstract
systems It would be difficult to claim that any of the models we have discussed
goes far towards explaining [the observed differences in growth rates of countries and periods]. (Hahn and Matthews 1964, 88890)
Clearly, these models are contrary to F53. Whether Hahn and Matthews
were influenced by F53 (which they did not cite) is hard to say, as is
whether the apparently widespread acceptance of their conclusion owes
much to F53.
6.2
Monetarism
F53 provided monetarists with a battering ram for their assault on the
prevailing Keynesian orthodoxy. At that time the wealth effect of falling
interest rates on consumption was little known, and with exchange rates
being more or less fixed, the claim that changes in the money supply could
substantially affect income appeared to require an (at that time) seemingly
implausibly high interest elasticity of investment. Friedman responded that
while it would be desirable to trace the channels by which money affects
income, this is not necessary. Instead, one can document numerous cases
in which changes in nominal income have followed exogenous changes in
money, and combine this fact with the general explanation from price
theory, that when there is an exogenous increase in the supply of one
137
asset (money) the demand for the other asset (goods) increases. The other
leading variant of US monetarism, the work of Brunner and Meltzer, too,
emphasized testing by implications, even though they provided a fairly
elaborate theoretical discussion.17
Without the focus on prediction that F53 advocated monetarism would
not have achieved its popularity. Most economists would then have agreed
with Frank Hahns (1971, 61) objection that: Friedman neither has nor
claims to have a monetary theory. His strong and influential views are not
founded on an understanding of how money works, but on what his
empirical studies have led him to believe to have been the course of
monetary history. In principle, it is obvious that the demand for
money, and hence its velocity, depends upon the expected yields of all
competing assets. Yet, unlike the Yale school, Friedman disregarded
them because he believed that including them did little to improve the
theorys predictions.
The monetarists case is also bound up with their rejection of the
mixture of introspection and casual empiricism that was a prominent
part of then prevailing Keynesian thinking. Examples are Keyness consumption function, his intuition that the main factors generating changes
in GDP are changing expenditure motives rather than changes in the
money supply, and the empirical significance of the speculative demand
for money. To successfully challenge Keynesian theory monetarists had to
show that what seemed commonsensical and intuitively obvious was an
insufficient basis for macroeconomics.18
The need to distinguish sharply between normative and positive elements played little role in the debate. Although normative issues permeated
policy discussions, both sides failed to draw sufficiently clear distinctions
between normative and positive issues. Still another Friedmanian theme,
the need for interplay between theoretical and empirical work, may have
played a substantial role in the rise of monetarism, since it was well suited to
17
18
Brunners (1969) view on the role of assumptions was similar to Friedmans.

That expenditures depend more on the incentives to spend than on money holdings
seems obvious to middle-class people (such as most economists) who think of expenditures primarily as consumption. It may not be so obvious to the rich who are busy with
asset reallocation, or to the poor who are severely cash constrained. From the late 1930s to
the early 1950s Keynesians attached much importance to speculative liquidity preference
without any evidence that it is empirically significant. The issue of plausibility of assumptions vs. accuracy of predictions also plays an important role in disputes about monetary
policy. Until the application of time-inconsistency theory one of the two central points in
this debate was the Keynesian argument that one can infer from the plausible assumption
that central bankers are well intentioned and intelligent, that they will pursue reasonable
policies. Monetarists responded by citing instances of perverse policy, that is instances
where the Keynesian theory predicted badly.
138
Thomas Mayer
what monetarists were doing. Moreover, the monetarists interplay of

theory and empirical work may have made monetarism more popular by
providing a welcome contrast to the bifurcation of much of the Keynesian
research effort between theorists who did not test their theories, and applied
economists who focused on forecasting.
6.3
New classical theory
The substantive message of new classical theory, rapid market clearing

and real business cycles, has won a foothold in macroeconomics, but not a
victory. But its methodological message, that is, its reductionist dogma,
has. Despite Friedmans own rejection of new classical theory, some
economists credit (or blame) F53 for it, since it justifies the use of
unrealistic assumptions, such as highly flexible wages. As Wade Handss
chapter in this volume shows, within the science-studies dimension this is
so, and the victory of new classical methodology documents the influence
of F53. But this is more a defeat than a victory for Friedman, because the
new classicals misuse F53. By insisting that discussions of macroeconomic phenomena make no assumptions that are not firmly grounded
in microeconomics they treat assumptions as an integral and crucial part
of the analysis, while to Friedman, assumptions are background scenery
that can be varied, depending upon the play currently on stage. Moreover,
when Friedman objected to testing by the realisticness of assumptions, he
did not do so in favor of a priorism, but in favor of predictive testing. And,
although the new classicals often present predictive tests, they give the
impression that these tests are tagged on more for the sake of decoration
than for the heavy lifting: Theory ahead of observation is their slogan. I
conjecture that no economist has ever accepted rational-expectations
theory primarily because he or she found the empirical evidence
convincing.
On the issue of Marshallian versus walrasian economics, new classical
economists are obviously not Friedmanian. It is unclear to what extent
they adhere to his rejection of casual empiricism, in part, because using
elaborate econometrics is no talisman against casual empiricism, and in
part because it is unclear whether the estimates inserted into calibrated
models are derived from an objective survey of the literature or selected
because they happen to give results consistent with new classical theory.
Friedmans insistence on distinguishing positive from normative elements
is again not relevant. But the interplay between theory development and
intimate knowledge of the data that Friedman recommended is relevant
and, it seems missing in new classical theory.
139
Sources of influence
Although, as previously discussed, several factors independently impelled

economics in the directions advocated in F53, I believe (though I certainly
cannot prove it) that F53 played a significant role in changing economics.
One reason for this belief is the merit of its thesis. Methodologists are right in
calling it philosophically naive and confused, and yet it is a much better
guide for practicing economists than its two main rivals at the time, a priorism and naive empiricism. A second reason is its persuasive style which
presents seemingly audacious, and hence stimulating, statements with an air
of sweet reasonableness. Third, we have Friedmans achievements in substantive economics, often grounded in a methodology which, if not exactly
that of F53, bears a reasonably close relation to it. Whether the association of
F53 with the marginalist controversy helped or hindered is hard to say.
Marginalists had a strong incentive to welcome F53. But institutionalists
could dismiss it as enemy propaganda, without appreciating the large olive
branch Friedman extended to them by his opposition to walrasian economics. Friedmans political views may also have induced some economists to
oppose F53. But I suspect that this was only a minor factor. Finally,
Friedmans personality for some reason, seems to provoke either ardent
admiration or sharp opposition; the net effect of that is hard to gauge.
8
Conclusion
Since 1953 economics has changed along the lines F53 recommended.
However, other factors could also have accounted for these changes,
particularly for the shift away from testing by the realisticness of assumptions. At the same time, some of these factors could also explain why
Friedman seems to have failed in his opposition to walrasian economics.
But it is certainly possible that without F53 the walrasian tide might have
been stronger, so that even on this issue F53 may have been influential.
How much of the observed changes in economics one attributes to F53
and how much to these other factors depends, in part, on ones acceptance
of the Stiglerian and postmodernist view that economists theory choice is
governed by their self-interest. But even if it is, F53 could still have been
influential by providing a convenient rationalization for, and thus accelerating, the changes that occurred. To what extent it did so is a matter of
intuitive appraisal rather than of solid evidence, and I feel qualified to
make such an appraisal only with respect to macroeconomics.
Although the immediate inspiration for, and the examples given in, F53
are microeconomic, F53 seems to have been more influential in macroeconomics since so much of modern microeconomics consists of building
140
Thomas Mayer
models that explore the implications of various assumptions (see also

Williamsons chapter in this volume). Perhaps F53s greater influence
on macroeconomics is due, in part, to Friedmans own shift of emphasis
towards macroeconomics, or to macroeconomics being traditionally
more policy oriented, and hence more aware of the need for a
Marshallian approach and for correct predictions.
Within macroeconomics F53 probably played an important perhaps a
very important role in the ready acceptance of the LCH, PIT, and the
rational-expectations consumption function. And it did play a major role
in the monetarist counter-revolution, and in that way in the synthesis of
monetarism and new Keynesianism that underlies the currently prevailing
macroeconomics. Eventually the hard facts of experience would have
prevailed over the Keynesianism of the 1960s and 1970s even without
F53, but it would have taken them longer. And perhaps new classical
theory, too, would not have made the big time were it not for F53s
warnings against testing by the realisticness of assumptions, and against
reliance on casual empiricism. It is even possible that this was the most
important influence of F53. That would be ironic.
In microeconomics I suspect, but that may just reflect my lack of
familiarity with it, that F53 did influence primarily Chicago economists,
and perhaps work on public choice theory. But by its stress on the
explanation of specific phenomena it may also (probably indirectly) have
influenced work on asymmetric information.
References
Allen, Clark (1954). Review of Milton Friedman Essays in Positive Economics.
Southern Economic Journal, 10, January, 3949
Backhouse, Roger (2002). Economic models and the role of informal scientific
methods. In Uskali Mki (ed.), Fact and Fiction in Economics. New York:
Cambridge University Press, 20213
Baumol, William (1954). Review of Milton Friedman Essays in Positive Economics.
Review of Economics and Statistics, 36, November, 4625
Blaug, Mark (1980). The Methodology of Economics. Cambridge: Cambridge
University Press
Boulier, Brian and Robert Goldfarb (1998). On the use and nonuse of surveys in
economics. Journal of Economic Methodology, 5, June, 122
Brunner, Karl (1969). Assumptions and the cognitive quality of theories.
Synthese, 20, 50125
Colander, David (1991). The lost art of economics. Journal of Economic
Perspectives, 6, Summer, 1918
Dasgupta, Partha (2002). Modern economics and its critics. In Uskali Mki (ed.),
Fact and Fiction in Economics. Cambridge: Cambridge University Press, 5789
141
Duesenberry, James (1949). Income, Saving and the Theory of Consumer Behavior.
Figlio, David (1994). Trends in the publication of empirical economics. Journal of
Economic Perspectives, 4, Summer, 17987
Friedman, Milton (1949). Discussion. American Economic Review, 39, May, 1969
(1950). Wesley C. Mitchell as an economic theorist. Journal of Political Economy,
58, December, 46395
(1953). Essays in Positive Economics. Chicago: University of Chicago Press
(1957). A Theory of the Consumption Function. Princeton: Princeton University
Press
Friedman, Milton and Simon Kuznets (1945). Income from Independent Professional
Practices. New York: National Bureau of Economic Research
Friedman, Milton and Anna Schwartz (1963). A Monetary History of the United
States. Princeton: Princeton University Press
Gurley, Jack and Edward Shaw (1960). Money in a Theory of Finance. Washington,
DC: Brookings
Hahn, Frank (1971). Professor Friedmans views on money. Economica n.s., 39,
February, 6180
Hahn, Frank and Robin Matthews (1964). The theory of economic growth: a
survey. Economic Journal, 74, December , 779902
Hammond, J. Daniel (1990). McCloskeys modernism and Friedmans methodology: A case study with new evidence. Review of Social Economy, 48,
Summer, 15871
(1996). Theory and Measurement. Cambridge: Cambridge University Press
(1998). Friedman, Milton. In J. Davis, D. W. Hands and U. Mki (eds.),
Handbook of Economic Methodology. Cheltenham: Edward Elgar
Harsanyi, John (1995). Games with incomplete information. American Economic
Review, 85, June, 291393
Hayek, Friedrich von (1992). The pretence of knowledge. In Assar Lindbeck (ed.),
Nobel Lectures, Economic Sciences, 19691980. Singapore: World Scientific
Hirsch, Abraham and Neil De Marchi (1990). Milton Friedman. New York:
Harvester Wheatsheaf
Hoover, Kevin (2001). The Methodology of Empirical Macroeconomics. Cambridge:
(2004). Lost causes. Journal of the History of Economic Thought, 26(2), 14964
Hutchison, Terence (1954). Review of Milton Friedman Essays in Positive
Economics. Economic Journal, 64, December, 7969
Keynes, John M. (1936). The General Theory of Employment, Interest and Money.
New York: Prentice Hall
Keynes, John N. (1963). The Scope and Method of Political Economy. New York:
Augustus Kelly
Klein, Lawrence (1950). Economic Fluctuations in the United States, 19211941.
New York: Wiley
Leontief, Wassily (1992). Structure of the world economy. In Assar Lindbeck
(ed.), Nobel Lectures, Economic Sciences, 19691980. Singapore: World
Scientific
Lerner, Abba (1944). The Economics of Control. London: Macmillan
142
Thomas Mayer
Mayer, Thomas (1993). Friedmans methodology of positive economics: a soft

reading. Economic Inquiry 31, 213223
(1995). Doing Economics, Cheltenham: Edward Elgar
(2001). Misinterpreting a failure to confirm as confirmation: a recurrent misreading of significance tests., Working paper, www.econ.ucdavis.edu
(2004). The influence of Friedmans methodological essay. University of California,
Davis, Working paper no. 041
Modigliani, Franco (2001). Adventures of an Economist. New York: Texere
Musgrave, Alan (1981). Unrealistic assumptions in economic theory. Kyklos, 34,
(3), 37787
Newman, Peter (1954). Review of Milton Friedman Essays in Positive Economics.
Economica, 21, August, 25960
Reder, Melvin (1982). Chicago economics: permanence and change. Journal of
Economic Literature, 20, March, 138
Schneider, Erich (1955). Review of Milton Friedman Essays in Positive Economics.
Weltwirtschaftliches Archiv
Snowdon, Brian and Howard Vane (1999). Conversations with Leading Economists.
Cheltenham: Edward Elgar
Soper, C. S. (1954). Review of Milton Friedman Essays in Positive Economics.
Economic Record, 30, November, 3236
Sugden, Robert (2002). Credible worlds: the status of theoretical models. In
Uskali Mki (ed.), Fact and Fiction in Economics. Cambridge: Cambridge
University Press
Tinbergen, Jan (1992).The use of models: experience and prospects. In Assar
Lindbeck (ed.), Nobel Lectures, Economic Sciences, 19691980. Singapore,
World Scientific
Vickrey, William (1954). Review of Milton Friedman Essays in Positive Economics.
American Economic Review, 44, June, 397400
Waldman, Michael (2003). Durable goods theory for real world markets. Journal
of Economic Perspectives, 17, Winter 2003, 13154
Did Milton Friedmans positive methodology

license the formalist revolution?
D. Wade Hands
A man who has a burning interest in pressing issues of public policy, who
has a strong desire to learn how the economic system really works in
order that that knowledge may be used, is not likely to stay within the
bounds of a method of analysis that denies him the knowledge he seeks.
He will escape the shackles of formalism, A far better way is to try to
derive theoretical generalizations to fit as full and comprehensive set of
related facts about the real world as it is possible to get.
(Friedman 1946, 631)
Introduction
Milton Friedmans 1953 paper The methodology of positive economics

(hereafter F53) is clearly the best-known and most cited work in twentiethcentury economic methodology. F53 often exhausts the practicing economists exposure to the methodological literature. As the philosopher Daniel
Hausman put it: It is the only essay on methodology that a large number,
perhaps a majority, of economists have ever read (Hausman 1992, 162).
There is a vast methodological literature surrounding F53, but
this chapter will not approach the essay from the perspective of any of
these familiar methodological positions. For example, what follows
will not: (1) summarize Friedmans argument in any detail;1 (2) try to
situate Friedmans essay within the existing philosophical, particularly
*
Earlier versions of this paper were presented in Milton Friedmans Methodology Paper
after 50 Years at the ASSA meetings in Washington, DC, January 2003 and in the
Philosophy Department Colloquium at the University of Quebec at Montreal, April
2003, as well as at the EIPE conference Positive Economics: Milton Friedmans Essay
at 50 in Rotterdam, December 2003. A shortened version, with the title Did Milton
Friedmans methodology license the formalist revolution? was published in Journal of
Economic Methodology, 10 (2003), 50720. I would like to thank Roger Backhouse, Robert
Leonard, Uskali Mki, Tom Mayer, Robert Nadeau, and numerous other attendees of the
above presentations for helpful comments. Errors and omissions are of course solely my
responsibility.
There are numerous summaries available in the secondary literature, including Blaug
1980/1992, Caldwell 1982/1994, Hands 2001, and Hausman 1992.
143
144
D. Wade Hands
philosophy of science, literature;2 (3) make the case for a particular interpretation of key words like assumptions, realism, prediction, etc;3 (4)
try to reconcile (or critique) the relationship between Friedmans methodological advice and his actual scientific practice;4 (5) evaluate the adequacy
of Friedmans stated methodology as an economic methodology. What
follows is broadly concerned with F53, but it will not address any of the
traditional methodological questions associated with Friedmans famous
essay.
What I will do is summarize and evaluate two conflicting interpretations
of F53 that have emerged within the recent methodological literature;
in particular, I will address a disagreement regarding the relationship
between the advice contained in Friedmans essay and the ascent of
formalism in economics. I will not take any substantive (new or existing)
position on Friedmans essay, but I will take a position on the role that F53
played in licensing the so-called formalist revolution. I will leave the
debate over whether one ought to do economics in the way suggested by
Friedmans essay, as well as various questions about the true essence of
Friedmans position, to methodologists with greater expertise in such
subjects. I will focus on the more naturalistically inclined question of the
role that Friedmans ideas played in the ascendancy of a particular set of
socially held beliefs about the (only) acceptable way to do research within
the science of economics.
I will focus my discussion on three influential contributors to this
ongoing debate: Mark Blaug, Terence Hutchison, and Thomas Mayer.
Blaug and Hutchison have argued repeatedly that F53 did in fact license
the formalist revolution, while Mayer has argued precisely the opposite;
the formalist revolution was (at least in part) a result of not following
Friedmans methodological advice. Juxtaposition of the views of these
two sets of authors on the question of Friedmans essay raises important
questions about the relationship between how one interprets F53 and how
one views its impact on the profession. For as we will see, all three parties
agree about essentially every methodological issue surrounding the
famous essay they have fundamentally the same conception of formalism; they all agree that it became the dominant approach to economic
2
3
4
Although Bolands (1979) instrumentalist reading is the most popular of such philosophical reconstructions, Mki (1989, 1992, 2000, this volume) and Hoover (2004 this
volume) offer persuasive realist interpretations. A realist reading is also supported by
Friedmans emphasis on the importance of novel facts.
Although I consider Mkis (1989, 1992) suggestion that we replace Friedmans realism
with realisticness to be a good one.
Although I would admit that I find the most detailed of such efforts Hirsch and De
Marchi (1990) to be relatively unpersuasive.
Did Milton Friedmans methodology license the revolution?
145
theorizing in the post-World War II era and that it was generally a bad
thing; they agree in principle that the right way to do economics involves
less mathematical abstraction, more and more serious empirical testing,
increased focus on the real facts of economic life, and a greater emphasis
on economic policy and yet they disagree sharply about the impact of
Friedmans essay on the rise of formalism.
On the formalist revolution
The formalist revolution has received increasing attention from those

working in the history and philosophy of economics during the last
few years. While debate rages about the causes and many of the specific
characteristics of the formalist revolution, there does appear to be a
consensus regarding the broad general features and approximate timing
of this fundamental change in the dominant form of economic theorizing.5 There was an identifiable difference between the representative
research paper appearing in the most prestigious economics journals in
the 1930s versus the 1970s; and it is equally clear that the later work was
more formal more mathematical, more analytical, less historical, less
institutional, more standardized, and more narrow regarding admissible
priors than the economic theorizing of an earlier period.
Although I agree with Roy Weintraub and others6 that a serious
archeology of formalism (Weintraub 1998, 1844) would require a careful, and deeply textured, historical investigation involving a sensitivity to
the subtle variations among the different versions of formalism that have
existed within the history and philosophy of mathematics; attention to the
evolving and overlapping epistemic visions at work in science, social
science, and mathematics; as well as an examination of the socially constructed character of all of the relevant disciplinarities I do not believe
that such an historical archeology is required here. The current project is
simply to investigate two different, rather starkly conflicting, answers to
the question of whether Friedmans methodology licensed the formalist
revolution: Blaug and Hutchison claiming that it did, and Mayer claiming
that it did not (and could actually have prevented it). If these three authors
had substantially different views of the essential features of the formalist
revolution, then it would be necessary to ferret out those differences in
5
Hutchison (2000, 16) seems to give Benjamin Ward (1972) credit for initially discussing
formalism in economics, but the Friedman quote at the beginning of this chapter shows
that it goes back far beyond the early 1970s.
This literature includes Backhouse (1998), Golland (1996), Mirowski (1986, 2002),
Punzo (1991), Weintraub and Mirowski (1994), and Weintraub (2002).
146
D. Wade Hands
order to determine if their different interpretations of formalism were

responsible for their different views of the role of F53. But there are no
such differences. All three authors are concerned with roughly the same
features of postwar economic theorizing and they define, and date, formalism in basically the same way. Since all three authors share the same
notion of formalism, the variation in their views about Friedman cannot
be explained by the variation in their views about formalism, and thus the
only characterization of formalism relevant to the current investigation is
their (shared) view. The question of a serious archeology of formalism can
wait for another context.
Blaug, Hutchison, and Mayer all essentially agree about what formalism
is/was, roughly when it hit economics, which subfields within economics
are the best exemplars of the formalist influence, and whether formalism
was a good thing for the discipline or not (it was not). They also seem to
agree, though perhaps less strongly, about the essential methodological
features of a better, less-formalist, approach to economic theorizing, and
which economists come closest to living up to these quite different standards. The most relevant texts on these issues are Blaug (1994, 1997, 1998,
2002, 2003), Hutchison (1992, 2000), and Mayer (1993, 1995). It is
useful to note that while the arguments of these three authors will be the
main subject in what follows, they are certainly not alone in their agreement about the what, when, which, and whether of formalism; recent
commentators as disparate as Tony Lawson (2003) and Deirdre
McCloskey (1994) also agree on these same basic issues.
According to the three authors mentioned above the formalist turn
clearly involved the increased use of mathematics, abstraction, and deductive modes of reasoning. Theorems and proofs replaced systematic argumentation; success came to be measured almost exclusively in terms of
the development and application of new techniques; and the form of a
theoretical argument came to take precedence over its economic content.
According to Blaug, formalism involves the worship of technique (Blaug
2002, 36) and giving absolute priority to the form of economic theories
rather than their content (2002, 34); for Hutchison it is the method of
rigorous, logical deduction from (what have been claimed to be) a small
number of self-evident principles (Hutchison 2000, 17) and the substitution of fantasy content for realistic, or relevant, content (2000, 18).
For Mayer formalism is abstract theory that is concerned with high-level
generalizations, and looks toward axiomization (Mayer 1993, 7) and
the criteria by which formalists judge economic research are, to a large
extent, those of mathematicians (Mayer 1993, 25). Formalism transformed economics from an informal field, mainly using a natural language and relatively simple techniques, to a more formal field that
147
converses much of the time in the language of mathematics, and puts a

great stress on the use of the latest techniques (Mayer 1995, 73).
While formalism is clearly linked to mathematics, all three authors
recognize that the formalist mode of argumentation does not actually
require the use of mathematical analysis. All three cite Ricardo as an
example of a proto-formalist (Blaug 2002, 36; Hutchison 2000, pp. 3,
17; Mayer 1993, 7). It is also clear that none of the three are against the
general idea of using mathematics as a tool in economic analysis; mathematical symbolism and/or an occasional derivative are not the problem.
Now formalism is not the same thing as formalization or mathematization
because it is possible to express a theory mathematically and even axiomatically
without necessarily degenerating into formalism, which simply means giving top
priority to the formal structure of modelling irrespective of its content; it is to
prefer rigour and precision to relevance, and elegance and logical coherence of
analysis to practical implications; it is to make a fetish of theory and to deride a
vulgar appeal to the real world. (Blaug 1994, 131)
Even though mathematics is neither necessary nor sufficient for formalism, the rise of mathematical economics, and particularly ArrowDebreu
general equilibrium theory, during the 1950s and 1960s clearly represented the high-water mark for formalism in economics and it set the tone
for the most prestigious economic theorizing during the latter half of the
twentieth century. All three authors have very negative things to say about
the impact of Walrasian general equilibrium theory; Mayer is not quite as
harsh as Blaug and Hutchison, but he still considers it an example of
formalism rather than his preferred empirical-science economics. They
all agree that economics took a seriously wrong turn and Arrow and
Debreu (1954) pointed the way. A few of their remarks are worth quoting
in detail.
If we can date the onset of the illness at all, it is the publication in 1954 of the
famous paper by Nobel laureates Kenneth Arrow and Gerard Debreu; it is this
paper that marks the beginning of what has since become a cancerous growth in
the very centre of microeconomics. (Blaug, 1997, 3)
If there is such a thing as original sin in economic methodology, it is the worship
of the idol of mathematical rigour, more or less invented by Arrow and Debreu
(1954) and then canonized by Debreu (1959) in his Theory of Value five years later,
probably the most arid and pointless book in the entire literature of economics.
(Blaug 2002, 27)
In the 1950s and 1960s the formalist revolution met with very little critical
resistance. In fact, the high confidence of leading mathematical abstractionists
and formalizers in the impregnability of their achievements continued unabated
through the 1970s and much of the 1980s, especially with regard to their flagship,
148
D. Wade Hands
the new ultra-abstract and extremely refined version of general equilibrium

analysis (Hutchison 2000, 20)
Not only do they agree about the pernicious effect of ArrowDebreu

general equilibrium theory, they also agree about the succeeding theorizing that most clearly followed in its formalist footsteps. There are many
formalist successors, but new classical macroeconomics and rational
expectations theory are perhaps the clearest examples. Given Mayers
interest in policy-driven macroeconomics, it is not surprising that he
argues that the new classical coup dtat is a striking attempt to impose
a formalist regime on what was previously largely an empirical science
domain (Mayer 1993, 80) and that the goal of new classical theory is to
develop, not theories, but theorems (1993, 90). But the other authors
also echo his remarks: Unfortunately, macroeconomics in recent years
has fallen prey to empty formalism in which the principal goal of theorizing seems to be analytical elegance and not a better grasp of practical
problems and hence greater control of economic events (Blaug 2002,
30). It is important to note that such criticism is not restricted to
economic theorizing with the same political-economic slant as rational
expectations theory. Blaug (2002, 278) blasts Sraffa; Hutchison blames
his old teacher Joan Robinson (ostensibly a formalist even in her
Keynesian and Marxian periods [Hutchison 2000, 25n13]); and, as
we will see below, Mayer actually defends the methodological practice of
mid-century Keynesians. Formalism seems to be scattered all over the
political-economic landscape, and for these three authors it is equally
bad wherever it appears.7
In addition to agreement about the methodological disease and those
most responsible for the contamination, there also seems to be a consensus about a better economic methodology what economists should do
instead of practicing formalism as well as the economic exemplars for
this proper methodological practice. Blaug and Hutchison have long
argued that economics should be a real SCIENCE, and that the proper
way to do that is to follow the strict falsificationist rules associated with
the philosophy of Karl Popper. Blaug, of course, has often veered off the
falsificationist path in the direction of Lakatos, but even there his methodological focus has remained fixed on empirical progress, the aspect of
Lakatoss work that emphasized empirical testing almost as much as
7
Backhouse (1998) lists three types of formalism in economics axiomatization, mathematization, and methodological. The formalism that concerns these authors undoubtedly
involves all three of Backhouses categories, but the first, axiomatization, is clearly the main
culprit.
149
Poppers bold conjectures and severe tests.8 Mayer endorses a similar

position that he calls empirical science economics, a methodology that
is less concerned with the philosophy of Karl Popper, but is quite similar
in its emphasis on empirical testing and the necessity of confronting
economic theories with economic reality.
In contrast to formalist economists, there are the empirical science economists,
who take as their model the natural sciences; not only physics, but also the less
highly developed sciences, such as geology and biology. These fields, too, use
mathematics, but only as a convenient tool. They are much less concerned with
mathematical virtues, and more concerned with empirical testability, and with the
extent to which a theory enhances our understanding of what we actually observe.
(Mayer 1993, 25)
In addition to these similar views about the proper scientific method, there
also seems to be agreement about which economists best characterize this
good empirical practice. They all point to Alfred Marshall as an exemplar of
the right, non-formalist, way to do economics. Hutchison claims that
Marshall had very subtle but incisive methodological views (Hutchison
2000, 4) and doubts that any other leading economist has ever had quite
such broad and penetrating insights on fundamental methodological issues
as the nature of economic theorizing (2000, 290), while Mayer argues that
his distinction between formalist and empirical science economics is quite
similar to the distinction between Marshallian and Walrasian economics,
since Marshallians are much more in the empirical science camp than are
Walrasians (Mayer 1993, 35). They also consider Keynes and a few others
to be methodological exemplars, but the Marshallian focus on economics as
a practical tool for solving policy problems in the ordinary business of life
makes his approach the paradigm case of non-formalist economic theorizing. It is useful to note that Friedman has also consistently cited Marshall
as an (perhaps the best) example of the proper methodological approach,
and that Friedmans Marshallianism is supported by many later methodological commentators (Hoover, this volume, for example).
Finally, there is the fact that all three of these authors praise the other
two for their methodological insights. While this is not surprising for
Blaug and Hutchison who have, and agree that they have, essentially the
same methodological position, it also seems to be true for Mayer as well.
Mayer praises both Blaug and Hutchison (1992, 3943) for their analysis
of formalism and general equilibrium theory, and Hutchison (2000, 21)
returns the favor. Blaug even goes so far as to say that his view and Mayers
8
I will not summarize the well-known positions of Hutchison and Blaug. The interested
reader can examine Hutchison (1938, 1992, 2000) and Blaug (1980/1992) or the summaries in secondary sources such as Caldwell (1982/1994) or Hands (2001).
150
D. Wade Hands
are essentially the same; Mayer relies less on the philosophical literature,
but the final result seems to be the same.
Thomas Mayer (1993) has recently published a study of Truth versus Precision in
Economics. He eschews all discussion of formal methodology but he also denies
that an invisible hand always impels economists to develop the best kind of
economics; in general, he argues for an empirically oriented economics against a
formalistic one. As far as I am concerned, he is on the side of the angels and his
book is full of illuminating illustrations of how many modern economists have
sacrificed relevance for rigour, technique for substance. (Blaug 1994, 131)
Friedmanic license
It is clear that Blaug, Hutchison, and Mayer fundamentally agree about

economic methodology; they agree about what is wrong, who does/did it
most wrong, how to do it right, who does/did it most right, and they even
praise each others views. Given this agreement, they should have similar
views regarding key contributions to the field such as Friedmans essay,
right? Well actually, no; as suggested above, they disagree substantively
about the question of Milton Friedmans methodology, in particular the
question of whether F53 was responsible for whether it licensed the
formalist revolution.
Since they agree about so much, one might suspect that they disagree
about Friedman because they have different interpretations of what
Friedman said in his famous essay, but that is not the case. Although
the methodological literature is rife with various interpretations of
Friedmans methodology, these three economists essentially agree about
the content of Friedmans essay: the only test of an economic theory is the
accuracy of its predictions, and the realism of its assumptions does not
matter.9 The disagreement is not about Friedmans main methodological
message, but rather about the impact of that message on the profession.
According to Blaug and Hutchison, the formalist revolution was based on
turning away from the facts, the real facts, of economic life and focusing
instead on abstract modeling deduced from empirically ungrounded
assumptions. Their position is that by asserting that the truth, or the realism,
of the assumptions is not important, Friedmans methodology openly
licensed the anti-empirical practice of formalist economics. Friedman of
course also emphasized prediction and the importance of verifiable empirical implications, but they argue that once the assumptions do not matter
9
There is of course much more to Friedmans essay than this one simple sentence, but it is
the key, and generally agreed upon, message. See Friedman (1953) or the secondary
sources cited in note 1 for more details.
151
cat was out of the methodological bag, the profession was free to go speeding
down the formalist road.
The tendency with regard to abstraction towards anything goes was further
encouraged by the doctrine, none to clearly expounded in Milton Friedmans
famous essay The methodology of positive economics, according to which the
unrealism of assumptions need, and should, not be questioned, provided the
conclusions or predictions were tested. (Hutchison 2000, 193)
What runs through all this is the licence that Friedmans methodology of positive
economics gave economists to make any and all unrealistic assumptions, provided only that their theories yielded verifiable implications. But even if one grants
for the sake of argument that assumptions do not need to be descriptively accurate,
Friedman failed to insist that they do need to be robust, that is, capable of being
relaxed without fatal damage to the model they underpin. (Blaug 2002, 30)
Although Blaug and Hutchison both cite a number of other, primarily

sociological and profession-specific factors that helped to bring about the
formalist revolution such as the Americanization of academic economics
and the rise of publish-or-perish Friedmans methodological message is
clearly one of the most important forces.
Mayers reading of the impact of Friedmans positive methodology is of
course quite different. Mayer supports not only Friedmans methodology,
but also most aspects of his monetarist economics. In fact Mayer has spent
much of his life furthering the Friedman macroeconomic program;
including vigorously defending it against later developments in rational
expectations and new classical macroeconomics that others consider to
be the offspring of Chicago monetarism. One way to interpret Mayers
writings on economic methodology during the last decade or so is to view
this work as the explicit defense and further elaboration of Friedmans
methodological position that Friedman himself never provided.
According to Mayer, the profession started its path down the formalist
road when economists turned away from the methodological injunctions
in Friedmans essay. Although Mayer often sides with monetarism on
questions of macroeconomic theory and policy, he considers the methodological behavior of both sides of the Keynesianism versus monetarism debate to be fundamentally sound. The two sides of the debate
clearly differed about theory (the elasticity of money demand, for example), the empirical evidence (the stability of the consumption function,
for example), and policy (monetary versus fiscal policy, for example),
but they did not differ fundamentally about the grounds on which one
should conduct a scientific debate about theory, evidence, and policy.
Both sides were doing empirical science economics they were using
the best available empirical evidence in an effort to identify general
152
D. Wade Hands
macroeconomic principles and regularities that could be used as the basis

for effective policy. This positivistic methodology followed Friedmans
prescriptions, and found an exemplar in his monetarist practice, but it was
not unique to the monetarist side, since Keynesian methodology is more
or less similar to monetarist methodology (Mayer 1995, 76). Both sides
shared a common, and for Mayer a proper methodological stance, a
stance that was best described and defended in Friedmans essay. Mayer
argues that while there were many sociological and professional factors
that contributed to the transformation of the dominant style of theorizing
within economics, the mistake could have been prevented by listening to,
and following, Friedmans methodological advise. F53 did not license the
formalist revolution; in fact it could have prevented it.
the rise of monetarism was fostered by the spread of the then new positivistic
methodology among economists. This methodology combined the insistence of
traditional theory on the rational behavior of maximizing agents with the insistence
of institutionalists, such as Wesley C. Mitchell, on explaining observed phenomena. It was largely, but by no means entirely, the work of Milton Friedman.
(Mayer 1995, 74)
License? We dont need no stinking license!
If these differing views about whether F53 licensed the formalist revolution do not stem from different interpretations of Friedmans essay, the
formalist revolution, or even the proper way to do scientific economics,
then what is the source of the disagreement? I would like to suggest
the difference stems from fundamentally different views regarding the
function and role of methodological inquiry. Crudely put, Blaug and
Hutchison approach economic methodology more from the viewpoint
of, and with the values of, professional philosophers of science; while
Mayer approaches it more from the perspective of a practicing economist.
Since this is a relatively strong statement that conflicts with how the
authors themselves view the situation, it will require some elaboration.
Let me begin with Mayers case since it is not quite as controversial, and
then move on to Blaug and Hutchison where my interpretation is diametrically opposed to the way the authors view their own positions.
Mayer grounds his interpretation of the rise of formalism, and the
corresponding demise of good policy-driven economics, in what is essentially an economic analysis of the behavior of those within the economics
profession during the post-World War II period. Mayers most explicit
statement of the causes of the formalist turn are contained in chapter 2 of
Mayer (1993); the chapter is titled Economists as economic agents:
towards a positive theory of methodology. In this chapter he does not
153
offer the type of epistemic advice that characterizes most philosophy of

science, but rather provides an economic analysis, specifically a public
choice analysis, of the relevant episode in the history of economic thought.
His argument is simply that what happened in economics was a market
failure in the market for economic knowledge. Various institutional
changes during the immediate postwar period led the economics profession to look inward, rather than outward, for both inspiration and praise.
The professionalization of academic economics; the standardization
of graduate education; and the consolidation of an invisible college of
editors, referees, and grant-funding agencies, all contributed to the establishment of a market for academic research in economics that had, and
still has, much in common with the market for government services
(Mayer 1993, 11). The result is a situation where academic economists
by and large, write for each other and thus the usual market discipline
does not exist (1993, 10). The people responsible for determining and
enforcing the standards are the same people that must work within the
confines of those standards.
The market for economic research is therefore self-referential; we write papers to
please ourselves and each other. This would be consistent with maximizing the
publics welfare if we evaluated papers entirely objectively by their social utility,
without being biased by our self-interest. But the assumption that producers
disregard their self-interest is one that we would be loath to make about others.
(1993, 12)
Although one might criticize the details of Mayers public choice story, my
purpose here is neither to challenge the adequacy of his narrative nor to
speculate about how the story would be different if it were told from a
different explanatory perspective within the social sciences. The issue here
is not the details of Mayers particular story, but the general tack or style to
methodological questions his approach exhibits. His style is grounded in a
particular social science public choice theory and for that reason it is far
less likely to blame an observed set of methodological practices on bad, or
easily misinterpreted, methodological rules. The behavior of the profession, the choices that were made rather than others that were not made,
can be explained, but from Mayers point of view they are to be explained
in essentially the same way that one would try to explain any other set of
social behaviors/choices. Looked at from the perspective of public choice
theory, a social organization that is based on the self-interested behavior of
individual agents is very likely to be inefficient (epistemically or economically) when those choices are not constrained by the forces of a competitive market. In particular, if there is imperfect competition in the
market and suppliers determine demand, then inefficiency will be the
154
D. Wade Hands
(perfectly understandable) consequence. The profession had other available choices they could have listened to what Friedman had to say about
methodology but they chose not to for good (good in the sense of being
able to be explained by social science) reasons. The onus of the inquiry is
to explain the path that was taken, explain the observed social behavior,
rather than fret about finding the perfect set of rules. This is not to say that
social scientists are never interested in what appears on a particular tablet
of rules; it is just not one of the first explanatory issues that come to mind.
The more obvious issues requiring social explanation seem to be why the
rules were or were not accepted by the crowd, and what were the social
consequences of those actions. Friedmans essay does not provide a very
good explanation in this sense: that is, it does not provide a very good
social explanation of the causes and consequences of the social phenomenon of the formalist revolution.10
My claim is that neither Blaug nor Hutchison would be particularly
troubled by the fact that Friedmans essay is not a very good social
explanation for the rise of formalism, because they are not really interested
in social explanations the explanation of why, or how, or for what
interests, certain methodological rules came to be accepted they are
more interested in the philosophical question of whether the rules are true
or right. The priority is on the content of the tablet, not on the social
explanation of its acceptance, rejection, or revision. The implicit presumption is that if it is true, if the methodological tablet contains the
correct rules, then they will be accepted. This is essentially a philosophers approach to questions of either epistemology or ethics. Let the
philosophers write the rules correctly what exactly is the true, the
good, and the beautiful? and the really important job is done. The rest
is mere mundane social explanation; leave it to the pocket-protector
crowd.
It is important to emphasize that neither Blaug nor Hutchison would
endorse such an attitude if the issue were economic policy. For good
10
It is useful to note that social explanation-based approaches are increasingly becoming the
norm, rather than the exception, in the methodological literature on formalism in economics. This is true of Weintraub (2002), but there are many others as well. For example, both
Tony Lawson and Deirdre McCloskey, two economists who hold quite different views
about both methodology and economics, have recently examined the question of the
formalist revolution: Lawson (2003, ch. 10) and McCloskey (1994, chs. 10 and 11). Both
of them offer social explanations of the phenomenon (rather than philosophical homilies)
and both couch their explanations in the same type of explanatory framework they would
employ in the explanation of any other social phenomena. In this sense in their general
approach to the question of formalism (not the answers they provide) both Lawson and
McCloskey approach formalism more like Mayer than like Blaug and Hutchison.
155
economic policy they both argue that one should take into consideration
the behavior of those conducting the policy, the response of the agents that
will react to it, and the fact that such responses can never be worked out
in formalist-inspired detail; actual economic science should be situationsensitive and will generally not produce universal results. There is no
good policy independent of the context of its intended, perhaps messy,
economic application. This is the local, small-scale, applied, Marshallian
focus that Blaug and Hutchison both advocate in economics: the empirical, case-by-case method of the leading authorities from Adam Smith
onwards (Hutchison 2000, 333). Positing abstract and universal economic theories that are not based on the real economic context is precisely
the problem with Walrasian formalism. Hutchison, in particular, argues
that effective policy can never be based on pristine optimization, but
will always be imprecise, rule-of-thumb, and conditional. Such policydriven economics is still serious science; even though it is obviously
remote from the Utopian, fantasy levels, traditionally assumed to obtain
throughout the economy by pure theorists in their blackboard exercises
about optimal or maximizing equilibria (which have provided such a
seriously misleading basis for the discussion of real-world policy making)
(Hutchison, 2000, 21516). So when one is doing economics, the notion
of the right rules is not independent of context, agency, and collective
response; in fact the whole idea of a right rule does not even make sense
independently of the relevant social context. On the other hand, when
Blaug and Hutchison are doing methodology the emphasis is entirely
reversed. No longer are things messy, rule-of-thumb, and contextsensitive; in the domain of methodology, the only worthwhile results are
abstract, universal, and context-transcendent rules for differentiating
legitimate science from illegitimate non-science/nonsense; contextspecific rules are necessary for economic policy, but epistemic policy
requires exceptionless universal norms.
This attitude, the attitude that methodology is about finding universal,
epistemically correct, rules for scientific inquiry, and not about the more
social science-based questions of how rules in fact become socially operative epistemic norms or how the norms change from one research site to
another is precisely (or at least has been until recently) the dominant
attitude of mainstream philosophers of natural science. According to
this view, there exists a special philosophical method, not grounded in
the social science or science generally, that can evaluate and pass judgment on the epistemic virtues or cognitive significance of various domains
of scientific knowledge. Once the rules for the epistemically proper
scientific method had been found they were not of course, but that is a
separate story then proper/good scientists would follow them and
156
D. Wade Hands
bad/non-scientists would not. The only role for social explanation in this
view of science is negative. If scientists did not follow the proper scientific
method, their failure could be explained by social forces. So the Lysenkoinspired period in Soviet biology is to be explained solely by social forces,
while Watson and Crick were just following the proper scientific method.
The BlaugHutchison way of doing methodology is a page right out of
the philosophy of science textbook.
It is important to emphasize that this philosophers-way characterization of the BlaugHutchison position is not at all how they would characterize their own view. In fact they characterize it in just the opposite way.
According to their interpretation of their own view, they are taking a
position that is closer to economics, and those who focus on more
descriptive or explanatory methodological issues are letting the philosophical piper play the tune. Although they would not put Mayer in this camp,
they spend a lot of time criticizing methodologists who take a more
contextual or sociological approach as opposed to searching for universally correct rules for being under the spell of avant-garde philosophy.
In Hutchisons words: With no respect for the history and objectives of
economics, valuable and vital distinctions are rejected by comprehensive
academic-philosophical decree, regardless of the special characteristics and
problems of economics and political economy, and of the job which
economists want to get done (Hutchison 2000, 205). As Blaug and
Hutchison see it, those who want to describe and/or explain how methodological rules become dominant, get reinforced, or emerge from the
self-interested actions of scientific agents are acting like philosophers,
while those (such as themselves) who want to proscribe abstract universal
rules of the one true scientific method are respecting the history and
objectives of economics. Frankly, I just do not get it; I just do not see how
they could possibly believe this to be the case. It seems so obvious that
philosophers have generally sought abstract universal truths, while economists (at least non-formalist economists) have acted like dutiful social
scientists and sought the best available answers to context-specific social
questions. While Marshall would undoubtedly agree with what Blaug and
Hutchison have to say about economics, he is surely spinning violently in
his grave when they talk about methodology.
Both Blaug and Hutchison are very critical of the argument offered
by Arrow, Debreu, Hahn, and others that abstract general equilibrium
theory, particularly the first and second fundamental theorems of welfare
economics, provides a formalization, and thus a defense, of the free
market and/or Adam Smiths invisible hand. They both poke fun at the
notion that somehow no one understood, or could defend, the free
market until Arrow and Debreu offered us their formal theorems based
157
on blatantly and even scandalously unrepresentative of any recognizable

economic system (Blaug 2002, 26). Historically persuasive arguments
for the free market for example those of Smith, Mill, and Marshall have
been based on the wishly-washy, case-by-case approach (Hutchison
2000, 319) and in no way come from, or are even enhanced by, the formal
rigorous results of ArrowDebreu theory. The real defense of competitive markets has not come from abstract universal characterization of
efficiency and perfect competition, but from case-by-case investigation
of real situations in actually occurring market economies; what GE
analysis has done to Smiths conjecture is to eviscerate it of real-world
content, and transform it into a piece of rigorous, empirically vacuous,
hyper-abstract analysis, based on a range of fantastic, science-fictional
assumptions (Hutchison 2000, 314).
What Blaug and Hutchison do not seem to notice is that their approach
to methodology is exactly like the ArrowDebreu approach to the invisible
hand. Arrow and Debreu are arguing that the only way to settle real-world
controversies about the efficiency of actual markets is first to obtain formal
theorems regarding the relationship between a universal concept of efficiency and abstract competitive equilibrium prices. Blaug and Hutchison
are entirely correct that the latter is neither necessary nor sufficient for the
former. What they do not seem to recognize is that they are saying essentially
the same thing about methodology; we cannot settle real-world controversies
about the efficiency of actual practices within the context of specific
economic research programs, until we first obtain a general characterization of the relationship between universal scientific laws and the
abstract empirical basis. As Blaug and Hutchison would say when the
same problem is presented in the economic (as opposed to epistemic)
context: pshaw. It is not the case that no serious science got, or gets, done
until the abstract questions of formalist philosophy of science get
answered. The only way that such purely philosophical questions have
ever been answered is within the context of a very abstract model of the
relationship between statements in the theoretical language and those
in the observation language involving formal correspondence rules, and
these formal answers (as Thomas Kuhn and an array of others have
shown) have absolutely nothing to do with the far messier activities of
real successful science. Such positivist-inspired models of epistemic efficiency bear exactly the same relationship to actual decision-making in real
science as ArrowDebreu proofs of existence of competitive equilibrium
prices bear to actual decision-making in real market economies: that is
none. We dont need no stinking formalist license, in either economic or
in epistemic policy.
158
D. Wade Hands
A short answer, a long answer, and conclusions
The previous section made the case that Mayers methodological

approach is grounded in social science; he employs just one particular
social science public choice theory but his mode of thinking and
argumentation comes from social science, not philosophy. Blaug and
Hutchison on the other hand, approach questions in economic methodology from a traditional philosophy of science point of view. While one
might criticize the details of Mayers argument, or his selection of public
choice theory, there is a consistency between his approach to methodology and his approach to understanding other aspects of social life.
Relative to Mayer, the approach of Blaug and Hutchison seems almost
schizophrenic. They engage in armchair philosophizing about universals
and absolutes when addressing methodological questions, and yet endorse
a brand of economics that claims to be empirical, case-by-case, and
context-specific. Worse yet, the same kind of formalized abstract intellectual exercises they dislike most in economics (exemplified by Arrow
Debreu general equilibrium theory) are just fine when the subject matter
is philosophy of science (exemplified by logical empiricism and Popperian
falsificationism); for them formalism is a bad way to approach questions of
economic efficiency, but a good way to address questions of epistemic
efficiency.
Even if one accepts my analysis of the differences between these
two approaches, it is still not clear how this bears on the question of
whether Friedmans methodology licensed the formalist revolution. So
did Friedmans economic methodology license the formalist revolution or
not? Is Mayer right, or Blaug and Hutchison? I have a relatively short
answer and then a longer answer that I will attempt to connect up to the
whole issue of the two different approaches I have elaborated.
The short answer is that Mayer is right; there does not seem to be a
convincing argument that Friedmans essay on economic methodology in
any way caused, or even substantially contributed to, the formalist revolution. Neither Debreu (1959), the canonical Arrow and Hahn (1971),
nor any other major work in Walrasian general equilibrium theory even
cite Friedman. More importantly, the leaders of the formalist revolution
were generally not sympathetic to either Friedmans brand of economic
theory or his political economy; as Philip Mirowski (2002) and others have
argued, Walrasian formalism germinated at Cowles by economists who
were generally left-leaning, Keynesians, and less than happy about their
relationship with the Chicago economics department in general and
Friedman in particular. Recall the Friedman quote at the beginning of this
chapter; it came from his 1946 review of Oskar Langes Price Flexibility and
159
Employment. Friedmans attack on Lange, like his ongoing criticism of the

Cowles approach in general, was based primarily on their commitment to
formalism in economic modeling: the theory provides formal models of
imaginary worlds, not generalizations about the real world (Friedman
1946, 618). He clearly disagreed with their support for Keynesian macroeconomic policies, but, as Mayer argued, the conflict between Friedmans
monetarism and Keynesian macroeconomics was more of a disagreement
about the significance of certain empirical relationships than about the
methodological question of whether the behavior of empirical aggregates
should play a role in macroeconomic theorizing. The whole idea that
Friedmans essay, with its emphasis on empirical prediction and practical
application, licensed people to write, and professional journals to accept,
papers whose main results were theorems deduced from the assumption
that an economy is defined by:
for each i = 1, , m a non-empty subset Xi of completely preordered by 5i for
each j = 1, , n a non-empty subset Yj or a point of (Debreu 1959, 75)
does seem to be, on the face of it, a bit ludicrous.

Clearly as Mayer (this volume), Hirsch and De Marchi (1990), and
many others have argued, Friedmans own work, particularly Friedman
and Schwartz (1963), is about as far from formalism as any research in
economics could possibly be. Of course one could always argue that even
though his own work was not formalist, his methodological essay still
might have licensed the formalism of others. But even if one only reads
the assumptions dont matter part of Friedmans essay and skips the
only successful empirical predictions do matter part, there still doesnt
seem to be any reason to believe that Friedmans essay was any more likely
to license formalism than the other popular methodological positions of
the period.
In particular, Samuelsons operationalism seems prima facie to be far
more amenable to the kind of licensing role that Blaug and Hutchison
assign to F53. Samuelsons weak axiom of revealed preference (1938,
1948) ostensibly rendered all of demand theory operationally meaningful, thus qualifying it as legitimate empirical science and licensing it
to be used as the cornerstone for the next four decades of formalist
exercises in excess demand-based Walrasian economics. In order to see
how Samuelsons methodology played such an important role it is necessary to go back to the period 193050 when the current textbook theory
of demand was being stabilized. Although everyone working on demand
theory during this period agreed that the theory should be grounded in the
empirical evidence, there was substantive disagreement about exactly
how/where that empirical evidence should make its presence felt. The
160
D. Wade Hands
most obvious approach, taking ones cues from philosophical interpretations of the natural sciences, would be to empirically test the theory that
appears at the end of the theoretical process. That is, start with a theory of
demand based on some version of individual optimization, aggregate over
the relevant individual agents, deduce various empirical implications of
the theory, and then finally test those implications against accepted empirical evidence. The problem was that this straightforward approach to
the empirical grounding of demand theory proved to be extremely problematic (see Hands and Mirowski 1998; Mirowski and Hands 1998).
An alternative approach and one that proved to be less problematic
involved putting the empirical in at the very front of the process. By
using Samuelsons weak axiom of revealed preference to empirically
ground the individual preferences that constituted the primitives of the
theory, the theorizing could go on happily without ever needing to test
the results at the end of the theoretical process. Once the empiricism
was in at the beginning, one was entirely free to deduce whatever
formal results one desired, never again confront these deduced results
with the evidence, and yet still remain on perfectly firm epistemic ground.
Samuelsons operationalism thus conveniently provided just the empirical
licensing that the formalist revolution needed. Friedmans methodology
does not work so neatly; Friedman would keep asking those pesky questions
about testing the theory at the end of the process. Is it any wonder that
Arrow and Hahn (1971) and other key formalist texts did repeatedly cite
Samuelsons theory of revealed preference and never cited Friedmans
essay. It should also be noted that the weak axiom played a key role in
many of the most important theorems about the uniqueness and stability of
Walrasian competitive equilibrium. Taken as an ensemble, Samuelsons
methodology and his revealed preference theory neatly solved (or at least
seemed to solve at the time) a myriad of problems associated with Walrasian
formalism. Clearly if any methodology provided a license for formalism, it
was Samuelsons, not Friedmans.
I would also note that similar remarks can be made about other methodological positions in the literature at the time. While it is hard to make the
case that a strict falsificationist reading of Karl Poppers philosophy of
science would license formalist theorizing in economics (falsificationism
has its own problems, but turning economics into existence theorems is not
one of them), it is important to recognize that the most self-consciously
Popperian authors of the day were Klappholz and Aggasi (1959) and their
critical rationalist and non-falsificationist brand of Popperian methodology
would allow for various types of non-empirical, including perhaps purely
mathematical, criticism. I believe the case for Samuelsons operationalist
methodology is much stronger, but the point is simply that there were any
161
number of methodological positions that could have licensed the formalist

revolution a whole lot more effectively than F53.
Well, so much for the short answer. What does this short answer have to
do with my long story about the differences between the methodological
approaches of Mayer on the one hand and Blaug and Hutchison on the
other? Perhaps the key point in the long answer can be stated rather
briefly. I have tried to make the case in Hands (2001) and elsewhere that
philosophical ideas about the character of scientific knowledge in economics (or other sciences) is not independent of how one thinks about
social activity, including economic activity, more generally. Sometimes
this relationship is recognized and explicit, but more often it is not. In fact
there are good reasons to try to separate, and thereby purify, the two. It
may well be that how a particular culture X characterizes their deity super-X
reflects, among other things, the social relations within culture X; and yet
drawing attention to this fact would hardly be a good career strategy for a
prospective spiritual leader. The word is most powerful and persuasive when
it comes from elsewhere (preferably above), and keeping them (the gods)
and us (social life) separate is often key to the whole normative endeavor.
The point is that this is as true for the epistemically normative (how one
ought to do science) as for the ethically normative (how one ought to act
toward others). The approach of Blaug and Hutchison, and for that matter
most of traditional philosophy-of-science-inspired economic methodology,
has been to use philosophy as the them from which the rules come. This is
what I have called the shelf of scientific philosophy view of economic
methodology. Now (and finally getting to the point about Friedmans
essay) this type of methodology is much more likely to find undesirable
behavior to be grounded in bad methodological teachings. Since good
economics follows the rules set down by philosophers of scientific knowledge, theoretical behavior that seems to be bad must be because practitioners were given the wrong rules. On this scenario, methodology really
matters and bad practice must be because of bad methodology. Hence if
economics took what seemed to be a wrong turn, then some methodology
must have licensed it. F53 is the most famous paper in economic methodology, therefore it must be responsible for the wrong turn.
Mayers approach is more willing to admit that how one views economic and social life can have an impact on how one views scientific
knowledge. If one comes from a discipline where competitive markets
produce efficiency, and one observes inefficiency, then it seems natural to
start looking for the market imperfections that caused it. If one can explain
how certain social beliefs (such as formalism is the way to do economic
research) emerged as an equilibrium solution from the self-interested
behavior of rational agents under reasonable external constraints, then
162
D. Wade Hands
one has an explanation for the belief. This answer will undoubtedly be less
persuasive among those outside the community of economists, those who
explain the emergence of socially held beliefs in other ways, but down here
on the ground in day-to-day professional life that is not really a problem.
Such a vision, or any vision grounded in social science, will be far less
likely to blame the licenser. Methodological rules matter if they are
grounded in the disciplinary form of life, but such rules are only one
small piece of any explanation of why the particular social result occurred.
For Mayer, Friedmans methodological ideas were grounded in his activities as an economist and those that engaged in formalist economics did so
for reasons, but those reasons had much more to do with the social
conditions and constraints the relevant economic theorists faced than
with any of the particular words in Friedmans methodological essay.
Thus Mayer is less likely to view F53 (or for that matter any other essay
on methodology) as licensing formalism than Blaug and Hutchisons
more philosophy-driven view.
I would like to close by admitting that I have probably overstated
my case: both the case that Blaug and Hutchison are examples of
methodology-as-viewed-from-philosophy, and also that Mayer has a
naturalistic and context-specific approach to methodological questions.
Blaug and Hutchison do discuss various sociological, and even economic,
forces that contributed to the professions change in theoretical style; and
Mayer does defend Friedmans methodology philosophically as a properly
scientific way of doing economics. But while there are certain ways in
which both views leak a bit in the opposite direction, I contend that the
main point remains; even with their occasional nod to social forces, Blaug
and Hutchison are doing methodology in the spirit of the philosophy of
science; and Mayer, while retaining the philosophical high ground for
Friedmans methodology, is conducting his methodological inquiry primarily as applied social science. And I claim these differences provide the
best explanation for their difference of opinion about whether Friedmans
methodology did or did not license the formalist revolution.
References
Arrow, Kenneth J. and Gerard Debreu (1954). Existence of an equilibrium for a
competitive economy. Econometrica, 22, 26590
Arrow, Kenneth J. and Frank H. Hahn (1971). General Competitive Analysis. San
Francisco: Holden-Day
Backhouse, Roger (1998). If mathematics is informal, then perhaps we should accept
that economics must be informal too. The Economic Journal, 108, 184858
163
Blaug, Mark (1980). The Methodology of Economics: Or How Economists Explain.

(1992). The Methodology of Economics: Or How Economists Explain, 2nd edition.
(1994). Why I am not a constructivist: confessions of an unrepentant popperian.
In R. E. Backhouse (ed.), New Directions in Economic Methodology. London:
Routledge, 10936
(1997). Ugly currents in modern economics. Policy Options, 38
(1998). Disturbing currents in modern economics. Challenge, 41, 1134
(2002). Is there really progress in economics? In S. Boehm, C. Gehrke,
H. D. Kurz, and R. Sturn (eds.), Is There Progress in Economics?. Cheltenham:
Edward Elgar, 2141
(2003). The formalist revolution of the 1950s. Journal of the History of Economic
Thought, 25(2), 14556
Literature, 17, 50322 [reprinted with minor revisions as chapter 2 of Boland
1997]
(1997). Critical Economic Methodology: A Personal Odyssey. London: Routledge
Caldwell, Bruce J. (1982), Beyond Positivism: Economic Methodology in the
Twentieth Century, 1st edition. London: George Allen and Unwin
(1994). Beyond Positivism: Economic Methodology in the Twentieth Century, 2nd
edition. London: Routledge
Debreu, Gerard (1959). Theory of Value. New Haven: Yale University Press
Friedman, Milton (1946). Lange on price flexibility and employment: a methodological criticism. American Economic Review, 36, 61331
(1953). The methodology of positive economics. In Essays in Positive Economics.
Chicago: University of Chicago Press, 343
Friedman, Milton and Anna Schwartz (1963). A Monetary History of the United
States 18671960. Princeton: Princeton University Press
Golland, Louise Ahrndt (1996). Formalism in economics. Journal of the History of
Economic Thought, 18, 112
Hands, D. Wade (2001). Reflection Without Rules: Economic Methodology and
Contemporary Science Theory. Cambridge: Cambridge University Press
Hands, D. Wade and Philip Mirowski (1998). Harold Hotelling and the neoclassical
dream. In R. Backhouse, D. Hausman, U. Mki, and A. Salanti (eds.), Economics
and Methodology: Crossing Boundaries. London: Macmillan, 32297
Hausman, Daniel M. (1992). The Inexact and Separate Science of Economics.
Theory and Practice. Ann Arbor, MI: University of Michigan Press
Hoover, Kevin (2009). Friedmans methodological stance: causal realism (this
volume)
Hutchison, Terence (1938). The Significance and Basic Postulates of Economic
Theory. London: Macmillan
(1992). Changing Aims in Economics. Oxford: Blackwell
(2000). On the Methodology of Economics and the Formalist Revolution.
Cheltenham: Edward Elgar
164
D. Wade Hands
Klappholz, Kurt and Joseph Agassi (1959). Methodological prescriptions in economics. Economica, 26, 6074
Lange, Oskar (1944). Price Flexibility and Employment. Bloomington, IN: Principia
Press
Lawson, Tony (2003). Reorienting Economics. London: Routledge
Mki, Uskali (1989). On the problem of realism in economics. Ricerche
Economiche, 43, 17697
Methodology, 10, 171195
(2000) Kinds of assumptions and their truth: shaking an untwisted F-twist.
Kyklos, 53, 30322
(2009). Unrealistic assumptions and unnecessary confusions: rereading and
rewriting F53 as a realist statement. This volume
Mayer, Thomas (1993). Truth versus Precision in Economics. Aldershot: Edward Elgar
(1995). Doing Economic Research: Essays on the Applied Methodology of Economics.
Aldershot: Edward Elgar
(2009). The influence of Friedmans methodological essay. This volume
McCloskey, D. N. (1994). Knowledge and Persuasion in Economics. Cambridge:
Mirowski, Philip (1986). Mathematical formalism and economic explanation. In
P. Mirowski (ed.), The Reconstruction of Economics. Boston: Kluwer-Nijhoff,
179240
(2002). Machine Dreams. Economics Becomes a Cyborg Science. Cambridge:
Mirowski, Philip and D. Wade Hands (1998). A paradox of budgets: the postwar
stabilization of American neoclassical demand theory. In M. S. Morgan and
M. Rutherford (eds.), From Interwar Pluralism to Postwar Neoclassicism [supplement to HOPE vol. 30]. Durham, NC: Duke University Press, 26992
Punzo, Lionello F. (1991). The school of mathematical formalism and the
Viennese circle of mathematical economists. Journal of the History of
Samuelson, Paul A. (1938). A note on the pure theory of consumers behaviour.
Economica, 5, 6171
(1948). Consumption theory in terms of revealed preference. Economica, 15,
24353
(1947). Foundations of Economic Analysis. Cambridge, MA: Harvard University
Press
Ward, Benjamin (1972). Whats Wrong with Economics? New York: Basic Books
Weintraub, E. Roy (1998). Controversy: Axiomatisches Miverstndnis. The
Economic Journal, 108, 183747
(2002). How Economics Became a Mathematical Science. Durham, NC: Duke
University Press
Weintraub, E. Roy and Philip Mirowski (1994). The pure and the applied:
Bourbakism comes to mathematical economics. Science in Context, 7,
24572
Appraisal of evidence in economic

methodology
Melvin W. Reder
Though not the only topic of Milton Friedmans Methodology of positive

economics (F53) the subject of this chapter is a very important theme of
that essay. To understand the influence of that essay it is necessary to view it
as a section of a major thread in the history of economic thought. The
thread is more easily identified than characterized precisely: starting with
Adam Smiths The Wealth of Nations and after making juncture with
Utilitarianism, the thread emerged in the late nineteenth century as neoclassical economics and after much further elaboration has become (in the
late twentieth century) strongly associated with new classical economics.
The thread identifies with what Schumpeter (1954) called a preanalytic
vision or, alternatively, a particular frame for perceiving the functioning of
an economy. Let us call this frame The Invisible Hand.
Two dominant characteristics of the frame are an emphasis upon (i) the
advantages of specialization and division of labor and (ii) the role of voluntary exchange in facilitating their achievement. Following the thread and
the associated frame is one way of organizing a history of economic thought
and it is in this context that I interpret F53. In short, I consider F53 to be an
important part of a long intellectual tradition that has undergone intermittent and continuing change without losing its essential character.
Notoriously F53 has been subject to differing interpretations. The one
offered here has been partly anticipated in the literature, though it may
contain some new wrinkles. However, I am not sure on the latter point
and make no claims to originality. As I interpret it, F53 constitutes an
(important) link between the earlier neoclassical literature, notably
Marshalls Principles, and the post-1960 new classical writings that feature
an insistence on explicit microfoundations as a condition of acceptability for any piece of economic theory. As will be seen, what constitutes a
Although this chapter has benefited greatly from the comments of David Colander, Kevin
Hoover, Uskali Maki, and Thomas Mayer, the author retains sole responsibility for whatever
errors of substance or infelicities of expression remain.
165
166
Melvin W. Reder
proper microfoundation of a theory is intimately related to what would

constitute acceptable evidence in appraising its validity.
Section I locates F53 in the thread of economic thought associated with
the Invisible Hand with special attention to the criteria for appraising the
validity of an economic theory. Section II considers the relation of the
criteria discussed in section I to some other criteria that have been proposed. Section III instantiates the criteria discussed in the preceding
sections by reference to a recent attempt to obtain empirical information
bearing (adversely) on the neoclassical theory of wage behavior. The final
section summarizes the preceding argument and discusses the reasons for
the survival of the Invisible Hand frame and its relation to the positivenormative distinction upon which F53 places heavy emphasis.
I
Persisting methodological dispute(s) between neoclassical economists
and their new classical successors on one side and various opponents
(Keynesians, Institutionalists, devotees of Behavioral Finance, etc.) on
the other have been a facet of a disagreement about how economic problems should be framed.1 For the first group (for brevity, neoclassicals) an
economic problem is framed as a situation in which each of a set of
resource-constrained but interacting individuals attempts to maximize his
or her utility. The level of utility attainable by any individual is limited both
by his or her resource endowment and a (publicly known) technology that
specifies the set of attainable outputs that the resource endowment can
generate. For most problems output is taken to include exchanges at
market-determined prices which are exogenous to any given individual:
these prices function as additional resource constraints to each individual.
A major topic of the history of economic thought from 1776 to about
1920 was refinement of this frame to make it consistent with the theory of
market price. A major application of this (latter) theory was the competitive industry, the output of which was treated as a homogeneous product
produced by a large number of independent firms, each firm identified
with an individual owner assumed to be a constrained maximizer of (his or
her own) utility of which profit was taken as an indicator. The firm was
assumed to be a price taker with the prices of the output and of each input
generated by (competitive) markets. The output quantity of the industry
and the utilization quantity of each of its inputs were assumed to be equal
to the sum of the analogous quantities of its constituent firms.
1
The concept of a frame is used in the sense of Tversky and Kahneman (1974) from whom
the concept has been borrowed.
167
While this is a roughly accurate, albeit oversimplified, description of the

theory of the competitive industry as of (say) 1920, no competent economist believed that any real-world industry satisfied the assumptions
specified. However, varying from one writer to another, they would
argue especially in applications as though they believed the departures
of reality from this theory were not sufficiently important to warrant
abandonment of the Invisible Hand frame. While occasionally the incongruity of the theory of the competitive industry with the characteristics of
real-world industries was recognized in economic literature, prior to 1920
the difficulties were papered over rather than confronted. The main
attempt at reconciling the theory of the competitive industry with some
theory of the firm was made in Marshalls Principles where the conceptual
Representative Firm was introduced.2
The difficulties associated with the notion and use of the Representative
Firm served as a major point of departure for the efforts of the 1920s to
rectify the seeming incompatibility of the price-taking, profit-maximizing
firm with the theory of the competitive industry. These efforts culminated
in the emergence of an explicit neoclassical concept of a competitive firm
that was a price-taking, profit maximizing entity. But the development of
this theory clearly revealed the descriptive inadequacy of such a firm as a
representative of the constituents of observable competitive industries. As
a result, the neoclassical concept of a competitive firm was widely rejected
as unrealistic (i.e. descriptively inaccurate) almost at first sight, with the
result that in the early 1930s alternative concepts (e.g. Monopolistic
Competition and Imperfect Competition) were developed in which the
firm was portrayed as a price-setting, quantity- and quality-choosing entity
although (usually) still considered to be focused on profit making. Such
concepts allowed for the possibility that firms might exploit tradeoffs
among posted selling prices, quantities sold, product characteristics
(qualities), terms of payment and delivery, etc., to enhance profits, with
the neoclassical competitive firm nesting within any of them as a special
case where all such tradeoffs were limited to zero. Because they permitted
consideration of such tradeoffs, which were obviously of very frequent
occurrence, it was claimed that such concepts were more realistic than
the neoclassical competitive firm.
The economic history and intellectual spirit of the times (1930mid
1950s) allowed the claim to greater realism of non-competitive theories to
prevail against the weak arguments offered in support of the neoclassical
2
The term industry refers generally to aggregates of independent decision entities whose
productive activity reflects the impact of common forces (e.g. geographic areas, use of
common inputs) and is not restricted to producers of a common product.
168
Melvin W. Reder
competitive theory. For a time roughly two decades the failure of the
competitive theory to allow for the possibility that observed behavior
might reflect exploitation of the aforementioned tradeoffs was sufficient
to make it generally unacceptable. It was as an attempt to buttress this
rejection that testimony of economic actors (e.g. business managers) was
introduced as support for the claim of unrealism of the assumptions of the
competitive theory. In large part, F53s attack on the relevance of such
testimony to appraisal of the competitive theory was in the nature of a
rebuttal.
Briefly to review F53s attack, consider an economic analyst attempting
to choose between two explanations of price setting and output determination among a group of firms assumed to be competitors in the sale of
output and the purchase, hire or rent of inputs. Let one of these explanations be the competitive theory and the alternative any member of a class
of theories whose common characteristic is that they permit tradeoffs
among prices, quantities, qualities, and contract terms. The competitive
theory is similar to, though not identical with, the theory of an idealized
competitive industry as presented in contemporary textbooks.
In the idealized competitive industry inputs and outputs are homogeneous: this is consilient with the notion that as of any given time all units
are bought or sold at a single price with the identity of the transacting
parties a matter of indifference to all concerned. Moreover, the characteristics of the constituent firms are irrelevant to the analysis of their
pricequantity behavior so that the firm is a convenience of speech; a
redundant notion without independent function. However, for realworld application it is necessary to recognize that both firms and competitive industries have differentiating characteristics that somehow must
be taken into account. Organized exchanges aside, the output of a realworld competitive industry is a congeries of differing products, sold on a
variety of contract terms, delivered in various ways, etc. An analogous
remark applies to each of its constituent firms.
To apply the conceptual structure of an idealized competitive industry
to measurement of such a collection of heterogeneous products and
contract terms requires decisions of classification and weighting as discussed below. Deferring discussion of details, the result of such statistical
editing is the production of price and quantity numbers referring to an
entity called an industry that are designed to reflect the impact of the
forces usually operating through costs that bear similarly upon (most
of) the constituent firms.
The argument of F53 was designed to focus attention upon the operation of these common forces: i.e. the intent was to apply the competitive
model of the idealized industry to the analysis of (real-world) relations
169
among stylized prices and quantities of stylized industries, ignoring

(abstracting from) variations in these relations among constituent firms.
The focus of theories of imperfect and/or monopolistic competition was
the opposite: i.e. these theories highlighted the options of individual firms
in the setting of sale prices, product characteristics, and terms of transaction. Implicitly, Friedman argued that when the empirical counterparts
of the price and quantity variables of the idealized industry were properly
defined and measured, usually the relations among them would be found
similar to those alleged by earlier neoclassical economists and robust to
whatever additional information could be gained from detailed study of
interfirm differences in the relation of price to marginal cost resulting
from differences in selection of product characteristics, contract terms,
etc. From this it followed that a sound research strategy would abstract
from such differences (without denying their existence) and concentrate
upon the relations among industry variables. This contention was denied
by devotees of Imperfect/Monopolistic Competition.
Industry variables and reports of firms

As already indicated, the facts of industry prices and quantities are
obtained from reports furnished in one way or another from constituent firms and aggregated to obtain numbers referring to the industry.
Consequently, whatever information about prices and quantities might be
obtained from managers of individual firms would also be contained in the
reports from which the industry facts were derived. Thus, when consistent
with what is reported about industries, statements of managers about
prices and quantities made in interviews with researchers is redundant.
When not consistent such statements are false.
Let me amplify: published reports of industry prices and quantities refer
to the behavior of an industry, but they are constructed from reports of
individual firms. Of necessity the reports of firms to a survey involve
various sorts of editing, some by the reporting firm and very often by the
compiler of the aggregate as well. Editing by the compiler is required by
the need to combine interfirm differences in qualities of outputs and
inputs, technology, business practices reflected in the manner of price
quotation, etc. Moreover, because reporting firms must fit the particulars
of their activities into categories prescribed by the compiler of industry
data, the reporting process inevitably does violence to the concepts for
which the categories were designed. Still further, often the compiler must
adjust the reports so as to avoid letting weight shifts among firms
generate results that do not reflect industry-wide movements in variables
170
Melvin W. Reder
of theoretical interest. Added to these editing effects there are reporting

errors, simple mistakes, and, occasionally, deliberate falsifications.
I can think of no basis for a general statement of the comparative
reliability of a firms report that is encapsulated in a number purporting
to reflect the state of an industry-wide variable, and what its manager
would report, one-on-one, to a researcher. However, depending on its
format, an individually tailored inquiry (e.g. an interview) might elicit
information about the details of a firms business practices, especially
about its rules of thumb and overall strategy, that generally could not be
obtained from a standard-form questionnaire administered to a number
of firms.
Obviously, the quality of the information that can be elicited through an
interview is heavily dependent upon the knowledge and understanding of
the individual(s) that speak for the firm in communicating with the investigator. At one extreme the spokesperson may not grasp the purpose of the
interrogation and furnish virtually no information, while at the other he or
she may become an active participant in the study, formulating original
questions and criticizing proposed interpretations of the responses.3 But
while the information obtainable from in-depth interviews and/or questionnaires designed for specific firms may serve as the basis for case
studies of individual firms, they will not help in analyzing the behavior
of industries.
This is because the analysis of the pricequantity relations of a competitive industry focuses upon changes in conditions of demand or supply
that impact similarly upon all firms in the industry, while the focus of a
case study is upon changes that alter the position of a studied firm (e.g. its
market share) relative to that of competing firms.
In providing information to an investigator, typically the spokesperson
of a firm highlights material that bears upon the firms relative position,
leaving in the background determinants of the firms response to developments common to all firms in the industry. While an increment to information about a particular firm may be relevant to an economic history of
that firm, or of value to a prospective investor, it is not relevant to an
appraisal of the application of the theory of a competitive industry to the
firms (real-world) industry.
In short, if a report emerging from an interview with a firms manager is
consistent with the theory of a competitive industry it may be accepted as
credible albeit redundant; but if not consistent it should be judged either
as false or, if not false, as reflective of behavior idiosyncratic to the firm
3
A classic example of collaboration between an analyst and a manager is furnished by

Herbert Simon and Chester I. Barnard. See Simon (1947).
171
and/or to a particular time period (i.e. as a disturbance). This is not the

place to defend or attack such a procedural rule, but to identify it and
describe its methodological function, which is to separate propositions
that are compatible with the Invisible Hand frame from those that are not.
Accordingly, the rule advises one to treat reports of price-setting or wagesetting behavior that are indicative of failure to maximize profits, or of
alteration in the relation of price to marginal cost (or of the price of an
input to the value of its marginal product), as meriting zero weight in
appraising the validity of propositions of price theory. This is not to say
that Friedman denied that such departures from pure competition might
occur, but that F53 recommended that such statements should be ignored
(abstracted from) in formulating and testing theories about market behavior, except where there are genuine monopolies.
The underlying rationale for recommending such a procedure is that it
provides a better guide to the formulation of successful theories about
economic phenomena as judged by the criteria advocated in F53 than
some alternative procedural rule that allowed a significant role for variations in the relation of price to marginal cost among firms. The procedure
is recommended with a caveat against applying it where a monopoly is
present. Without going into details, let me assert that acceptance of the
Invisible Hand frame is associated with belief that in a given situation it is
(i) almost always possible for properly trained economists to make commonsense judgments on which model (competition or monopoly) is to be
applied and (ii) that such judgments will cluster sufficiently well as to
suggest the existence of a consensus.
F53 and the theory of the firm
In F53, idiosyncratic behavior of a firm that cannot be reconciled with
price-taking/profit-maximization was considered to be the reflection of a
transitory stochastic disturbance to any pricequantity relationship in
which the record of such behavior might be incorporated. Thus, inter
alia, F53 was an attack on the then current fashion of hedging the application of propositions about pricequantity relations among industry
variables with qualifications stemming from the effects of idiosyncratic
variations in the relation of price to marginal revenue among its constituent firms. The intention was to reduce concern of economic theorists
with the details of individual firm behavior, in effect expelling its study to
Business History and/or other disciplines. One consequence of this was to
diminish the value of information about the idiosyncratic characteristics of
particular firms and its main source, individualized reports from their
managers.
172
Melvin W. Reder
However, this effort has been only partially successful. The success is
reflected in the continuing search for relations among aggregate variables,
with observations on individual entities (e.g. firms) serving as data points.
Often empirical workers engaged in such search are disdainful of detailed
accounts of the behavior and intentions of particular individuals.
Nevertheless the theory of the firm, an imperiled embryo in 1953, has
since become a vigorous and independent branch of economics.
Although always present among reflective businessmen, interest in
explaining the behavior of firms had long been repressed among economic
theorists by the unchallenged assumption that competitors must be pricetakers. While the Monopolistic Competition Revolution freed them of
their inhibitions on this point, it offered little guidance beyond the insight
that the implications of pure competition were unrealistic and entailed
drastic oversimplifications of the problems faced by the management of a
competitive firm.
While theoretical analysis of such problems begins with Coases (1937)
seminal paper, The nature of the firm, the subject did not take hold
among economists until the 1970s with the development of theories about
transaction costs, incentive payments, etc. As these ideas were symbiotic
with new developments in game theory they had the effect of marginalizing the theory of monopolistic competition among students of firm
behavior. Another, and more important effect, was greatly to increase the
interest of (many) economists in obtaining detailed information about
individual firms.
The best often the only possible source of such information is the
manager(s) of these firms. The most important and difficult questions
concern not so much the details of behavior as the strategies, intentions
and objectives that motivate the behavior. Managers can and do discourse
at length on these subjective matters, sometimes persuasively.
Notoriously, the problem of accepting their statements at face value is
that they reflect the preconceptions of the speakers and the effects that the
speakers desire their statements to have upon auditors. Nevertheless,
because of the paucity of alternative sources, such statements (often
embellished with self-serving accounts of pertinent events) get factored
into the histories of firms.
To the sceptical suggestion ignore the talk, just look at the record of
behavior (aka the facts), the answer lies in the need for consilience as
discussed in section II. The importance of consilience, especially between
accounts of the behavior of aggregates and the alleged intentions of
component individuals, is rooted in the immanent objective of all economics to relate the intentions of individual actors to their (often unintended) consequences.
173
II
Although not the only criterion of theoretical merit that it recognizes, F53
places great stress upon capability of making correct predictions (including retrodictions). This view is consilient with its argument that the
validity of an economic theory depends upon the conformity of its implications with the observed behavior of prices and quantities to which the
realism (descriptive accuracy) of its assumptions is irrelevant. But while it
is generally accepted that predictive capability is an important criterion for
judging the merit of any theory, there has been considerable dissent as to
its importance relative to other criteria. As the reader may surmise,
opinions about how the evidence offered in support of a theory should
be appraised is closely related to the criteria by which the merit of the
theory is to be judged.
While I shall not attempt a general discussion of the criteria to use in
appraising theories, our topic requires that we consider some criteria other
than predictive capability. Though logical consistency is an obvious candidate, I have little to say about it beyond pointing to its obvious importance.
The criterion that I shall stress is, I think, underemphasized in F53: consilience of a theory with other beliefs, especially those associated with other
theories, that have wide acceptance. (Other beliefs include the oftmentioned beliefs that arise from intuition and/or introspection.) In
Consilience, E. O. Wilson (1998) argues strongly for contribution to the
unity of knowledge as an evaluative criterion for theories in any subject.
Wilson is especially critical of the failure of economics, especially the
theory of individual choice, to establish proper connections with theories
in psychology covering the same subject. However, after some strong
though not altogether accurate criticism of its procedures, Wilson concedes that We are better off if the economists speak than if they remain
silent (1998, 197). While it is obvious that the implied subject of such
allegedly beneficial speech is public policy, Wilson does not offer any
explicit rationale for this concession.
I submit that the implicit rationale for the concession is belief that
general knowledge of relevant facts about the behavior of economies and
of the ideas present and past about their functioning, together with
their visible efforts to advance this knowledge, makes economists a source
of valuable information. However, Wilsons concession is weak: i.e. economists should be invited to speak but in competition with other speakers,
and not necessarily to be believed.
An important reason for the weakness of the concession, though surely
not the only one, is lack of consilience of what economists say about
choice behavior with the dicta of psychology and, indirectly, with biology
174
Melvin W. Reder
and so forth. This lack of consilience is a major reason, though perhaps

not the only one, for Friedmans desire to make prediction about the
behavior of aggregate variables prices and quantities the primary
criterion of theoretical validity, despite the non-realism nonrealisticness pace Uskali Mki of the associated theory of individual
choice. All that Friedman, and neoclassicals generally, require of a theory
of individual choice is that it be logically compatible with the implications
of theoretical propositions about the behavior of prices and related (aggregate) quantities.
Consequently, the testimony of individual economic actors about their
preferences and motivations is to be given zero weight in appraising
propositions of price theory: i.e. considered redundant where it is compatible with the theory, and irrelevant where it is not. But if consilience
with psychology is also to be considered a criterion of theoretical merit,
such summary rejection will not do. In 1953 experimental economics had
hardly commenced, so it is not surprising that its embryonic activities
escaped Friedmans attention. But a half century later Nobel awards and a
growing stream of books and articles, both by economists and psychologists, have made it impossible to avoid considering experimental evidence
bearing upon the choice behavior of individuals. By no means has all of
this evidence been adverse to the viewpoint of F53, but because some of it
has been or could be interpreted so as to make it seem adverse it would
be well to remark on the problems facing economists and psychologists
concerned with the apparent lack of consilience with this burgeoning field.
As this is beyond the purview of the present chapter, I shall limit
discussion to one salient example: the effect on choice behavior of framing
a given prospect as a gain or a loss. As usually interpreted, the conventional definition of rationality implies that an individuals choice among a
set of alternative sources of utility will depend upon (inter alia) the
contents of the various sources and not upon the manner in which the
sources are presented. While it is often recognized that incompleteness of
information may lead individuals to make wrong choices (i.e. choices
incompatible with the assumption of rationality), the general neoclassical
view is that such errors will be transient and/or will cancel out so that their
net impact upon the observed relations of pricequantity aggregates will
be negligible.
However, the experiments of (psychologists) Tversky and Kahnemann
(1974) showed that by framing an identical utility prospect as the result of a
gain or a loss, laboratory subjects could be induced systematically to
choose the former in preference to the latter. While it is not to the present
purpose to appraise these findings, it may be noted that they have received
wide acceptance among psychologists and experimental economists and
175
were cited in the Nobel Prize awarded to Kahneman (2003). Therefore

for the sake of the argument I shall assume the findings to be valid, and
focus upon the problem(s) of integrating them with neoclassical theory. A
convenient point of departure for discussing the framing effect of shocks
to the determinants of individual choice and the effect of such shocks (or
lack of it) upon the relation between prices and aggregate quantities is the
cluster of issues associated with money illusion.
In Keyness General Theory of Employment (1936) it was asserted that
individual workers would accept reductions in real wages provided that
their money wages were not reduced (framed as a non-loss), but would
resist reductions in money wages even if they were offset by decreases in
consumer prices (framed as a loss): i.e. aggregate labor supply was alleged
to be influenced by money illusion. Whether true or false, such an allegation of asymmetry in the reaction of labor supply to inflation and deflation
is incompatible with the tenets of neoclassical price theory. Its occurrence
forces economists to choose between accepting as valid the reported outcome of experiments about the effects of framing on individual behavior
on the one hand and maintaining adherence to the neoclassical theory of
labor supply on the other.
Let us consider the making of such a choice as an exercise in Bayesian
decision-making. Allow the neoclassical theory of labor supply to be the
hypothesis and experimental results about the effect of framing on choices
among utility source bundles to be the sample. Whether the hypothesis
should be altered by the results of the experiment will depend upon the
characteristics of the sample, and upon the confidence reposed in the
validity of the hypothesis. For the application at hand assume the sample
characteristics to be such as to warrant rejection of the hypothesis, or at
least to require amendment of it. However, assume that confidence in the
validity of the hypothesis together with the difficulty of finding an acceptable alternative (including amended versions as alternatives) make rejection unacceptable.
To a non-human seeker of truth, this dilemma is not critical: it
would simply accept its inability to reach a decision as transitory and
continue research. However, most economists, along with many noneconomists, would join Wilson in saying that it is better that economists
speak out (decide on the hypothesis) in an agenda-determined time frame
rather than remain silent. But to speak out to accept the hypothesis
despite its lack of consilience with the findings of other sciences requires
that economists must reject the relevance and/or the validity of (some of)
what the other sciences say and men of affairs believe about pricesetting and resource allocation. Though not stated explicitly, F53 contains a powerful exhortation to minimize concern about (lack of)
176
Melvin W. Reder
consilience of the propositions of neoclassical economic theory with the

opinions of outsiders.
Without pursuing the matter I would suggest that, for any science,
appraisal of theories should involve balancing the desirability of consilience with the results obtained in neighboring fields against the urgency
of applying the findings in its own. But while this suggestion expresses
little more than common sense, it is often forgotten in discussions of
methodology.
III
In the multifaceted controversies surrounding F53, it is easy to identify
and label the doctrines of one major party: neoclassical economics. But
while the opposing ideas are more diffuse, they are unified on one point:
the frame of the economy associated with neoclassical economics the
Invisible Hand is misleading. The various opponents of the neoclassical
frame include partisans from other disciplines and, among economists,
Keynesians, institutionalists, etc., with each category covering a considerable diversity of opinion.
However, one point on which all Keynesians concur is that an economy
may vary in the degree to which it utilizes available resources, and that the
unutilized resources respresent waste. Basically, neoclassicals deny this
and have offered a variety of subtle arguments to explain away the
appearance of waste due to unemployment that is alleged to be involuntary. For better than half a century the validity of such arguments has been
strongly contested, with a critical issue being the involuntariness of the
undisputed fact of non-employment. As will be seen, the issue turns upon
whether, and (if so) how, involuntariness can be identified and the associated quantity measured.
Is the notion of an involuntary decision about anything an oxymoron, or can it be interpreted so as to make sense? If the latter, can
testimony of decision-makers be considered as a source of evidence and, if
so, how should it be weighed against (potentially) conflicting evidence
inferred from observable behavior? As some potential answers to such
questions are obviously incompatible with adherence to the Invisible
Hand, adherents of that frame have good reason to reject them as an
acceptable source of information.
But if unemployment on some basis is judged to be involuntary, a
further question arises: why dont employers offer unemployed workers
lower wages than they are paying their current employees? For the
moment, I shall merely remark that the only possible source from which
answers to such a counterfactual question could come must be statements
177
of individuals who have the capability of setting wages: it is impossible to

draw inferences from behavior in which they did not engage. Bewley
(1999) gives an extensive report on an effort to elicit statements from
employers on this matter. But before discussing the methodological
implications of this report, we need to discuss the troublesome concept
of involuntary unemployment and its correlative, sticky nominal wages.4
Involuntary unemployment and sticky wages

From the publication of The General Theory to the present, the concept of
involuntary unemployment has been a source of confusion rooted in the
dual purpose of the book. Keynes wanted both to displace the neoclassical
vision of the economy which precludes involuntary unemployment, and
to persuade neoclassical economists that his alternative vision was
superior.
Had he been concerned only with the first objective, the difficulty with
involuntariness could have been avoided easily. Given technology, the
stock of capital goods and the real wage rate, the level of aggregate output
could have been assumed plausibly to depend solely upon the level of
effective demand. Then, given a fixed number of job seekers, finding the
quantity of unemployment would have been a matter of arithmetic.
The analogy with a game of musical chairs is striking: assuming there
are more players than chairs when the music stops, some participants will
have no seats, with the number unseated determined in advance as the
difference between the number of seats and the number of players.
Clearly, being without a seat would be involuntary. While the identity of
the players who were to be unseated would not be explained, neither did
The General Theory purport to identify which workers were to be unemployed, but only their quantity.
While something like this musical chair analogy had popular currency
during the second quarter of the last century, and even now is not extinct,
it is far too crude to appeal to most economists, among whom Keynes was
no exception. Both to convince others, and probably himself, he had to
devise a story that would reconcile an economy being in a state of equilibrium although there were desirous sellers of labor services who were
unable to transact at the going wage. The main story presented in The
General Theory (there were several) ran in terms of workers being resistant
4
Another interesting and painstaking effort to obtain information from economic actors about
thought processes relevant to their price-making behavior is reported in Blinder et al. (1998).
Although the findings in the report are germane to the argument of this chapter, they raise so
many collateral issues that (reluctantly) I have decided not to comment upon it.
178
Melvin W. Reder
to reductions in money wages even though real wages remained

unchanged: i.e. the explanation suggested the presence of money illusion
and was at least first cousin to a TverskyKahneman framing effect. I pass
over the details of the explanation because it never really caught on even
among Keynesians, and Keyness own attachment to it is uncertain.
For roughly a quarter of a century the labor market side of The General
Theory was left in a kind of limbo, with the process of wage determination
obscured by treating both money wages and labor supply as exogenous.
However, this state of affairs was highly unsatisfactory to a new generation
of economists, including adherents of both the Keynesian and the neoclassical traditions, who demanded (as a condition of acceptability) that
economic theories should have sound microfoundations.
There has been a plurality of candidates for the role of microfoundation
provider. One of these is a development and extension of the neoclassical
frame, frequently identified as new classical.5 The distinguishing characteristic of new classicalism is insistence on the fulfillment of all implications of the assumption that (in equilibrium) markets clear with resource
owners optimizing: neoclassicals sometimes fudged this requirement. As I
see it, the principal difference between new classicals and neoclassicals is
rooted in the problem of discussing the behavior of individuals in a
situation of market disequilibrium, while continuing to assume they are
always optimizing. Without going into details, new classicals proposed to
solve this problem by assuming that individuals always maximized
expected utility on the maintained assumption that expectations were
rational. Although the problem of establishing microfoundations arises
in all markets, a major point of reference has been the labor market and the
interpretation of the concept of involuntary unemployment.
What the various approaches to microfoundations other than the new
classical share is rejection of the neoclassical frame. Two papers having
much in common, in addition to salience and merit, are the Nobel
lectures of Akerlof (2002) and Stiglitz (2002) which I shall use as exemplars of the other than new classical approach to involuntary unemployment and labor market behavior. While the new classical approach to
microfoundations of labor markets and involuntary unemployment has
been presented in various ways, for convenience I shall focus upon one of
the earliest and still one of the best Lucas and Rapping (1969). Leaving
aside other aspects of labor supply I shall concentrate upon the effects of
intertemporal planning by individuals and households.
An excellent summary and critique of this development is Hoover (1988).
179
In essence, The General Theory sought to frame fluctuations in employment as resulting from fluctuations in effective demand with above normal non-employment identified as involuntary unemployment: possible
fluctuations in labor supply were ignored. In contrast, Lucas and Rapping
sought to frame fluctuations in employment as the result of fluctuations in
labor supply, sometimes (but not always) caused by fluctuations in aggregate effective demand. While the Keynesian conception of unemployment
is single period with no attention paid to intertemporal relations of labor
supply, Lucas and Rapping consider individuals to be multiperiod utility
maximizers, whose plans allow for the possibility of interperiod substitution of labor supply.
In any given period decisions about labor supply are conditional upon
the state of current employment opportunities (real wage rate).6 Should
current employment opportunities be less favorable than an individual
had anticipated, he or she might find it advantageous to substitute further
education, a vacation, engaging in home repair, etc., for current gainful
employment. Such intertemporal substitution of employment would be
coordinated with intertemporal deviations from an asset accumulation
plan: e.g. drawing down of savings, increased borrowing, reduction of
current consumption. Conversely, regardless of the current state of
employment opportunities, previous periods in which labor market states
had been more favorable than anticipated would lead an individual to
work less and/or reduce his or her rate of asset accumulation in order to
increase his or her rate of consumption. Still further, in multiperson
households, shock-induced deviations from an intertemporal household
plan might induce crossperson as well as intertemporal substitutions of
time. However, because the costs of exercising options to alter intertemporal plans of labor supply may appear excessive, some workers may avoid
them by accepting lower wages and/or inferior jobs.
Implicit in the LucasRapping story is the assumption that shockcaused deviations from intertemporal plans may be anticipated and
planned for. Indeed, the implicit self-insurance that such plans envisage
might conceivably be supplemented by purchased insurance as proposed
by Robert Shiller (2003). But the uncertainty associated with intertemporal plans cannot be insured against. Occurrence of a shock always raises
questions as to whether the regime upon whose continuation such plans
Lucas and Rapping use real wage rate as a synonym for net advantages of employment.
So defined, the real wage rate covers fringes of all kinds, especially the prospect of an
increased probability of future employment as a result of present employment. While this
definition presents problems, a discussion of them would take us far afield.
180
Melvin W. Reder
were based might have undergone an adverse change, and if so what the
new regime might be and what reformulation of such plans it would entail.
The psychic cost of the entailed uncertainty is greatly enhanced by the
fact that because an individuals life cycle is finite and affected by hysteresis it is often infeasible to defer study or adventures of one kind or
another for later realization. To this should be added the high cost of
overcoming the moral hazard associated with lending to an impecunious
loan seeker. Together, these obstacles to intertemporal substitution of
labor supply can provide a rationale for rejecting the new classical viewpoint as a frame for viewing involuntary unemployment.7
But The General Theory nowhere invokes such psychic costs as a rationale for deploring involuntary unemployment. Left implicit, although
clearly present, is the claim that involuntary unemployment is the obverse
side of unused productive capacity and that the potential output of such
capacity (valued at market prices) can serve as a measure of the resulting
loss. For the sake of argument, assume that while this claim might hold for
a short period window (say six months), as the window lengthens to two
years the effect of intertemporal substitution of employment would begin
to affect the level of potential output: i.e. initial subnormal employment
would be (at least partially) offset by later supernormal employment.
Thus, viewed through a sufficiently long window, lost output due to a
major depression might be offset more or less by the effects of a
supernormal supply of labor from workers struggling to make up for the
potential earnings previously lost. Conversely, an unusually large or prolonged boom might lead to earlier retirements, longer vacations, etc. by
workers who had already earned and saved enough to satisfy the aspirations embedded in their multiperiod supply plans.
I conjecture that in viewing the economic history of the past century
most economists would agree that during particular short periods where
aggregate output and employment fell below their respective potentials
the shortfalls were partly (or even wholly) offset by unusually large supplies of productive services, earlier or later. Moreover, they would also
agree that because measuring the relevant magnitudes is very difficult,
summary characterizations should be tentative. However, in choosing a
frame for viewing the normal state of a capitalist economy, many of them
would desert the ranks of the tentative and become more or less committed adherents to either the Invisible Hand frame or the frame of the
Keynesians.
7
The reader will note the relation of the above remarks to Frank Knights (1921) distinction
between risk and uncertainty.
181
For reasons to be given, adherence to the Invisible Hand frame is

congenial to framing fluctuations in employment as the result of fluctuations of labor supply rather than of labor demand. This does not imply that
neoclassicals deny that adverse shocks to aggregate labor demand ever
occur. What the framing preference reflects is belief that the response of
individuals to shocks to labor demand favorable or adverse is to vary
the current quantity of labor supplied with offsetting variations occurring
subsequently. Individuals who are unusually averse to such intertemporal
substitution, or to whom it would be unusually costly, can avoid it by
accepting reductions in their current real wage including acceptance of
less attractive jobs. Such a view of non-employment precludes the possibility of using it as a basis for inferring a loss of potential output over any
substantial period of time.
By contrast, the Keynesian frame enables one to consider employment
levels that are less than some exogenously given labor supply as an
indicator of wasted potential output. That is, the Keynesian framework
provides a rationalization for neglecting the possibility that current underemployment might cause an increase in future output by increasing future
labor supply. Such neglect makes it possible to sum (across periods)
the losses of potential output inferred from allegedly involuntary
unemployment.
The policy implications of preferring one of these frames to the other
are obvious: the Keynesian frame highlights current gains in aggregate
output due to Full(er) Employment while the Invisible Hand implies
that such gains are illusory and reflect intertemporal shifts of resources.
Similarly, the Keynesian frame facilitates pointing to the benefits of policies that accelerate industrialization, while the Invisible Hand highlights
the costs of going faster than market forces would (seem to) warrant.
While not explicitly adopting the Keynesian frame, the view of the
functioning of a capitalist economy and the attitude toward public policy
associated with that frame are reflected clearly in the Nobel lectures of
George Akerlof (2002) and Joseph Stiglitz (2002). In revealing a shared
aversion to the Invisible Hand, both lectures cite trivialization or outright
denial of the occurrence of involuntary unemployment as a source of
complaint. However, they do not cling to Keyness concept of involuntary
unemployment, but introduce a quite different though now widely
accepted definition. As stated by Akerlof:
The existence of good jobs and bad jobs makes the concept of involuntary
unemployment meaningful: unemployed workers are willing to accept, but cannot
obtain, jobs identical to those currently held by workers with identical ability. At
the same time, involuntarily unemployed workers may eschew the lower paying or
lower-skilled jobs that are available A meaningful concept of involuntary
182
Melvin W. Reder
unemployment constitutes an important first step forward in rebuilding the foundations of Keynesian economics. (2002, 415).8
While this statement is congenial to commonsense, it has the effect of

seriously changing the meaning of involuntary unemployment. What is
a good job? I suggest that the term has reference to a consensus view (in
a given society) of what would be appropriate for one of its members of
given characteristics (age, education, job experience, geographical location, etc.) to accept. On the Akerlof definition the amount of involuntary
unemployment depends critically upon a societys consensus definition of
the kinds of job offers that individuals may not refuse on pain of sanctions
being invoked. In judgmental societies a given level of effective aggregate
demand might generate much less involuntary unemployment than in
more tolerant societies. Indeed, if it were required that in order to escape
the stigma of being voluntarily unemployed an individual was required to
accept job offers where compensation was solely by results without a
guarantee of minimum payment, involuntary unemployment would
cease to exist on the Akerlof definition.
As his lecture indicates, Akerlofs concept of involuntary unemployment dovetails with the notion of efficiency wages. This notion presents
the employer as a price (wage) setter and not as the taker of a market- (or
sometimes a union-) determined wage rate. In the context of this notion
the presence of involuntary unemployment raises the question of why
employers dont reduce the wages that they set. Answering this question
is the subject of a monograph by Truman Bewley (1999). (Bewleys wage
concept is the commonsense nominal wage and not the LucasRapping
real wage which would be quite likely to fall in the presence of excess labor
supply.) Bewleys concern is with the nominal wages of employees who
retained their jobs in the presence of excess labor supply without experiencing a wage cut.
Bewley did not study the recorded history of what such workers were
paid (as F53 would recommend), but instead asked employers (or their
representatives in charge of compensation decisions) about why they had
not reduced the wages they did offer, or why they had failed to reduce
them more than they had, contrary to the recommendation of F53. In the
early 1990s Bewley interviewed over 300 wage setters in the New
England region adjacent to Yale University where there had been a severe
recession. Because of the conditions prevailing in the local labor market, a
maintained (though unspoken) assumption in most of the interviews was
8
The idea of a job applicant having identical ability with a current job holder raises
methodological problems about the notion of specific human capital. While such problems
are not always trivial, I shall refrain from discussing them here.
183
that the proposal of a (temporary) wage cut would not have provoked an
immediate quit. Yet such cuts were rarely proposed when circumstances
required a reduction in payroll: the typical response was to lay off
workers, or dismiss them, but only rarely to reduce their rates of pay
and even less frequently to give workers an option of some combination
of shorter hours and reduced rates of pay.
The striking feature of the interviews was the failure of most managers
to invoke labor market considerations as an explanation of their wagesetting behavior. While the reasons given for their behavior included a
variety of considerations (e.g. fairness to the employees, avoidance of
worker resentment that might cause future shirking or quits when labor
market conditions became more favorable to employees, maintenance of a
pleasant atmosphere in the workplace). Only rarely did the expressed
sentiments make reference to conditions in the labor market. As Bewley
points out, the reasons for wage decisions given by managers run strongly
parallel to the views expressed by personnel experts, social psychologists,
sociologists, and other academic non-economists.9
Bewley summarizes the reasons given for not reducing wages under the
rubric of employer desire to maintain morale: i.e. wages dont fall during
recessions because employers dont wish to damage employee morale.
Worker and/or trade union resistance is not invoked as a causal factor in
explaining the failure of employers to exploit excess labor supply to reduce
wages. This implies that in a recession most employed workers cannot
reduce their probability of being laid off or dismissed by accepting a wage
cut because their employers do not wish to lower wages for fear of
adversely affecting morale. For the same reason, employers would not
respond to offers by unemployed workers to accept lower wages than
those paid to current job holders of equal productivity. In short, unemployed workers cannot use the tools of a competitive market accepting
lower wages to increase their probability of finding (or retaining) a job
and therefore their unemployment should be considered involuntary.
This state of affairs would seem to be inconsistent with the idea that
employers seek to maximize expected profits and Bewley so interprets it,
constructing a theory alternative to profit maximization in which
employer wage-setting behavior is made subject to a morale-maintenance
constraint. (While the details of this theory are of substantive interest, they
have no direct bearing on the present discussion so I shall refrain from
9
While Bewley does not stress the point, his interviews reveal a tendency (frequently
remarked elsewhere) toward enhanced employer willingness to exploit excess labor supply
to reduce wages when pressed to improve cash flow (e.g. threatened action by creditors), as
well as lessened resistance of workers to such exploitation.
184
Melvin W. Reder
comment.) Nevertheless, it is possible to interpret employer concern for

maintenance of morale in a manner consistent with pursuit of maximum
expected profits.
To make such an interpretation it is necessary for the concept of specific
human capital to be accorded greater salience than Bewley or most other
authors give it. Specific human capital (pace Gary Becker) reflects skill
and/or knowledge embodied in a worker, but useful only to a specific
employer: hence it can be destroyed by a job separation, to the loss of both
worker and employer. Typically, specific human capital is created by investments (of time and resources) by both parties implying the expectation of
both to a multiperiod return. Consequently, a disgruntled employee might
(rationally) retain a job despite dissatisfation with his or her employers
behavior and respond by withholding some effort (accordingly reducing
productivity), but not enough to provoke the employer to dismiss him or her
and thereby lose that employees share of the joint investment.
Being aware of this possibility a rational employer might, subject to the
constraint of expected present value maximization, forego potential gains of
current quasi-rent made possible by transitory excess labor supply and refrain
from reducing nominal wages. However, he or she might also ( la Lucas
Rapping) seek out other components of the real wage that could be reduced
without causing as much loss of morale as would result from a reduction in
nominal wages. Perceptive readers will note that concern for maintaining
morale could serve as a rationalization of the reported reluctance to engage in
temporary layoffs as well as an impediment to temporary reductions of
nominal wages. (While the morale constant tradeoff of layoffs against wage
cuts is an interesting topic, it is not relevant to the argument of this chapter.)
In considering the concern of employers with maintenance of morale, it
helps to remember that a rational employer considers an investment in
specific human capital as similar (in many respects) to an investment in
plant or equipment. Customarily such investments may be maintained
in depressed periods despite the considerable cost of doing so. For specific
human capital such cost takes the form of higher wages than an imaginary
spot labor market for workers of equivalent human capital would require.10
What new classicals might term maintenance of specific human capital, Akerlof, Bewley et al., along with many others (business persons,
personnel specialists, psychologists, and sociologists) would describe as
concern with fairness, honoring implicit promises and other praiseworthy
objectives that are incompatible with the profit maximization entailed by
10
A critical reader might object to the concept of a market for specific human capital. While I
believe that I could defend use of the concept, I recognize the difficulty but eschew the
lengthy diversion that such a defence would entail.
185
the Invisible Hand. This difference in perceptions of long-term employer

objectives is one consequence of using the Invisible Hand frame rather than
some alternative. A second consequence is the interpretation of nonemployment as involuntary unemployment and a waste of resources rather
than a realization of a stochastic process affecting the intertemporal allocation of time and output. Yet a third consequence is refusal to accept
explanations for counterfactual behavior (why possible courses of action
not taken by economic actors) are rejected as rationalizations of behavior
seeming to violate economic laws associated with the Invisible Hand. This
chapter argues that concern for about the first two of these consequences is
a major reason for the methodological decision reflected in the third.
IV
In seeking to influence the views of other people, economists and other
social scientists attempt to induce them to think in a particular way about
whole classes of problems, many of which cannot be specified in advance.
Many of the most important problems concern public policy. Although far
from the whole of the matter, the frame that one uses is an important
element of ones method of thinking about a particular class of problems. In addition to the frame (or frames) that an individual uses, he or she
usually has one or more theories and/or prejudices.
Neoclassical economics is associated with a variety of theories and
(possibly) a plurality of frames. However, in this article I have been
concerned with one particular neoclassical frame, the Invisible Hand,
which is embedded in F53 and exhibited in most of Friedmans other
writings as well. As reiterated here, and innumerable times in the literature for well over half a century, that frame presents economic problems
as involving choices or decisions made by individuals who are subject to
constraints of endowed resources and seek to maximize some objective
function which (loosely) can be identified as expected utility. In the
process of maximization, individuals engage in mutually agreeable transactions with others in which the terms of trade (prices) are determined by
competitive markets and are accepted as givens by all parties concerned.
The competitive market and the associated method of price determination are very important characteristics of this frame: it is these properties that lead to the identification of market-determined prices
with alternative marginal costs. This identification generates many of
the non-intuitive implications of neoclassical theories that make the
Invisible Hand frame intellectually appealing, and motivates efforts to
encourage its adoption and utilization.
186
Melvin W. Reder
As stated in section I, Friedman is noted (inter alia) for his insistence that
a major determinant of the quality of a theory is the extent to which it
facilitates useful predictions. But what he urges (in F53 and elsewhere) is
not the collection of unrelated theories, each possessing great power to
make useful predictions, but adoption of a specific frame for a set of theories
that share certain important characteristics. In effect he is telling economists that (in non-monopoly situations) the best way to find successful
theories is to restrict search to theories that are compatible with the Invisible
Hand frame. By implication Friedman holds that it is members of this class
of theories that have the best chance of yielding useful predictions.
As already indicated, economics has no salient criterion for theory
selection that is alternative to what is proposed in F53. However, during
the past half century there has been increasing emphasis on the importance of achieving consilience between the implications of theories accepted in economics and those accepted in psychology and/or sociology that
purport to explain the actions of individuals and firms by allowing for the
possibility of non-rational behavior.
But to achieve consilience with such theories entails abandonment, or
at least drastic alteration, of the Invisible Hand frame. A striking example
of attempts to compel such alteration is provided by the Nobel lectures of
Akerlof and Stiglitz, discussed in section III. The arguments of these
lectures dovetail nicely with the findings of Bewleys interviews with
wage setters in the 1990s, which are strikingly similar both in style and
results to the empirical research that F53 sought to discredit.
At first blush, the combined effect of the arguments of Akerlof and
Stiglitz and the reports of Bewleys interviews with wage setters would
seem to limit the applicability of the Invisible Hand framework. However,
as argued in section III, the contrast between that framework and the
opposing Keynesian vision is very much blurred when intertemporal
substitution of labor is considered. Indeed, the issues involved in interpreting a given spell of aggregate unemployment as an irreparable loss of
potential output rather than an intertemporal substitution of labor involve
so much conjecture on both sides that it is hard to understand the intensity
with which this, and similar disputes, are pursued.
The explanation lies in the importance attached to maintaining the
intellectual viability of the alternative frames for viewing a competitive
economy that are associated with alternative interpretations of unemployment. The importance attached to maintaining the viability of a frame
is due to the influence that its acceptance is believed to exercise on the
views that individuals take of the propriety of governmental intervention
in the functioning of an economy. Over the decades Friedman has many
times insisted upon the separation of his views on positive economics (as
187
in the title of F53) from those on normative economics, especially on

public policy. And this posture is shared widely among economists: not
only among his many disciples, but also among his opponents.
To some extent I also share it. Nevertheless, I suggest that it is no
accident that the overlap between adherence to the Invisible Hand
frame and advocacy of laissez-faire though not complete should be
very substantial. Acceptance of the Invisible Hand leads one to look for
amelioration of shortages either of supply or demand to the workings of
markets rather than to quantity adjustments directed by a communal
authority. It also sensitizes one to hitherto unexploited opportunities for
using the instrumentalities of the market, thereby reducing the scope for
governmental actions, and the seeming need for its regulative activity.
Conversely, rejection of the Invisible Hand is likely to arise from perception of instances where it appears to have failed. Such perceptions
often are not deduced from widely applicable theories but are inspired by
observation of particular cases, either personally or through reports of
affected persons, and often are associated with ad hoc remedies. When
such remedies conflict with one or another implication of neoclassical
economic laws, as they often do, their intellectual support must come
from testimony of affected parties often based on (what they believe to be)
their first-hand experience.
The history of economic thought, especially in the last half century,
strongly suggests that the best obtainable empirical evidence manipulated
by the best available econometric techniques is inadequate to persuade
adherents of a particular frame to support a policy that is uncongenial to
that frame. Hence the importance attached to the selling of a conceptual
frame that is conducive to acceptance of policies that one thinks one will
find congenial.11
References
Akerlof, George A. (2002). Behavioral macroeconomics and macroeconomic
behavior. American Economic Review, 92, 41133
Bewley, Truman (1999). Why Wages Dont Fall During a Recession. Cambridge,
MA: Harvard University Press
11
While I do not say that economists typically choose theories, frames, etc. to rationalize
preconceived policy conclusions, I believe that, varying from one economist to another,
theory may inspire policy, and vice versa. Careful sorting out of the positive from the
normative comes at the stage of presentation sometimes consciously, sometimes not
but after the stage of inspiration.
188
Melvin W. Reder
Blinder, Alan S. et al. (1998). Asking About Prices. New York: Russell Sage
Foundation
Coase, Ronald H. (1937). The nature of the firm. Economica, 4, 386405
Hoover, Kevin D. (1988). The New Classical Macroeconomics: A Sceptical Inquiry.
Oxford: Blackwell
Kahneman, Daniel (2002). Maps of bounded rationality: psychology for behavioral economics. American Economic Review, 93, 144975
Keynes, John Maynard (1936). The General Theory of Employment, Interest and
Money. New York: Harcourt, Brace and Company
Lucas, Robert E. and Leonard A. Rapping (1969). Real wages, employment and
inflation. Journal of Political Economy, 77, 72154
Schumpeter, Joseph A. (1954). History of Economic Analysis. New York: Oxford
University Press
Shiller, Robert J. (2003). The New Financial Order. Princeton, NJ: Princeton
University Press
Simon, Herbert A. (1947). Administrative Behavior. New York: Macmillan
Stiglitz, Joseph E. (2002). Globalization and its Discontents. New York: W. W.
Norton
Tversky, Amos and Daniel Kahneman (1974). Judgment under uncertainty:
heuristics and biases. Science, 185, 112430.
Wilson, Edward O. (1998). Consilience. New York: Alfred A. Knopf
The politics of positivism: disinterested

predictions from interested agents
David Teira Serrano and Jess Zamora Bonilla
Politics in The methodology
Of the six sections composing The methodology of positive economics

(F53), the first (The relation between positive and normative economics) is apparently the least discussed in the F53 literature, probably as a
result of it being not only the shortest section, but also the least relevant for
the issue of realism. In view of Milton Friedmans subsequent career as a
political preacher, one wonders whether this first section ruled it in the
way the other five directed Friedmans scientific performance. After all,
the role of prediction in defining positive economics was already advanced
therein: when an economist predicts, his or her results are independent
of any particular ethical position or normative judgments. This is also
why positive economics is a politically relevant discipline: as long as the
differences about economic policy among disinterested citizens derive
only from different predictions about the economic consequences of
taking action, these differences could be eliminated by the progress of
positive economics. Our plan in this chapter is to present, in the first place,
the role of political motivations in the development of Friedmans methodological stance. As we will discuss in section 2, Friedman was involved
in the policy-making process right from the beginning of his professional
career, and was able to experience at first hand the relevance of economic
predictions in generating a consensus not only among politicians or the
public opinion, but among the profession itself. Conversely, Friedman
could also appreciate how difficult it was to reach a consensus on a
particular policy when the economists disagreed on its practical consequences. In this respect, as we will see, F53 attempted to guarantee the
political efficiency of economic research. However, the sociological turn
in science studies raises the question on what basis can we deem a
prediction neutral? Is it simply that economists produce these positive
Our joint research was funded by the Urrutia Elejalde Foundation and by the grant BF200304372.
189
190
predictions disinterestedly, even while deeply engaged in political debates?

In section 3, we analyze how Friedman himself produced predictions
immediately before F53 was drafted, and how this procedure lies at the
core of his Marshallian approach, which he contrasted to the Walrasian
strategy on the grounds of its higher political relevance. Yet, from a
sociological viewpoint, it is precisely this Marshallian strategy that seems
most objectionable: the way Friedman deals with theoretical concepts in
economics in order to obtain predictions makes them particularly susceptible to manipulation by a not so disinterested economist. We will see that
F53 does not provide any defense whatsoever to counteract this.
Finally, we argue that the knowledge produced by economists can
only gain the trust of lay audiences if the latter know that the activity of the
former is constrained by an appropriate methodological contract (Zamora
Bonilla 2002), and we suggest that F53 can be best understood as an
attempt to persuade the economics profession to adopt a certain methodological contract, an agreement which ensures that supporting economic research is a rational choice for citizens and politicians (section 4).
Nevertheless, our own case will not be long, as a brief article we add by way
of conclusion makes it much better than we did (section 5).
2
Predicting for the policy-makers
Whereas only a minor section of F53 was devoted to normative issues, a

significant part of Friedmans professional experience between 1935 and
1945 took place in Washington, DC, quite close to the political arena. We
are going to examine the various contexts in which Friedman came to
experience the connection between prediction and politics, which, in our
opinion, provide the drive behind F53s first section. We discuss
Friedmans first stay in Washington in the New Deal years, also paying
attention to the debate on the political relevance of statistical economics in
the early 1930s. We will also consider how Friedman developed a position
of his own early in the 1940s, while appointed at the Treasury. By the end
of the decade, it had become a standard view among the profession if we
are to judge by an American Economic Association (AEA) report that
Friedman coauthored in 1947. Finally, we exemplify the role Friedman
assigned to prediction as a consensus-generating tool by means of an
analysis of two of the papers included in his Essays in Positive Economics.
Let us then start in 1935. A growing demand for young economists
sustained by New Deal policies gave the young Friedman an opportunity
to enter a depressed labor market (Friedman and Friedman 1998, 60).
Paradoxical though it may seem now, he was appointed at an agency
intended to promote economic planning at the highest levels of
The politics of positivism
191
government, the National Resources Committee (NRC) (Brinkley 1995,

24550). Under the supervision of Hildegarde Kneeland, Friedman joined
an NRC task force assigned to coordinate the Study of Consumer
Purchases (SCP), a wide-ranging statistical survey intended to quantify
consumer incomes and expenditure in the United States in the period
19356 (NRC 1938, 1939). Or, in other words, to provide a statistical
answer to the debate over the causes of the Great Depression which heated
up the 1932 presidential campaign. At its roots Roosevelt had argued
then lay the maladjustment between industrial production and the purchasing power of would-be consumers, and the way out consisted simply in
raising the latter (Brinkley 1995, 6972; Balisciano 1998,1605). Six years
after Roosevelts election, the NRC income report was printed. It still
addressed the underconsumption issue in its preamble (NRC 1938,1),
even if the policies so far implemented had worked to the detriment of
consumers (Brinkley 1995, 71). Though credited among the contributors,
Friedmans memoirs skip the political implications of the SCP, focusing
instead on the statistical challenges involved therein (Friedman and
Friedman 1998, 616).
New Deal policies required government intervention in the American
economy on a scale never seen before. In need of figures (Duncan and
Shelton 1992, 322), the American administration multiplied its statistical
production, and most economists were appointed with a view to conducting empirical research, their lack of technical qualification notwithstanding
(Hotelling 1940, 458). As a result of the year he spent in Columbia with
Harold Hotelling, Friedman was certainly familiar with the subtleties of the
methods introduced by Ronald Fisher in the 1920s, but sampling design
was a newborn discipline, almost unknown to most federal agencies.1 The
SCP thus gave him the opportunity to face the challenges of applying
theoretical statistics to the design and implementation of a large-scale
survey (Kneeland et al. 1936; Schoenberg and Parten 1937). Indeed, the
SCP gave none other than Jerzy Neyman the opportunity to devise a double
sampling procedure to correct the one applied by the NRC team, to which
he was exposed during his stay at the Department of Agriculture Graduate
School in 1937. Friedman himself posed the problem during one of
Neymans lectures.2 In sum, New Deal planning fostered the completion
1
On the introduction of mathematical statistics in the USA in 1933, see Stigler (1996). The
use of mathematical statistics by Federal agencies in that same year is discussed in Duncan
and Shelton (1992, 321). For an overview of the development of sampling techniques, see
Seng (1951) and Desrosieres et al. (2001). For an extended discussion of the methodological challenges herein involved as Friedman could have met them, see Teira (2007).
Cf. Neyman (1938, 328). On the context, see Reid (1998, 1378) Friedmans contribution is mentioned in p. 148. For an extended discussion, see Teira (2007).
192
of Friedmans statistical training, both on a practical and theoretical level. It

is quite plausible that he had then become aware of the political relevance of
statistical economics. However, it is also possible that he already knew that
statistical economics could also make economists differ.
In 1932, Roosevelt recruited three Columbia professors as campaign
advisers. Among the members of this Brain Trust, as they came to be
known, was an economist, Rexford Tugwell, who had proclaimed the year
before in the AEA meeting the end of the laissez-faire era (Tugwell 1932).
According to Tugwell, it was federal government responsibility to adjust
production and consumption through price control, securing the purchasing power of salaries. Moderate voices demanded a mere indicative
planning, an ineffective solution for Tugwell since it did not eliminate
uncertainty (Tugwell 1932, 85).
It was certainly the end of free entrepreneurial activity, as Frank Knight
Milton Friedmans revered teacher had conceived in his 1921 classic
Risk, Uncertainty and Profit.3 Economic planning required perfect information, which was, for Knight, unattainable: the decision of each economic
agent is essentially unique and thus unpredictable. Individual freedom was
at stake if the planners were to reach their goals (Knight 1932, 448).
In response to those who viewed economics as a mere technique of
prediction and control, Knight warned that only the short-term view of the
price problem, as contained in static economics, is to be deemed scientific.
Unfortunately, static economics did not allow for far-reaching predictions as
required by planners, whereas those derived from institutional approaches to
economics amounted to nothing more than sheer philosophy of history
(Knight 1924, 146). This was hardly good news to announce in the times of
the Great Depression: still an undergraduate student in Chicago, Friedman
could attest to the reactions it aroused in 1932 when Knight delivered his
tongue-in-cheek lectures on The Case for Communism before an overreactive audience (Friedman and Friedman 1998, 37).
By the end of the 1930s, in sum, Friedman had already experienced how
predictions were requested by economists with a normative purpose
(namely, state intervention) and the objections raised against economics
assuming such a role. As for his own position concerning statistics and
economic planning, little is known; but even in 1938 Aaron Director could
make jokes about Friedmans very strong New Deal leanings authoritarian to use an abusive term (apud Friedman and Friedman 1998, 81).
In the 1940s, Friedmans stance began to take shape. By 1942, he was a
self-proclaimed thorough Keynesian: the terms money or monetary policy
3
On this particular point, see Pradier and Teira (2000, 2002).
193
seldom appeared in his writings (Friedman and Friedman 1998, 113).

The year before, he had been summoned again to Washington, this time
appointed as the Principal economist at the Treasurys division of tax
research (Friedman and Friedman 1998, 10525). Having completed
the draft of his dissertation at the NBER that same year, Friedman was
an accomplished expert in both income and consumption studies, a
profile highly demanded again by the Federal Government. Hitlers
Germany had already proven how menacing state intervention could be
if economic planning were taken to its furthest consequences. Roosevelts
liberals were now inclined to intervene through more compensatory tools,
in Alan Brinkleys words: i.e. a combination of Keynesian fiscal measures
and enhanced welfare-state mechanisms (Brinkley 1995, 15464). In a
crude war context, stabilizing consumption required active taxing policies
that prevented inflation. Friedmans role at the Treasury obliged him to
take part in the real policy-making debate, which involved frequent congressional testifying, providing consulting services to many a senator,
speech drafting, and so forth. Friedman argued for a demand-oriented
approach to income analysis, which conferred theoretical significance to
the available statistical data (e.g. Shoup et al. 1943, 11153) an approach
already exemplified in his own dissertation: see below section 3.
By the end of the 1940s, the two main theses later stated in F53s first
section were already explicit in Friedmans texts: statistical economics was
affirmed as a politically relevant discipline, not only for being instrumental in
the policy-making process, but also for its virtue as a consensus-generating
device. We should take into account that this was not a minority view among
the profession by then, as may be noticed in a report delivered by the AEA
Committee on Public Issues in 1950. Friedman and three coauthors4 were
assigned to present to a general audience the state of the art concerning the
problem of economic instability. One of the first topics they addressed was
the relevance of economics in dealing with inflations and recessions, its
forecasting limitations notwithstanding. Taking sides against Knights indictment on institutional economists, the AEA committee proclaimed:
A second misconception is contained in the proposition that effective stabilization
policies can be designed only if we understand fully all the causes of fluctuations.
Partial knowledge can be very useful for deciding how to act. (Despres et al.
1950, 512)
Even if the calculations involved in planning were out of the reach of any
statistical economist, the policy-maker could still benefit from concrete
4
Together with Emile Despres, Albert Hart, and Paul Samuelson. The committee had been
appointed in 1947.
194
forecasts of the evolution of certain economic variables. On the basis of

this partial knowledge, the profession could reach a consensus on which
policies were most convenient to the country (Despres et al. 1950, 505).
That is, they deserved to be trusted by the public opinion, their usual
discrepancies notwithstanding (511).
The connection between partial knowledge, predictions and consensus
among the profession was even more explicit in another paper, by
Friedman alone, on the issue of stability published the year after the
committee formed. In A monetary and fiscal framework for economic
stability (Friedman [1948] 1953a) he argued for an economic agenda
which would lead to the attainment of a number of consensual values such
as freedom, efficiency, and economic equality (Friedman [1948] 1953a, 134),
on which not many disinterested citizens would disagree. If the profession
could be persuaded first of the convenience of the agenda the greatest
common denominator of many different proposals , it may be presented
as a minimum program for which economists of the less extreme shades
can make common cause (Friedman [1948] 1953a, 135). Friedmans
agenda consisted of the following four core points:
The essence of this fourfold proposal is that it uses automatic adaptations in the
government contribution to the current income stream to offset, at least in part,
changes in other segments of aggregate demand and to change appropriately the
supply of money. (Friedman [1948] 1953a, 139)
The case for this proposal was thoroughly presented two years later in
The effects of a full-employment policy on economic stability: a formal
analysis (Friedman 1951). Full-employment policies were usually analyzed on the basis of a four-variable model: consumption (C), income (Y),
investment (I), and government expense (G).
Y CI G
C f Y
The income level Y0 that will bring about full employment depends
directly on government expense, which must counterbalance I. Stability
is therefore the stability of income Y. The discussion of the effects of
discretional government action proceeds in a way inspired by the theory
of statistics rather than economic theory (Friedman [1948] 1953a, 121),
more precisely by the statistical analysis of variance.5 Income at t Z(t) is
decomposed into two variables, the first one X(t) measuring the level
Friedman (1949, 951), from which the idea evolved (Friedman [1948] 1953a, 121n.)
195
attained without a full-employment policy, the effects of which are measured by the second one Y(t). The analysis of income stability would thus
work analogously to that of variance:
For X or Z the variance measures the fluctuations in income in the absence or
presence of a countercyclical policy. For Y the variance may be regarded as
measuring the magnitude of the countercyclical action taken. (Friedman [1948]
1953a, 123)
In this way, it would be possible to analyze in which conditions the

variance of Z is less than the variance of X, obtaining Ys optimal value.
Therefore, Friedman concludes, the variance of Y and its coefficient of
correlation with X allow us to classify any full-employment policy according to its effects on economic stability. Yet, statistics would not allow any
planner to control the cyclical variations at will. Quite the opposite: only
automatizing the response, Friedman argues, would minimize the delay
between the variation to counteract and the action which would counteract it or, in statistical terms, to approximate the value of the correlation
between X and Y to the desired 1. Therein lies, according to Friedman,
the positive superiority of a monetary and fiscal program based on automatic reactions over discretional policies, beyond their normative convenience to each particular citizen.
These two papers thus exemplify the role that F53s first section
assigned to prediction in the generation of public consensus about particular policies: as long as the divergence lies in the policy to be chosen, it
may be resolved on a purely consequentialist basis through statistical
analysis. Given that positive economic predictions are independent of
any normative commitment on the part of their authors, they deserve to
be trusted by any disinterested citizen when a decision on the policy to adopt
is at stake.6 In spite of its brevity, F53s first section reflects thus a
significant part of Friedmans professional experience, a real methodological concern about the role an economist could play in promoting
public policies without losing his scientific integrity. The question we
are going to address in the next section is how F53 contributes to such a
goal.
Yet, although unmentioned in F53, attention should also be paid to the fact that professional
consensus among economists on the most convenient policies is required for any general consensus to be reached (Friedman [1948] 1953a; Despres et al. 1950). Or conversely, as Rose
Friedman put it: It is not difficult to see why different layman might have different
predictions about the economic consequences of certain actions when the experts disagree (Friedman and Friedman 1998, 217). For laymen and experts alike, the way to
attain consensus is the same one: predicting. This topic will be taken up again in section 4.
196
How to disagree on predictions
Between meeting Hotelling in Columbia (1933) and returning to Chicago

(1946), Friedmans statistical qualification grew incessantly: on the
practical side, he extended his training first at the NRC (19357) and
later at the NBER (193740), where he collaborated with Simon Kuznets
while completing his dissertation; on the theoretical side, he was also able to
benefit from his two-year appointment at the Statistical Research Group
(19435).7 Friedman also engaged in a collective effort to renew the teaching of statistics in American higher education, a task he undertook first
during his brief stay in Wisconsin and later on in Chicago, where he
contributed to the creation of a separate Department of statistics (Olkin
1991, 1257). Friedman was therefore extremely well qualified not only to
advocate predictive theories, as he did in F53, but to settle a methodological
standard on what should count as a good prediction. He did not do this,
however, either in his 1953 paper or afterwards. The methodological
dilemma that arises therein may be stated as follows: can we really base a
consensus, both public and professional, on predictions without any criterion to assess them? F53 proves to be extremely effective in discriminating
politically relevant vs. irrelevant economic theories, once we assume that a
predictively empty theory is eo ipso politically useless. However, lacking
such a criterion, the particular predictive procedure that Friedman had in
sight seems to be particularly capable of manipulation today, suspicious as
we have become after the sociological turn in the study of science. In order
to show it, let us examine first how Friedman achieved predictions.
By way of example, let us focus on a NBER joint monograph, part of
which was also submitted to Columbia as a doctoral dissertation by
Friedman. Friedman and Simon Kuznets finally published it under the
title Income from Independent Professional Practice in 1946. Its methodological relevance was emphasized by Friedman himself, who declared to
E. B. Wilson in 1946 that its chapters 3 and 4 were his most successful
pieces (Stigler 1994, 1200): the balance between demand theory and
statistical analysis exemplified a particular combination of the approaches
to economics developed in Chicago and Columbia which were to distinguish Friedmans own style. Besides, the study was explicitly aimed at
reaching conclusions relevant for public policy, educational in this case
(Friedman and Kuznets 1945, v). Friedmans expertise in income analysis
was now applied to discern whether professional workers constitute a
noncompeting group; i.e. whether their number (and hence their
7
For further details, see Friedman and Friedman (1998, 6776, 12547).
197
income) was exclusively determined by the relative attractiveness of

professional and nonprofessional work, or rather by the number of prospective students who count on the particular resources required to
pursue the career in question (Friedman and Kuznets 1945, 93).
Among the five professions considered, medicine and dentistry provide
particularly suitable data. Given that both professions require similar
abilities and training, we should expect the prospective practitioners to
choose between them mainly on the basis of their respective level of
return (Friedman and Kuznets 1945, 123). Did the differences in average income between doctors and dentists correspond thus to the equilibrium level, or do they rather point to an entry restriction such as the
licenses imposed by the American Medical Association?
In our opinion, the most salient methodological feature in these two
chapters lies in the way demand categories are redefined so that predictions can be derived therefrom. Instead of searching for an empirical
counterpart that corresponds to its theoretical definition, Friedman
restates demand theory so that it corresponds with the statistical variables
at hand: e.g. careers would be chosen according to the average expected
income as reflected in the arithmetic mean (Friedman and Kuznets 1945,
65, 145); prices and expected income would approximate the commodities offered by doctors and dentists, in the absence of a definite unit to
measure them (155); regression analyses performed separately on the
1934 and 1936 data for per capita income of both professions, their
number per 10.000 population and general per capita income (16173)
would do as proxy to discern whether the difference between their average
incomes lies at the equilibrium level. Though admitting that none of these
variables would account for the behavior of any particular individual,
Friedman and Kuznets argue that they may nevertheless explain the
behavior of the group of prospective entrants as a whole (96). In short,
even if there is not a perfect correspondence between theoretical concepts
and the data, it is preferable to relax our definition of the former so that
predictions may be derived.
Indeed, the alternative at hand (rigorous definitions without an empirical counterpart) seemed to Friedman methodologically undesirable.
Friedman opted for avoiding schizoid concepts, i.e. those thoroughly
competent in the field of deductive analysis but utterly incompetent for
quantitative analysis as he claimed in a joint paper with Allen Wallis
from the same period (Wallis and Friedman 1942, 176). The more
mathematical constraints are imposed by the definition of a theoretical
concept, the less value it will have for the organization of empirical data
(Wallis and Friedman 1942, 186). This sort of taxonomic theorizing
was precisely the mark of the Walrasian approach, which was canonically
198
criticized in Friedmans 1946 review of Lange: no empirical counterpart

will correspond to those abstract taxonomies in the data (Friedman
[1946] 1953a, 2869).
The core of the dichotomy between Walrasian and Marshallian
approaches seems to lie in the way theoretical concepts are defined:
Walrasian taxonomies are opposed to Marshallian filing cases. Whereas
theoretical concepts in the former were defined by purely logical means,
producing entirely abstract taxonomies without correspondence in the
data, in the latter they were redefined with a view to obtaining a classification of the data. We may better understand what this classification
issue had to do with predictions if we take into account Friedmans
predictive technique of choice: linear regression enhanced by the analysis
of variance.
It was probably Hotelling who introduced Friedman to this technique
in Columbia, given that he counted among the few in the USA who were
then abreast of Ronald Fishers recent breakthrough in statistical inference (Friedman 1992, 2131; Stigler 1996). According to the teaching of
Fisher,8 for a regression curve to be traced the first step consisted in
classifying data in a contingency table, so that every observed figure was
filed under a given variable. On the basis of this classification, contingency
correlations and eventually regression may be calculated. Therefore
very strict definition of the variables would make unfeasible the classification of the data, thus preventing predictions from being obtained.
Handling variables in a more Marshallian spirit would allow the economist to render a theory predictively more fruitful.
Let us return now to the opening question of this section: in view of all
this, why did Friedman not set up a standard in order to assess predictions? Part of the answer may be that such a standard could have affected
the very definition of the variables involved, restricting its predictive
capacity. On the other hand, it seems as if F53 was aimed more at
discriminating between predictive and non-predictive approaches than
at characterizing good predictions. For Friedman, choosing between
Marshall and Walras was not an entirely methodological issue: it
depended also on the purpose for which the theory is constructed and
used (Friedman 1953b, 90).
Part of Friedmans purposes was to derive politically relevant predictions, as we saw in section 2. In this respect, a rough prediction is better
than none: even if the theoretical relevance of the variables analyzed by
8
In this respect, an extension of Galton and Pearsons approach, also adopted by Hotelling
later on. Allen Wallis still used it in his 1956 manual (Wallis and Roberts 1956). For a
superb analysis of this intellectual tradition, see Armatte (1995).
199
Friedman and Kuznets was at most conjectural, they did not hesitate in
blaming the difference observed in the average income between medicine
and dentistry on the medical licenses.9 Conversely, if policy conclusions
were always to rest on a prediction, Walrasian approaches were inherently
useless for the policy-maker, as Friedman often pointed out.10
Irrespective of its purported precision as a demarcation criterion, prediction proves to be extremely discriminating when it comes to the political
efficiency of an economic theory.
However, in the aftermath of the science wars, the mere claim that such
a criterion may pick out theories being at once positive and politically
useful is clearly suspicious. Friedmans definitional strategy called into
question a central claim in our methodological received view: theoreticity.
In order to guarantee that each theoretical term had a clear empirical
(positive) counterpart, rigorous definitions were called for. From this point
of view, it is not strange that Walrasian axiomatizations have been so often
appreciated by philosophers of science. However, when rigor in definitions is relinquished with a view to facilitating predictions, the sociological
challenge to the neutrality of scientific classifications seems most threatening. According to a tradition dating back to Durkheim and Mauss,11
our most basic classifications of sense-data are inescapably biased by
social interests, since the concepts we use to this effect are rooted in our
daily practice. There is no disinterested ground on which to base scientific
concepts. Now, in our opinion, even those who disagree with David Bloor
should concede that Friedmans methodology does not provide such a
disinterested ground. Quite the contrary: for those economists whose
daily practice consists in obtaining predictions with a political view, it is
essential that the definition of theoretical concepts is loosened so that the
data they are provided with can be classified and predictions follow. Yet,
once the rigor of the definition is relaxed, who could guarantee that it will
not be manipulated in order to obtain interested predictions?
By way of example, let us recall that the conclusions of Friedman
and Kuznetss analysis were called into question by the American
Medical Association. Apart from a few years delay, it took a Directors
10
11
The analysis is necessarily conjectural and our quantitative results are only a rough
approximation. But the problem is real, and a rough approximation seems better than
none (Friedman and Kuznets 1945, viii).
E.g. Friedman ([1946] 1953a, 300; [1947] 1953a, 316)
After all, categorization is a central concern in many a sociological approach to science: the
starting point was E. Durkheim and Marcel Mausss 1903 essay Primitive classification.
Bloor (1982) vindicated its relevance for the Strong program. Philip Mirowski introduced this
tradition in the sociology of economic knowledge in his More Heat than Light (1989, 412).
200
comment by Carl Noyes (Friedman and Kuznets 1945, 40510) to be

added before the NBER accepted the publication. The notes major
concern was to state certain reservations over the scientific validity of
the results obtained in chapters 3 and 4. More precisely, to argue that it
could have been different, had the authors introduced additional distinctions (e.g. specialists vs. general practitioners) in their analysis of the data:
i.e. different classifications. In this respect, Friedmans methodological
theses on prediction as the real mark of value-free science fail to provide
an adequate defense against objections over the possibility of manipulating the classification of the data according to the particular predictions
you want to derive therefrom. How much precision and conformity with our
experience should a prediction render to rule out any possible bias?
Friedman was obviously not unaware of this particular bias Ronald
Fisher himself had the occasion to warn him when the Friedmans visited
him in Cambridge, just a year after the publication of F53 (Friedman and
Friedman 1998, 244). We may wonder why he left this objection unanswered. For the moment, the only available clue is the one provided by
Rose Director Friedman:
I have always been impressed by the ability to predict an economists positive views
from my knowledge of his political orientation, and I have never been able to
persuade myself that the political orientation was the consequence of the positive
views. My husband continues to resist this conclusion, no doubt because of his
unwillingness to believe that his own positive views can be so explained and his
characteristic generosity in being unwilling to attribute different motives to others
than to himself. (Friedman and Friedman 1998, 21819)
That is to say, a principle of mutual methodological trust, which like

most ideals is often honored in the breach (Friedman 1967, 5). In the
end, the production of positive predictions of political use would require a
sort of tacit normative commitment on the part of the economist: to serve
the public interest, albeit self-interestedly (Friedman 1986, 9). The
obvious question is why? A tacit consensus may have still existed early in
the 1950s among a cohort of American economists who had been
attracted to the profession in the lower reaches of a big depression, to be
subsequently recruited by the government to fight it (Friedman and
Friedman 1998, 34). However, it is dubious whether this consensus still
exists today. As we pointed out in section 2, Friedmans claim was quite
the opposite: consensus on the best economic policy available, both
among professionals and general audiences, was to be grounded on
predictions, so that its normative neutrality should be guaranteed right
from the beginning. If we still expect economics to play such a consensual
role in our public sphere, we should somehow restate this previous tacit
201
consensus among the profession.12 A reexamination of F53 in a contractarian light might provide several clues to achieving it.
Making the contract explicit
Let us then reason as if. What if we reread Friedmans methodological

case as if it were a plea addressed to his fellow economists in order to gain
the trust of the public opinion on the results attained by the profession?
The problem would then center around how economists can signal to
their lay fellow citizens that the advice provided by the former is scientifically sound. Stated in terms of the economics of information, Friedmans
plea may be thus restated as the economists part in a contract (as agents)
with the consumers of their science (the principals). The parties to the
contract find themselves in an asymmetrical situation: those who produce
knowledge can better judge the true quality of their output (or their own
capabilities as producers) than those who consume it. Therefore, the
demanders desideratum would be that the contract be designed in such
a way as to oblige the suppliers to work efficiently in the production of
items (of knowledge) having the highest possible utility for them.
The principals main difficulty is that they are often incapable of controlling by themselves whether the contract is being complied with by the
agents or not, or to what extent it is. Even more, they are unable to design
by their own means something like an optimum contract: ironically, the
principals would perhaps need to hire an economist to design an optimum
contract for them. Due to the inability of the scientists to gain the trust of
the principals, the former would be forced to set up the contract as an
agreement among themselves, which would serve as the constitution of their
discipline. Under its rule, the consensus they reach and/or their past
technical success would do as signals of epistemic quality. However, this
also opens the door for moral hazard problems: since the consensus
among experts is the only signal that users may often have of the quality
of researchers or of the knowledge produced by them, the losers in the
race for a discovery could simply deny that real knowledge has been
attained by other colleagues, preventing as a result the constitution of
that consensus.
As we saw in section 2, throughout the 1930s and the 1940s, Friedman
often found himself acting as an agent for many different principals: first
as a member of the NRC task force researching consumer incomes and
expenditures for the Roosevelt administration and later as the Principal
12
For further details about our discussion of Friedmans methodology, see Teira 2004.
202
economist at the Treasurys division of tax research. It may be said that in

this case the principals trust in the use of statistical economics had already
been gained at least a decade before Friedman came to Washington for
example, during the Hoover administration (cf. Barber 1985, 89).
However, as we pointed out before, different views on the political use
of statistical economics were in dispute among the profession in the early
1930s, a debate which was not completely closed by the time the AEA
commission addressed the problem of economic instability late in the
1940s. The fact that Roosevelt could have initially ignored Keynes himself
may indicate that choosing the right economist was still difficult, and the
disagreement among the profession certainly did not contribute to simplify the choice. Both the AEA report and Friedmans parallel paper
coincided in the need for a professional consensus that ensured that the
best policy could be implemented. Furthermore, Friedman was arguing
for a particular procedure to reach it, i.e. judging each alternative according to its predictive reliability. His 1953 proposal of a methodological
constitution for the economic profession took this rule as its cornerstone.
Once this rule was adopted, economics would turn into a positive
endeavor in our terms, a trustworthy alternative for every principal.
A contractarian assessment of Friedmans proposal requires further
refinements of our notion of a social scientific contract. Basically, the
aim of such a contract is to render compatible, on the one hand, the
incentives of the people supporting scientific research (i.e. the desire for
useful knowledge), and, on the other, the incentives of researchers themselves. These are generally considered to be seeking recognition in return
for their discoveries, as well as for other sorts of privileges that may result
therefrom. There are two possible sources of divergence between the
researchers interests and those of their customers: first, it may be that
the opportunity arises for the scientist to obtain recognition without
producing real discoveries; second, their research domains may be really
worthless for the customers. But the incentives of the researchers themselves may also be in mutual conflict, not only because of the competition
to attain better results, but also because scientists may simply opt for
denying public recognition to their rivals even when they believe that
they really deserve it. A well-designed social scientific contract is, then,
a set of norms that make researchers behave in an honest way toward
those who support the production of science (producing useful knowledge) and toward their colleagues (e.g. acknowledging their discoveries
when appropriate). Though different scientific communities or schools
may have different social contracts, their essential and common elements should be the following ones:
203
(a) A set of methodological rules, determining which actions (in particular,

which assertions) a researcher is entitled or committed to perform;
these norms are of an inferential nature, since they connect the previous
entitlements and commitments of a researcher with a specific action
which can be seen as their consequence. Needless to say, this type of
norms has been the main traditional topic of the philosophy of science.
(b) A set of scoring rules, which, formally understood, is a function transforming the entitlements and commitments acquired by a researcher
into publicly expressed scientific merit (or his/her score); these
norms serve basically to define the significance of a scientific item as
well as the competence of a researcher.
(c) Lastly, a set of resource allocation rules, which determine what and how
many scientific resources are to be apportioned to a researcher
according to his or her absolute or relative merit; violations of these
rules can also modify a researchers score if he or she is considered
responsible for the misallocation.
Traditional views of the social contract of science initially had an
almost absolutely naive conception of the second and third types of rules.
As to the former, it was assumed that scientists could be relied on to assess
the score of their colleagues; concerning the latter, the deference of the
public toward the scientists decision about what projects deserved to be
funded was simply presupposed. Friedman was probably not far from
sharing this positivist view (cf. Friedman 1986, 8). However, after the
social turn in science studies, trust in science seems to require an explicit
commitment to scoring and allocation rules, if scientists and nonscientists are to reach a reasonably efficient and self-supporting equilibrium in the game of science.
In this respect, the conclusions of section 2 seem devastating. It may
happen, for instance, that a quack succeeded in deceiving public opinion
by producing apparently good, yet theoretically irrelevant predictions
(Friedman 1991, 36). If scoring rules existed, scientific economists
would penalize the deceiver, but that would require in turn that a clear
standard to assess predictions were settled. After our discussion in section
2, we know that there is no such standard. Friedman seems rather to opt
for relying exclusively on the personal integrity of each economist in
producing predictions, on a purely reciprocal basis. As for the orientation
of lay audiences in funding economic research, Friedman seems to doubt
its very possibility,13 his previous methodological plea notwithstanding.
13
E.g.: There is no satisfactory solution to the dilemma posed by the propositions: (1) there
is a body of positive economics that can yield reliable predictions of the consequences of
change; (2) there are experts in positive economics; (3) differences about the desirability
204
Friedman rewritten?
Fortunately, other methodologists are now facing this challenge. By way

of example and conclusion, we would like to transcribe here a singular
piece we came across while surfing the internet some time ago. The article,
signed by a certain Pierre Menard, was posted on an old site on Friedman,
no longer active as a matter of fact, the domain was canceled soon
afterwards, and replaced by the following: www.coldbacon.com/writing/
borges-quixote.html. However, it deserves attention from a contractarian
viewpoint, as it rewrites a substantial part of Friedmans 1953 paper in
precisely the spirit of our reconstruction. Stat rosa pristina nomine
The economics of a positivist methodology
1.
The relation between positive and normative economics Confusion
between positive and normative economics is to some extent inevitable.
The subject matter of economics is regarded by almost everyone as vitally
important to himself and within the range of his own experience and
competence; it is the source of continuous and extensive controversy
and the occasion for frequent legislation. Self-proclaimed experts
speak with many voices and can hardly all be regarded as disinterested.
The conclusions of positive economics seem to be, and are, immediately
relevant to important normative problems, to questions of what ought to
be done and how any given goal can be attained. Laymen and experts alike
are inevitably tempted to shape positive conclusions to fit strongly held
normative preconceptions and to reject positive conclusions if their normative implications or what are said to be their normative implications
are unpalatable.
It seems that positive economics should in principle be independent of
any particular ethical position or normative judgments, for it deals with
what is, not with what ought to be. Its task, it has been sometimes
argued, is to provide a system of generalizations that can be used to make
correct predictions about the consequences of any change in circumstances, and so, its performance should be judged by the precision,
scope, and conformity with experience of the predictions it yields. In
short, positive economics should be an objective science, in precisely
the same sense as the physical sciences are assumed to be. Of course, the
of governmental policies often reflect different beliefs about the consequences of the
policies conclusions of positive economics rather than different values; (4) there is
no simple litmus test by which a citizen can decide who is an expert and who is a quack;
yet (5) even though the patient is incompetent to choose the physician, there is no
alternative in a free society. (Friedman 1975, p. x).
205
fact that economics deals with the interrelations of human beings, and that
the investigator is himself part of the subject matter being investigated in a
more intimate sense than in the physical sciences, raises special difficulties
in achieving objectivity at the same time that it provides the social scientist
with a class of data not available to the physical scientist. But, after all,
surely neither the one nor the other makes a fundamental distinction
between the goals and methods of the two groups of sciences.
More importantly, normative economics cannot be independent of
positive economics. Any policy conclusion necessarily rests on a prediction about the consequences of doing one thing rather than another, a
prediction that must be based on positive economics. Governments wanting to improve the living conditions of their citizens, or wanting at least to
maximize the chances of being re-elected by favoring a majority of their
constituency, need to be confident about the likely consequences of the
policies that experts are recommending. It is then both in the interest of
the economist and of the politician to have a corpus of substantive
hypotheses that are successful when judged by its predictive power. On
this basis, the judgment has been ventured that currently in the Western
world differences about economic policy among citizens who try to be
disinterested derive predominantly from different predictions about the
economic consequences of taking action differences that could perhaps
be eliminated by the progress of positive economics rather than from
fundamental differences in basic values, differences about which individuals ultimately face a choice between fighting or showing tolerance. For
example, in the debate about the so called basic income, no one seems
to deny that the ultimate goal is to attain an acceptable standard of living
for all, but strong differences arise when predictions on whether this goal
would be attained or not with the application of that policy, and at what
costs, begin to be drawn. The same takes place in numerous other examples. Hence, the argument follows, achieving consensus about the relevant consequences of each economic policy would reduce to a high extent
the dissensus existing about alternative courses of political action.
This is, of course, a positive statement to be accepted or rejected just
on the basis of empirical evidence, but this evidence could only be
achieved if positive economics advanced as much as to generate a
degree of consensus on empirical predictions much higher than what
has been attained until now. On the other hand, an apparent consensus
about economic predictions can derive either from an appropriate
scientific methodology, or from any political strategy to silence those
making different predictions, and the same dilemma arises regarding a
possible consensus about the normative implications of those predictions.
If this judgment is valid, it means that the normative relevance of positive
206
economics (i.e. the usefulness of its empirical predictions for the fostering
of our social values) can only be assessed through an analysis of the
conditions under which economic knowledge is generated, discussed,
and applied, in order to understand whether the actual conditions can be
expected to generate a corpus of predictively useful generalizations about
economic phenomena, or whether some amendments of those conditions
are both desirable and feasible.
2.
Positivist economics Any normative judgment presupposes some
particular values, for there is no such thing as an absolutely impartial
evaluation. When we describe someone as disinterested, we simply
mean that, in the assessment of positive as well as of normative propositions, he is adopting a moral perspective which does not only take into
account his own private interest and opinions, but also the welfare and the
opinions of other people. However, since different individuals may obviously have different moral values, the judgments of disinterested evaluators need not coincide, although a higher degree of agreement can be
expected than in the case of interested agents. The first essential point
in making any normative pronouncement as experts is, then, to openly
declare the moral values which underlie that judgment. In the argument
elaborated here, these values reduce to the following: the point of view that
must be adopted in order to evaluate the methodological practices of economists is
essentially the point of view of the citizens. After all, policy-makers have to use
the best available economic knowledge in order to promote the welfare of
the members of the political community. Under a democratic regime, this
demand transforms easily into a necessity, for those parties most capable
of putting into use the right economic knowledge will tend to be the
winners in the political competition. The task left for us, economists, is
then that of supplying those theories, models or hypotheses which can
be most efficient in the formulation of right economic policies, i.e.
policies that actually promote the interests of the citizens of a democratic
society.
As a methodological ideal, the positivist conception of science is certainly more appropriate for orienting the practice of economics toward
that goal than the methods which, according to a common interpretation,
are customarily employed in our discipline. Committing surely a blatant
oversimplification, current economics is usually described as divided into
two camps: on the one hand, mainstream economics would basically be
organized as a kind of disguised mathematics, in which the production of
theorems is seen as a much more important and praiseworthy job than the
discovery of regularities about the empirical world. Of course, the formal
models those theorems are about are usually devised as abstract
207
representations of possible economic circumstances, but not much effort

is put the criticism follows into discussing how one can know whether a
concrete empirical situation is better represented by means of a specific
formal model or by a different one. So, the practical usefulness of this
model-building activity is in general rather low, save perhaps in producing
Nobel Prize winners. On the other hand, a heterogeneous bunch of
heterodox approaches would permanently complain about the lack of
realism of the models produced by mainstream economists; unfortunately, the positive contributions of these critical schools consist either in
mere descriptive explorations of the ontology to which the orthodox
models fail to respond, or in the development of formal models presumably more realistic but in the end no more efficient in the discovery of
true empirical regularities.
From a positivist point of view, the common mistake of these two
approaches, at least in their unadulterated forms, would be the presumption that truth is the basic value of scientific research, instead of predictive
capacity. Mathematical economists glorify theorems because, as purely
formal statements, these are necessarily true, whereas heterodox economists criticize the models from which those theorems are derived because
their assumptions are clamorously false. But, clearly, what makes a theorem an interesting one is not just its truth, for all of them are equally true
but not equally relevant, and likewise, the practical relevance of a model
does not depend on the realism of its assumptions, but only on the
accuracy of its empirical predictions.
In general, economists, lacking the capacity to experiment with most of
the parcels of reality they study, have traditionally developed a method
which is based on introspection and casual observation for guaranteeing
the truth of the premises from which their argument starts, and on logicomathematical deduction for guaranteeing that those arguments are truthpreserving, and hence, that their conclusions are also true. But we may
doubt that, from the point of view of practical relevance, this strategy is
optimal. For, in order to successfully apply economic knowledge to
reality, the empirical accuracy of the conclusions is much more important
than the descriptive truth of the premises, and hence, a theory with false
assumptions (such as simplifications and idealizations) but from which
many accurate conclusions follow is much more useful than a theory with
only true axioms but from which just a couple of trivialities can be derived,
even if the first theory has a few false empirical implications as well.
Truly important and significant hypotheses will be found to have
assumptions that are wildly inaccurate descriptive representations of
reality, for a hypothesis is significant if it explains much by little, that is,
if it abstracts the common and crucial elements from the mass of complex
208
and detailed circumstances around the phenomena to be explained and

permits valid predictions on the basis of them alone. In general, both the
process of abstraction (for constructing hypotheses) and the process of
prediction (for testing their validity) demand that the theory is mathematically well organized. Mathematization is just an economical tool for
achieving a great deductive power with a tractable set of premises; it
must not be pursued for its own sake, and less still has it to be hoisted as
the dominant criterion for judging the validity of an economic theory. In
the end, predictive success must be the ultimate criterion, if economics
wants to become a socially relevant discipline.
Of course, all this does not mean that political pressures must be the
exclusive, or even the most important guide of economic research. As in
any other field of science, sheer curiosity is often the most powerful
stimulus toward discovery. The truly significant point is that the method
employed for selecting the best theories or models, even if this selection
is carried out with no interference from politics, should favor those
hypotheses with the highest predictive success. Only if economists follow
this positivist method will they advance toward the ultimate practical
goal of their science, which is the development of a theory that yields valid
and meaningful (i.e. not truistic) predictions about phenomena not yet
observed, and will they also satisfy their intellectual desire of revealing how
superficially disconnected and diverse phenomena are manifestations of a
more fundamental and relatively simple structure.
3.
Some implications for an economics of economics The abstract
methodological issues we have been discussing in the past section serve
to describe the ideal form economic knowledge should have, ideal from
the point of view of the common citizens, who, in a sense or another, are
supporting economic research. We have referred in passing to some ways
economics is usually practiced, which, all our gross simplifications considered, do not seem to correspond to this ideal. We have not examined to
what extent the spread of these wrong economic methods depend on
the specific academic organization of inquiry within the field of economics. Perhaps all intellectual practices have some kind of internal life
which makes some ideas grow no matter what the institutional setting
where they disseminate is. Indeed, past proponents of a positivist method
for economics have put the question as if all that mattered were the logical
persuasiveness of their proposals, and not so much the interests of the people
which the proposals affect. In this sense, they have actually preached with
the voice of the methodologist, rather than from the attitude of a positivist economist like the one they pretended to be.
209
However, the assumption of single-minded pursuit of self-interest by

individuals in competitive industries (and the academia can be taken to
be such an industry) has worked very well in a wide variety of hypotheses, and it is therefore likely to seem reasonable to us, economists, that it
may work in this case as well. The epistemologist or methodologist, on the
other hand, is accustomed to a very different kind of model for explaining
the spread and the working of scientific ideas, in part because he uses a
different body of indirect evidence (the logical connections between
ideas, rather than the phenomena of interdependent choice economics is
usually about), but of course, neither the evidence of the economist nor
that of the epistemologist is conclusive: the decisive test of any hypothesis
is just whether it works for the phenomena it purports to explain. So, I
propose to analyze the question about the means to establish an optimum economic method from the point of view of economic theory itself.
Seen from this economic viewpoint, a fundamental problem for the
instauration of an appropriate economic method is that of asymmetric
information: the people for whom the right method is appropriate (the
citizens) are not those who can better appreciate what the power of each
alternative method is (the economists). The desideratum for the
demanders of economic knowledge (the citizens/principals) would be
that of designing a contract so that the suppliers (the economists/agents)
worked efficiently in the production of items of knowledge having the
highest possible utility for the former. The principals main difficulty,
however, is that they often cannot control by themselves whether the
contract is being obeyed by the agents or not, and usually they even ignore
the issue of how to design something like an optimum contract; they
have trusted some economists to design it, and this clearly leads to a
vicious circle, for how are we to know who are the good economists in
the first place?
We also need to take into account the fact that the citizens/principals do
not form a homogeneous group: a few of them have much more political
power than the rest, and all have interests, values, and beliefs that are in
conflict with those of many others. Therefore, collective action problems
arise as well. For example, perhaps the people who had the capacity to
enforce an optimal contract in economics would prefer that no real
empirical progress were made in this field, in order to keep using the aura
of mathematical certainty of some branches of economics to justify before
the public some policies designed to favor the most privileged. This
possibility well documented in the history of our discipline and in
general the plurality of interests existing within our societies, suggests that
this plurality should be somehow reflected within the optimal contract.
210
In this situation, the economists/agents are forced, if they want to gain the
principals trust, to establish the contract as an agreement among themselves, as a disciplinary contract, so to speak, and to use this consensus as a
signal of the quality of their output. Unfortunately, this also opens the
door for moral hazard problems: since the consensus among experts is
often the only signal that users may have of the quality of a researcher or of
the knowledge produced by her, the losers in the race for a discovery
could simply deny that real knowledge has been attained by other colleagues, preventing as a result the constitution of that consensus. The
possibility also exists that a majority of the agents sign a contract which
is efficient for them, but not for the citizens. In principle, the interests of the
economists may be incompatible with those of the citizens for two reasons
at least: first, scientists may sometimes have an opportunity to attain their
goals without producing real discoveries (for example, by inventing their
data), and second, they can devote their effort to fields or problems which
are of no obvious value for the people supporting science (for example, by
looking just for elegant equilibrium solutions). The disciplinary contract of scientific research has to be seen, then, as a set of norms which
would ideally make researchers behave in an honest way toward the
citizens (producing useful knowledge) and toward their colleagues (e.g.
acknowledging their discoveries when it is due).
I want to suggest that a good task for future economists will be to devise
some alternative, hypothetical disciplinary contracts for the practice of
economics, exploring their properties both analytically (by way of game
theory, economics of information, social and public choice, etc.) and, as far
as possible, empirically (by way of institutional and experimental economics,
as well as history of economics), and subjecting these proposals to serious
discussion and criticism not only by the economists themselves, but also by
political or civilian organizations. To some extent, the result of this discussion
would count as an empirical test of the positive predictions of those hypothetical contracts about the normative judgments they would induce. Many
economists will certainly doubt that this task is necessary at all, for they are
persuaded that, in the long run, academic competition leads by itself to the
adoption of a right disciplinary contract, as much as political competition
tends to select those parties who choose the right economic advisers. But
the progress experienced during the last fifty years in our knowledge of how
to efficiently solve peoples main economic problems has been so tiny, and
the reasons are so clear namely that, realistically, both academic and
political competition can lead to stagnation rather than to progress that it
would be worth the effort to devise a new disciplinary contract for economics.
P. Menard
211
References
Armatte, M. (1995). Histoire du modle linaire. Formes et usages en statistique
et conomtrie jusquen 1945. Unpublished doctoral dissertation, EHESS
Balisciano, Mrcia L. (1998). Hope for America: American notions of economic
planning between pluralism and neoclassicism, 19301950. In Mary S.
Morgan and Malcolm Rutherford (eds.), From Interwar Pluralism to Postwar
Neoclassicism, annual supplement to History of Political Economy, 30. Durham,
NC: Duke University Press
Barber, W. J. (1985). From New Era to New Deal. Herbert Hoover, the Economists, and
American Economic Policy, 19211933. New York: Cambridge University Press
Bloor, D. (1982). Durkheim and Mauss revisited: classifications and the sociology
of knowledge. Studies in History and Philosophy of Science, 13(4), 26797
Brinkley, A. (1995). The End of Reform: New Deal Liberalism in Recession and War.
New York: Vintage Books
Despres, E., M. Friedman, A. Hart, P. Samuelson, and D. Wallace (1950). The
problem of economic instability. American Economic Review, 40, 50538
Desrosires, A., E. Lie, M. Mespoullet, and E. Didier (2001). Sampling Humans.
Berlin, Max-Planck-Institut fr Wissenschaftsgeschichte
Duncan, J. and W. Shelton (1992). US government contributions to probability
sampling and statistical analysis. Statistical Science, 7(3), 32038
Friedman, Milton (1946). Lange on price flexibility and employment. American
Economic Review, 36, 61331. Reprinted in Essays in Positive Economics
(1953a)
(1947). Lerner on the economics of control, Journal of Political Economy, 55,
40516. Reprinted in Essays in Positive Economics (1953a)
(1948). A monetary and fiscal framework for economic stability. American
Economic Review, 38, 24564. Reprinted in Essays in Positive Economics (1953a)
(1949). Rejoinder to Professor Friedmans proposal: comment. American
(1951). The effects of a full-employment policy on economic stability: a formal
analysis. Reprinted in Essays in Positive Economics (1953a)
(1953a). Essays in Positive Economics. Chicago: University of Chicago Press
(1953b). The methodology of positive economics. In Essays in Positive
Economics. Chicago: University of Chicago Press, 343
(1975). Theres No Such Thing as a Free Lunch. LaSalle, IL: Open Court
Publishing Co.
(1967). Value judgments in economics. In S. Hook, Human Values and
Economic Policy. New York: New York University Press, 8593. Reprinted
in K. Leube (ed.), The Essential Friedman. Stanford: Hoover Institution, 38
(1986). Economists and economic policy. Economic Inquiry, 24(1), 110
(1991). Old wine in new bottles. The Economic Journal, 101, 3340
(1992). Do old fallacies ever die? Journal of Economic Literature, 30, 212932
Friedman, Milton and Rose Friedman (1998). Two Lucky People. Memoirs.
Chicago: University of Chicago Press
Friedman, Milton and Simon Kuznets (1945). Income from Independent Professional
Practice. New York: National Bureau of Economic Research
212
Hammond, J. Daniel (1993). An interview with Milton Friedman on methodology. In Bruce C. Caldwell (ed.), The Philosophy and Methodology of
Economics, vol. 1. Aldershot: Edward Elgar, 21638
Hotelling, H. (1940). The teaching of statistics. Annals of Mathematical Statistics.
11, 45771
Hotelling, H., W. Bartky, W. E. Deming, M. Friedman, and P. Hoel (1948). The
teaching of statistics [a report of the Institute of Mathematical Statistics committee on the teaching of statistics]. Annals of Mathematical Statistics, 19, 95115
Kneeland, H., E. Schoenberg, and M. Friedman (1936). Plans for a study of the
consumption of goods and services by American families. Journal of the
American Statistical Association, 31, 13540
Knight, F. H. (1924). The limitations of scientific method in economics. In
R. Tugwell, (ed.), The Trend in Economics. New York: Alfred A. Knopf,
22967. Reprinted in F. Knight (1935), The Ethics of Competition and Other
Essays. New York and London: Harper & Brothers, 10547
(1932). The newer economics and the control of economic activity. Journal of
Political Economy, 40(4), 43376
(1991). The case for communism: from the standpoint of an ex-liberal. In
W. J. Samuels, (ed.), Research in the History of Economic Thought and
Methodology [Archival Supplement]. Greenwich, CT: JAI Press, 57108
Mirowski, Philip (1989). More Heat Than Light: Economics as Social Physics. New
York: Cambridge University Press
National Resources Committee (1938). Consumer Incomes in the United States.
Washington, DC: United States Government Printing Office
(1939). Consumer Expenditures in the United States. Washington, DC: United
States Government Printing Office
Neyman, J. (1938). Contribution to the theory of sampling human populations.
Journal of the American Statistical Association, 33, 10116. Reprinted in A
Selection of Early Statistical Papers of J. Neyman. Cambridge: Cambridge
University Press, 1967
Olkin, I. (1991). A conversation with W. Allen Wallis. Statistical Science, 6(2),
12140
Pradier, P. C. and D. Teira (2000). Frank Knight: le risque comme critique de
lconomie politique. Revue de Synthse, 121(4), 79116
(2002). Frank Knight y los positivistas. In Wenceslao Gonzlez et al. (eds.),
Enfoques filosfico-metodolgicos en economa. Madrid: FCE, 10741
Reid, Constance (1998). Neyman from Life. New York: Springer Verlag
Schoenberg, E. and M. Parten (1937). Methods and problems of sampling presented by the urban study of consumer purchases. Journal of the American
Statistical Association, 32, 31122
Seng, Y. P. (1951). Historical survey of the development of sampling theory and
practice. Journal of the Royal Statistical Society, 114, 21431
Shoup, C. S., M. Friedman, and R. P. Mack (1943). Taxing to prevent inflation;
techniques for estimating revenue requirements. New York: Columbia University
Press
Stigler, S. (1994). Some correspondence on methodology between Milton
Friedman and Edwin B. Wilson. Journal of Economic Literature. 32, 1197203
213
(1996). The history of statistics in 1933. Statistical Science, 11(3), 24452

Teira, David (2004). Azar, economa y poltica en Milton Friedman. Biblioteca
Virtual Miguel de Cervantes, Alicante
(2007). Milton Friedman, the statistical methodologist. History of Political
Economy, 39(3), 51128
Tugewell, R. (1932). The principle of planning and the institution of laissez faire.
American Economic Review, 22(1), 7592
Wallis, W. A. and M. Friedman (1942). The empirical derivation of indifference
functions. In O. Lange et al. (eds.), Studies in Mathematical Economics and
Econometrics. Chicago: University of Chicago Press, 1758
Wallis, W. A. and H. V. Roberts (1956) Statistics a New Approach. New York:
Free Press
Zamora Bonilla, J. (2002). Scientific inference and the pursuit of fame: a contractarian approach. Philosophy of Science, 69, 3003
Part 4
Theoretical context: firm, money, expected utility,

Walras and Marshall
Friedmans 1953 essay and the marginalist

controversy
Roger E. Backhouse
Introduction
The marginalist controversy of the 1940s was arguably one of the debates
through which modern microeconomic theory that underlying Paul
Samuelsons neoclassical synthesis was established (Backhouse 2003).
Milton Friedmans The methodology of positive economics (1953)1
played a major role in the resolution of that controversy and many of the
ideas on which the essay was based arose out of it. The marginalist controversy, narrowly defined, took place in 19467, sparked by the challenge
to marginal theory made by Richard Lester, a labor economist. Section 2
outlines this controversy, picking out aspects that are relevant to understanding Friedmans essay. However, this controversy formed part of a
much broader, more complicated debate over price theory and the theory
of the firm, drawing on and responding to the full-cost pricing debates of
the 1930s (see Lee 1984; Lee and Irving-Lessman 1992; Mongin 1992;
Hausman and Mongin 1997). It also fed into further debates on the theory
of the firm that took place in the first half of the 1950s. These are discussed
in sections 3 to 5. Section 6 then considers the way in which Friedman
responded to the challenge posed by this controversy, drawing on his
correspondence with Don Patinkin over the period 1948 to 1950. This
leads, in section 7, into a discussion of F53 in relation to these controversies.
The marginalist controversy of 19467
Lester (1946) approached the subject of marginalism as a labor economist

aware of a large gap between the way the labor market was treated in price
theory and in macroeconomics.2 In the former, labor demand depended
on the wage rate and the marginal product of labor but in the latter it was
1
2
Referred to simply as F53 from now on.

It has been argued that labor economics remains a more empirical field than many others in
economics. See Oswald and Trostel (2000).
217
218
Roger E. Backhouse
determined by the level of aggregate demand. Lester sought to bridge this

gap by drawing together a wide variety of evidence on the way the labor
market worked. He started his article by drawing attention to questionnaire evidence that businessmen saw output as more important than wage
rates in their employment decisions. However, this was arguably the least
important amongst the pieces of evidence that he cited. (1) He cited a
wide variety of evidence showing that average variable cost decreased up
to full capacity, after which it rose very sharply. (2) The Oxford economists had established that businessmen were ignorant of the elasticity of
demand. (3) Because wage rates differed across regions in the USA, the
imposition of minimum wages after 1939 had caused wages to rise differently in different industries. Lester found that, where wages had risen
most, so too had employment. Minimum wages had not had the effect on
employment that marginalist theory predicted. (4) He asked businessmen
how they would respond to rises in wages (a phenomenon of which they
had recent experience), finding that employment would not be reduced:
instead, firms would improve efficiency, introduce labor-saving machinery, and increase their sales effort. His conclusion was that, for a wide
variety of reasons, businessmen did not adjust employment in relation to
wages and productivity in the way that marginal theory required.
The challenge to marginalist theory was taken up by Machlup (1946),
the source of some of the arguments in F53. However, Machlups article
was much more than a response to Lester: it was a wide-ranging defense
of marginalism and an attack on the empirical research that had been
undertaken to refute it. He sought to clarify the meaning of marginalist
theory, which he claimed was much misunderstood, and he criticized both
Lesters questionnaire evidence and his statistics. Differences between
economists and accountants language meant that detailed interviews
were needed to back up questionnaires and to probe the possibility that
ethical beliefs or the law influenced the way in which businessmen
rationalized their output decisions. Statistical work on the relationship
between wage rates and employment was nearly useless because
we have no way of eliminating the simultaneous effects of several
other significant variables, especially those of a psychological nature
(Machlup 1946, 548).
Lester (1947) chose to reply to Machlup and to Stigler (1947), on
minimum wage rates, eliciting rejoinders from both of them. Beyond
this, Lester did not respond, though his position was defended against
Machlup and Stigler by Oliver (1947). One of the things revealed by
Lesters reply is the extent to which he believed that theory should be
empirically based.
219
At the heart of economic theory should be an adequate analysis and understanding

of the psychology, policies and practices of business management in modern
industry. (Lester 1947, 146)
Awareness of institutional differences led him to criticize marginalists for

treating all markets as though they were the same.
Reasoning about labor markets as though they were commodity markets seems to
be an important explanation for erroneous conclusions on such matters as minimum wages. (Lester 1947, 146)
This was in complete contrast to the view expressed by Machlup

(1947, 519).
Economics in a narrow sense is confined to such aspects of conduct as can be
explained with reference to the principles of maximizing satisfaction, income, or
profit. Under definitions of this sort any deviations from the marginal principle
would be extra-economic.
It is impossible to imagine Lester making such a remark. For Machlup, it

clearly established the presumption that maximizing behavior should be
assumed unless there were overwhelming evidence that a phenomenon
could not be explained in this way. In contrast, Oliver (1947, 379),
supporting Lester, suggested that normal rules of debate implied that,
if business men say that they think in average-cost terms, then, if the burden of
proof rests on anyone at all (and it should not in economic discussion), it rests on
the marginalists who do not believe the business man, rather than on the economic
iconoclasts who (with modifications of varying importance) take him at his word.
Oliver provided a list of phenomena that, he claimed, were inconsistent

with marginalism and argued that the onus was on marginalists to provide
evidence that was consistent with their theory and not with any other.
Lester emphasized the quality of his evidence: he claimed that his questionnaire results were replicable, that results could be cross-checked and
that they could be tested against other studies. Machlup, he contended,
argued for a subjective interpretation of profit maximization (maximization given perceptions of costs and revenues) but when it was shown that
businessmen did not think this way, he disputed the evidence as inconsistent with his prior beliefs.
Lesters empiricism also emerged in his exchange with Stigler. His main
criticism of Stiglers analysis was that he did not understand how wages
were determined, how labor markets operated, or how business decisions
were made (Lester 1947, 142). Stigler had ignored the wealth of experience with minimum-wage laws over thirty or forty years, giving his paper
a pre-World War I flavor (Lester 1947, 143). Lester, supported by
Oliver, was thus arguing that theories should, somehow, be testable and
220
Roger E. Backhouse
that they should be tested. Their contention was that Machlup was
defending marginalism by interpreting it so broadly that it was irrefutable,
and that Stigler was simply ignoring evidence.
Stigler (1947, 1545) accused Lester of being confused about what he
meant by marginal analysis. Did he mean that the logical structure of
formal theory was invalid, or did he just question the theorys empirical
content? Use of the phrase marginal analysis, he contended, suggested
the former. More serious (given that any reader of Lester would know he
was criticizing the empirical relevance of marginalism) was a confusion
over whether or not perfect competition was being assumed. Stigler
(1947, 155) took Lester to task for claiming that marginal productivity
theory depended on the assumptions that agents were price-takers
and that monopoly power was absent. Yet, rather than citing Robinson
(1933) or Chamberlin (1933), surely standard works by that time, to draw
attention to the issue of imperfect competition, he poured scorn on Lester
for criticizing the stupid assumption, that no economist had ever held,
that an employer in an imperfect market would reduce his price to
increase his sales if it increased his short-run profit. The result was that
the issues of profit maximization and perfect competition were not clearly
distinguished.
This exchange provides the immediate background to F53, which later
took up several of Machlups arguments. Though it concerned the relevance of profit maximization, many other issues were involved, such as
attitudes toward imperfect competition and the criteria by which theories
should be appraised. These can be understood only by considering earlier
debates. Because of their variety, they have to be considered separately,
though there were, of course overlaps.
The Cambridge and Harvard responses to Marshall
To understand developments in the theory of the firm and market structure from the 1920s to the 1950s, it is essential to start with Alfred
Marshalls theory, as developed in his Principles of Economics ([1890]
1920) and Industry and Trade (1919).3 The most important feature of
his theory was that it was not an equilibrium theory: it was an attempt to
analyze a world in which technical progress led to continuous change; in
which knowledge was limited and organizational forms were changing
(see Raffaelli 2003, 2004). He criticized the use of static mechanical
analogies, preferring the use of biological ones, such as that of the trees
3
OBrien (1984) provides the best discussion of this. See also OBrien (1983 and 1990).
221
in the forest. Every tree in a forest was changing, but the forest might
remain the same (cf. Marshall [1898] 1925).
Confusion has arisen because although Marshall used mathematical arguments, he did not confine himself to things that he could
express using mathematics. Exact but limited knowledge needed to be
combined with broader estimates that rest on uncertain foundations
(Marshall 1919, 676). He drew an analogy with navigating a ship in waters
with uncertain currents: knowing the ships precise location was impossible, but it was important to ensure that further errors were not introduced. In his theory of the firm and the market, profit maximization
provided the exact theory of the firm, and competitive supply and demand
provided the theory of price. However, these did not constitute his theory,
for they were merely the skeleton around which he hung a more realistic
theory that encompassed many factors that were not covered by the
mathematics. On the demand side, these factors included changes in
human character; on the supply side, the effects of time, imperfect knowledge, technical change, and pricing policy.
Marshalls firms were unable to sell as much as they wished at the
prevailing market price (competition was not perfect) but neither did
they face a given demand curve: they engaged in marketing, which
might account for a large proportion of total costs and of capital, and
they might use temporary price cuts as a way to shift the demand curve
outwards. Increasing returns to scale were widespread and were hard to
distinguish from technical progress: returns to scale might not be reversible, with the result that cost curves depended on time as well as on
output. In addition, firms were continuously evolving, changing the way
they were organized and the shape of their cost curves. Even if the industry
were in equilibrium (in the sense that its price and output were not
changing) individual firms would typically not be in equilibrium: they
would be entering, exiting, and changing the scale of their operations. The
representative firm was not the average firm but the firm that was relevant
in a particular context. Supply and demand were used as conceptual tools
even though Marshall did not confine himself to competitive equilibrium
analysis. For Marshall (1919, 512), general propositions in regard
to either competition or monopoly are full of snares: the most extreme
forms of competition aimed to undermine competition and had to be
restrained.
Faced with Marshalls industrial analysis, economists moved, during
the 1920s and 1930s, in different directions. Though the labels are an
oversimplification, the main ones that need to be considered can be called
the Cambridge, Harvard, institutionalist, and Oxford approaches. There
were also many other writers whose work fits into none of these categories,
222
Roger E. Backhouse
such as Friedrich Hayek, John Hicks, or Fritz Machlup.4 In brief, the

Cambridge approach was to develop a static equilibrium theory that
eliminated the inconsistencies that Marshalls theory was perceived to
contain. The Harvard approach was also to develop equilibrium theory,
though in a more Marshallian way, taking account of industrial developments that were less important when Marshall wrote his Principles. The
institutionalist and Oxford approaches involved collecting evidence on
short-run price-setting behavior and building theories that might account
for this behavior. Though it is the non-marginalist theories developed
by the last two groups that formed the immediate background to the
LesterMachlupStigler exchange, to understand the controversy and
Friedmans response to it, it is important to realize that at least four
approaches to the firm and market structure were being debated.
The Cambridge approach was rooted in the so-called cost controversy of the 1920s during which a number of criticisms were raised
concerning Marshalls price theory. John Clapham (1922) accused
Marshalls categories of increasing and decreasing returns industries of
being empty boxes: no one had established which industries fell into
which category. More important, Piero Sraffa (1926) raised two objections to Marshallian theory. The first was Marshalls failure to solve the
problem of making increasing returns to scale compatible with perfect
competition. Assuming economies of scale that were external to the firm
but internal to the industry might be theoretically acceptable but they were
the class of returns most seldom to be met with (Sraffa 1926, 540).5 His
second objection was the incompatibility of partial equilibrium analysis
with decreasing returns to scale. In the long run, costs could rise with
output only if factor prices rose, violating the ceteris paribus assumption
needed for partial equilibrium analysis. Sraffa suggested that these problems be solved by moving toward a theory of monopoly. Alongside this
were criticisms of the Marshallian theory of the firm in particular of the
representative firm. Marshalls representative firm was a way to handle
the problems arising from having many types of firm in an evolutionary
context. If markets were viewed in terms of static equilibrium it was, not
surprisingly, taken to be the average firm and was arguably redundant.
Lionel Robbins (1928) and Pigou (1928) both attacked Marshalls
concept, paving the way for the replacement of Marshalls realistic,
evolutionary theory with a theory of static equilibrium.
4
5
For discussion of other approaches see OBrien (1983, 1984).

He was clearly not convinced by Marshalls extensive work on industrial districts, where he
found extensive evidence for such returns.
223
The marginal revenue curve (e.g. Harrod 1930) and the theory of imperfect competition (Robinson 1932, 1933) emerged out of this destruction of
Marshallian price theory. Marginal revenue was implicit in Marshalls (and
Cournots) first-order conditions for equilibrium under monopoly, but
around 1930, economists introduced the label, enabling it to be placed
against marginal cost. They also noted the point (known to Cournot) that if
the demand curve were not horizontal, there would be no supply curve:
supply would depend on the elasticity of demand as well as on price. The
geometry of marginal revenue, marginal cost, average revenue, and average
cost was provided by Robinson, establishing the theory of the firm found in
modern introductory textbooks. She labeled the general case, encompassing both monopoly and competition, imperfect competition. Firms
faced downward-sloping demand curves and faced a large number of
competitors, but freedom of entry and exit eliminated monopoly profits,
pushing the firm to a point where its demand curve was tangential to the
downward-sloping portion of its average cost curve. On the assumption that
downward-sloping demand curves were the result of advertising, not substantive differences between products, she drew conclusions about the
waste and inefficiency caused by having output lower than that at which
average cost was minimized, and about the exploitation of labor caused by
the real wage being less than labors marginal product. In this sense she was
working in the Pigovian welfare economics tradition.
The central figure in the Harvard approach to the theory of the market
structure is Edward Chamberlin. Though his work has often been bracketed with Robinsons, its content and background were very different.
Frank Knights Risk, Uncertainty and Profit (1921) had provided a very
strict definition of perfect competition, involving perfect information.
John Maurice Clarks The Economics of Overhead Costs (1923) had explored
the implications of overhead costs for business decision-making. There
was also the influence of Allyn Young, his PhD supervisor at Harvard.
Chamberlin, in the thesis that became The Economics of Monopolistic
Competition (1933), was concerned not to repair or replace Marshallian
theory but to extend it to encompass features of business life that were
barely developed when Marshall wrote his Principles in the 1880s, notably
advertising and product differentiation. His firms chose not just price and
output, but also advertising (selling costs) and the degree of product
differentiation. He also devoted considerable attention to oligopoly.
Chamberlin did not think, like Robinson, in terms of monopoly versus
competition. His operated with a far richer menu of market structures,
classifying them in at least three dimensions: product differentiation;
advertising expenditure; and the number of firms. Monopolistic competition was simply one of many market structures in a range that included
224
Roger E. Backhouse
different types of oligopoly. Taking account of the fact that knowledge

might not be perfect, he distinguished between pure and perfect
competition, a distinction that did not appear in Robinsons work. He
developed a theory with two demand curves for each firm, one which
applied if other firms matched a price change, the other if they did not.
Using these two curves he was able to explain how price might be determined under oligopoly, and the processes whereby markets adjusted to
changes in the underlying conditions. In these processes, firms would
adjust not just price and output but also selling costs and product
differentiation.
The Harvard approach to the theory of the firm and market structure
may have touched the Cambridge one with the theory of imperfect/
monopolistic competition but just as it had different origins it led in
different directions. The most important was the project of extending
Chamberlins theory into a general theory of industrial structure (Mason
1939; Bain 1942, 1956). The number of dimensions to industrial structure was increased to include durability of the product, the degree of
vertical integration, and other aspects of market structure. This was
assumed to influence firms conduct, and hence economic performance.
This established an empirical research agenda looking at relations
between industrial structure and measures of performance across industries. Though rooted in Chamberlins work, Mason and Bain laid the
foundations for a thoroughly empirical analysis of industrial organization that was capable of a range of theoretical interpretations.
4
Institutionalism and Oxford
During the 1930s, in response to the Great Depression, economists

questioned whether prices were in practice more rigid than conventional
theory would suggest. In the United States, Gardner Means (1935) drew a
sharp distinction between market prices (as in the market for wheat or
cotton) and what he called administered prices. These were set by firms
and maintained for considerable periods of time. They might occur in
either competitive or monopolistic markets. He found evidence that
goods could be divided into two clear categories: in one prices changed
frequently (nine times or more every year) and in the other very infrequently (less than three times a year). A quarter of all prices changed less
than once every ten months. This was far more stability than could be
expected if all prices were subject to the vagaries of supply and demand. Its
relevance for the depression was that output fell most rapidly in markets
with stable prices, and held up in markets where prices were flexible. The
ratio of flexible to rigid prices was highly correlated with the index of
225
industrial production in the late 1920s and early 1930s (Means 1936, 27).
His policy conclusion was not that businesses should be broken up to
make prices more flexible (this would drastically reduce wealth), but that
monetary policy should be used to keep (flexible) market prices in line
with those prices that were more stable.
Meanss work was part of a much broader interest in pricing policies
in the United States. The National Recovery Act (1933) led to the production of much statistical data, generating a wide range of responses to
what was discovered.6 Inflexibility of prices was believed to be important
and Roosevelt established a Cabinet committee on price policy in 1934.
Walter Hamilton undertook a series of investigations on prices, and Means
continued his work after 1935 as Director of the National Resources
Committee. Price policy was very much in the air in the 1930s. Most of
the empirical work was very much in the institutionalist tradition, which
spurned thinking about abstract individuals in favor of grounding theory
firmly in empirical data (cf. Rutherford 1999).
In Britain, parallel work was done by the Oxford Economists Research
Group (OERG). Harrod (1939, 12) introduced this by pointing out that
the theory of the firm presumed full knowledge, an assumption that was
formally incorrect. There was thus scope for empirical work to establish
whether or not firms short-run behavior deviated systematically from
short-run profit maximization. He wanted to establish the principle that
there was scope for the method of direct question (1939, 3). The OERG
set out to question businessmen about their knowledge of costs and
demand, and the methods by which they set prices in the short run.
The results of these surveys, described by Hall and Hitch (1939, 18)
was that businessmen overwhelmingly followed what they called fullcost pricing.
An overwhelming majority of the entrepreneurs thought that a price based on full
average cost (including a conventional allowance for profit) was the right price,
the one which ought to be charged.
The reasons why Hall and Hitch believed that firms followed this policy
were that they did not have the information needed to maximize short-run
profits and, more important, that many markets were oligopolistic. In
such markets, understandings about prices might have evolved (some
trade bodies published price lists that could be followed) or they might
involve basing prices on the full-costs of a dominant firm. Firms feared to
lower price below full-cost because they feared competitors would cut
their prices too, or because they believed demand to be very inelastic.
6
See Lee (1984, 1109ff) for a discussion of this evidence and responses to it.
226
Roger E. Backhouse
They did not raise prices because they expected competitors would take
business away from them. They formalized these beliefs in the kinked
demand curve, the origins of which can be found in the work of
Chamberlin and Robinson.
Full-cost pricing, as presented by Harrod, Hall, and Hitch, presented
only a limited challenge to conventional theory, taken to be the Chamberlin
model of monopolistic competition. They were aware that full-cost pricing
was consistent with profit maximization if the mark-up bore the appropriate
relationship to firms perceived elasticities of demand. More important,
they did not question that profit maximization was an appropriate assumption for the long run. Harrod (1939, 3) wrote that the profit motive was
securely grounded: Any one seeking seriously to challenge this would be
regarded as a hopeless sentimentalist.7 Phrases such as taking goodwill
into account when setting prices implied concern for longer-run profits
(Hall and Hitch 1939, 18). Moreover, even in the short-run, they were very
circumspect in their treatment of theoretical results derived from profit
maximization and seeing themselves, to some extent, as filling the gaps.
For example, under oligopoly, all theory could say was that price would be
somewhere between the pure polypoly price (taking no account of other
firms responses) and maximization of joint profits (Hall and Hitch 1939,
30). Empirical work was needed to establish whether the assumption of
perfect knowledge, known to be incorrect, led to results that were systematically wrong.
From controversy over markets to controversy

over marginalism
In response to these concerns, there was, in the early 1940s, an explosion

of empirical work on cost curves and firms pricing policies. Because there
were so many approaches to the problem, we find several different debates
going on. For example, exponents of the Harvard approach, which sought
to relate price policy to variables describing industrial structure, were
skeptical about the value of Meanss correlations of price-stickiness with
fluctuations in output. Thus Mason (1939, 656) criticized the measures
of price used by Means, claiming that they failed to take account
of discounts and quality changes. At the other extreme, he criticized
Walton Hamilton (1938) for going too far in relating business decisions
to specific circumstances: it required knowing what Messrs Ford and
Firestone dreamed in the night before the morning of their big tire deal
7
So much for Marshalls economic chivalry!
227
(Mason 1939, 61). The Harvard approach, which sought to relate price
policies to specific indicators of market structure, lay in between these
approaches.
On Harvards second front, Mason (1939, 62) was critical of theorists
in what I have called the Cambridge approach: The extent to which the
monopoly theorists refrain from an empirical application of their formulae is rather striking.8 In a footnote to this sentence, he added,
presumably alluding to Robinson (1933, 1),
Some theorists, pursuing their analysis on a high plane, refer to their work as tool
making rather than tool using. A toolmaker, however, who constructs tools
which no tool user can use is making a contribution of limited significance.
Some knowledge of the use of tools is probably indispensable to their effective
fabrication.
(This is a remark that one could easily imagine coming from Marshalls
pen.9) When confronted with theory, he looks with favor towards
institutionalism.
The theory of oligopoly has been aptly described as a ticket of admission to
institutional economics. It is to be regretted that more theorists have not availed
themselves of this privilege. (Mason 1939, 645)
Harrods remarks, cited above, fit into this same general category, of
economists seeking to establish the legitimacy of empirical research to
augment the results of theoretical inquiry.
After the near simultaneous publication of The Economics of Imperfect
Competition and The Economics of Monopolistic Competition, there were
debates over definitions of concepts and how monopoly power might be
measured. Machlup (1937, 1939) was a contributor to these, seeking to
establish consistent terminology. He supported neither the Cambridge
nor the Harvard approaches; citing John Hicks in support, he argued that
monopolistic competition was less important than was generally thought.
However, he defended the theory against empirical attack. He dismissed
the value of questionnaire evidence, citing his own experience in business
(Machlup 1939, 2334). He argued that his partners, who were ignorant
of economic theory, did not reach conclusions that were significantly
different from his. His partners, he contended, behaved as sensible
economists suppose them to act even though they often rationalized
their decisions in a misleading way. The result was that an investigator
using questionnaires would have reached mistaken conclusions. He went
so far as to refer to an average-cost superstition, explicitly leaving it
8
His examples were Richard Kahn and Abba Lerner.
Cf. Pigou (1925, 427, 437).
228
Roger E. Backhouse
open whether this referred to non-profit-maximizing businessmen, or to

economists who believed that businessmen behaved in this way. However,
unlike in his later work, he defended aspects of the Harvard approach.
Examining the many dimensions of market structure would no doubt
show that many of these were irrelevant, but it was worth knowing that.
He went on (1939, 236) to provide a long list of things that government
could learn from the monopolistic competition literature.
In his 1939 paper, Machlup made some of the criticisms of empirical
and questionnaire research that were later to emerge in the marginalist
controversy proper. However, in 1939 they were side issues. The validity
of marginalist methods was but one amongst many issues in a debate that
was primarily about market structure and price determination. Similar
remarks could be made about the contributions of the OERG: Harrod
explicitly endorsed profit maximization and saw empirical research as
establishing the theorys applicability. Hall and Hitch presented full-cost
pricing as supplementary to standard theory, and were ambiguous about
whether it conflicted with profit maximization. It was only after Lesters
challenge in 1946 that the main issue came to be seen as marginalism
versus a non-marginalist alternative. After the marginalist controversy of
19467, the debate over the theory of the firm took on a different character. Debate over markets was transformed into debate over marginalism.
This was precisely the time when Friedman began working on F53.
The outcome of the debate, to which F53 contributed in no small
measure, was to marginalize the full-cost approach. Several economists
took up the task of developing the idea of full-cost pricing from a decision
rule that described firms pricing policies into an alternative to marginal
analysis. For example, in the United States, Wilford Eiteman (1945,
1947, 1948; cf. Lee 1984, 111618) put forward a critique of marginalist
theory based on the structure of firms costs. In multi-product firms,
marginal costs of individual products were not identifiable. Even when
they were, the typical pattern was for costs to fall up to full-capacity,
meaning that for a wide range of revenue functions there would be no
point below full capacity at which marginal cost and marginal revenue
would be equal. There would be a discontinuity in the marginal cost curve
at full capacity, and firms would produce at this point. Eiteman and
Guthrie (1952) used a questionnaire to ask businessmen about their
costs, finding that the overwhelming majority of the 366 firms that
responded believed that costs declined, reaching a minimum either at or
shortly before full capacity output, a result they thought inconsistent with
marginal analysis.
In Britain, the OERG, after being disbanded during the war, continued
its work. The key figure was Philip Andrews. He moved from the surveys
229
used by Hall and Hitch to detailed studies of individual industries: rayon,

dominated by Courtaulds, and the boot and shoe industry. On the basis of
this work, he developed a theoretical alternative to Chamberlinian
monopolistic competition theory (1949a, b; see Lee and Earl 1993,
134; Lee and Irving-Lessman 1992). He saw himself as moving away
from Chamberlins theory back to a Marshallian conception of the equilibrium of the industry. Industries typically comprised firms of different
size, with different cost conditions, and a major factor behind pricing
decisions was potential entry from firms that produced in related markets.
His studies of technology in both synthetic fabrics and footwear persuaded him that capital and labor had to be used in relatively fixed
proportions, causing costs to be constant up to full capacity. Firms typically aimed to produce at less than full capacity, not because of falling
marginal revenue but because they designed plant so as to allow some
spare capacity beyond their expected level of output (machinery might
break down, and over time they would hope for an increase in sales).
Prices would be set in terms of normal costs, similar in conception to
full costs. Because firms were concerned with the long run, they attached
great value to goodwill, and did not maximize short-run profits. Though
Andrews never fully developed his theory (Lee and Earl 1993, 223), he
was, unlike Hall and Hitch, clearly offering an alternative to marginalism.
Andrews had several supporters amongst more traditional Marshallians,
but the predominant reaction was one of hostility (Lee and IrvingLessman 1992).
However, whilst the case against marginalist theory was being developed in several directions, the evidence in favor of full-cost pricing was
also being reinterpreted in such a way as to defuse controversy and to rally
support round more orthodox, Chamberlinian theory. The major contributor here was Richard Heflebower (1955). This paper was presented
at a conference at Princeton University in June 1952 and offered a careful
survey of much of the literature, both American and British (Stigler,
Machlup, and Bain were present, but the main critics of marginalism
were not). He was considered by most economists to take the full-cost
arguments very seriously, but came down on the side of arguing that the
evidence was consistent with marginalist theory.10 He was skeptical about
the questionnaire evidence and argued that industry studies did not support full-cost conclusions. He concluded that profit margins were in
practice much less rigid than they appeared at first sight to be and that
firms deviated from their normal margins when they faced a competitive
10
The paper is useful as a source of references to the earlier literature.
230
Roger E. Backhouse
reason to do so (1955, 3835). This suggested that there was no inconsistency between full-cost pricing and marginalism.
Heflebower saw merit in full-cost pricing under some circumstances.
Under oligopoly, where rationality did not define a unique price, full-cost
pricing was persuasive and there was evidence that firms had followed
such policies during the 1930s. Full-cost pricing also made sense in
industries where price decisions had to be made very frequently, as in
industries such as metal casting, metal stamping and forging. In these, it
was important to have a formula that could be used to calculate prices that
might vary from one order to the next. Similar considerations applied in
retailing, though here he noted that the same retailer might charge very
different margins on different products. Margins on products were not the
same as overall margins. He accepted Harrods point that full-cost pricing
might be a way to achieve long-term profit maximization.
However, despite accepting many of the full-costers claims, in the last
resort he came down against full-cost pricing being important. In the
tradition of Chamberlin, Mason, and Bain, he emphasized the need to
take account of the peculiar circumstances of individual markets, expressing skepticism about any general rule. Good theories, he believed, had to
be historically situated and they needed to be dynamic. But rather than
seeing full-cost theories as a way of dealing with dynamics, he emphasized
the need to explain pricing in terms of the evolution of market structures.
A satisfactory theory must go beyond explaining the cost-price output relations for
a given product and must be prepared to appraise the conditions of choice among
other variables and the economic results of that choice. This requirement, like the
others set forth here, is not introduced for the sake of descriptive realism but
because it is essential for better prediction. (1954, 122)
This was the Harvard structureconductperformance approach,

expressed using different language. Conditions of choice meant market
structure and other aspects of the situation in which firms operated; economic results meant performance. Against the full-costers he argued the
need for a theory of why firms behaved as they did, going beneath observed
behavior to its causes; against Chicago (and presumably Friedman), he
argued that market structure had to be considered because it affected
predictions, not simply for the sake of being realistic.
Friedmans early reactions to the

LesterMachlupStigler exchanges
These controversies form the background to F53, which Friedman started

writing soon after the LesterMachlupStigler exchanges. Don Patinkin
231
had been a PhD student at Chicago, but left in 1948 to take a position
at the University of Illinois, after which he moved to Jerusalem. Their
correspondence, therefore, begins in 1948.11 In this correspondence,
Patinkin repeatedly criticizes Friedman for not following his own empiricist methodology and Friedman responded to Patinkin in much more
detail than he responded to Lester. This means that Friedmans letters to
Patinkin show clearly how he responded to empiricist claims such as
Lester and others were making.12
In the first letter, after some paragraphs on debt and Friedmans political philosophy, Patinkin turned to methodology.
Now too, some questions on your insistence that our theory should consist only of
statements that people act as if. You once used this argument to explain to me
that you felt the criticism made of the Chamberlin analysis of imperfect competition (viz. that the business man does not know the shape of the curves postulated
by the theory) was invalid, since all that has to be shown is that the business man
acts as if he knows these curves. On this I agree with you insofar as you restrict
yourself to the implication of the as if hypothesis which can be checked up
without referring to the assumption. For example, one such implication of the
theory is that profits are zero. This can be checked. But what about the implication
that the businessman will operate at the point where marginal costs and revenues
are equal? I see no way of checking up on this implication. The same problem
holds also for the position which substitutes subjective curves for objective
ones. (Patinkin to Friedman, October 25, 1948).13
Patinkin is here making the point that theoretical implications must be

testable, reminiscent of the LesterOliver objection that Machlup broadened marginalism to the point of tautology. However, it is worth noting
that Patinkin recalls having heard Friedman use the as if argument
against Chamberlin, not the full-costers. Friedmans reply refers to the
article that he was trying to write and which later became F53.
I have been trying to write some kind of an article on the general methodological
problem you raise, and maybe I ought to neglect that. I dont regard the statement
that marginal costs and revenues are equal as in any way an empirical implication
of an economic hypothesis. It is simply a restatement of the hypothesis or, if you
will, a mathematical deduction from it. It seems to me the kinds of implications
that one needs to check are of a different order. Thus, from the hypothesis that an
11
12
13
For a broader perspective on this correspondence, see Leeson (1998) and Backhouse
(2002). See also Hammond (1996) and Hirsch and De Marchi (1990).
Lester is rarely mentioned by name, but there is no doubt that they are discussing this
controversy. In one letter, Patinkin even offered Friedman a specimen examination
question, based on data on motives for saving, that exactly parallels Lesters questionnaire
on what determined employment.
All quotations from the FriedmanPatinkin correspondence are taken from the Don
Patinkin Papers, held in the Economists Papers Project, Duke University.
232
Roger E. Backhouse
entrepreneur will operate at the point at which marginal costs and revenues are
equal, one can infer that a change which is known to have increased his costs but not
to have increased his revenues will tend to reduce output. Thats an empirical
implication that one can check, and so on. Again, by substituting subjective curves
for objective ones, it is possible to take all the content out of a hypothesis and make it
not a hypothesis at all but a pure tautology. For example, if one says that he behaves
in such a way as to equate marginal costs and marginal revenue for subjective curves
which are subject to change at any time, then of course no observable phenomena
would be inconsistent with that hypothesis, hence the hypothesis has no content and
is of no interest. (Friedman to Patinkin, November 8, 1948)
Friedman saw MR=MC as equivalent to profit maximization. He ruled out

the idea of MR=MC as a means of making decisions: he implies that
applying it to decision-making would mean adopting a subjective interpretation of revenue and cost, rendering the theory untestable. In his next
letter, Patinkin pursued this point, asking whether it would constitute a
test of the theory to ask whether businessmen undertook projects only
when the extra profit they believed they would generate exceeded the extra
costs they expected to incur (Patinkin to Friedman, December 13, 1948).
A month later (January 11, 1949) he switched to the issue of firms giving to
charity: enlarging the concept of profit maximization to encompass public
esteem and prestige, he suggested, rendered the concept of profit meaningless. He drew the conclusion that in social science, we deal with subjects
who can express themselves in subjective ways. You run into the whole
problem of whether we can believe what a person says. This is the problem
that is obviously absent from the physical sciences. He thus objected to
Friedmans extending physical-science methodology to social science.
This letter elicited a clear statement from Friedman on two points: that
a theory is designed to explain only a certain class of problems, and the
role of introspection. Given the importance of these points they are worth
quoting in full.
I dont think there is any essential difference between us on the experiment about
testing the theory of maximizing profit. The point you make is, as I see it, that
maximizing profit cannot appropriately be considered the sole or exclusive concern of corporations. That is, not all observable behavior of corporations can be
rationalized as a consequence of strict single-minded profit maximization. This is
undoubtedly correct. I have no quarrel with you. The point is that the profit
maximization theorem or hypothesis does not say this. What it does say is that
there exists an important class of phenomena (not all phenomena) which can be
rationalized in terms of a single-minded profit-maximizing entrepreneur. Your
experiment does not provide any test of this. Thus, to use my favorite example, the
fact that a feather did not fall a particular distance in the time that would be
predicted by the vacuum formula is not universally applicable. It does not tell
you that the formula may not be applicable over a wide range of circumstances.
233
Similarly with respect to your point about physical sciences and social sciences. It seems to me that the two are fundamentally essential so far as methodology is concerned. The existence of human beings does not change any of the
fundamental methodological principles. It has two rather different effects. The
first of these is that it provides a new source of information namely, introspective observation. This information is extremely important and relevant and is
the one important advantage we get out of the fact that our science deals with
human beings. It seems to me the role this additional information can play is in
suggesting fruitful hypotheses, that is, it enters at the stage of deriving hypotheses. The second way that the existence of human beings affects our science is
that it introduces a fundamental element of indeterminateness. This is a point
that Knight has always made and I think quite correctly. In the physical sciences
the observer and the thing observed are separable and distinct. In the social
sciences they are the same. Thus in the social sciences the enunciation of a
result or a law for predicting the stock market perfectly may make that law
invalid. For example, if one were to state a technique for predicting the stock
market perfectly, you know that if that technique were believed in it would then
become wrong. Thats the real difficulty raised by human beings. (Friedman to
Patinkin, January 18, 1949)
Introspection comes in as a source of hypotheses: it is not relevant for

testing them. Patinkin (January 22, 1949) took up the point about marginalist theory referring only to an important class of problems. He argued
that it was important to specify exactly what properties that class of problems will have. Otherwise you are getting into a tautology. Unfortunately
for us, Patinkin then visited Chicago where, presumably, their discussion
continued face-to-face.
A year later (February 4, 1950), Patinkin returned to his concerns about
the conclusions reached by Machlup and Stigler.
Ive also reread the 19467 AER LesterMachlupStigler controversy on marginal
productivity. It has left me with mixed reactions, On the one hand, Lester had
made some definite empirical investigations. Admittedly, his technics were crude.
But I definitely do think that Machlup and Stigler dispense with his finding much
too glibly. Theres something else thats worrying me: Im afraid Machlup proves
too much. For him marginal productivity can explain everything, and hence (does
this sound familiar?) it can explain no thing. In particular, I should like to ask
Machlup your question: under what circumstances would you reject the marginal
productivity theory. I dont know if he could produce any (can you?); but even if he
did, I would be willing to bet that faced with such circumstances he would plead
special factors. The dammed thing is that its almost impossible to set up a
critical experiment in economics.
March 5, 1950
My God! Is that how long this letter has been lying here unfinished? In the interim
many things have happened. But before I get on to them, Id like to finish up the
previous discussion.
234
Roger E. Backhouse
What I was going to say is that one thing Stigler said in that marginal productivity argument sent shudders through me. That was his statement on the bottom
of p. 156 (AER, March 1947) that even he would not accept a certain result even
if Lester obtains it from 6,000 metal working firms. A similar statement is at
the bottom of p. 99, first column, in Stiglers review of Survey of Contemporary
Economics (JPE, April, 1949) where Stigler summarily dismisses an observation
of Reynolds on wage differentials. I dont know whos right or wrong on this; but it
does seem to me that Stigler must present some empirical evidence before sounding off like that. And in this respect I completely agree with Lester; we have had
some experience already on minimum wage legislation. Isnt it a sad reflection on
the sad state of economics that Stigler can write a straight textbook analysis of
minimum wage legislation without even feeling himself obligated to make any
empirical surveys. Stigler is just an example of what everyone does including
yours truly. I am merely bemoaning the type of training we receive that doesnt
make us horrified at casual empiricism. Thats all for the moment, though Id
like to get your reaction.
Here we find Patinkin putting to Friedman precisely the points made by

Lester, by way of criticism of Machlup and Stigler: that Machlup rendered
the theory tautological and that Stigler turned his back on evidence.14
Friedman (March 21, 1950) agreed with Patinkin, clearly differentiating
his position from those of Machlup and Stigler.
P. S. (2) Re LesterMachlupStigler controversy I agree thoroughly with you.
I say, a plague on both their houses. Lester et als criticism is beside the point, and
Machlups and Stiglers replies are beside point. Lester asks: Do businessmen
behave the way theorists say they do; M and S answer yes or if not, well change
our statement so as to empty it of content. Right point is: economic theorists do
not (or should not) say anything about how businessmen behave or how they make
their decisions. They should say: the consequences of such and such a ch[an]ge
can be predicted by treating bus. men as if they did such and such. Test of that
theory is whether predictions are correct, not what bus. men say they do. So Lester
is asking foolish question and M and S giving foolish answer. My best example
[word illegible] to illustrate point is to say that Lesters method of investigation is
14
In a letter to Martin Bronfenbrenner (February 4, 1950), Patinkin expressed himself even

more strongly. Referring to Lesters empirical attempt, he wrote: One cannot merely
wave it away with the hand, as do Machlup and Stigler. Furthermore, I think these
gentlemen especially former prove too much. According to their formulation, the
marginal productivity theory proves everything, and hence nothing. In particular, I should
like to ask Machlup what empirical findings he would take as evidence that the marginal
productivity was incorrect. I am sure that what would happen is that he would not be able
to cite a case, and even if he did, if he were ever confronted with empirical evidence that
this case had actually happened, he would find some reasons for calling it special circumstances and hence not refuting the theory. Stigler is even more doctrinaire. His statement
that certain of Lesters empirical findings he would not accept even if presented with
6,000 cases sends shudders down my spine. Is that the way a supposedly empirical
science is supposed to be presented? [Some spellings have been corrected.]
235
like trying to figure out causes of longevity by asking old people to what factors they
attribute their longevity. (Friedman to Patinkin, March 31, 1950, handwritten)
Friedman thought Lesters questions were misconceived because they

reflected a misconception about the way theory should be used; and
Machlup and Stigler were wrong to dismiss evidence so completely.
This exchange clarifies Friedmans rejection of any idea that theory
should be trying to explain business decisions. Such decisions depended
on subjective factors and were therefore untestable. His reference
(January 18) to Knights principle of indeterminacy suggests that he
thought this a fundamental problem, which helps account for his consistently held view that the attempts of Lester and others to answer this
involved asking foolish questions.15 Friedmans comment that there was
a class of problems for which marginalist theory provided the answer
was completely beside the point: few of the full-costers would have
disagreed. Their aim was to find a way to solve problems that it could
not answer.
F53 as a response to the marginalist controversy
Friedman (F53, 16) used the 19467 marginalist controversy as the

motivation for what he described as a more extensive discussion of
methodological principles involved than might otherwise seem appropriate. In his brief discussion of the controversy, as in his letter to
Patinkin, he criticized both sides, Machlup and Stigler as much as Lester.
The articles on both sides of the controversy largely neglect what seems to me
clearly the main issue the conformity to experience of the implications of
marginal analysis and concentrate on the largely irrelevant question of whether
businessmen do or do not reach their decisions by consulting schedules, or curves,
or multivariable functions showing marginal cost and marginal revenue (F53, 15;
emphasis added).
Friedman is here criticizing both those (Lester) who questioned the

relevance of marginal analysis and those (Machlup and Stigler) who
supported it. By focusing on the 19467 marginalist controversy, in
which Machlup and Stigler ignored these issues, Friedman was led to
focus on one among many issues involved in the controversy over full-cost
pricing.
Friedmans explicit criticisms of anti-marginalist writings came later in
his essay (F53, 313). He echoed Machlups dismissal of questionnaire
15
If fundamental uncertainty played such an important role in Friedmans thinking, it is

worth noting the parallel with Keynes.
236
Roger E. Backhouse
evidence but his main focus was on the desire for a realistic theory rather
than one that predicted better than an alternative theory for as wide a
range of phenomena. One would not guess from this dismissal of this work
that behind the controversy lay the goal of explaining the large amount of
evidence on the behavior of prices during the 1930s and 1940s. In this
literature was much work that could be construed as finding a theory that
could predict firms price policies (understood to refer to the prices set by
firms in relation to relevant circumstances). Full-costers claimed that fullcost pricing had predictive power, whereas their opponents claimed it had
not. Friedman is not explicitly dismissing full-cost so much as rejecting
one (but only one) of the arguments used to support it.
However, although Friedman motivated F53 in this way, it was both
more and less than an attack on the full-cost pricing literature. It was more
in that he had other targets; it was less in that he addressed only a few fullcost arguments. Another important target, arguably his main one, was the
theory of monopolistic competition. He objected to the claim made by
Chamberlin (1957) and Triffin (1940)16 that monopolistic competition
might prove the basis for a more general theory of value. The theory,
he claimed, possesses none of the attributes that would make it a truly
useful general theory (F53, 38). It was useful in that it refined Marshalls
theory, thereby enrich[ing] the vocabulary available for describing industrial experience. His most substantial criticism was that whilst it might be
useful in analyzing firms or even the economy as a whole, it could not offer
any analysis of the industry. Once one admitted product differentiation,
every firm produced a unique product.
The theory of monopolistic competition offers no tolls for the analysis of an
industry and so no stopping place between the firm at one extreme and general
equilibrium at the other. It is therefore incompetent to contribute to the analysis of
a host of important problems: the one extreme is too narrow to be of great interest;
the other, too broad to permit meaningful generalizations. (F53, 39)
The solution, he contended, was to adopt Marshalls practice of dividing

firms into industries, such that the similarities among the firms in each
group were more important than the differences among them (F53, 35).
Here, it is important to him that Marshall did not assume perfect competition, even when using supply and demand to analyze a competitive market.
Marshall, it is said, assumed perfect competition; perhaps there once was such a
thing. But clearly there is no longer, and we must therefore discard his theories.
The reader will search long and hard and I predict unsuccessfully to find in
Marshall any explicit assumption about perfect competition or any assertion that
16
And to some extent Robinson (1933) in her last chapter, A world of monopolies.
237
in a descriptive sense the world is composed of atomistic firms engaged in perfect

competition Marshall took the world as it is; he sought to construct an engine
to analyze it, not a photographic reproduction of it. (F53, 34)
Thus he judges the Cambridge and Harvard approaches both to be misconceived. The Cambridge economists were wrong to suggest that in
exposing logical inconsistencies in the theory of competitive equilibrium,
they were exposing problems with Marshalls theory, for he did not
assume perfect competition. The Harvard economists were wrong to
assume that because product differentiation, advertising expenditure, and
so on were more realistic, their theories were better than Marshalls
theory. Thus Friedman was rejecting both branches of the new establishment in value theory, without claiming that the world was perfectly
competitive.17
Hausman and Mongin (1997, 266) suggest that full-cost theories
aroused controversy because the exact interpretation of fundamental
theory was not yet settled. This is correct, though arguably it understates
the variety of theoretical and empirical approaches that existed. From the
1920s and 1940s, economists engaged in wide-ranging discussions
about the theory of the firm and market structure, approaching the problem from different angles. Some developments were theoretically driven,
resolving apparent inconsistencies in Marshalls theory. Others were
driven by observations about business practices and institutional details,
supplemented very extensively by statistical investigations. The result
was a complex variety of approaches, with debates across many of these.
Imperfect/monopolistic competition was seen very differently in Cambridge
and Harvard. Full-cost ideas and price stickiness were discussed by exponents of this new establishment in value theory, who saw it as filling a gap in
the theory of oligopoly, as well as by institutionalists (Means and Hamilton),
and by Marshallians (Andrews). The distinction between Marshallian
theory and perfect competition was rarely spelled out.18
Against this mixture of approaches, some compatible with each other
and others not, Friedman was defending an almost unique position. He
did not accept either the Cambridge or the Harvard version on imperfect/
monopolistic competition (though he accepted that their formal analysis
of monopoly was useful for certain problems) or the theoretical arguments
used to defend full-cost pricing. On the empirical front, he was skeptical
about the relevance of most of the evidence being accumulated for
administered and full-cost pricing, and about the Harvard economists
17
18
He also rejected Walrasian general equilibrium theory as represented by the Cowles

Commission, but that is a different story.
An exception was Shove (1933).
238
Roger E. Backhouse
attempt to explain prices in terms of industrial and market structure.

However, he also rejected the anti-empirical stance of the Austrians,
Machlup and Stigler. Ironically, he was in some ways closest to Andrews,
who shared his commitment to a Marshallian approach (and his sympathy
towards free enterprise). His argument about as if reasoning was particularly important because it enabled him to attack on several sides simultaneously. Much of the marginalist controversy and the debates out of which
it arose had been concerned with explaining how economic outcomes
were actually reached. Dismissing this entire problem made much of this
literature irrelevant, greatly narrowing the scope of economic theory.19
Though he remained Marshallian in many ways, in this respect Friedman
was moving theory further away from Marshall.
References
Andrews, P. W. S. (1949a). Manufacturing Business. London: Macmillan
(1949b). A reconsideration of the theory of the individual business. Oxford
Economic Papers 1, 5489
Backhouse, Roger E. (2002). Don Patinkin: interpreter of the Keynesian revolution. European Journal of the History of Economic Thought, 9(2), 186204
(2003). The stabilization of price theory, 192055. In Warren J. Samuels, Jeff
E. Biddle, and John B. Davis (eds.), A Companion to The History of Economic
Thought. Oxford: Blackwell
Backhouse, Roger E. and David Laidler (2004). What was lost with ISLM.
History of Political Economy 36, annual supplement, The History of ISLM,
ed. M. de Vroey and K. D. Hoover, 2556
Bain, Joe S. (1942). Market classification in modern price theory. Quarterly
(1956). Barriers to New Competition. Cambridge, MA: Harvard University Press
Chamberlin, Edward (1933). The Theory of Monopolistic Competition. Cambridge,
MA: Harvard University Press
(1957). Toward a More General Theory of Value. New York
Clark, John Maurice (1923). The Economics of Overhead Costs. Chicago: University
of Chicago Press
Clapham, J. H. (1922). Of empty economic boxes. Economic Journal, 32(127),
30514
Eiteman, Wilford J. (1945). The equilibrium of the firm in multi-process industries. Quarterly Journal of Economics, 59(2), 2806
(1947). Factors determining the location of the least-cost point. American
Economic Review, 37(5), 91018
(1948). The least-cost point, capacity and marginal analysis: rejoinder.
American Economic Review, 38(5), 899904
19
A similar narrowing took place in macroeconomics. See Backhouse and Laidler (2004).
239
Eiteman, Wilford J. and Glenn E. Guthrie (1952). The shape of the average cost
curve. American Economic Review, 42(5), 8328
Friedman, Milton (1953). The methodology of positive economics. In Essays on
Positive Economics. Chicago: University of Chicago Press, 343
Hamilton, Walton (1938). Price and Price Policies. New York: McGraw Hill
Hammond J. Daniel (1996). Theory and Measurement: Causality Issues in Milton
Friedmans Monetary Economics. Cambridge: Cambridge University Press
Harrod, Roy (1930). Notes on supply. Economic Journal, 40, 23241
(1939). Price and cost in entrepreneurs policy. Oxford Economic Papers 2 (old
series), 111
Hausman, Daniel M. and Phillippe Mongin (1997). Economists responses to
anomalies: full-cost pricing versus preference reversals. History of Political
Economy, 29 (annual supplement), 25572
Heflebower, Richard B. (1954). Toward a theory of industrial markets and prices
American Economic Review, Papers and Proceedings, 44(2), 12139
(1955). Full cost, cost changes and prices. In G. J. Stigler (ed.), Business
Concentration and Price Policy. Princeton: Princeton University Press
Theory and Practice. Brighton: Harvester Wheatsheaf
Knight, Frank Hyndman (1921). Risk, Uncertainty and Profit. Boston: Houghton
Mifflin
Lee, Frederic S. (1984). The marginalist controversy and the demise of full-cost
pricing. Journal of Economic Issues, 18(4), 110732
Lee, Frederic S. and Peter Earl (eds.) (1993). The Economics of Competitive
Enterprise: Selected Essays of P. W. S. Andrews. Cheltenham: Edward Elgar
Lee, Frederic S. and J. Irving-Lessman (1992). The fate of an errant hypothesis: the
doctrine of normal-cost prices. History of Political Economy, 24(2), 273309
Leeson, Robert (1998). The early PatinkinFriedman correspondence. Journal of
the History of Economic Thought, 20(4), 43348
Lester, Richard A. (1946). Shortcomings of marginal analysis for wageemplyment problems. American Economic Review, 36, 6382
(1947). Marginalism, minimum wages, and labor markets. American Economic
Review, 37(1), 13548
Machlup, Fritz (1937). Monopoly and competition: a classification of market
positions. American Economic Review, 27(3), 44551
(1939). Evaluation of the practical significance of the theory of monopolistic
competition. American Economic Review, 29(2), 22736
(1946). Marginal analysis and empirical research. American Economic Review,
36(4), 51954
(1947). Rejoinder to an antimarginalist. American Economic Review, 37(1),
13754
Marshall, Alfred ([1890] 1920). Principles of Economics, 8th edition. London:
Macmillan
(1925). Mechanical and biological analogies in economics. In A. C. Pigou (ed.),
Memorials of Alfred Marshall. London: Macmillan. Originally published in
Distribution and exchange. Economic Journal, 8 (1898), 3759
(1919). Industry and Trade. London: Macmillan
240
Roger E. Backhouse
Mason, E. (1939). Price and production policies of large-scale enterprise.

American Economic Review, Papers and Proceedings, 29(1), 6174
Means, Gardiner C. (1935). Price inflexibility and the requirement of a stabilizing
monetary policy. Journal of the American Statistical Association, 30(90), 41013
(1936). Notes on inflexible prices. American Economic Review, 26(1), 2335
Mongin, Phillippe (1992). The full-cost controversy of the 1940s and 1950s: a
methodological assessment. History of Political Economy 24(2), 31156
OBrien, D. P. (1983). Research programmes in competitive structure. Journal of
Economic Studies 10(4), 2951. Reprinted in OBrien 1994
(1984). The evolution of the theory of the firm. In Frank H. Stephen (ed.),
Firms, Organization and Labour. London: Macmillan. Reprinted in OBrien
1994
(1990). Marshalls industrial analysis. Scottish Journal of Political Economy,
37(1), 6184. Reprinted in OBrien 1994
(1994). Methodology, Money and the Firm: The Collected Essays of D. P. OBrien,
vol. I. Cheltenham: Edward Elgar
Oliver, Henry M. (1947). Marginal theory and business behavior. American
Oswald, Andrew and Philip Trostel (2000). What can recent labour research teach
us about macroeconomics? In Roger E. Backhouse and Andrea Salanti (eds.),
Macroeconomics and the Real World, vol. 2: Keynesian Economics, Unemployment
and Policy. Oxford: Oxford University Press, 11344
Pigou, A. C. (ed.) (1925). Memorials of Alfred Marshall. London: Macmillan
(1928). An analysis of supply. Economic Journal, 38, 23857
Raffaelli, Tiziano (2003). Marshalls Evolutionary Economics. London: Routledge
(2004). Whatever happened to Marshalls industrial economics? European
Journal of the History of Economic Thought, 11(2), 20929
Robbins, Lionel (1928). The representative firm. Economic Journal, 38, 387404
Robinson, Joan (1932). Imperfect competition and falling supply price. Economic
Journal, 42, 54454
(1933). The Economics of Imperfect Competition. London: Macmillan
Rutherford, Malcolm. (1999). Institutionalism as scientific economics. In Roger
E. Backhouse and John Creedy (eds.), From Classical Economics to the Theory of
the Firm: Essays in Honour of D. P. OBrien. Cheltenham: Edward Elgar,
22342
Shove, Gerald F. (1933). The imperfection of the market: a note. Economic
Journal, 43, 11324
Sraffa, Piero (1926). The laws of returns under competitive conditions. Economic
Journal, 36, 53550
Stigler, George J. (1946). The economics of minimum wage legislation. American
(1947). Professor Lester and the marginalists. American Economic Review, 37(1),
1547
Triffin, Robert (1940). Monopolistic Competition and General Equilibrium Theory.
Whitaker, John K. (1996). The Correspondence of Alfred Marshall, Economist. 3 vols.

Oliver E. Williamson
Inasmuch as all theories, not just the neoclassical, start with the existence
of firms (Arrow 1999, vi), since the theory of the firm figures prominently
in both Milton Friedmans essay on The methodology of positive economics (F53) and my own research agenda, and since we are all closet
methodologists, I responded with alacrity to the invitation to prepare a
paper on F53 as it relates to the theory of the firm. My remarks are
organized in four parts.
I begin with what I take to be the main message of the essay: most
economists are and should be engaged in the study of positive economics.
What I regard as overreaching parts of the essay are discussed in section 2.
Post-1953 developments in the theory of the firm are sketched in section 3.
Concluding remarks follow.
The main message
The first ten pages of the F53 contain the main message. Specifically, the
task of positive economics is to provide a system of generalizations that
can be used to make correct predictions about the consequences of any
change in circumstances. Its performance is to be judged by the precision,
scope, and conformity with experience of the predictions it yields
(F53, 4). Additionally, simplicity and fruitfulness are important criteria
in evaluating alternative theories (F53, 10). As Friedman subsequently
remarks, Most phenomena are driven by a very few central forces. What a
good theory does is to simplify, it pulls out the central forces and gets rid
of the rest (Snowdon and Vane 1997, 196). Indeed, a fundamental
hypothesis of science is that appearances are deceptive and that there is a
way of looking at or interpreting or organizing the evidence that will reveal
superficially disconnected and diverse phenomena to be manifestations of
Comments on earlier versions by Jack Letiche and Robert Solow, by participants at
the Friedman (1953) conference in Rotterdam, and by Claude Menard are gratefully
acknowledged.
241
242
a more fundamental and relatively simple structure (F53, 33). But

whereas Friedman puts it in the singular a way and asimple
structure I would put it in the plural: there are ways and simple
structures (on more of which later).
His argument that a theory is more fruitful the more precise the
resulting prediction, the wider the area within which the theory yields
predictions, and the more additional lines for further research it suggests
(F53, 10) is instructive, recognizing that there may be tradeoffs among
these three. Also, although all would-be theories must expect to come
under criticism, you dont beat something with nothing but must offer
something in its stead (Kuhn 1970, 77). Awaiting a rival theory, the
continuous use and acceptance of [a] hypothesis over a long period, and
the failure of any coherent, self-consistent alternative to be developed and
to be widely accepted, is strong indirect testimony to its worth (F53, 23).
As between normative and positive theory, F53 tells us that (1) more
economists should concern themselves with real phenomena rather than
idealizations and (2) all would-be positive theories should stand up and be
counted by making predictions and submitting these to empirical testing. This last is the way by which to sort the sheep (hard-headed theories)
from the goats (fanciful constructions).
Overreaching
F53 is unarguably controversial witness that we are still debating the

merits fifty years later. Partly this may be because rhetorical purpose is
often served by exaggeration, but there is also reason to believe that the
author revels in controversy. (The latter interpretation is suggested by
Friedmans remark that If you want to get controversy about one of your
articles, write something which will be attacked and then dont answer the
attackers because it opens a field day (Snowdon and Vane 1997, 203).)
The merits of the central message notwithstanding, Friedman is also given
to overreaching.
2.1
Assumptions
Truly important and significant hypotheses will be found to have assumptions

that are wildly inaccurate descriptions of reality, and, in general, the more significant the theory, the more unrealistic the assumptions. (F53, 14; emphasis added).
Wildly inaccurate? Unrealism is a virtue? To be sure, Friedman subsequently observes that the relevant question to ask about the assumptions of a theory is not whether they are descriptively realistic, for they
243
never are, but whether they are sufficiently good approximations for
the purpose at hand (F53, 15), with which few would disagree. And
he further admits to a tradeoff if confronted with a theory that is known
to yield better predictions but only at a greater cost (F53, 17). If, for
example, the more fine-grained predictions of a theory that works out of
less extreme assumptions imposes added data collection costs, the gains
may be worthwhile for only a subset of circumstances. Indeed, one of
the challenges to extreme assumptions such as perfect gas laws or zero
transaction costs is to specify the circumstances under which these can
be presumed to work (F53, 18). Who could disagree? But if that is the
argument, why the bombastic introduction?
2.2
Alternative theories
Friedman observes that

The abstract methodological issues we have been discussing have a direct bearing
on the perennial criticism of orthodox economic theory as being unrealistic as
well as on the attempts that have been made to reformulate theory to meet this
charge [C]riticism of this type is largely beside the point unless supplemented
by evidence that a hypothesis differing in one or another of these respects from
[orthodoxy]yields better predictions for as wide a range of phenomena. (F53,
301; emphasis added)1
But why the insistence on wide range? What if a rival theory provides
better predictions only for some phenomena? What if it calls our attention to new phenomena? As posed by Friedman, the contest between
textbook orthodoxy, which is the product of many years of development,
and upstart contenders, which have not had the benefit of successive
extensions, applications, and refinements, is needlessly protective of the
former. (As Thomas Kuhn observes, the early versions of most new
paradigms are crude [1970, 156], often heuristic models [1970, 184].
Clumsiness [is] inevitable in a first venture [1970, 33]. Successive refinements thereafter take shape, sometimes with the benefit of a new
mathematics.)
Thus although Friedman subsequently takes the position that Science
in general advances primarily by unsuccessful experiments that clear the
ground (Snowdon and Vane 1997, 296), F53 has a much narrower view
of the enterprise. As previously remarked, he speaks of a way and a simple
structure, both in the singular (F53, 33).
1
Even more restrictive would be an insistence that rival theories yield better predictions for
the same range of phenomena.
244
2.3
Theories of the firm (singular)
The 1930s witnessed the appearance of Chamberlins theory of monopolistic competition, which aspired to close a gap between Marshalls polar
cases of monopoly and perfect competition. On Melvin Reders reading of
F53, monopolistic competition theory, with its emphasis on descriptive
realism, posed a serious threat to the neoclassical theory of firm and
markets. Rather than defend neoclassical theory by affirming the descriptive accuracy of the perfect competition model, Friedman argued that
the primary criterion for judging the merit of any model is its capacity for
generating correct and substantively important predictions (Reder 2003,
528). Restoring confidence in the validity of neoclassical theory [was]
high among Friedmans objectives (Reder 2003, 528).
Friedman took the position that not only was monopolistic competition theory lacking in tools (F53, 39), but Marshallian theory which
worked out of two ideal types of firms: atomistically competitive firms,
grouped into industries, and monopolistic firms (F53, 35) was altogether adequate.2 Everything depends on the problem; there is no
inconsistency in regarding the same firm as if it were a perfect competitor
for one problem, and a monopolist for another (F53, 36). Earlier he
explains that Each occurrence has some features peculiarly its own
The capacity to judge that these are or are not to be disregarded can be
learned but only by experience and exposure in the right scientific
atmosphere, not by rote (F53, 25). Again, by referring to the right
scientific atmosphere, Friedman appears to scant the possible benefits
of pluralism.
Friedmans efforts to buttress the neoclassical theory of the firm by
invoking economic natural selection are also instructive. His famous as
if arguments on the positioning of leaves on trees as if each leaf sought
to maximize the amount of sunlight it receives (p. 19); and that an expert
billiard player makes shots as if he knew the complicated mathematical
formulas that would give the optimum (F53, 21; emphasis in original)
are introduced to support the hypothesis that under a wide range of
circumstances individual firms behave as if they were seeking rationally
to maximize their expected returns (F53, 21; emphasis in original), since
otherwise they will not be viable (F53, 22).
Correct as far as it goes. But why not name the circumstances under
which the maximization of expected returns hypothesis is problematic, as
recommended by Tjalling Koopmans (1957, 141):
2
Chamberlin versus Chicago issues have been addressed at length elsewhere. See George
C. Archibald (1961, 1987), Robert Kuenne (1967), and Richard Lipsey (2001).
245
Such a change in the basis of economic analysis would seem to represent a gain in
realism attributable to a concern with the directly perceived descriptive accuracy of
the postulates. It would lead us to expect profit maximization to be most clearly
exhibited in industries where entry is easiest and where the struggle for survival is
keenest, and would present us with the further challenge to analyze what circumstances give to an industry that character. It would also prevent us, for purposes of
explanatory theory, from getting bogged down in those refinements of profit
maximization theory which endow the decision makers with analytical and computational abilities and assume them to have information-gathering opportunities
such as are unlikely to exist or be applied in current practice. It seems that nothing
is lost, and much may be gained, in thus broadening the postulational basis of
economic theory.
To delimit the adequacy of the profit maximization hypothesis along these

lines is not apostasy, but progress.
3
Theories of the firm (plural)
3.1
The atmospherics
F53 makes no reference to the heady atmospherics of the 1950s, which

was a decade during which the social sciences were rushing toward rigor.
Paul Samuelsons Foundations of Economic Analysis had been published in
1947. John von Neumann and Oskar Morgensterns Theory of Games and
Economic Behavior, first published in 1944, went into its third edition in
1953. Kenneth Arrows Social Choice and Individual Values was published in 1951. And exciting new general equilibrium treatments of
welfare economics were taking shape (Arrow 1951b; Debreu 1954).
The Cowles Commission (then at the University of Chicago, but soon
to leave for Yale) had become the mecca of mathematical economies.
The contiguous social sciences also had mathematical ambitions. As
Herbert Simon put it,
Mathematics has become the dominant language of the natural sciences not
because it is quantitative a common delusion but primarily because it permits
clear and rigorous reasoning about phenomena too complex to be handled in
words. This advantage of mathematics over cruder languages should prove to be of
even greater significance in the social sciences, which deal with phenomena of the
greatest complexity. (1957, 89).
The idea that success in the social sciences would be realized by the
mathematization of everything was in the air.3
3
Interestingly, Simon subsequently revised his views and recommended that the economics
profession be more attentive to the needs of the many students who view mathematical
tools with distrust and deplore the necessity of devoting their research time to formalisms
246
As between positive and normative, much of the research during

this era was predominantly normative. As discussed below, needless
errors of public policy have normative origins. Possibly Friedmans 1953
essay was intended (in part) to forestall these developments. Be that as it
may, the view that rival theories do not coexist but collide and that there is
one right way discourage a plurality of instructive constructions.
In fact, the self-imposed limits of the theory of the firm as production
function to which a profit maximization purpose was ascribed were highly
restrictive. Faced with the anomalies posed by non-standard and unfamiliar
contracting practices and organizational structures, industrial organization
economists from both Harvard (barriers to entry) and Chicago (price discrimination) appealed to monopoly purpose to explain these phenomena.
A problem with such explanations is that the requisite market power
preconditions to exercise monopoly purpose were often missing. More
generally, the one-string banjo of monopoly reasoning was often humbug
(Coase 1972, 67):
if an economist finds something a business practice of one sort or another
that he does not understand, he looks for a monopoly explanation. And as in this
field we are very ignorant, the number of ununderstandable practices tends to be
very large, and the reliance on a monopoly explanation, frequent.
Absent an alternative lens, monopoly reasoning ran amok. As Justice

Stewart put it in his dissenting opinion in Vons Grocery (1966), the
sole consistency that I can find in [merger] litigation under Section 7
[is that] the Government always wins. An antitrust steam roller had been
loosed. Public policy toward business was careening out of control
(Williamson 1985, ch. 14).
Insistence on a single, all-purpose theory of the firm was at the heart of
the problem. As would subsequently become apparent, it is a mistake to
confuse the firm of [orthodox] economic theory with its real-world namesake. The chief mission of neoclassical economics is to understand how
the price system coordinates the use of resources, not the inner workings
of real firms (Demsetz 1983, 377). Theories of the firm that regard the
inner workings of real firms as consequential have since moved beyond
technology (production functions) to make provision for organization.
Prior preoccupation with monopoly purpose has given way to a larger set
of purposes, to include efficiency, in the process.
that they regard as mainly sterile (Simon 1997, 90). Assar Lindbeck puts it somewhat
differently: university teachers and researchers of economics should assume a greater
responsibility for transmitting knowledge and understanding of real-world problems,
including common sense rather than dwell on simple classroom exercises, with oversimplified and often unrealistic assumptions (2001, 32).
247
The argument is not that the orthodoxy lens of choice is wrong but that
it is not all-purpose. There are other instructive perspectives. As Avinash
Dixit has recently put it (1996, 9):
the neoclassical theory of production and supply viewed the firm as a profitmaximizing black box. While some useful insights follow from this, it leaves some
very important gaps in our understanding and gives us some very misleading ideas
about the possibilities of beneficial policy intervention. Economists studying business and industrial organization have long recognized the inadequacy of the neoclassical view of the firm and have developed richer paradigms and models based
on the concepts of various kinds of transaction costs. Policy analysis stands to
benefit from opening the black box and examining the actual workings of the
mechanism inside.
3.2
Managerial and behavior theories
The large size of modern firms has often resulted in diffuse ownership. Is it
reasonable under such circumstances to assume that the managers of a
modern corporation will choose to operate it in the interests of the owners
(Berle and Means 1932, 121)? Or might the management operate such
corporations in a fashion that sometimes compromised ownership interests?
The obvious way to respond to this possibility was to take exception
with the assumption of profit maximization. If managers had the latitude
to tilt the operations of the firm in pursuit of managerial purposes, then
restate the objective function of the firm to reflect these added purposes.
That simple change aside, the firm would still be described as a production function and the same marginal apparatus could be employed. Upon
postulating a different (more realistic) objective function to which managerial purposes were ascribed, the ramifications of managerialism could
be derived and contrasted with those of profit maximization.
Albeit a topic of ongoing discussion, models of managerial discretion did
not appear until almost thirty years after Berle and Means highlighted the
separation of ownership from control issue in 1932. The first of these was
William Baumols model (1959) postulating that large firms maximize revenues subject to a minimum profit constraint. Robin Marris (1964) shortly
thereafter advanced a growth maximization hypothesis. And I expressed the
objective function of the firm as a utility function in which managerial
staff and emoluments as well as profits appeared (Williamson 1964).
Because all three of these models described the firm as a production
function, all remained within the resource allocation tradition. Greater
realism for size preference (Baumol), growth preference (Marris), and
expense preference (Williamson) notwithstanding, few novel ramifications for public policy emerged. To be sure, competition in product and
248
capital markets were both viewed as good, in that these served as a check
upon managerial discretion, but such competition was also desired for
standard resource allocation reasons. Also, although lump-sum taxes
induced changes in factor proportions in the expense preference model
that differed from neoclassical predictions and it could further be shown
that the intertemporal management of slack had business cycle ramifications (Williamson 1968), qualitative responses by firms to changes
in tax rates, interest rates, wages, investment subsidies and the like
[turned out to be] pretty much the same, regardless of what firms are
assumed to maximize (Solow 1971, 319). Thus even though firms with
different objectives [will differ quantitatively], if their qualitative response
is more or less insensitive within limits, then objective function differences do not much affect the research agenda (Solow 1971, 319).
Put differently, so long as the firm is described as a production function,
it does not matter much, at least qualitatively, which of several plausible
objective functions is chosen. In that event, considerations of parsimony
support continued use of the profit maximization setup.
The behavioral theory of the firm (Cyert and March 1963) took a
different tack. Rather than emphasize realism in motivation (as in the
managerial theories), behavioral theory focused on realism in process.
The purpose was to explain (predict) price and output decisions of the
firm with process-oriented models in which internal decision rules and
routines were featured. Albeit successful at a very microanalytic level of
analysis (e.g. predicting department store pricing to the penny (Cyert
and March 1963, ch. 7)), the theory did not engage issues of resource
allocation more generally. Also, although the behavioral theory of the firm
devotes two pages (2956) to antitrust policy, the discussion focuses
entirely on uncertainty avoidance, is wholly devoid of specific applications, and has had no discernible impact on antitrust enforcement.
Evolutionary economic theory aside, which also adopted the routine as
the unit of analysis (Nelson and Winter 1982), the main influence of
behavioral theory has been in the field of organization theory where it
has been massive rather than economics.
3.3
The lens of contract
What I refer to as the lens of contract approach to economic organization

breaks away from the resource allocation paradigm (prices and output;
supply and demand) to examine economic organization from a different
partly rival but mainly complementary perspective. To be sure, contractual approaches to economic organization had been under discussion for a
249
long time (Condillac 1776). It was not, however, until the 1970s that
operationalization got under way in a concerted fashion.
The lens of contract approach to economic organization was prefigured
by John R. Commons, who had long contested the all-purpose reliance on
the efficient resource allocation paradigm. But there was more than mere
criticism in Commons. As against simple market exchange between faceless buyers and sellers who meet for an instant to exchange standardized
goods and services at equilibrium prices (Ben-Porath 1980, 4), Commons
had an abiding interest in going concerns and reformulated the problem
of economic organization as follows: the ultimate unit of activity must
contain in itself the three principles of conflict, mutuality, and order. This
unit is a transaction (1932, 4). Not only does Commons move to a more
microanalytic unit of analysis (the transaction), but his three principles
prefigure the concept of governance in that governance is the means by
which to infuse order, thereby to mitigate conflict and realize mutual gain.
This latter turns out to be a recurrent theme in the evolving contractual
account of the purposes served by complex contract and organization.
Coases classic paper, On the nature of the firm (1937), also raised
related concerns. Three lapses in the orthodox theory of firm and market
organization are especially noteworthy: (1) the distribution of transactions
between firm and market were taken as given, whereas these should be
derived; (2) going beyond production costs, there was a need to recognize
that transaction cost differences were often responsible for the choice of
one mode rather than another; and (3) orthodoxy had no good answers for
the puzzle of what is responsible for limits to firm size.
Coases subsequent critique of the market failure literature in his
equally famous paper on The problem of social cost (1960) identified
additional lapses of logic. Upon reformulating the tort problem (or, more
generally, the externality problem) as a problem of contract, he showed
that externalities vanished when the logic of zero transaction costs is
pushed to completion. As Coase put it in his Nobel Prize lecture (1992,
717; emphasis added):
Pigous conclusion and that of most economists using standard economic theory
was that some kind of government action (usually the imposition of taxes) was
required to restrain those whose actions had harmful effects on others (often
termed negative externalities). What I showed was that in a regime of zero
transaction costs, an assumption of standard economic theory, negotiations between
the parties would lead to those arrangements being made which would maximize
wealth and this irrespective of the initial assignment of property rights.
Kenneth Arrows examination of The organization of economic activity:

issues pertinent to the choice of market versus nonmarket allocation
250
(1969) likewise made a prominent place for transaction costs, both in

general and with reference to vertical integration. The general argument is
this (Arrow 1969, 48; emphasis added):
I contend that market failure is a more general condition than externality; and both
differ from increasing returns in a basic sense, since market failures in general and
externalities in particular are relative to the mode of economic organization, while
increasing returns are essentially a technological phenomenon.
Current writing has helped to bring out the point that market failure is not
absolute; it is better to consider a broader category, that of transaction costs, which in
general impede and in particular cases completely block the formation of markets
[T]ransaction costs are the costs of running the economic system.
Organizational considerations now take their place alongside technology,

which had previously been treated as determinative. Upon recognizing
that organization matters, transaction cost differences, as between internal organization and market exchange (where both are now regarded
as alternative modes of contracting), have obvious ramifications for vertical integration: An incentive for vertical integration is replacement of the
costs of buying and selling on the market by the costs of intrafirm transfers;
the existence of vertical integration may suggest that the costs of operating
competitive markets are not zero, as is usually assumed by our theoretical analysis
(Arrow 1969, 48; emphasis added).
The need to place the study of positive transaction costs onto the agenda
was clearly posed. Adding a perfunctory transaction cost term to production cost or utility function expressions would not, moreover, suffice. If, as
James Buchanan puts it, mutuality of advantage from voluntary exchange is
the most fundamental of all understandings in economics (2001, 28), and if the
purposes of complex exchange are frequently misconstrued by orthodoxy,
then a science of exchanges approach (Buchanan 2001, 28) to economic
organization is awaiting development.
As perceived by Buchanan, the principal needs for a science of exchange
were in the field of public finance and took the form of public ordering:
Politics is a structure of complex exchange among individuals, a structure
within which persons seek to secure collectively their own privately defined
objectives that cannot be efficiently secured through simple market
exchanges (1987, 246; emphasis added). Inasmuch as the preconditions
for simple market exchanges are not satisfied when problems of collective
choice are posed, a new calculus of consent, so to speak, was needed
(Buchanan and Tullock 1962; Brennan and Buchanan 1985). The field of
public choice took shape in response to the perceived needs.
Public ordering is not, however, the only or even the predominant way
of dealing with complex market exchange. On the contrary, huge numbers
Science of
choice
251
Orthodoxy:
scarcity and
resource allocation
Constitutional economics
Economics
Public
ordering
Science of
contract
Ex ante
incentive alignment
Private
ordering
Ex post
governance
Figure 9.1 The sciences of choice and contract
of private-sector transactions do not qualify to be described as simple

market transactions between faceless buyers and sellers. Given that
anti-competitive interpretations for non-standard and unfamiliar contracting practices and organizational structures are frequently bankrupt,
and since mutuality of advantage is the fundamental purpose of exchange,
why not interpret the governance of contractual relations as an effort to
implement the Commons Triple of conflict, mutuality, and order?
Complex contracting and organization would thus be construed mainly
(but not exclusively) as self-help efforts by the immediate parties to a
transaction to align incentives and craft governance structures that are
better attuned to their exchange needs. The study of private ordering
(with reference to industrial organization and microeconomics more
generally) thus takes its place alongside public ordering.
Figure 9.1 sets out the main distinctions. The initial divide is between
the science of choice (orthodoxy) and the science of contract. The former
corresponds to the ideal transaction in both law and economics namely,
simple market exchange to which the resource allocation paradigm applies
and competition prevails. The science of contract, by contrast, applies
to complex exchanges for which non-market supports are required.
As discussed above, complex contracting then divides into public (constitutional economics) and private ordering parts, where the second is
split into two related branches. One branch deals with ex ante incentive
alignment (mechanism design, agency theory, the formal property rights
literature), often with reference to efficient risk bearing. The second
252
features the ex post governance of contractual relations (contract implementation, with emphasis on the mitigation of contractual hazards).
Interestingly, Chester Barnards book on The Functions of the Executive
(1938) is prominently featured in both the ex ante incentive alignment
and ex post governance approaches to economic organization. Thus the
recent book by Jean-Jacques Laffont and David Martimort on The Theory
of Incentives credits Barnard with being the first [to] attempt to define a
general theory of incentives in management (2002, 11). Even more important, in my judgment, is Barnards insistence that adaptation is the central
problem of economic organization, where the adaptations to which Barnard
had reference were those of a conscious, deliberate, purposeful kind
(1938, 4) accomplished through administration. Taken together with the
autonomous adaptations to changes in relative prices to which Friedrich
Hayek (1945) gave prominence, the stage was set, as it were, for the
combined study of markets (autonomous adaptations) and hierarchies
(cooperative adaptations).
The incentive branch has been principally concerned with the employment relation whereas the paradigm problem for the governance branch is
vertical integration. The latter is the make-or-buy decision in intermediate
product markets. The manner in which this is decided has obvious
ramifications for the boundary of the firm. More generally, a large number
of contractual phenomena turn out to be variations on the very same
transaction-cost economizing themes that surface when vertical integration
is examined in comparative contracting terms. Vertical market restrictions
(customer, territorial, and franchise restrictions), hybrid modes of organization, the organization of labor, the choice between debt and equity, the
assessment of regulation/deregulation, the use of public bureaus, and the
like are all matters to which the logic of transaction-cost economizing has
been usefully applied. Not only do many of the predictions of transactioncost economics differ from orthodoxy, but the data have been broadly
corroborative.
Whereas the conceptual framework out of which transaction-cost economics works differs from orthodoxy (in that it is a contractual construction with emphasis on ex post governance), it is nevertheless in agreement
with F53 that predictions attended by empirical testing are the cutting
edge. The key message of the methodology of positive economics thus
reaches beyond orthodoxy. Yet the methodology to which transactioncost economics subscribes is even closer to that of Robert Solow (2001).
Describing himself as a native informant rather than as a methodologist,
Solows terse description of what one economist thinks he is doing
(2001, 111) takes the form of three injunctions: keep it simple; get it
right; make it plausible.
253
He observes with reference to the first that the very complexity of real
life [is what] makes simple models so necessary (2001, 111). Keeping
it simple entails stripping away the inessentials and going for the jugular.
The object is not to explain everything but to narrow the focus to central
features and key regularities (some of which may previously have been
scanted). Getting it right includes translating economic concepts into
accurate mathematics (or diagrams, or words) and making sure that further
logical operations are correctly performed and verified (Solow 2001, 112).
But there is more to it than that. A model can be right mechanically yet be
unenlightening because it is imperfectly suited to the subject matter. It can
obscure the key interactions, instead of spotlighting them (Solow 2001, 112;
emphasis added). If maintaining meaningful contact with the key phenomena of interest (contractual or otherwise) is valued, then rigor that sacrifices
meaningful contact poses a tradeoff, especially if more veridical models
yield refutable implications that are congruent with the data.
This last brings me to a fourth injunction: derive refutable implications to
which the relevant (often microanalytic) data are brought to bear. Nicholas
Georgescu-Roegen had a felicitous way of putting it: The purpose of
science in general is not prediction, but knowledge for its own sake, yet
prediction is the touchstone of scientific knowledge (1971, 37). There is a
need to sort the sheep from the goats when rival theories are at an impasse.4
Inasmuch, however, as the immediate prospect for a unified, allpurpose theory of the firm is very doubtful, there may be several sheep
from which to choose depending on the specific problems for which
good answers are wanting. In the event, pluralism has much to recommend it. As Simon puts it, I am a great believer in pluralism in science.
Any direction you go has a very high a priori probability of being wrong, so
it is good if other people are exploring in other directions perhaps one of
them will be on the right track (1992, 21) to which I would add that
several investigators might be on rights tracks (plural).
Conclusions
Very few books, and even fewer papers, are celebrated on their fiftieth
anniversary. There is no question but that F53 was and remains influential mainly because the central message, at least among pragmatically
oriented economists, is fundamentally correct. Would-be theories, in
4
The fruitfulness to which F53 (33) refers is also pertinent in comparing theories: theories
that have wide application, in that many phenomena turn out to be variations on a theme,
are to be preferred to theories of narrow scope, ceteris paribus.
254
economics and throughout social sciences, need to stand up and be

counted by making predictions and inviting empirical tests.
Interestingly, although the theory of the firm was once criticized for the
paucity of empirical work (Holmstrom and Tirole 1989, 126), I would
argue that the critics have the empirical cart before the theoretical horse.
What is sorely needed in the theory of the firm literature are interesting
and testable predictions. As such predictions began to appear, empirical
research mushroomed.5 I am persuaded that Melvin Reders advice that
economists focus on the construction of relatively simple theories that
have some prospect of accounting for the gross facts (1999, 298) is
apposite.
References
Archibald, George C. (1961). Chamberlin versus Chicago. Review of Economic
Studies, 29(1), 228
(1987). Monopolistic competition. In J. Eatwell et al. (eds.), The New Palgrave
Dictionary of Economics, vol.3. London: Macmillan, 5314
Arrow, Kenneth J. (1951a). Social Choice and Individual Values. New York: John
Wiley
(1951b). An extension of the basic theorems of classical welfare economics.
In Jersey Neyman (ed.), Proceedings of the Second Berkeley Symposium on
Mathematical Statistics and Probability. Berkeley, CA: University of California
Press, 50732
(1969). The organization of economic activity: issues pertinent to the choice of
market versus nonmarket allocation. In The Analysis and Evaluation of Public
Expenditure: The PPB System, vol. 1. US Joint Economic Committee, 91st
Congress, 1st Session. Washington, DC: US Government Printing Office,
3973
(1999). Forward. In Glenn Carroll and David Teece (eds.), Firms, Markets, and
Hierarchies. New York: Oxford University Press, viiviii
Bain, Joe (1956). Barriers to New Competition. Cambridge, MA: Harvard University
Press
Barnard, Chester (1938). The Functions of the Executive. Cambridge: Harvard
University Press (fifteenth printing, 1962)
Baumol, W. J. (1959). Business Behavior, Value and Growth. New York: Macmillan
Ben-Porath, Yoram (1980). The F-connection: families, friends, and firms and
the organization of exchange. Population and Development Review, 6 (March),
130
Berle, Adolph A. and Gardner C. Means, Jr. (1932). The Modern Corporation and
Private Property. New York: Macmillan
5
To be sure there were lags. By the year 2000, however, there were over 600 published
empirical papers in transaction-cost economics alone (Boerner and Macher 2001).
255
Boerner, Christopher and J. Macher (2001). Transaction cost economics: a review

and assessment of the empirical literature. Unpublished manuscript, Haas
School of Business, University of CaliforniaBerkeley
Brennan, Geoffrey and James Buchanan (1985). The Reason of Rules. Cambridge:
Buchanan, James (1987). The constitution of economic policy. American Economic
Review, 77 (June), 24350
(2001). Game theory, mathematics, and economics. Journal of Economic
Methodology, 8 (March), 2732
Buchanan, James and Gordon Tullock (1962). The Calculus of Consent. Ann
Arbor, MI: University of Michigan Press
Condillac, Etienne Bonnot ([1776] 1821). Le commerce et le gouvernement. In
Oeuvres completes de Condillac, vol. 4, Paris: Briere
Coase, Ronald H. (1937). The nature of the firm. Economica n.s., 4, 386405
(1960). The problem of social cost, Journal of Law and Economics, 3 (October),
144
(1972). Industrial organization: a proposal for research. In V. R. Fuchs (ed.),
Policy Issues and Research Opportunities in Industrial Organization. New York:
National Bureau of Economic Research, 5973
(1992). The institutional structure of production. American Economic Review, 82
(September), 71319
Commons, John R. (1932). The problem of correlating law, economics, and
ethics. Wisconsin Law Review, 8, 326
Cyert, Richard M. and James G. March (1963). A Behavioral Theory of the Firm.
Englewood Cliffs, NJ: Prentice-Hall
Debreu, Gerard (1954). Valuation equilibrium and Pareto optimum. Proceedings
of the National Academy of Social Sciences, 40, 58892
Demsetz, Harold (1983). The structure of ownership and the theory of the firm.
Journal of Law and Economics, 26 (June), 37590
Dixit, Avinash (1996). The Making of Economic Policy: A Transaction Cost Politics
Perspective. Cambridge, MA: MIT Press
Friedman, M. (1953). The methodology of positive economics. In Essays in
Positive Economics. Chicago: Chicago University Press
(1997) [in Snowdon and Vane (1997)]
Georgescu-Roegen, Nicholas (1971). The Entropy Law and Economic Process.
Hayek, Friedrich (1945). The use of knowledge in society. American Economic
Review, 35 (September), 51930
Holmstrom, Bengt and Jean Tirole (1989). The theory of the firm. In Richard
Schmalensee and Robert Willig (eds.), Handbook of Industrial Organization.
New York: North Holland, 61133
Koopmans, Tjalling C. (1957) Three Essays on the State of Economic Science. New
York: McGraw-Hill
Kuenne, Robert E. (ed.) (1967). Monopolistic Competition Theory: Studies in
Impact : Essays in Honor of Edward H. Chamberlin. New York: John Wiley
Kuhn, Thomas S. (1970). The Structure of Scientific Revolutions, 2nd edition.
256
Laffont, Jean-Jaques and David Martimort (2002). The Theory of Incentives.

Princeton: Princeton University Press
Assar Lindbeck (2001). Economics in Europe. CESifo Forum, Spring, 313
Lipsey, Richard (2001). Successes and failures in the transformation of economics. Journal of Economic Methodology, 8(2), 169201
Marris, Robin (1964). The Economic Theory of Managerial Capitalism. New York:
Free Press.
Nelson, Richard R. and Sidney G. Winter (1982). An Evolutionary Theory of
Economic Change. Cambridge, MA: Harvard University Press
Reder, Melvin W. (1999). Economics: The Culture of a Controversial Science. Chicago:
University of Chicago Press
(2003) Remarks on The methodology of positive economics. Journal of
Economic Methodology, 10, 52730.
Samuelson, Paul A. (1947). Foundations of Economic Analysis. Cambridge, MA:
Harvard University Press
Simon, Herbert (1957). Models of Man. New York: John Wiley
(1992). Economics, Bounded Rationality, and the Cognitive Revolution. Brookfield,
CT: Edward Elgar
(1997). An Empirically Based Microeconomics. Cambridge: Cambridge University
Press
Snowdon, Brain and Howard Vane (1997). Modern macroeconomics and its
evolution from a monetarist perspective. An interview with Professor
Milton Friedman. Journal of Economic Studies, 24(4), 192222
Solow, Robert (1971). Some implications of alternative criteria for the firm. In
Robin Marris and Adrian Wood (eds.), The Corporate Economy. Cambridge,
MA: Harvard Universtiy Press, 318342
(1997). How did economics get that way and what way did it get? Daedalus,
126(1), 3958
(2001). A native informant speaks. Journal of Economic Methodology, 8 (March),
111112
United States v. Vons Grocery Co. (1966). 384 U.S. 270, 301 (Stewart, J.,
dissenting)
Von Neumann, John and Oskar Morgenstern (1953). Theory of Games and
Economic Behavior. Princeton, NJ: Princeton University Press
Williamson, Oliver E. (1964). The Economics of Discretionary Behavior: Managerial
Objectives in a Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall
(1968). A dynamic-stochastic theory of managerial behavior. In Almarin
Phillips and Oliver Williamson (eds.), Prices: Issues in Theory, Practice, and
Public Policy. Philadelphia: University of Pennsylvania Press
(1985). The Economic Institutions of Capitalism. New York: Free Press
10

Jack Vromen
Introduction
Friedmans selection argument in his 1953 essay The methodology of

positive economics (F53) received little attention in economic methodology. There may be several reasons for this. Friedmans exposition of the
argument is so short and hidden in his lengthy defense of marginalism that
it might easily have escaped the attention of economic methodologists.
Alternatively, methodologists may have noticed the argument, but may
have considered it to be too insignificant to Friedmans overall argument,
too obviously just a rhetoric trick to savior marginalism, or simply as too
obviously flawed to merit any discussion. Whatever the reason for this
relative neglect by methodologists, the argument has not gone unnoticed
in the economics literature. On the contrary, economists of different
stripes and persuasion have taken up the argument and have bent it to
their own purposes. Just the fact that this happened by itself warrants
taking a closer look both at Friedmans original statement of the argument
and the use that other economists subsequently made of it.
What exactly is there in Friedmans selection argument that has
been found worthwhile to comment and elaborate upon? One of the
aims of this chapter is to point out that different answers are given to
this question. Different commentators have different ideas about what it is
in Friedmans argument that warrants further consideration. As we shall
see, the differences are vast. Different commentators concentrate on
different aspects of the argument. Different commentators come up with
different interpretations of the argument and of what the argument is meant
to establish. And, last but not least, different commentators draw different
conclusions as to what implications the argument has for economic theorizing. The chapter argues in particular that the implications for economic
theorizing that most commentators impute to Friedmans argument may
well be quite different from the ones Friedman intended. In this sense, this
is a tale about unintended consequences of Friedmans selection argument.
257
258
Jack Vromen
The argument contextualized
Friedmans selection argument is part of a larger argumentative strategy to

defend marginalism in general, and the profit-maximization assumption in
the theory of the firm in particular, against critiques leveled by so-called
antimarginalists. Based on responses to questionnaires, antimarginalists
such as Hall and Hitch (1939) in the UK and Lester (1946) in the USA
argued that marginal costs and revenues were no part of the decisionmaking of businessmen. Instead businessmen were allegedly observed to
follow rules of thumb like the full-cost pricing rule. The antimarginalists
concluded from this that the theory of the firm had to be discarded and had
to be superseded by a theory with more realistic behavioral assumptions.
Friedmans selection argument was specifically devised to bring home the
point that even if antimarginalists were right that businessmen follow fullcost pricing rules, this does not jeopardize the usefulness and applicability
of the (marginalist) theory of the firm. More precisely, the argument was
meant to show that when it comes to assessing a theory of the firm it does
not matter what decision rules and decision procedures businessmen
follow. What matters is what businessmen actually do, not their motives,
reasons, deliberations, and the like for doing so, let alone what they say they
do (or what they themselves believe they are doing).
Friedman does not need many words to present his selection argument.
Strictly speaking, Friedman dedicates no more than just one paragraph in
the whole text of F53 to the exposition of the argument.
Confidence in the maximization-of-returns hypothesis is justified by evidence of a
very different character. This evidence is in part similar to that adduced on behalf
of the billiard-player hypothesis unless the behavior of businessmen in some way
or other approximated behavior consistent with the maximization of returns, it
seems unlikely that they would remain in business for long. Let the apparent
immediate determinant of business behavior be anything at all habitual reaction,
random chance, or whatnot. Whenever this determinant happens to lead to
behavior consistent with rational and informed maximization of returns, the business will prosper and acquire resources with which to expand; whenever it does
not, the business will tend to lose resources and can be kept in existence only by the
addition of resources from outside. The process of natural selection thus helps
validate the hypothesis or, rather, given natural selection, acceptance of the
hypothesis can be based largely on the judgment that it summarizes appropriately
the conditions for survival. (F53, 22)
Presumably Friedman was the first to have developed this type of selection
argument in defense of the theory of the firm. Although a similar argument appeared earlier in Alchian (1950), it seems that the first draft of
Friedmans essay (which already contained the selection argument) was
259
written before Friedman (as an editor of the Journal of Political Economy;

Hammond 1991) came across Alchians paper. In his essay Friedman
notes that his own presentation of the argument is independent in origin
of Alchians.1 Some ten years later Becker (1962) published a similar
argument. Compared to Alchians and Beckers versions of the argument,
which are full-paper articulations of the argument, Friedmans is shortest
and most cryptic. This might explain why in several discussions of the
argument, Friedmans version is given short shrift or is even not mentioned at all. In his wide-ranging overview of earlier attempts in economics
to import biological models, for example, Hirshleifer (1977) does not
contain a single reference to F53. There is long discussion of Alchian
(1950), Enke (1951), Penrose (1952), Simon (1955), and Winter (1964,
1971), but F53 is not mentioned once.
But Friedman was not the first to bring evolutionary considerations to
bear on the antimarginalism controversy. In a somewhat forgotten paper
Harrod (1939) already contains a few interesting evolutionary speculations.
Other lines of thought suggest themselves. It may be that certain procedures, of
which application of the full cost principle is one example, are thrown up purely by
chance in the first instance and survive by a process akin to natural selection in
biology. New business procedures would then be analogous to new mutations in
nature. Of a number of procedures, none of which can be shown either at the time
or subsequently to be truly rational, some may supplant others because they do in
fact lead to better results. Thus while they may have originated by accident, it
would not be by accident that they are still used. For this reason, if an economist
finds a procedure widely established in fact, he ought to regard it with more respect
than he would be inclined to give in the light of his own analytic method. (Harrod
1939, 7)
The point Harrod is trying to make is different from, if not opposite to

Friedmans. Friedmans argument was meant to persuade fellow economists that the issue of what the immediate determinants of firm behavior
are is irrelevant for appraising the usefulness of the theory of the firm. By
contrast, Harrods point is that evolutionary considerations should caution economists not to be too readily dismissive of decision procedures
observed in firms. The fact that some procedures and rules are observed
widely may testify to their superiority to competing procedures and rules
in prior selection processes. Whereas Friedman provided an evolutionary
argument to show the irrelevance of the empirical findings of the
1
The full footnote (14) reads: This example, and some of the subsequent discussion,
though independent in origin, is similar to and in much the same spirit as an example and
the approach in an important paper by Armen A. Alchian, Uncertainty, Evolution, and
Economic Theory, Journal of Political Economy, LVIII (June, 1950), 21121.
260
Jack Vromen
antimarginalists, Harrod sought to provide an evolutionary rationale for

their relevance. In section 5 we shall see that in some relevant respects
Nelson and Winters evolutionary theory can be interpreted as a further
development and articulation of Harrods antimarginalist evolutionary
line of thought.
It is not likely that Friedman himself thought that in his promarginalism case much hinged on his selection argument. For immediately after having presented his selection argument, Friedman continues
as follows: An even more important body of evidence for the
maximization-of-returns hypothesis is experience from countless applications of the hypothesis to specific problems and the repeated failure of its
implications to be contradicted (F53, 22). Especially if we take Friedman
to mean here that the countless applications of the hypothesis provide
strong indirect testimony to its empirical record (F53, 23), this of course
connects with the general methodological point Friedman is making in his
essay. Economic theories should be appraised not on the basis of the
realisticness of a theorys assumptions, but on the basis of the empirical
accuracy of the theorys testable implications (or predictions). To some
extent this makes the selection argument redundant in Friedmans
defense of the hypothesis. Whether or not the selection argument is cogent
and convincing, in the last instance the testable implications of the theory
should be borne out by the facts. This suggests that Friedman holds that if
the empirical record of the theory were poor, the selection argument by
itself would never provide sufficient evidence to accept the theory.
Even with hindsight it is difficult to tell to what extent F53 helped
restore or sustain the confidence of fellow economists in the marginalist
theory of the firm in the face of the antimarginalist assaults. At any rate,
the marginalist theory of the firm arguably did survive the antimarginalist
assaults. Arguably this is all Friedman wanted to accomplish. So, we could
say, Friedmans main mission has been accomplished. Case closed. For
Friedman this means that economists now had their hands free to get on
with their job and to move on to other, more interesting theoretical and
empirical issues. This is indeed what many economists actually did. But
there are also a few economists (and commentators) who singled out
Friedmans selection argument for further consideration. This is surely
not what Friedman intended and hoped for.
3
Evolutionary reinterpretations of economic theory
under a wide range of circumstances individual firms behave as if they were

seeking rationally to maximize their expected returns (generally if misleadingly
called profits) and had full knowledge of the data needed to succeed in this
261
attempt; as if, that is, they knew the relevant cost and demand functions, calculated
marginal cost and marginal revenue from all actions open to them, and pushed
each line of action to the point at which the relevant marginal cost and marginal
revenue were equal given natural selection, acceptance of the hypothesis can be
based largely on the judgment that it summarizes appropriately the conditions for
survival. (F53, 212; emphasis added)
Here we come across Friedmans notorious as-if methodology. What

Friedman wants to make clear is that the theory of the firm is not committed to a particular view on the actual immediate determinants of firm
behavior. In particular, the theory is not wedded to the view that businessmen actually conduct the marginalist calculations that the theory ascribes
to them. The only thing the theory is committed to, Friedman argues, is
that businessmen actually display the observable behavior predicted by
the theory. Thus understood, the theory does not tell us anything about
the actual determinants of firm behavior. This makes one wonder what
are, or what could be, actual determinants of firm behavior. If it is not
necessarily marginalist deliberations that make businessmen behave in the
way they (allegedly) do, then what does ensure that they behave in the way
they (allegedly) do? Friedmans selection argument can be taken to imply
that it is ultimately natural selection in competitive markets that see to
this. From here it is a small step to argue that what is suggested in the
argument is that the theory of the firm should be reinterpreted along
evolutionary lines. Appearances nothwithstanding, the theory is not really
about motives, deliberations, and calculations of businessmen that make
firms and industries behave the way they (allegedly) do. Instead it is about
natural selection that makes firms and industries behave the way they
(allegedly) do.
Friedman leaves no doubt that he (of course) believes that businessmen do not actually and literally solve the system of simultaneous equations that the theory of the firm assumes they do. The hypothesis that firms
maximize expected returns is similar in this respect, Friedman argues, to
the hypothesis that leaves around a tree maximize the sunlight they receive
and to the hypothesis that expert billiard players solve complicated mathematical equations to calculate the best shot possible.2 The latter example
of the expert billiard player was already introduced in Friedman and
Savage (1948) to make the point that objections that the assumption
(that expert billiard players solve mathematical equations before making
their shots) is a patently unrealistic description of the behavior of expert
2
Although Friedman consistently talks about hypotheses, it is to be doubted that the

examples he discusses can meaningfully be called hypotheses. Henceforth I shall use
assumption instead of hypothesis.
262
Jack Vromen
billiard players, though natural and understandable, are not strictly

relevant (Friedman and Savage 1948, 297). Friedman and Savage argue
that unless he was capable in some way of reaching approximately the
same result as that obtained from the formulas, he would not in fact be
likely to be an expert billiard player (1948, 298). Recall that Friedman
himself observes that this is similar to the evidence that justifies confidence in the maximization-of-returns assumption: unless the behavior of
businessmen in some way or other approximated behavior consistent with
the maximization of returns, it seems unlikely they would remain in business for long (F53, 22).
But is the evidence adduced on behalf of the expert billiard player
assumption really analogous to that adduced on behalf of the maximization
of returns assumption? In the case of the maximization-of-returns assumption, competition for resources is taken to be causally responsible for the
(alleged) fact that firms display behavior approximating maximization of
returns. In the case of the expert billiard player, is Friedman envisaging
some similar sort of contest ensuring that the behavior of winners approximates perfect play? This is not clear. If Friedman has such a contest in
mind, then it seems it does not follow that the winning billiard players
necessarily play almost perfect games. A billiard player who made no
mistakes in locating the balls, in solving the relevant mathematical implications, and in implementing the according shots would terminate each game
in one turn. Even world champions do not succeed in doing so. Thus the
behavior of winners of billiard contests may fall far short of perfect play.
Alternatively, Friedman may simply be fleshing out what we understand by
an expert billiard player: by an expert billiard player we mean someone
who is capable of reaching approximately the same results as that obtained
by the formulas. But if this reading is indeed intended by Friedman, it may
well be true that such an expert billiard player does not (and will not) exist.
At any rate, both readings boil down to pretty much the same: there need
not be any approximately perfectly playing billiard player around. Even if
there were cutthroat competition, the behavior of winners may deviate not
just slightly but significantly from perfect play.
What about the other example that Friedman takes to be similar to the
maximization-of-returns assumption, that of leaves trying to maximize the
amount of sunlight? Friedman argues that we of course know that leaves
do not deliberately or consciously seek anything. Yet on the basis of the
assumption we can explain (the quotation marks are Friedmans) that
trees tend to have a higher density on the side that receives most sunlight.
What is actually happening here is that there is a process of passive
adaptation to external circumstances going on. The consequences of
this selection or adaptation process are the same as those that would
263
obtain if leaves were actually to maximize the amount of sunlight: trees

have a higher density of leaves on the side that receives most sunlight.
It is tempting to argue that in one respect at least the leaves-around-thetree assumption is more similar to the maximization-of-returns assumption
than the expert-billiard-player assumption. Unlike in the expert-billiardplayer assumption, in both the leaves-around-the-tree assumption and the
maximization-of-returns assumption Friedman seems to believe that something like natural selection is actually driving the process, purportedly
leading it to outcomes as predicted by the respective assumption. The
disanalogy here of course is that whereas leaves do not actually deliberate
and consciously seek anything, billiard players try to make the best shots
and businessmen try to maximize profits. Note that Friedman is not denying that businessmen try to maximize profits. What he denies is that in their
decision-making businessmen do exactly the things that the theory of the
firm ascribes to them. But Friedman considers the whole issue of what are
the actual immediate determinants of firm behavior the mental processes
businessmen go through before making decisions or displaying behavior
(which includes their goals and motives) to be irrelevant for the issue at
hand. It can safely be assumed that firms behave as if businessmen go
through all the mental operations ascribed to them in the maximizationof-returns assumption because only firms that do so survive natural selection in competitive markets.
Satz and Ferejohns radical reinterpretation: rational choice theory

as an external-structuralist theory
But if this is at the heart of Friedmans argument, then is not the correct
interpretation of the maximization-of-returns assumption that it refers to
natural selection in competitive markets? If in competitive markets the
maximization-of-returns assumption summarizes appropriately the conditions of survival (F53, 22), and if these conditions are independent of
immediate determinants of firm behavior, as Friedman argues, then can it
be maintained that the assumption is really about the immediate determinants of firm behavior? Or should we conclude that this is only apparent
and that the assumption is really about selection pressure in competitive
markets? This is the line taken by Satz and Ferejohn (1994). Satz and
Ferejohn argue that rational-choice explanations are most plausible in
settings in which individual action is severely constrained.3 For example,
3
Satz and Ferejohn do not cite empirical support for this claim. There does seem to be some
experimental support for the claim that rational choice theory does better in competitive
environments. See, for example, Smith (1991).
264
Jack Vromen
the theory of the firm has been a far more productive explanatory theory
than the theory of consumer behavior, Satz and Ferejohn argue, because
consumers face less competitive environments than firms. Rationalchoice explanations are argued to gain their explanatory power from
features of the agents environment rather than from the mental states or
the psychological make-up of agents. This leads Satz and Ferejohn to
plead for a radical reinterpretation of rational-choice theory.
The received interpretation is that rational-choice theory is a psychological theory, dealing with what is going on internal to agents, consistent
with methodological individualism. By contrast, the interpretation Satz
and Ferejohn favor is one in which rational-choice theory appears as an
external-structuralist theory that is about external pressures that environments impose on agents. Satz and Ferejohn remark that evolutionary
biology, which views nature as a selective structure, provides the crucial
paradigm for their discussion (Satz and Ferejohn 1994, 81). They believe
that there are social analogues to natural selection acting as powerful
selective mechanisms. One such analogue is an environment of competitive capital markets acting as a selector on firms (1994, 79). On Satz and
Ferejohns reinterpretation, rational-choice theory carries no a priori
commitments to a particular mental causal mechanism internal to agents
(1994, 86). The assumed psychology of the agent is rather an entirely
imputed one: preferences are derived on the basis of an agents location
in a social structure. This is why the interest of maximizing profits can be
imputed to firms in competitive markets. Even though there can be many
different motives and deliberations behind pursuing and attaining maximum profits (1994, 79), given their position in a competitive market it
makes sense to ascribe the motive to make maximum profits to firms.
Binmore: memes as immediate determinants of behavior
A similar evolutionary reinterpretation of expected-utility theory is advocated by Binmore (1994). Binmore argues that leading utility theorists
have long abandoned a literal interpretation of their theory. Utility theory
does not explain observable behavior by referring to antecedently and
independently identified preferences (or utility functions), expectations,
and risk profiles of the agents in question. It is rather the other way
around. Preferences, expectations, and risk profiles of agents are inferred
from their observable choice behavior. In doing so, by identifying the
agents revealed preferences, expectations, and risk profiles, and by
assuming that the agents maximize their expected utility, their observable
behavior is rationalized. The assumption really is that agents behave, as
Binmore puts it, as though they maximize their expected utilities thus
265
derived. This of course is very much in line with Friedman and Savage
(1948), for example, and with Savages influential subjective interpretation of expected-utility theory in general.
Binmore goes one considerable step further by suggesting not only that
what agents do, how they behave, is the outcome of processes of sociocultural evolution, but also that the latter view reflects the currently
dominant or prevailing interpretation of expected-utility theory.
Stigler seems to me a windmill at which it is pointless to tilt. Not even in Chicago
are such views given credence any more. I suspect that this is largely because a
paradigm is now available that preserves optimization as the underlying mechanism, but no longer sees economic agents as omniscient mathematical prodigies.
The new paradigm regards evolutionary forces, biological, social, and economic,
as being responsible for getting things maximized
mechanisms exist which could lead to behavior on the part of homo sapiens
which mimics that of homo economicus, without requiring that homo sapiens necessarily thinks the same thoughts as those attributed to homo economicus. At his
fireside, homo sapiens will tell different stories from homo economicus, and neither
type of story need correlate very well with the actual reasons that the species came
to behave the way it does. (Binmore 1994, 1920)
The issue at hand here is whether homo economicus provides a useful model
to social phenomena. Stiglers stance was that where self-interest (narrowly conceived) and ethical values conflict, self-interest prevails.
Binmore argues that it is no longer assumed that homo economicus, current
economic theorys depiction of human agents, is solely preoccupied with
his own interests narrowly conceived. Modern homo economicus is an
individual that is consistent in seeking to attain whatever his goals may
be. Binmore maintains that nowadays most proponents of (and commentators on) expected-utility theory hold that sociocultural evolution sees to
it that inconsistencies in human behavior are eventually weeded out. It is
this belief, Binmore argues, that is behind the retention of homo economicus
in current economic theory.
Binmores contention that it is not omniscience of agents, but evolutionary forces that are responsible for getting things maximized is in line
with Friedmans selection argument. But his further suggestion that the
actual reasons for people to behave the way they do are memes is not in line
with Friedmans argument. Following Dawkins (1976),4 Binmore thinks
4
In his controversial accounts of biological and cultural evolution, Dawkins makes it

perfectly clear that his own talk about selfish genes and memes are convenient shorthands.
I expressed all these reasonings in terms of decisions of individual animals but, to repeat,
this is just shorthand. What is really going on is that genes for maximizing grandchildren
become more numerous in the gene pool. The world becomes full of genes that have
successfully come down the ages (Dawkins 1995, 137).
266
Jack Vromen
that believing that our behavior is molded by especially socioeconomic

forces commits one to believe also that our behavior is determined by the
memes, the units of socioeconomic evolution, that we, as their vehicles,
are infected with: to an observer, it will seem as though an infected agent
is acting in his own self-interest, provided that the notion of self-interest
is interpreted as being whatever makes the bearer of a meme a locus for
replication of the meme to other heads (Binmore 1994, 20; emphasis in the
original). While Friedman prefers to remain silent or agnostic about the
immediate determinants of behavior, Binmore contends that memes
are immediate determinants of individual behavior.
Aumann: the evolutionary interpretation of the Nash equilibrium
is the fundamental one
Besides Binmore, Aumann is one of the few game theorists to have
explicitly addressed methodological issues.5 Perhaps more than anyone
else, Robert Aumann has sought to provide rationalistic foundations of
solution notions in game theory, notably of the Nash equilibrium. Among
game theorists, Aumann is well known for his staunch defense of the
rationality postulate and of his rejection of weaker rationality assumptions. Indeed he has been criticized for taking such an unyielding stance
by other prominent game theorists, such as the Nobel laureate Reinhard
Selten. In his famous 1985 paper What is game theory trying to accomplish? Aumann argues that the Nash equilibrium embodies the most
important and fundamental idea of economics, that economic agents act
to maximize their utility (Aumann 1985, 43). Recently, in an interview
with Eric van Damme, Aumann argues that besides this rationalistic
interpretation, the Nash equilibrium also allows for an evolutionary interpretation. What is more, Aumann argues that the evolutionary interpretation is more fundamental than the rationalistic one (Aumann 1997, 18).
Let us quote the relevant passages unabbreviated:
When there is a formal relationship between two interpretations, like the rationalistic (conscious maximization) and evolutionary interpretations of Nash equilibrium, then one can say, oh, this is an accident, these things really have nothing to
5
A more integral comparison between Friedman and Aumann would be interesting but has
to await for another occasion. Suffice it to say here that although Aumann favors a type of
formalistic economic modeling that Friedman presumably disavows, and although
Aumanns predictive accuracy of models should be subordinate to the comprehension
that they yield, Aumann is remarkably close to Friedmans methodology on other points.
Aumann also holds that assumptions do not have to be correct, that the usefulness rather
than the truth of theories is important, and that one can tell the usefulness of some solution
concept, for example, from the extent to what it is actually used by economists.
267
do with each other. Or, one can look for a deeper meaning. In my view, its a
mistake to write off the relationship as an accident.
In fact, the evolutionary interpretation should be considered the fundamental
one. Rationality Homo Sapiens, if you will is a product of evolution. We have
become used to thinking of the fundamental interpretation as the cognitive
rationalistic one, and of evolution as an as if story, but thats probably less
insightful.
Of course, the evolutionary interpretation is historically subsequent to the
rationalistic one. But that doesnt mean that the rationalistic interpretation is the
more basic. On the contrary, in science we usually progress, in time, from
the superficial to the fundamental. The historically subsequent theory is quite
likely to encompass the earlier ones. (Aumann 1997, 18)
A superficial glance at this quote might suggest that Aumanns stance

towards the notion of Nash equilibrium is similar to Satz and Ferejohns
radical reinterpretation of rational-choice theory. Both seem to argue that
the rationalistic interpretation of Nash equilibrium and rationalchoice theory respectively, albeit preceded by the evolutionary interpretation, is not the fundamental interpretation. Since evolutionary pressures
and processes rather than processes of rationalistic conscious deliberation do the bulk of the explanatory work, it is the evolutionary interpretation that is the fundamental one. But appearances are deceiving here.6
In calling the evolutionary interpretation of Nash equilibrium the fundamental one, Aumann does not want to convey the belief that convergence to Nash equilibria is due to evolutionary mechanisms rather than to
conscious maximization. In this he differs from Binmore. Binmore argues
that to the extent that people act rationally, they do not do so because of
omniscient maximizing. They do so because their actual reasons are the
products of evolutionary mechanisms. Upon closer inspection, it turns
out that what Aumann wants to convey here is that rationality, or rather
the capacity for human beings to act rationally, is itself a product of
evolution. The evolutionary interpretation is fundamental in the sense
that prior (or antecedent) evolutionary processes have laid the foundation
for current maximizing behavior.
Aumann argues that paying attention to its evolutionary origins enables
us to see that it is the function of our capacity for conscious rational utility
maximization to adapt flexibly to complex environments (Aumann 1997,
21). Following Mayrs (1961) distinction between proximate and ultimate
causes of behavior, Aumann posits that there are in principle two different
6
Aumann seems to be wrong in suggesting that the evolutionary story is often treated as the
more facile, as if one. In all the invocations of natural selection, or evolutionary forces,
we have come across thus far the cognitive rationalistic interpretation provides the as if
story.
268
Jack Vromen
explanations for our behavior. Conscious rational utility maximization is

an answer to the question how we arrive at our behavior.7 But if we want to
understand why we behave the way we do, we must be looking at the
evolutionary origins of this capacity or mechanism for conscious rational
utility maximization. The general idea here, that ancient evolutionary
processes have resulted in conscious utility maximizing as an immediate
determinant of human behavior, has recently been worked out in more detail
by, for example, Robson (1996, 2001a, 2001b) (see Vromen 2003 for a
critical discussion). Although it is not clear whether Aumann believes that
conscious rational utility maximization means that individuals actually go
through the calculations ascribed to them in utility theory, it is clear that
Aumanns fundamental evolutionary interpretation of Nash equilibrium involves a particular view on the immediate determinants of
behavior. In this he differs from Friedman.
Let us take stock. We have seen that there is no shortage of radical
reinterpretations of standard economic theory along evolutionary lines.
Yet we have also seen that the reinterpretations shift in subtle, but significant ways. The only thing Friedman wanted to argue, it can be maintained, is that in spite of all the antimarginalist criticisms there is nothing
wrong in continuing the ongoing practice of applying the maximizationof-expected-returns assumption in the theory of the firm. Friedman saw
no need for further reflections on the issue of whether his selection
argument invites an evolutionary reinterpretation of utility theory.8 He
neither saw a need to identify immediate determinants of behavior. Yet
this is what the authors discussed here undertake. Satz and Ferejohn argue
that if evolutionary pressures in competitive markets are to carry the
explanatory burden, as Friedman suggests, then the theory is best understood as a theory about the outcomes or results of natural selection in
competitive markets. Binmore can be understood as adding to this a
hypothesis about proximate causes, or actual immediate determinants as
Friedman calls them, of human behavior. Whereas Friedman seems to
prefer an agnostic stance towards these, and argues that there is no need to
identify them, Binmore suggests that, the self-understanding of people
notwithstanding, memes are in control of their behavior. Aumann, finally,
argues that conscious maximizing can be resuscitated as the proximate
cause of human behavior on evolutionary grounds.
7
8
Aumann does not argue that we always engage in conscious rational utility maximization:
to the extent that we do calculate, it may be a result of evolution (Aumann 1997, 32).
Friedman would approve of the license that the evolutionary reinterpretations provide for
continued use of equilibrium analysis. But he presumably would have less sympathy for
formalistic uses of (Nash) equilibrium analysis in game theory.
269
Evolutionary game theory: checking the validity

of the argument
Thus far it has been tacitly assumed that the evolutionary processes
envisaged in the reinterpretations indeed do lead to the results and outcomes that standard economic theory predicts. Is this assumption warranted? Let us first look once again at the relevant crucial passage in
Friedmans argument.
given natural selection, acceptance of the hypothesis can be based largely on the
judgment that it summarizes appropriately the conditions for survival. (F53, 22;
emphasis added)
Friedman suggests that the evidence that natural selection in competitive markets provides for the continued use of the maximization-ofreturns hinges crucially on the judgment that the assumption at least
roughly gets the conditions for survival right. If firms that do not attain
maximum returns can survive natural selection, then the argument
breaks down.
So the judgment that the maximization-of-returns assumption summarizes approximately the conditions for survival must be correct for the
selection argument to carry any weight in favor of the assumption. Is it
correct? Friedman apparently believed that no further argument or examination is needed to see that the judgment is correct. Others saw this
differently. Friedmans selection argument incited many attempts to
examine and scrutinize this judgment critically. Not surprisingly, different
examiners with different persuasions came to different conclusions. The
disagreement here is not only due to differences in persuasion, predilections, and the like of the examiners, however. It is also due to the fact that
it is not perfectly clear what the judgment is actually about. What exactly is
claimed? Note, for example, that the conditions for survival is ambiguous. Are the conditions necessary or sufficient? Is it claimed that attaining
maximum returns is necessary for firms to survive? Or is the claim that
attaining maximum returns is sufficient for firms to survive? The former
claim is stronger than the latter. The former claim rules out a possibility
that the latter claim leaves open: firms earning less than maximum returns
may survive. Most examiners seem to assume that Friedman is making the
stronger claim that only firms earning maximum returns can survive.9
Many commentators have rightly observed that what Friedman is really
arguing is not that the theory of the firm gets phenomena at the level of
9
I think there is some textual evidence favoring this interpretation, as Friedman argues that
firms that happen not to earn maximum returns will tend to disappear.
270
Jack Vromen
individual firms right, but that the theory gets phenomena at the aggregate
level of markets and industries right. An example of the type of ultimate
claim Friedman is making is presented by Alchian (1950): the average
labor/capital ratio in some industry rises if the (relative real) price of labor
falls ceteris paribus. This suggests that the claim that only firms earning
maximum returns survive is just an intermediary step in making and
substantiating the latter ultimate claim (Langlois 1986; Vromen 1995).
Sometimes the intermediate claim Friedman is making is discussed under
the label of as-if rationality, or of as-if maximizing (Kandori 1997). The
ultimate claim Friedman is making is sometimes discussed under the label
of as-if optimality. As Alchians example makes clear, what is at stake is
whether competitive markets lead to productive efficiency. More generally, what is at issue, it seems, is whether competitive markets lead to
Pareto-optimality.10
In order to examine or scrutinize Friedmans claims systematically,
some formal structure is needed to model natural selection explicitly.
This again leaves room for different approaches. What formal structures
or techniques to apply? Several structures and techniques have been
applied (Matthews 1984; Witt 1986; Hansen and Samuelson 1988;
Schaffer 1989; Hodgson 1999, to mention just a few). Of all the attempts
made to scrutinize Friedmans claims systematically, two stand out in
terms of perseverance and faithfulness to his selection argument, I think:
those by Sidney Winter (1964, 1971, 1975) and by Lawrence Blume and
David Easley (1992, 1995, 1998). For reasons that will become apparent,
I shall devote a whole section (section 5) to Winters (and Nelson and
Winters) work. In their examination of Friedmans selection argument,
Blume and Easley adopt an evolutionary general equilibrium model. In
their model, if factor demands and supply evolves, prices (the firms
environment) also change.11 Blume and Easley show that although natural selection in markets favors profit maximization, equilibria need not
converge to competitive (Pareto-optimal) equilibria. They argue that
The weak link in the natural selection justification for the normative
properties of competitive markets is not the behavioral hypothesis of profit
maximization but the implication from profit maximization to Pareto
10
11
Note that although Friedmans essay is about positive economics, these are normative
issues.
Blume and Easley note that this allowance for endogenous price change differentiates
their approach from that of Nelson and Winter (1982). In passing, it is interesting to see
that in their examinations many mainstream economists (such as Blume and Easley)
refer to the work of the heterodox economists Nelson and Winter. This suggests that the
neglect of heterodox economists by orthodox ones may not be as massive as is sometimes argued.
271
optimality (Blume and Easley 1998, 2). Blume and Easley furthermore
argue that if capital markets are added, natural selection does not favor
profit maximization. Introducing capital markets in the picture may disturb the process in which the growth rate of firms is determined solely by
the portion of their revenues (earnings) that are invested in them. The
latter process, Blume and Easley argue, is the one Friedman has in mind
when referring to natural selection. Firms have to earn their own
resources and the higher their relative revenues, the more they invest
and expand.
Are Pareto-optimal Nash equilibria invariably selected?
The conclusion that there is more support for Friedmans intermediate
claim than there is for his ultimate claim seems to be drawn also in
evolutionary game theory.12 Arguably, evolutionary game theory has
become the prevailing modeling strategy to scrutinize claims and
arguments like the one Friedman is advancing (for useful overviews, see
Weibull 1995 and Kandori 1997). Apparently, many believe that evolutionary game theory naturally lends itself for such examinations. There
nowadays seems to be a consensus among game theorists that evolutionary game theory by and large supports Friedmans selection
argument. This might explain why many game theorists seem to believe
that Binmore and Aumann have a point in their proposals to reinterpret
standard economic theory along evolutionary lines. But before we conclude prematurely that evolutionary game theory supports Friedmans
argument and claims, let us have a closer look at what exactly it is that
evolutionary game theory does and does not establish.
Let us concentrate on George J. Mailaths (1998) Do people play Nash
equilibrium? Lessons from evolutionary game theory, since that paper
provides an accessible and (at that time) comprehensive overview of the
sorts of results obtained in evolutionary game theory that bear on
Friedmans selection argument. In the Introduction Mailath remarks
that a common informal argument is that any agent not optimizing in
particular, any firm not maximizing profits will be driven out by market
forces (1998, 1347). Mailath argues that evolutionary game theory is
suitable both to render the argument formal and to check its validity. One
of the things Mailath sets out to do in his overview is to find out to what
extent evolutionary game theory supports the claim made in the argument. For Mailath this claim, properly understood, is that agents
12
In Vromen (1995) I argued that Friedmans intermediate claim, if true, would validate his
ultimate claim. The work discussed here calls this into question.
272
Jack Vromen
eventually converge to play Nash equilibrium.13 The conclusion Mailath

arrives at is that evolutionary game theory gives qualified support for this
claim: In a range of settings, agents do (eventually) play Nash. There is
thus support for equilibrium analysis in environments where evolutionary
arguments make sense (1998, 1348).14
What does it mean to claim, as Mailath does, that evolutionary game
theory provides qualified support for the central position that the notion of
Nash equilibrium takes in non-cooperative game theory? It effectively
means that any stable state that the evolutionary processes as depicted in
evolutionary game theory can converge to is a Nash equilibrium. Or, to
use slightly more precise and technical parlance, any asymptotically
stable rest point of an evolutionary dynamic is a Nash equilibrium
(Mailath 1998, 1371). If some evolutionary dynamic, such as the socalled replicator dynamic,15 drives a process in the direction of a rest
point (a state that, once reached, is left unchanged by the dynamic), and
if this point is asymptotically stable (which means that any path that starts
sufficiently close to the rest point converges to that rest point), then this
outcome of the evolutionary dynamic is (and cannot be but) a Nash
equilibrium. This might be understood as implying that any enduring
state evolutionary processes (as depicted in evolutionary game theory) can
result in must be a Nash equilibrium. This result certainly is reassuring for
(Nash) equilibrium analysis. The main reason why Mailath argues that
evolutionary game theory provides support for equilibrium analysis is that
in many asymmetric games (games in which agents play different roles),
the Nash equilibria available are not asymptotically stable rest points.
Mailaths conclusion here seems to bear more on Friedmans intermediate claim than on his ultimate claim. Play of Nash equilibrium is
often seen as the embodiment of the idea that economic agents are
rational (Aumann 1985, 43). What lessons, if any, are there to be drawn
from evolutionary game theory about Friedmans ultimate claim? The
asymptotically stable rest points identified by evolutionary game theory
need not be optimal. In games in which there are multiple equilibria, for
example, it is possible that evolutionary processes get stuck into a Nash
equilibrium that is Pareto-dominated by another Nash equilibrium. This
13
14
15
In his unpublished PhD dissertation, Nash (1950) himself suggested a populationstatistical interpretation of his own equilibrium concept (Leonard 1994).
See also Cressman (2003) who reports that this is dubbed the Folk Theorem of Evolutionary
Game Theory by Hofbauer and Sigmund (1998).
The replicator dynamic basically says that the growth rate of a strategys share in some
population is proportional to the degree in which it does better than the populations
average. What is said here also holds for more general dynamics such as monotone
dynamics.
273
means that some (and perhaps all) agents would have been better off
(without there being someone who is worse off) had they all played their
parts in the other Nash equilibrium that did not materialize. So the fact
that in (symmetrical) games evolutionary dynamics produce Nash equilibria does not imply that agents will eventually be best off. In general it
may crucially depend on the initial state of the population what asymptotically stable rest point some evolutionary process will converge on. If the
evolutionary process starts in the basin of attraction of a Paretodominated equilibrium, it may remain locked into it. The fact that evolutionary game theory can account for this phenomenon of path dependence
is generally welcomed by evolutionary game theorists as one more important and attractive feature of their theory (see Mailath 1998, 1360 and
1368).16
Thus evolutionary game theory seems to lend more support to
Friedmans intermediate claim than to his ultimate claim. If evolutionary
processes converge to enduring, stable rest points then these are Nash
equilibria. This is taken to support Friedmans intermediate claim that
only rational, optimizing behavior survives. Friedmans ultimate claim,
that productive efficiency and Pareto-optimality are thereby safeguarded,
is in general not supported. Asymptotically stable rest points that the
enduring states evolutionary processes converge on may be Paretodominated Nash equilibria.
Are we still discussing Friedmans original claims?

One might rightly wonder, however, whether it really is Friedmans intermediate and ultimate claims that are addressed in evolutionary game
theory. One has to take quite some liberty in interpreting Friedmans claims
for evolutionary game theory to bear on them. It seems that Friedmans
argument has tacitly been turned into another one and that it is at most this
transformed argument that gets support from evolutionary game theory.
This transformation involves three aspects: mechanisms, conditions, and
results. In Friedmans argument the alleged mechanism is natural selection, which refers to a process in which the relatively most profitable firms
expand their market share at the expense of less profitable firms. In current
evolutionary game theory the driving mechanism is learning. Individual
agents are assumed to learn either individually or socially in a boundedly
16
See also Young (1996). It has to be noted, however, that the evolutionary game theorists
who embrace path dependence have also tried to show that at least in some games with
strict equilibria, path dependence need not be forthcoming in the ultra-long run (see
Foster and Young 1990; Kandori, Mailath, and Rob 1993).
274
Jack Vromen
rational way. Evolutionary game theory is about individuals and populations, not about firms and industries. This difference in mechanisms stipulated should not be exaggerated, however. The monotone dynamics mostly
posited in evolutionary game theory seems to be compatible with
Friedmans rudimentary ideas about the nuts and bolts of natural selection. In both the central idea is that the share or frequency of some type of
individual (whether it is a firm or a human agent) tends to be proportional
to its relative performance.
The differences with respect to the other two aspects, conditions and
results, may be less reconcilable. As to conditions, Friedman seems to
have competitive markets in mind. Competitive markets refer to a
parametric setting or context in which the behavior of the one firm (or
individual) does not affect the profitability (or the payoffs) of others.17
The profitability of each firm depends only on its own activities (Hansen
and Samuelson 1988). By contrast, game theory by definition deals with
strategic settings in which the payoffs obtainable by the one individual at
least partly depend on the behavior displayed by others. Like theories
about perfect competition, evolutionary game theory deals with large
populations. Populations are so large that the share of each individual in
the population is negligible. But mostly the assumption is that interactions
are pairwise. What the opponent does directly affects the payoffs that one
can obtain. For some evolutionary game theorists this is sufficient reason
to call the sorts of situations they are analyzing non-competitive (Banerjee
and Weibull 1995).
Friedmans intermediate claim is that the result or outcome of natural
selection is that eventually only firms survive that earn maximum
returns. The only survivors display behavior that can be said to be rational
in the sense that it coincides with the choices that a fully informed,
faultlessly calculating maximizer would make. Evolutionary game theory
points out that, with a few exceptions, evolutionary processes tend to
converge to Nash equilibria. Do individuals display optimal behavior
when they play a Nash equilibrium? The definition of Nash equilibrium
implies that they display optimal behavior if other individuals play their
parts in the equilibrium. If other players do not play their parts in the
equilibrium, the optimal behavior of an individual may also well be not to
play his or her part in the equilibrium. It is possible that individuals would
be better off in out-of-equilibrium situations. Furthermore, as Kandori
argues One prominent reason why irrational behavior may survive is that
17
To put it more accurately, the profitability of some firm does not depend directly on the
behavior of others. The behavior of others can affect the profitability of the firm indirectly,
however, via changed prices.
275
commitment to a suboptimal strategy pays in strategic situations (Kandori

1997, 258). In strategic situations, individuals that are, for example,
emotionally committed to play a particular fixed strategy may outperform
individuals that are not similarly committed (Frank 1988). Thus, pace
Mailath, it is questionable that demonstrating that the use of Nash equilibria is (almost fully) justified can count as a substantiation of Friedmans
intermediate claim.
There is ample reason to believe that Friedman would not be particularly fond of attempts made by evolutionary game theorists to check the
correctness of his judgment (or claims). For one thing he simply saw no
reason to do this. For another he would not have liked what probably is the
main reason underlying these attempts. Mailath seems to be looking for
the best possible justification for the standard practice in game theory to
treat Nash equilibrium as the solution concept. He argues that of all the
candidate justifications offered thus far learning seems to be the least
problematic (Mailath 1998, 1354). As evolutionary game theory is argued
to offer a particularly attractive approach to learning, he seems to present
this as an argument in favor of evolutionary game theory. Use of the Nash
equilibrium is not defended on Friedmanian grounds, namely that it
works well for some particular purpose at hand. Such an eye on usefulness
seems to be entirely lacking in Mailaths argument. Some evolutionary
game theorists seem to analyze dynamic models with no other purpose in
mind than to gain a better understanding of the properties of some
stipulated dynamic in some particular game.18 The issue of whether the
understanding thus gained has any relevance for a better understanding of
concrete phenomena in the real world does not seem to bother them at all.
This is the type of formalistic, blackboard economic theorizing, which
perhaps can be called Walrasian, that Friedman has little sympathy for.
Evolutionary economics: taking observed decision

procedures seriously
Let the apparent immediate determinant of business behavior be anything at all

habitual reaction, random chance or whatnot. Whenever this determinant happens to lead to behavior consistent with rational and informed maximization of
returns, the business will prosper and acquire resources with which to expand;
whenever it does not, the business will tend to lose resources and can be kept in
existence only by the addition of resources from outside. (F53, 22; emphasis
added)
18
It is clear that other economists working in this area, such as Nobel laureate Reinhard
Selten, have other aims in mind, such as a greater realisticness of game theory.
276
Jack Vromen
The emphasis is here put on the fact that Friedman argues that certain
results of natural selection obtain, no matter what apparent (or actual)
immediate determinants of firm behavior there are. Immediate determinants of firm behavior are argued to be non-consequential for selection
outcomes. Following Nelson and Winter (1982), evolutionary economists believe that this part of the argument is flawed. Immediate determinants of firm behavior do make a difference. What decision procedures
firms follow matters a great deal for what outcomes are selected.
Winter (1964) is well known as one of the first and one of the most
systematic and profound examinations of the selection argument. The
conclusion Winter arrives at is that to the extent that selection considerations support the use of traditional economic theory, they do so only under
special circumstances and for a restricted range of applications (Winter
1964, 268).19 In other words, what Winter establishes is that the support
that evolutionary considerations lend to traditional economic theory, and
notably the theory of the firm, is quite limited. Winter discusses several
circumstances under which selection considerations do not support
traditional economic theory. For example, he argues that a theory of the
firm, which is to yield refutable predictions about the behavior of individual
firms or aggregates of firms, should be distinguished from a theory of entry
and exit, which is to yield refutable predictions about the appearance and
disappearance of firms. When it comes to assessing the selection argument,
a theory of the firm should be supplemented by a theory of entry and exit.
Thus if an industry initially consists of profit-maximizing firms, these firms
may nevertheless be outcompeted by non-profit-maximizing firms that
have the compensating benefit of increasing returns to scale.20 The result
may be a mixed industry rather than an industry consisting only of profitmaximizing firms. Another condition to be met for evolution to produce an
industry solely consisting of profit-maximizing firms is that differential
profitability of firms should translate into a proportional differential growth
rate. If this condition is not met, if relatively profitable firms rest content
with paying out their extra profits in higher dividends while remaining at
the same scale of operation, for example, then less profitable firms will not
be driven out of business (Winter 1964, 244).
Winter furthermore argues that if information may be imperfect and
costly to gather, profit-maximizing firms need not be the ones surviving
natural selection: If such a situation obtains, reliance on rules of
thumb may, in a restricted range of normal situations, be more viable
19
20
See also The preceding section provides examples of combinations of initial conditions
and sets of possible organization forms for which this conclusion is not valid (248).
See Hodgson (1999) for a further discussion of this condition.
277
than profit-maximizing (1964, 258). In this vein, Winter argues that a

full-cost pricing rule might survive at the expense of a maximizing rule
which goes through extensive and expensive calculations to find the price
output combination that equated marginal cost and marginal revenue
(1964, 263). Thus again this presents an example of a situation in which
non-profit-maximizing firms need not be weeded out by natural selection. Note that this is reminiscent of Harrods (1939) idea that observed
business procedures, such as the full-cost pricing rule, are prevalent
because they supplanted others in a selection process.
Note also that profit maximizing is understood here as a particular,
time-consuming decision procedure. Winter assumes that what Friedmans
selection argument purports to establish is that only firms with profitmaximizing decision procedures survive natural selection. Strictly
speaking, this misrepresents Friedmans argument. He does not argue
that only firms with particular decision procedures survive natural
selection. What Friedman argues is that only firms that approximately
make maximum profits survive, no matter how they do this. It is possible
that firms that do not follow any decision procedure or decision rule can
make maximum profits (due to sheer luck, for example). As argued
earlier, for Friedman the issue of decision rules, or immediate determinants as he himself calls them, is essentially a non-issue. Whether or not
firms follow some fixed decision procedure or rule does not matter, if
they fail to make maximum profits they will be selected against.
Yet it is quite understandable why Winter reads Friedman in this way.
Winter states that any theory of selection requires the assumption that
individuals (in this case, firms) have fixed behavioral characteristics
(Winter 1964, 251). If we want to develop a theory of selection in order
to see what firms, or better, what types of firms, survive selection, then we
must assume that firms have some fixed behavioral characteristics.
Although it remains to be seen whether Winter is right in arguing that
without such an assumption there can be no theory of selection, it seems
he has a point here. To see this, let us get back to Friedmans argument. It
is easy to see that Friedmans argument that immediate determinants of
firm behavior do not matter is sound in some sort of Judgment Day
scenario. The argument is sound, that is, if at the end of the day it is
decided once and for all on the basis of the amount of resources that
firms themselves earned thus far what market share each of them gets.
Then, but only then, it does not matter whether the immediate determinant of the behavior of some firm is random chance, for example. The
only thing that counts then is indeed what resources the firm has earned at
the end of the day. And by pure chance or sheer luck it could be the case
that the firm happens to have made maximum profits. But this is not how
278
Jack Vromen
natural selection in competitive markets is supposed to work, even on

Friedmans own view. Note that Friedman is arguing that firms that do
not succeed in making maximum profits (returns) will tend to lose
resources. In other words, it is suggested that there are repeated selection
moments rather than one (one-shot) decisive selection moment. If
there is a sequence of selection moments, then what matters is how well
firms perform in subsequent periods in time. And it is highly unlikely that
the firm with the random chance determinant that was lucky enough to
have earned maximum profits in one period will also be as lucky in
subsequent periods.
This is exactly the point Winter is making: If the immediate determinants of behavior are habitual reaction, random chance, or whatnot,
there is no reason to believe that the firms which take actions consistent
with profit maximization at one time will also take actions consistent with
such maximization at all subsequent times (Winter 1964, 240). It could
be retorted here that although its probability may be close to nil, the
possibility cannot be ruled out that one and the same firm can be lucky
all of the time. But the more fundamental problem Winter is drawing our
attention to, I guess, is that if firms can change their behavior in unpredictable ways, it is hard if not impossible to trace what firms do relatively
better consistently, that is over extended periods of time. If we want to
know what firms perform relatively best in an ongoing process of economic natural selection, then we may be at a loss if firms change their
relevant operating characteristics faster than the rate at which they
undergo selection effects.
Nelson and Winters evolutionary theory in a nutshell
In any event, what started out as a forced move, as an assumption made
to meet the alleged requirements of selection theory, later grew into one of
the three cornerstones of Nelson and Winters evolutionary theory.
Winter (1964) coined the slightly misleading expression organization
form as the functional relationship between the state of the world and a
firms actions. Nelson and Winter (1982) talk of routines. Nelson and
Winter argue at great length that firm behavior is better understood in
terms of rigid routines than in terms of flexible, deliberate choice. Their
catchphrase routines as genes amply makes clear that they believe that in
certain relevant respects routines of firms are like genes of organisms. A
firms routines instruct or program the firms behavior in much the same
way as the genes of an organism instruct or program the organisms
behavior. Most of the time this works in an automatic fashion, without
the intervention of conscious volition. Nelson and Winter refer to
279
Friedmans example of the expert billiard player. Like the expert billiard
player, businessmen just figure out what to do. There is not much
deliberate choice involved in this. Routines are like genes also in another
respect. Just like genes, routines embody wisdom or knowledge acquired
and accumulated in the past.21 The type of knowledge involved here is
tacit knowledge. Moreover, it is know-how rather than knowing-that
that can be fully articulated.
A second cornerstone of Nelson and Winters evolutionary theory is
natural selection. Already in Winter (1964) it is clear that he does not
examine the workings and consequences of natural selection just to see
whether Friedmans argument is valid. Winter takes natural selection to
be one of the crucial forces working in competitive markets that determine
what firms and what industry structure prevail. In passing, we already
came across what Winter believes natural selection amounts to: firms
expand proportional to their profitability. Nelson and Winter set out to
model this process explicitly. In doing so they put into practice what Enke
(1951) and Koopmans (1957) pleaded for. Enke (1951) advances a
selection argument similar to Alchian (1950). But he draws a conclusion
from the argument opposite to the one Alchian draws: marginalism
should give way to what Enke called viability analysis. Arguably this is
exactly what Nelson and Winter want to develop. Koopmans (1957)
argues that if market selection is indeed the basis for the belief in profit
maximization, as Friedman argues, then that basis should be postulated
and not profit maximization, which it is said to imply in certain circumstances. In their evolutionary theory, Nelson and Winter follow up on this.
Nelson and Winter are clearly realists about their own evolutionary
economic theory.22 Or alternatively we can say that their preference for
an explicit evolutionary economic theory is ontologically grounded. They
prefer modeling selection processes in which routine-following firms are
involved explicitly over using equilibrium analysis in which maximizing
agents figure, because they believe that that gives a more adequate representation of what is actually going on in market economies. This longing
for realism (or realisticness) also shows in the third cornerstone in their
evolutionary theory. Following Schumpeter, Nelson and Winter argue
that endogenously engendered innovations are characteristic of capitalist
economies. They posit that firms are induced to search for better routines
in the face of adversity. Innovations are argued to be similar to mutations
21
22
To the extent that present circumstances differ markedly from the one in which routines
evolved, the resulting behavior can well be maladaptive.
For a more detailed discussion of this, see Vromen (2001).
280
Jack Vromen
in evolutionary biology. At first sight this seems to conflict with their belief
that firm behavior is routine. Routines as genes seems to imply that
mutations can only emerge as copying errors, or something of the sort.
But Nelson and Winter stress that rigid routines do not necessarily make
for rigid behavior. Routines may have a conditional structure allowing for
different behavior in different circumstances. Furthermore, firms may
engage in efforts to search for better routines if extant ones cease to
yield satisfactory results.
Nelson and Winter argue that their own evolutionary theory is superior
to orthodox theory. Their theory is superior not just because of its
alleged greater realism (or realisticness). They argue their evolutionary
theory is superior also because of its greater generality. Nelson and Winter
believe that their own evolutionary theory is more general than orthodox
theory in several respects. In its own domain of application, they argue,
orthodox theory demonstrably only covers limiting cases. Orthodox
theory correctly predicts what happens at the aggregate industry level
only under restricted conditions. Nelson and Winters evolutionary
theory is also argued to cover more ground than orthodox theory. It
addresses phenomena that fall outside the purview of orthodox theory,
such as Schumpeterian competition, technological innovation, and disequilibrium phenomena and processes. On top of that it is also more
general in that it aims at descriptive accuracy not just at the aggregate
level of industries and markets, but also at the level of individual firm
behavior. Friedman argues that a theory of the firm should only be concerned about empirical accuracy at the aggregate level of industries and
markets. Winter, by contrast, argues that ideally a satisfactory theory of
the firm should also be descriptively accurate about individual firm behavior. From a methodological point of view it is interesting to see that Winter
defends the superiority of a more general theory in this sense on Popperian
grounds. Winter explicitly refers to Poppers Logic of Scientific Discovery in
his argument that Friedmans auxiliary hypothesis is unacceptable that
the theory of the firm should only be tested against market phenomena,
and not against phenomena that pertain to the level of individual firms
(Winter 1964, 232). The empirical content of the theory of the firm would
be increased if testing against the latter type of phenomena were taken
seriously: We would certainly prefer a theory which predicted the market
behavior of firms and the results of direct study of their decision processes
to one which predicted only the former (1964, 233).
Nelson and Winter took up the challenge posed by Friedmans selection
argument, ran away with it, and ended up with something that seems to be
the opposite of what Friedman intended. Winter (1982) makes it abundantly clear that he rejects the use that Friedman wanted to make of his
281
selection argument.23 Instead of licensing the continued use of the maximization postulate, Winter argues that taking natural selection seriously
implies both that evolutionary processes in economies should be modeled
explicitly and that observed decision procedures should not always be
rationalized as instances of maximizing behavior. With their explicitly evolutionary theory Nelson and Winter appear as allies of the antimarginalists.
They seem to provide the sort of evolutionary underpinning of decision
rules such as the full-cost pricing rule that Harrod (1939) was hinting at.24
Furthermore, Nelson and Winter take realism (or realisticness) to be a
legitimate criterion for theory appraisal. They insist in particular on the
realisticness of behavioral assumptions in a satisfactory economic theory. In
all this Nelson and Winter seem to move away further from Friedmans
intentions than any of the other authors discussed in this chapter.
6
Conclusions
The interpretations of the proper implications of Friedmans selection

argument for economic theorizing that are discussed in this chapter have
been seen to differ not only from Friedmans own interpretation but also
from each other. Friedman maintained that the argument provided some
(but not conclusive) evidence in support of the marginalist theory of the
firm and left it at that. What the argument was meant to establish in
particular is that even if antimarginalist findings were reliable, i.e. that
businessmen follow particular decision rules and procedures in their
decision-making rather than that they contemplate marginal costs and
revenues, economists are justified in disregarding them. Some commentators argue that Friedman is right that standard economic theory can be
retained, but add to this that the argument suggests an evolutionary
reinterpretation of standard economic theory. Others qualify the claims
Friedman makes in his argument by arguing that standard economic
theory can be retained under some but not all conditions. Especially
Friedmans ultimate claim that natural selection in competitive markets
guarantees productive efficiency and Pareto-optimality has been shown
not to be tenable in general. Still others draw conclusions from evolutionary considerations that are opposite to the ones Friedman draws.
Explicit models of evolutionary dynamics that may help explain why
23
24
Winter prefers an interpretation that he names after Armen A. Alchian in which evolutionary processes, rather than maximization postulates are treated as fundamental. Winter
is trying to drive a wedge between Alchian and Friedman here that I think is not actually
there. Given his intervention in the roundtable discussion Alchian seems to agree.
Remarkably, Nelson and Winter (1982) do not mention Harrod (1939) in their list of
antecedents.
282
Jack Vromen
widely observed decision rules and procedures are prevalent should

supersede standard economic theory.
With the possible exception of the evolutionary reinterpretation, it is not
likely that Friedman was happy about the directions in which his commentators took his argument. It seems Friedmans argument had some sort of
boomerang effect. In the end his argument raised more worries and doubts
about the economic theory he favors than it was able to put to rest. That at
least some commentators may have misrepresented Friedmans argument
is not of much avail here. It also remains to be seen whether faithfulness in
representation should be a supreme goal in this case. Friedmans somewhat
cryptic exposition of the argument leaves considerable room for different
interpretations. More importantly, given the fact that Friedman himself
indicates that not too much hinges on the validity and cogency of the
argument, why would we believe that he is very careful and precise in his
exposition of the argument? Perhaps it does not matter too much whether
or not the interpretations of the implications of Friedmans argument for
economic theorizing discussed here get the gist of his argument right. They
have yielded insights that are interesting in their own right.
References
Alchian, Armen A. (1950). Uncertainty, evolution and economic theory. Journal
Aumann, R. (1985). What is game theory trying to accomplish? In K. J. Arrow and
S. Honkapohja (eds.), Frontiers of Economics. Oxford: Blackwell
(1997). On the state of the art in game theory: an interview with Robert Aumann
(taken by Eric van Damme). In W. Albers, W. Gth, P. Hammerstein,
B. Moldovonu and E. van Damme (eds.), Understanding Strategic Interaction:
Essays in Honor of Reinhard Selten. Berlin, etc.: Springer Verlag, 834
Banerjee, Abhijit and Jrgen Weibull (1995). Evolutionary selection and rational
behavior. In Alan Kirman and Mark Salmon (eds.), Learning and Rationality
in Economics. Oxford: Blackwell, 34363
Baumol, W. J. (1983). Book review of Nelson, R. R. and Winter, S. G. (1982).
Journal of Economic Literature, 21, 5801
Becker, Gary S. (1962). Irrational behavior and economic theory. Journal of
Political Economy, 70, 113
Binmore, Ken (1994). Game Theory and the Social Contract, vol. I: Playing Fair.
Cambridge, MA: MIT Press
Blume, Lawrence E. and David Easley (1992). Evolution and market behavior,
Journal of Economic Theory, 58, 940
(1995). Evolution and rationality in competitive markets. In Alan Kirman and
Mark Salmon (eds.), Learning and Rationality in Economics. Oxford:
Blackwell, 32462
283
(1998). Optimality and natural selection in markets. EconWPA, report-no: cae-001

Cressman, Ross (2003). Evolutionary Dynamics and Extensive Form Games.
Cambridge, MA: MIT Press
Dawkins, Richard (1976). The Selfish Gene. Oxford: Oxford University Press
(1995). Gods utility function. In River of Eden. London: Phoenix, 11155
Enke, S. (1951). On maximizing profits: a distinction between Chamberlin and
Robinson. American Economic Review, 41, 56678
Foster, Dean P. and H. Peyton Young (1990). Stochastic evolutionary game
dynamics. Theoretical Population Biology, 38, 21932
Frank, R. H. (1988). Passions within Reason. New York: W.W. Norton and Co.
Friedman, Milton and Leonard Savage (1948). The utility analysis of choices
involving risk, Journal of Political Economy, 56, 279304
Positive Economics. Chicago: University of Chicago Press, 343
Hall, R. L. and C. J. Hitch (1939). Price theory and business behavior. Oxford
Economic Papers 2
Hammond, J. D. (1991). Early drafts of Friedmans methodology essay. Unpublished
paper
Hansen, R. G. and W. F. Samuelson (1988). Evolution in economic games.
Journal of Economic Behavior and Organization, 10, 10738
Harrod, R. F. (1939). Price and cost in entrepreneurs policy. Oxford Economic
Papers 2, 111
Hirshleifer, Jack (1977). Economics from a biological viewpoint. Journal of Law
and Economics, 20, 152
Hodgson, Geoffrey M. (1999). Optimization and evolution: Winters critique of
Friedman revisited. In Evolution and Institutions: On Evolutionary Economics
and the Evolution of Economics. Cheltenham, UK, and Northhampton, MA:
Edward Elgar, 17595
Hofbauer, J. and K. Sigmund (1998). Evolutionary Games and Population
Dynamics. Cambridge: Cambridge University Press
Kandori, M., G. J. Mailath, and R. Rob (1993). Learning, mutation, and long run
equilibria in games. Econometrica, 61, 2956
Kandori, Michihiro (1997). Evolutionary game theory in economics. In David M.
Kreps and Kenneth F. Wallis (eds.), Advances in Economics and Econometrics:
Theory and Applications Seventh World Congress of the Econometric Society,
vol. 1. Cambridge: Cambridge University Press, 24377
Koopmans, Tjalling (1957). Three Essays on the State of Economic Science. New
York: McGraw-Hill
Langlois, Richard (1986). Rationality, institutions, and explanation. In Richard
Langlois (ed.), Economics as a process. Cambridge: Cambridge University
Press, 22555
Leonard, Robert J. (1994). Reading Cournot, reading Nash: the creation and
stabilization of the Nash equilibrium. Economic Journal, 104, 492511
Lester, R. A. (1946). Shortcomings of marginal analysis for wage-employment
problems. American Economic Review, 36, 6382
Mailath, George J. (1998). Do people play Nash equilibrium? Lessons from
evolutionary game theory. Journal of Economic Literature, 36, 134774
284
Jack Vromen
Matthews, R. C. O. (1984). Darwinism and economic change. Oxford Economic

Papers (Supplement), 46, 91117
Mayr, Ernst (1961). Cause and effect in biology. Science, 134, 15016
Nash (1950). Non-cooperative games. Unpublished PhD dissertation, Princeton
University
Nelson, Richard R., and Sidney Winter (1982). An Evolutionary Theory of
Economic Change. Cambridge, MA: Harvard University Press
Penrose, E. T. (1952). Biological analogies in the theory of the firm. American
Robson, Arthur J. (1996). A biological basis for expected and non-expected utility.
Journal of Economic Theory, 68, 397424
(2001a). The biological basis of economic behavior. Journal of Economic
(2001b). Why would nature give individuals utility functions? Journal of Political
Economy, 109, 90014
Satz, Debra and John Ferejohn (1994). Rational choice and social theory. Journal
of Philosophy, 7187
Schaffer, M. E. (1989). Are profit-maximizers the best survivors? Journal of
Economic Behavior and Organization, 12, 2945
Simon, Herbert A. (1955). A behavioral model of rational choice. Quarterly
Smith, Vernon L. (1991). Rational choice: the contrast between economics and
psychology. Journal of Political Economy, 99(4), 87797
Vromen, Jack J. (1995). Economic Evolution: An Enquiry into the Foundations of New
Institutional Economics. London: Routledge
(2001). Ontological commitments of evolutionary economics. In Uskali Mki
(ed.), The Economic World View: Studies in the Ontology of Economics.
Cambridge: Cambridge University Press, 189224
(2003). Why the economic conception of human behavior might lack a biological basis. Theoria, 18 (special issue on Darwinism and Social Science),
297323
Weibull, Jrgen W. (1995). Evolutionary Game Theory. Cambridge, MA: MIT Press
Winter, Sidney G. (1964). Economic natural selection and the theory of the
firm. Yale Economic Essays, 4, 22572
(1971). Satisficing, selection, and the innovating remnant. Quarterly Journal of
Economics, 85, 23761
(1975). Optimisation and evolution in the theory of the firm. In R. H. Day
and T. Groves (eds.), Adaptive Economic Models. New York: Academic Press,
73118
(1982). Oral contribution to general discussion: Evolution and human behavior. In R. P. Zerbe (ed.), Research in Law and Economics 4. Greenwich, CT,
and London: JAI Press Inc, 13165
Witt, Ulrich (1986). Firms market behavior under imperfect information and
economic natural selection. Journal of Economic Behavior and Organization 7,
26590
Young, H. Peyton (1996). The economics of convention. Journal of Economic
Perspectives, 10, 10522
11
Expected utility and Friedmans risky

methodology
Chris Starmer
Introduction
In this chapter my aim is to offer some reflection upon Friedmans

methodology, focusing in particular upon his work relating to the theory
of risky choice. With this objective in mind, I revisit two key papers in
which Friedman and Savage (1948, 1952) discuss the empirical support
for expected-utility theory (EUT).1 In these papers there are clear traces
of the methodological position set out in Friedmans 1953 essay (henceforth F53). For instance, a recurrent theme is that EUT should be judged
in terms of its predictive accuracy relative to its intended realm of application. At the same time, both papers are motivated by the scarcity of
direct empirical evidence, and in light of this each paper suggests some
remedy. My interest will focus especially on the second of these contributions, which argues that the normative appeal of EUT axioms can be
read as a source of indirect evidence for the hypothesis.
I will argue that this claim is methodologically problematic partly
because the argument set out in the joint 1952 paper fails to successfully
forge an explicit link from normative appeal of EUT axioms to predictive
success of the hypothesis. A key question addressed in this chapter is how,
if at all, such a link could be established. I suggest the issue is of more than
passing historical interest because there has been an apparent tendency in
modern economics for the selection of descriptively oriented theories to
be guided by normative criteria. Put differently, at times economists
behave as if they believe that normative appeal is a form of empirical
I am grateful to Robin Cubitt, Roberta Muramatsu, and Robert Sugden for helpful discussions leading to the development of this paper. I am grateful to Uskali Maki and participants
at the Conference The Methodology of Positive Economics: Milton Friedmans Essay at
50 at EIPE, Rotterdam, December 1213, 2003. I am grateful to the Leverhulme Trust for
financial support (award number: F/00 204/K).
1
For compactness I will refer to these two papers as FS48 and FS52 respectively.
285
286
Chris Starmer
evidence and I illustrate this with a particular example in the contemporary theoretical literature on risk.
Development of the expected-utility hypothesis
EUT is an extremely widely used piece of economic theory. In one respect

it is easy to see why. Risk is a pervasive aspect of the economic world and
for some time now, EUT has been the standard approach to modeling
decisions involving risks. But, although the theory dates as far back as
Bernoulli (1738), it did not find much favor with modern economists until
the 1950s. This may be partly explained by the fact that, in the form
presented by Bernoulli, the theory presupposes the existence of a cardinal
utility scale; an assumption which did not sit well with the drive towards
ordinalization during the first half of the twentieth century. FS48 suggests
that, prior to that, a firm (but mistaken) belief in the principle of diminishing marginal utility may have been a significant impediment to its acceptance.2 Whatever the precise historical reasons for disinterest, it is clear
that economists became much more interested in the theory following the
demonstration by von Neumann and Morgenstern (1947) that the
expected-utility hypothesis could be derived from a set of appealing
preference axioms.
It will be useful for what follows to provide a thumbnail sketch of key
ingredients in a typical axiomatization. A common approach has been to
model risky alternatives as prospects or lotteries. I will use bold face
letters to represent prospects and any prospect p = (p1,x1; pn,xn) is a
probability distribution over a fixed set of n consequences where p1 represents the probability of consequence x1 and so on.3 The expected-utility
hypothesis can then be derived from three axioms on preferences over
prospects: ordering, continuity, and independence. Ordering requires
there to be a complete and transitive pairwise ranking of prospects;4 the
2
3
4
The rejection of utility maximization as an explanation of choices among different degrees

of risk was a direct consequence of the belief in diminishing marginal utility. If the marginal
utility of money diminishes, an individual seeking to maximize utility will never participate
in a fair game of chance But this implication is clearly contradicted by actual behavior.
Marshall resolved this contradiction by rejecting utility maximization as an explanation
of choices involving risk. He need not have done so, since he did not need diminishing
marginal utility or, indeed, any quantitative concept of utility for the analysis of riskless
choices. (FS48, 280)
Take it throughout that, for all prospects, pi 0 for all i, and i pi=1.
More precisely, this axiom entails that: (1), for any pair of prospects q, r: either q > r or r
> q or both, where > represents the relation is (weakly) preferred to; and (2) for any
three prospect q, r, s, if q > r and r > s, then q > s.
287
addition of a continuity principle5 is then sufficient to guarantee that

preferences can be represented by some function V(.) which attaches a
real-value to every prospect.6 The independence axiom of EUT7 adds a
restriction of the following form. Consider two compound prospects
q = (q, p; s, 1p) and r = (r, p; s, 1p), where q is itself a probability
mixture of two prospects; it results in q with probability p otherwise s.
Similarly, r is a mix of r and s and notice that the probability of s is the
same (i.e. 1p) for both compound prospects. Independence implies that
the common component of the two compound prospects (in this case the
1p chance of s) is irrelevant to their relative ranking which should depend
purely on the ordering of the simple prospects q and r.8 Given this additional assumption, preferences can be represented by the familiar expected
utility function V(p)=ipi.u(xi), where p is any prospect, and u(.) is a
utility function defined over consequences.
Where lies the alleged appeal of this set of axioms or, equivalently, the
appeal of EUT so axiomatized? One, I think enlightening, way to reconstruct it is as follows. The standard method of modern economics has
been to understand behavior through the lens of optimization, that is, to
assume that agents are rational actors moved along by coherent and stable
preferences. Hence, to assume the existence of some preference function
over prospects V(p) is essentially to apply the standard method of economics to decisions over risks. But what justifies, from a normative point
of view, the added restriction of independence?
A classic answer is provided by Samuelson (1952), who argues firmly
that the type of independence condition entailed by EUT is itself a
compelling normative principle of rationality. The nub of the argument
is essentially this. Suppose you have to choose between q and r as
defined above and suppose, for the sake of illustration, that p=0.5 with
the compound lotteries resolved by a coin flip which determines the outcome to be s if tails comes up. Samuelson argues that, if tails does come
up, you wont care which option you chose (because you get s either way)
so you can safely ignore this common element of the prospects when
choosing between q and r, just as independence requires. Once it is
5
6
7
Continuity requires that for all prospects q, r, s, where q > r and r > s, there exists some
probability p such that there is indifference between the middle ranked consequence r and
the prospect (q, p; s, 1p).
V(.) represents preferences in the sense that V(p) V(q) if, and only if, prospect q is not
preferred to p.
Whenever I refer to independence principles later in this chapter, note that I have in
mind Expected Utility Independence conditions and not other types of independence
condition that feature elsewhere in economic theory.
More formally, the independence axiom of EUT entails that for all prospects q, r, s: if q >
r then (q, p; s, 1p) > (r, p; s, 1p), for all p.
288
Chris Starmer
conceded that the outcomes associated with tails can be ignored, it then
seems plain that an agent should choose between the compound prospects
on the basis of their ordering over the simple prospects q and r.
The argument for ignoring the tails outcome, and for (EUT) independence more generally, turns on the proposition that there can be no (rationally justifiable) complimentarity between the outcomes within a prospect
because they are mutually exclusive. If this is conceded, and independence
is accepted as an implication of rationality, then EUT has much more
significance than simply being just one amongst many possible models of
risk preference; EUT then has a claim to be interpreted as the logical
extension of rational economic analysis to the realm of risk. Whether or
not individual axioms of EUT can be defended as requirements of rationality has, of course, been a matter of much debate.9 Fortunately, we need
not enter these tricky debates because my primary concern will be to
examine what follows granting, for the purpose of the argument, that the
axioms of EUT can be taken as appealing principles of rationality.
To the extent that its axioms can be justified as sound principles of
rational choice to which any reasonable person would subscribe, the
axioms provide grounds for interpreting EUT normatively; that is as a
model of how people ought to choose. Some writings have placed emphasis on this normative interpretation of EUT. For example, Savage (1954)
presents what has become one of the most celebrated derivations of EUT
explicitly as an attempt to extend logical reasoning to situations of uncertainty. His primary aim is not to provide an empirical theory for predicting
human behavior, but instead to develop logical tools for deciding between
alternative courses of action:
Decisions made in the face of uncertainty pervade the life of every individual and
organisation. Even animals might be said continually to make such decisions, and the
psychological mechanisms by which men decide may have much in common with
those by which animals do so. But formal reasoning presumably plays no role in the
decisions of animals, little in those of children, and less than might be wished in those
of men. It may be said to be the purpose of this book, and indeed of statistics generally,
to discuss the implications of reasoning for the making of decisions. (Savage 1954, 6)
That said, Savage does accept that EUT may also have some potential
as a simple if crude empirical theory for predicting human behavior,
albeit in a suitably limited domain.10 It is to the interpretation of EUT
as an empirical theory that I now turn.
9
10
For the reader wishing to explore these debates a good place to start is Sugden (1991).
[EUT] can be interpreted as a crude and shallow empirical theory predicting the
behavior of people making decisions. This theory is practical in suitably limited domains,
and everyone in fact makes use of at least some aspects of it in predicting the behavior of
others (1954, 20).
289
EUT as an empirical hypothesis
In two now classic papers coauthored by Friedman and Savage (FS48,

FS52) and published shortly before Savages Foundations of Statistics
(1954) and F53 there is much more focus on the empirical interpretation
of EUT. Both papers interpret EUT explicitly as an empirical hypothesis
and present these papers as crude tests of it. The tests are crude, partly
because of the limited supply of direct evidence suitable for testing the
theory. Nevertheless, the two papers propose complementary strategies
for crude testing.
So how does EUT stand up to empirical scrutiny? In FS48 they argue
that, given a suitable specification of the utility function over wealth, the
theory can explain a set of stylized facts about how people respond to risk,
particularly in the context of gambling and insurance. They present a set
of five stylized facts to be explained and they propose a functional form for
the utility of wealth that is shown to be consistent with the presented facts.
A crucial feature of their proposal is that the utility function contains both
concave and convex segments. As such it allows the same individual to
express both risk-averse (insurance) and risk-seeking (gambling) behavior
one of the key stylized facts they set out to explain.
It is the argument of the second paper (FS52), however, that will be my
primary focus. This paper also presents EUT as an empirical hypothesis to
be evaluated relative to the evidence, but it brings a new and interesting
twist by introducing an explicit distinction between direct and indirect
evidence. Direct evidence comes from comparing the implications of
theory with observations generated within the intended domain of application, while indirect evidence relates to phenomena that are not in the
primary domain of interest. So, to the extent that EUT is intended to
explain market phenomena such as gambling and insurance decisions,
FS48 is in good part an evaluation of EUT on the basis of direct evidence.
In contrast, the main purpose of FS52 is to articulate indirect support
for EUT. An interesting feature of the argument is that it seeks to license
reading the normative appeal of EU axioms as a source of empirical
support. I will refer to this maneuver as the FS-twist. The argument goes
like this. There is, FS52 argue, indirect evidence rendering it plausible
that the hypothesis will continue to fail to be contradicted, in at least some
important domains (FS52, 466). The primary source of this evidence is
the plausibility of a set of postulates that are sufficient for derivation of
the hypothesis (FS52, 466).11 The postulates are a set of axioms
11
They also suggest as an aside that Coherence with rest of economic theory (FS52, 466)
may count as indirect evidence for EUT.
290
Chris Starmer
essentially equivalent to those discussed above. The question I wish to

pursue here is how the plausibility of postulates translates into
(indirect) empirical support for EUT? At first blush this seems slightly
mysterious, but some insight is provided by the following passage from
FS52:
In saying that these postulates are more plausible than the hypothesis to which they
are logically equivalent, we mean that the postulates immediately call to mind a
host of implications (or predictions from the hypothesis) susceptible to casual
observation. With respect to the class of phenomena to which these implications
relate, the hypothesis has had many opportunities for contradiction and has
repeatedly failed to be contradicted. The evidence is indirect because this is not
the class of phenomenon we are primarily interested in using the hypothesis to
predict (FS52, 466)
The suggestion that the postulates immediately bring to mind a host of

implications seems very reasonable at least partly because the axioms
make explicit particular formal properties of EUT which may be less than
transparently obvious from a statement of the hypothesis. For instance,
once I see and comprehend a statement of the independence principle, it
then becomes clear that it is implicit in EUT that common components of
prospects under consideration should be irrelevant to my choice among
those prospects. The argument, however, involves two further claims that
strike me as less obviously justified. One is the assertion that implications
which the postulates bring to mind are susceptible to casual observation, the second is the assertion that such casual observation has broadly
supported the hypothesis.
A difficulty arises in relation to both of these latter assertions because it
is not obvious what to understand by casual observation in this context.
Perhaps the most natural interpretation would be to think that casual
observations refer to instances of behavior thrown up by choices that
people happen to make in the world. But if we put aside, for the moment,
observations from specifically designed experimental tests of axioms, it is
doubtful that the world throws up much data bearing on the validity of
EUT axioms. An important reason for this is that the axioms are (for the
most part) propositions about consistency between choices, and as such
they do not typically tell us how individuals will choose in specific decision
problems. Because of this, it is far from obvious that casual observations of
decisions in the world generate rich opportunities for testing EUT axioms. For example, if we know that an agent prefers some prospect a over
b, and b over c, we can test transitivity if we can also observe their
preference between a and c. But it seems to me that casual opportunities
for collecting such observations are rather infrequent.
291
Another possibility is that casual observation is to be understood as

arising from introspective thought experiments, that is from personal
reflections of roughly the form: would I violate the transitivity principle
in this setting; or would I violate the independence principle in that
setting; etc., etc.? Such introspections, however, could hardly be regarded
as providing objective tests of what the agents behavior would be. If an
agent is, or tends to become convinced of the proposition that I would
not deliberately violate choice principle x, it seems that further introspections of the form would I violate x in this specific case can no longer
be considered independent observations. To the extent that the agent is
concerned to reason coherently, it seems the cognition that I would not
deliberately violate choice principle x will move them to conclusions
consistent with this, in thought experiments.
There is, however, another potential argument articulated in FS52 to
motivate the FS-twist. This works in two steps. The first is to argue that
the axioms are introspectively very appealing (FS52, 468) and seeks to
persuade the reader that each principle is not one he would deliberately
violate (FS52, 469).12 It seems reasonable to suppose that many people
would so endorse the principles. The second step is to assert that the
introspective appeal of the axioms is some reason for supposing that
people do actually tend to avoid flagrant violation of the principle
(FS52, 469).13 This second step appears to involve a questionable leap
from propositions about the intuitive appeal of choice principles to propositions about choice behavior. In order to make this leap explicit, consider
propositions P1 and P2:
(P1) the agent accepts axiom x as a principle they would not deliberately violate
(P2) the agent probably behaves in accordance with axiom x
As I read it, this part of the FS52 paper essentially asserts that because P1
holds, P2 is likely. But while P1 and P2 are not inconsistent, in the absence
of further premises, P2 is not implied by P1. P1 is a proposition about
normative judgments, while P2 is a proposition about behavior. In order
to generate the implication suggested by FS52 it is necessary to introduce
some premise linking normative beliefs with behavior. For example:
12
13
While in FS52 Friedman and Savage discuss each of the EU axioms, a key part of their
argument is concerned with justifying the independence axiom. This seems natural given
that this was and continues to be the most controversial element of the EUT system: it is
also the assumption that gives EUT most of its empirical content.
Although the claim here seems a moderate one given the qualification that normative
appeal is just some reason, I take it there is meant to be a substantive claim here given
that the main argument of the paper seeks precisely to use normative appeal as a source of
evidence.
292
Chris Starmer
(P3) agents rarely behave in contradiction with principles they believe they would
not deliberately violate
Notice, however, that it is not enough just to assume P3: for it to do the job
of converting normative appeal to evidence, it has to be empirically valid
or at least plausible. How would one assess whether this is an empirically
plausible claim? I can think of at least two possibilities. One would be to
refer to a theory of the choosing mind. If one could point to some model of
human mental processes which implied that normative beliefs govern
choice behavior, and show that there is satisfactory empirical support for
it, that would be one way to motivate a principle like P3. This strategy,
however, would involve a major departure from the as-if methodology
famously set out in F53.
It is clear that Friedman and Savage do favor an as-if interpretation of
EUT. For instance, F48 provides this methodological aside with respect
to evaluation of EUT:
An objection to the hypothesis just presented that is likely to be raised by many, if
not most, readers is that it conflicts with the way human beings actually behave and
choose. Is it not patently unrealistic to suppose that individuals consult a wiggly
utility curve before gambling or buying insurance, that they know the odds
involved in the gambles or insurance plans open to them, that they can compute
the expected utility of a gamble or insurance plan, and that they base their decision
on the size of the expected utility?
While entirely natural and understandable, this objection is not strictly relevant.
The hypothesis does not assert that individuals explicitly or consciously calculate
and compare expected utilities. Indeed it is not at all clear what such an assertion
would mean or how it could be tested. The hypothesis asserts rather that, in
making a particular class of decisions, individuals behave as if they calculated
and compared expected utilities and as if they knew the odds. The validity of this
assertion does not depend on whether individuals know the precise odds, much
less on whether they say that they calculate and compare expected utilities or think
that they do, or whether psychologists can uncover any evidence that they do, but
solely on whether it yields sufficiently accurate predictions about the class of
decisions with which the hypothesis deals. Stated differently, the test by results is
the only possible method of determining whether the as if statement is or is not a
sufficiently good approximation to reality for the purpose at hand. (FS48, 298)
This clearly has a great deal of resonance with the methodological position
Friedman sets out in F53: we should not count it against EUT that its
assumptions appear unrealistic; EUT is not to be interpreted as a model
of conscious human decision processes but as an as-if model with the
purpose of predicting behavior; as such, the only relevant test of the theory
is taken to be its predictive performance, though note the important caveat
that predictive performance is to be judged relative to the theorys domain
of application.
293
This as-if strategy entails that theories not be judged in terms of

whether they are defensible models of mental processes. So to invoke a
model of mental process as a defense of the theory would not seem to
provide an interpretation of the FS-twist which is consistent with F53 (or
FS48).
Another possibility would be to interpret P3 as a purely empirical (but
untheorized) principle. This, I suggest, provides an interpretation which is
more consistent with F53 but it is not unproblematic. Part of the problem
with this reading is that evidence from behavioral research runs against a
general claim that peoples behavior satisfies principles that most would
take to be normatively appealing. To support that claim, here are two
illustrations.
A.
Savage and the Allais paradox
A significant part of the debate about independence has centered on the

well-known Allais paradox. Consider the following action/state payoff
matrix where each row represents one of four risky actions (g1 to g4),
and columns represent possible states of the world (with probabilities
given at the top of each column). The numbers in the matrix are to be
read as the state contingent payoffs to each (in, say, 000s $).
Consider a choice between g1 and g2. The independence axiom implies
that since these two actions give the same consequence in the third state of
the world, that third state must be irrelevant to that choice. The same
argument applies to the choice between g3 and g4. Notice that if the third
column is blanked out, the choice between g1 and g2 is identical to the
choice between g3 and g4 (i.e. 0.11 chance of 500 vs. 01. chance of 2500).
Hence, independence implies that if g1 is (not) preferred to g2, g3 is (not)
preferred to g4. There is considerable evidence that many people faced
0.1
0.01
0.89
g1:
500
500
500
g2:
2500
500
g3:
500
500
g4:
2500
Choice 1
Choice 2
Figure 11.1 Action/state payoff matrix
294
Chris Starmer
with pairs of choices with this general structure choose g1 over g2 and g4
over g3 in violation of independence.
It is a matter of historical interest that Savage himself, in chapter 5 of
Foundations, conceded that he also violated independence when first
confronted with the Allais paradox. He reports that having once recognized the inconsistency, he reflected upon the situation and came to the
conclusion that his belief in the normative appeal of independence was
unshaken. Consequently, he determined that he must have made some
error in one of his initial choices. Re-examining the choices, Savage
applies independence, hence reducing both decisions to the choice
(0.11 chance of 500) vs. (01. chance of 2500). Then, Finally, consulting
my purely personal taste, I find that I would prefer the gift of $500,000
and, accordingly, that I prefer Gamble 1 to Gamble 2 and (contrary to my
initial reaction) Gamble 3 to Gamble 4 (Savage 1954, 103). On Savages
own account, his initial choice of g4 over g3 was an error which the
application of normative reasoning allowed him to detect and correct.
This account seems unproblematic in relation to the argument of S54.
In that context, Savage is explicitly concerned with developing a normative theory which may be used to police decisions. The fact that real
decisions may depart from the normative is a prerequisite for there to be
any interesting policing role for his axioms. However, the example provides a counterargument to the proposition that being normatively committed to a decision principle implies conforming behavior, and the
example seems especially compelling when the person concerned is an
eminent decision theorist violating a principle which is the most important
in his own normative theory.
B.
Framing effects
One possible defence against this example would be to suggest that

Savages mistake lay not in his choice, but in his adoption of independence
as normative. An alternative interpretation of the Allais paradox is that it
shows that independence is not a compelling normative principle (indeed,
this is Allaiss own reading). Someone who took this line would have the
option of arguing that the Allais paradox does not demonstrate any
inconsistency between compelling normative principles and choice
behavior. That argument, however, could be quickly dismissed by pointing to violations of description invariance. By description invariance I mean
the proposition that preferences between choice alternatives should not be
altered by redescriptions of the alternatives which leave the objective
choice set unchanged. I take this to be a very basic proposition implicit
in normative decision theories (so basic that it is rarely explicitly stated as a
295
proposition). There is, however, considerable evidence that very minor

changes in the presentation or framing of decisions can have dramatic
impacts upon the choices of decision-makers: such effects are failures of
description invariance. Here is one famous example due to Tversky and
Kahneman (1981) in which two groups of subjects call them groups I
and II were presented with the following cover story:
Imagine that the US is preparing for the outbreak of an unusual Asian disease, which is
expected to kill 600 people. Two alternative programs to combat the disease have been
proposed. Assume that the exact scientific estimate of the consequences of the program are
as follows:
Each group then faced a choice between two policy options. One group
chose between A and B:
If program A is adopted, 200 people will be saved.
If program B is adopted, there is a 1/3 probability that 600 people will be saved,
and a 2/3 probability that no people will be saved.
The other chose between C and D:

If program C is adopted, 400 people will be die.
If program D is adopted, there is a 1/3 probability that nobody will die,
and a 2/3 probability that 600 people will die.
The two pairs of options are stochastically equivalent. The only difference
is that the group I description presents the information in terms of lives
saved while the information presented to group II is in terms of lives lost.
Tversky and Kahneman found a very striking difference in responses to
these two presentations: 72 percent of subjects preferred option A to
option B, while only 22 percent of subjects preferred C to D. It seems
hard to deny that different behavior in the two problems is normatively
unsound. Yet similar patterns of response were found amongst groups of
undergraduate students, university faculty, and practicing physicians.
Real people it seems, do not in general behave consistently with normative
principles, even those that find very common assent.
Market selection and the FS-twist
Taken at face value, the evidence from simple choice experiments seems
to tell against the FS-twist. Yet, one might accept that it is possible to
construct experimental situations where people violate normative principles, but raise doubts about whether such violations would be common
features of the wider world. One reason for such doubt might be connected with Friedmans famous natural selection argument. In F53, he
296
Chris Starmer
argues that part of what may explain the success of the profit maximization
hypothesis is that market forces will promote the survival (demise) of firms
using strategies closest to (furthest from) profit maximization. The predictive success of the theory of profit maximization then relies partly on
features of market processes, not explicitly modeled in the economic
theories of firms and markets.
Analogous arguments could be constructed to suggest that EUT might
have better predictive success in a marketplace, relative to laboratory tests,
perhaps by virtue of the existence of market mechanisms that punish
deviations from it. Indeed, widely rehearsed money pump arguments
seem to have just this character. Such arguments typically work by suggesting that agents who fail to satisfy normative decision principles lay
themselves open to exploitation. Consider, for example, an agent who has
non-transitive preferences of the form q >r, r > s and s > q. Suppose
they are initially endowed with q. Repeatedly trading according to these
binary preferences could lead them to: (i) pay to swap q for s; (ii) pay to
swap s for r; and (iii) pay to swap r for q, and so on. This agent has then
been money pumped because they have paid for several trades only to find
themselves returned to their original endowment (minus some cash).
One might argue that if agents were exploitable in this way, we should
expect to see other people exploiting them. The lack of evidence of money
pumps in operation in the world might then be taken as evidence that on
the whole people must have preferences conforming at least roughly with
normative presumptions most of the time. I think these arguments are
unconvincing as attempts to suggest empirical support for EUT axioms
not least because careful theoretical analysis seems to defeat the claim
central to money-pump arguments that agents will be money-pumpable
unless their preferences satisfy EUT. Contrary to this view, Cubitt and
Sugden (2001) show that agents can have preferences which satisfy few, if
any, of the standard normative criteria, yet still be money-pump proof.
But even if money-pump arguments fail as a convincing blanket defense
of EUT, it is not entirely unreasonable to suppose that some market
mechanisms might actually promote conformity with EUT. While it is
currently hard to find much direct support for that conjecture, there is
evidence that at least some preference anomalies do appear to decay in the
context of some market mechanisms. One of the clearest examples relates
to contingent valuation methodology: a standard methodology for assigning money values to non-marketed goods. The two most widely used
valuation measures are willingness to pay (WTP) and willingness to
accept (WTA). Standard preference theory based on indifference curve
analysis implies that these two measures should generate similar valuations, but in practice WTA is typically much bigger than WTP. But
297
further research has shown that the gap between WTP and WTA tends to
be eroded as a consequence of exposure to some market mechanisms,
particularly the second-price variant of the Vickrey auction (see, for
example, List and Shogren 1999; Shogren et al. 2001).
While the existing literature suggests that more than one causal mechanism may lie behind the convergence of WTA and WTP in repeated
auctions, there is some basis for thinking that part of the explanation
involves subjects learning to behave more consistently with principles of
rationality. For example, one factor sometimes cited as a possible account
of the WTA/WTP disparity is that bidders may approach a market with a
tendency to behave cautiously by underbidding when buying and overasking when selling relative to their true values. Such a tendency may be
rationally justified in some market contexts, for instance in markets where
prices are negotiated through some process of haggling. In Vickrey auctions, however, revealing true value is a weakly dominant strategy and
bidders operating cautiously may miss valuable opportunities to trade
(e.g. by asking too high and then failing to sell at a price which is above
their value). A possible explanation of the convergence of WTP and WTA
is that missed opportunities of this sort cause agents to reflect on their
behavior and consequently adjust towards the optimal bidding strategy.
Some support for this explanation of convergence has been found in a
study conducted by Loomes, Starmer, and Sugden (2003).
In principle, then, selection reasoning is a genre of argument that might
work to forge, what seems to me, a missing link in the FS-twist argument
of FS52 by giving some account of how normatively appealing principles
may come to be manifest in market behavior even though they are widely
violated in simple choice experiments. While Friedmans use of this kind
of reasoning in F53 suggests that he may have been sympathetic to this
strategy, the text of FS52 provides no obvious indication that Friedman
and Savage had such a link in mind in this context. Even if they did, the
link would remain weak in the absence of more explicit (theoretical and/or
empirical) support for selection processes in markets specifically promoting behavior consistent with EUT.
5
Normative notions in descriptive dialogues
Let us take stock. In FS52 Friedman and Savage seek to argue that
because EUT can be restated in terms of normatively appealing axioms,
this provides a source of indirect support for the model as an empirical
theory of behavior. This is what I have called the FS-twist. I think that is a
questionable claim for at least three sets of reasons. First, the connection
from norm to decision, at least in FS52, seems mysterious. Second, while
298
Chris Starmer
it might be possible to articulate some connection, say by specifying a

model of decision process with that feature, my impression is that the
literature which does seek to model decision processes would typically
point in the opposite direction: in general, models of decision process
provide many varied reasons for thinking that behavior will deviate from
normatively appealing criteria.14 Thirdly, I have pointed to experimental
evidence against the presumption that normative appeal implies empirical
validity. At the end of the last section I did concede that selection
arguments might in principle provide a missing step. The link made in
that way, however, would remain an extremely tenuous one, absent a
clearer theoretical statement of it and an assessment of its empirical
plausibility.
The argument made by FS52 is of more than passing historical interest
however. Indeed, it seems to me that developments in the economic
theory of risk behavior, over recent decades, have been influenced in
significant ways by a widely presumed (but I think misconceived) connection between normative principles and actual behavior. For example,
since the late 1970s, numerous alternatives to EUT have been proposed
(largely motivated by empirical counterexamples such as Allaiss examples).15 In these developments, it seems clear that economists have paid
almost no attention to modeling phenomena that are inconsistent with
principles that most economists take to be normatively compelling. For
instance, although there is evidence that people violate principles like
monotonicity and description invariance, economists have not been
busy attempting to explain these findings.
Moreover, theories which have had the potential to explain violations of
normative principles have been heavily criticized in the literature. For
example, most would agree that the principle of monotonicity is a normatively appealing property in a theory of risky decision-making. Simply put,
monotonicity means that objective improvements to a prospect (e.g.
increasing at least some of its payoffs, holding others constant) should
not reduce its attractiveness. Some non-expected-utility theories, however, do predict failures of monotonicity, though in these cases the prediction has usually been a by-product of the model: an unintended
consequence of the theoretical structure adopted to explain some other
phenomenon.
14
15
For example, in Payne, Bettman, and Johnsons (1993) model, agents may draw on a wide
variety of heuristics based on tradeoffs between accuracy and effort. But while here
there is an economizing motive, there is no reason to expect the degree of coherence in
choice implied by EUT. Similarly, in Gigerenzer and Selten (2000), agents are conceived
of as adopting fast and frugal heuristics.
I review some of these developments in Starmer (2000).
299
One example is the model that extends EUT by attaching simple

decision weights to probabilities such that agents are assumed to maximize a function of the form: V(p) = w(pi)u(xi). This is essentially the
type of value function assumed in prospect theory of Kahneman and
Tversky (1979) and they show that the Allais paradoxes can be explained
by non-linearities in w(pi). But this strategy for explaining the Allais
examples has the side-effect that predicted choices among at least some
prospects will violate monotonicity. This has generally been regarded as a
fatal flaw sufficient to damn theories of this type. Arguments to this effect
have been made by Fishburn (1978), Machina (1982), and Quiggin
(1982), each of them heavyweights in the arena of choice theory.
No doubt conscious of these potential criticisms, Kahneman and
Tversky propose an editing heuristic which eliminates dominated options,
so long as they are detected (the dominance heuristic). This strategy for
inducing monotonicity, however, has attracted further criticism from
some economists. Quiggin (1982, 1993), for example, has criticized the
approach on two counts. First, he argues that by appropriate specification
of the preference function the dominance heuristic can be rendered
redundant. Second, he criticizes the Kahneman and Tversky strategy for
imposing monotonicity because it has the further spin-off effect that the
theory then admits violations of transitivity in pairwise choice. Quiggin
(1982, 327) describes this as an undesirable result.
What is the basis of this latter claim? It is not, I contend, based on sound
empirical evidence. First of all there is well-established evidence Tversky
(1969) produced some of the earliest that cyclical choice is a robust and
reasonably general phenomenon (for later evidence see Loomes et al.
(1989, 91)). It is true that the form of intransitivity predicted by prospect
theory is of an unusual kind and distinct from other forms of intransitivity.
That, however, presents a useful opportunity for testing a novel prediction
of prospect theory. Starmer (1999) reports an experiment which tests for
the specific form of intransitivity implied by prospect theory and finds very
strong evidence of it.
Notwithstanding this evidence, the economics literature continues to
be dominated by models built on transitivity and other empirically questionable but normatively attractive principles. In light of this, I am
inclined to the view that economists reactions to evidence, in the theory
of choice under risk, have been mediated by certain theoretical precommitments. In particular, a pre-commitment to preference theories
which satisfy normatively appealing criteria such as monotonicity, etc.
This pre-commitment has, in my view, delimited certain problems as
interesting (i.e. those that appeared at least potentially soluble in terms
300
Chris Starmer
of the pre-committed assumptions) and others as uninteresting or outside

the realm of economic inquiry.
If it is granted that such a pre-commitment has been at work, is that a
bad thing? I am not certain that it is. Some philosophers of science, among
them Kuhn and Lakatos, provide strong arguments for thinking that some
kinds of theoretical pre-commitments may be prerequisites for healthy,
ongoing, scientific inquiry. On a Lakatosian interpretation, for instance,
principles of rational choice like monotonicity and transitivity, might be
taken as part of the hard core of the research program of choice under risk.
What is questionable, however, is any tendency to mistake such a precommitment for evidence in favor of it.
Conclusion: reflections on positive economics
Friedmans 1953 essay is a remarkably rich and enduring piece. And,

while it is long, and complex, and deals with many subtle issues, it sets an
apparently clear line on certain methodological principles: positive economics is fundamentally empirical and should involve confronting theories with evidence from the world; the predictive ability of a theory (in its
intended domain of application) is the key test of a theorys success;
attempting to assess the realism of assumptions is not a meaningful test
of any theory.
While F53 is primarily an exercise in abstract normative methodology,
the papers which have been the primary focus of this discussion (FS48 and
FS52) are methodology in action in so far as they involve explicit attempts to
evaluate a particular theory, specifically, EUT. The contrast between the
abstract and applied methodology is both striking and interesting. The
abstract F53 essentially bids the economist, qua positive scientist, ask one
primary question of a proposed theory: how well does it fit the facts in its
intended domain of application? When it comes to the theory of risk, however, while Friedman and Savage explicitly endorse the key principles of
positive economics ( la F53), its application appears complicated by the
scarcity of a crucial ingredient: that is, direct evidence bearing on the
predictive success of EUT in domains of intended application. This looks
like a significant hurdle to get over if Positive Economics is to work in the
domain of risky choice behavior. I suggest that both FS48 and FS52 can
be read as attempts to respond to this, albeit in rather different ways. I
have focused on the argument of FS52 not least because that paper
involves strategies that might seem to present marked contrast with the
prescripts of F53, such as appeals to indirect evidence, to casual observation, to introspection, and to the intuitive appeal of axioms.
301
The simple fact that economics in action (that is the business of constructing and evaluating theories of economic behavior) turns out to be
more complex and messy than Positive Economics la F53 need not
necessarily undermine faith in its key methodological assertions. All must
surely accept that theorizing necessarily involves simplification and
abstraction from complex reality whether it be theory about the ideal
behavior of an economy, or theory about the ideal behavior of economists,
or whatever. Yet the persuasiveness of Friedmans methodology might be
reduced to the extent that his own economics in action were plainly at
odds with the prescriptions of F53.
An obvious question then is whether the FS-twist ultimately coheres
with the Positive Economics of F53? One difficulty in assessing this
arises from the fact that the FS-twist is not fully articulated in FS52:
that is, the paper does not provide a clear account of why we should
expect behavior to conform with normatively appealing axioms. I have
argued that there may be different ways of forging such a link and I have
discussed two particular possibilities: one way would be to develop a
theory of choosing mind, another would be to invoke a theory of
selection. I have suggested that the second of these routes would provide
a reading of FS52 which coheres much more naturally with F53. On this
reading, does the FS-twist then provide a satisfactory empirical defense
of EUT? I think not as it stands, because it begs a further question about
whether there are, in fact, selection mechanisms which operate in
domains of interest to promote conformity between behavior and
normative decision principles. I take it that this, in turn, is a question
susceptible to positive economic analysis and not something simply to
be taken on faith.
References
Appleby, L. and C. Starmer (1987). Individual choice under uncertainty: a review
of experimental evidence, past and present. In J. Hey and P. Lambert (eds.),
Surveys in the Economics of Uncertainty. Oxford: Blackwell
Bernoulli, Daniel ([1738] 1954). Exposition of a new theory on the measurement
of risk. Econometrica, 22, 236
Cubitt, R. and Sugden R. (2001). On money pumps. Games and Economic
Behavior, 37, 12160
Fishburn, P. C. (1978). On Handas New Theory of Cardinal Utility and the
maximization of expected return. Journal of Political Economy, 86, 3214
Positive Economics. Chicago: Chicago University Press
302
Chris Starmer
Friedman, Milton and L. Savage (1948). The utility analysis of choices involving
risk. Journal of Political Economy, 56, 279304
(1952). The expected-utility hypothesis and the measurability of utility. Journal
Gigerenzer, G. and R. Selten (eds.) (2000). The Adaptive Tool Box. Cambridge,
MA, and London: MIT Press
Kahneman, Daniel and Amos Tversky (1979). Prospect theory: an analysis of
decision under risk. Econometrica, 47(2), 26391
List, J. and J. Shogren (1999). Price information and bidding behavior in repeated
second-price auctions. American Journal of Agricultural Economic, 81, 9429
Loomes, G., C. Starmer and R. Sugden (1989). Preference reversal: informationprocessing effect or rational non-transitive choice? Economic Journal, 99,
14051
(1991). Observing violations of transitivity by experimental methods.
Econometrica, 59, 42539
(2003). Do anomalies disappear in repeated markets? Economic Journal, 113,
C15366
Machina, M. (1982). Expected utility theory without the independence axiom.
Neumann, J. von and O. Morgenstern (1947). The Theory of Games and Economic
Behavior, 2nd edition. Princeton: Princeton University Press
Payne, J., J. Bettman, and E. Johnson (1993). The Adaptive Decision Maker.
Quiggin, J. (1982). A theory of anticipated utility. Journal of Economic Behavior and
Organization, 3(4), 32343
(1993). Generalized Expected Utility Theory. Dordrecht: Kluwer
Samuelson, P. A. (1952). Probability, utility, and the independence axiom.
Savage, L. (1954). The Foundations of Statistics. New York: Wiley
Shogren, J., S. Cho, C. Koo, J. List, C. Park, P. Polo and R. Wilhelmi (2001).
Auction mechanisms and the measurement of WTP and WTA. Resource and
Energy Economics, 23, 97109
Starmer, C. (1999). Cycling with rules of thumb: an experimental test for a new
form of non-transitive behaviour. Theory and Decision, 46, 14158
(2000). Developments in non-expected utility theory: the hunt for a descriptive
theory of choice under risk. Journal of Economic Literature, 38, 33282
Sugden, R. (1991). Rational choice: A survey of contributions from economics
and philosophy. Economic Journal, 101, 75185
Tversky, Amos (1969). Intransitivity of preferences, Psychological Review, 76,
3148
Tversky, Amos and Daniel Kahneman (1981). The framing of decisions and the
psychology of choice. Science, 211, 45358
(1986). Rational choice and the framing of decisions. Journal of Business, 59(4),
pt. 2, 25178
12
Milton Friedmans stance: the methodology

of causal realism
Kevin D. Hoover
The God of Abraham; the methodology of Marshall1
The philosopher Bas Van Fraassen opens his Terry Lectures (published
as The Empirical Stance) with an anecdote: When Pascal died, a scrap
of paper was found in the lining of his coat. On it was written The God
of Abraham, Isaac and Jacob, not the God of the philosophers (Van
Fraassen 2002, 1). Pascals God talks and wrestles with men; Descartess
God is a creature of metaphysics. Analogously, with respect to the
Methodology of positive economics (F53) there are two Friedmans.
Most of the gallons of ink spilled in interpreting Friedmans essay have
treated it as a philosophical work. This is true, for example, for those
who have interpreted it as an exemplar of instrumentalism, Popperian
falsificationism, conventionalism, positivism, and so forth. And it is
even true for those critics, such as Samuelson (1963), whose credentials
as an economist are otherwise secure. Mayer (1993a, 1995, 2003) and
Hands (2003) remind us that Friedman was philosophically unsophisticated, and in the essay Friedman tried a fall with other economists, not
with philosophers. The origin of the essay was the quotidian practice of
economics, not abstract epistemology. To know the Friedman of the
economists, I propose to read the essay in light of Friedman (and
Schwartzs) A Monetary History of the United States, 18671960 (1963a)
and Money and business cycles (1963b), perhaps his most characteristic economic investigations.
My point is not that there is a particular philosophers position that
can be contrasted with a particular economists (or even Friedmans)
I thank Uskali Mki, Thomas Mayer, and the participants in the F53 Conference at the
Erasmus Institute of Philosophy and Economics, Rotterdam, December 1213, 2003 for
comments on an earlier draft. I also thank Ryan Brady for research assistance. J. Daniel
Hammond (1996) wrote an important book that examines Friedmans economic thought
from a causal perspective. While I intentionally did not reread that book in trying to
develop my own views somewhat independently, I could not help but be influenced by it.
303
304
Kevin D. Hoover
position. After all, Pascal surely recognized that there were a variety of
views of God among the philosophers. Nor would I suggest that philosophy has nothing valuable to say about, or learn from, Friedmans essay
nor Friedman about or from philosophy. It is rather a matter of approach.
Friedman cares most about doing economics and is largely innocent
of the interests of philosophers and, equally, of the distinctions and
nuances that philosophers routinely employ. If we either as economists
or philosophers insist on reading the essay as a philosophical work
addressed to philosophers, we must misread it.
The most infamous passage in Friedmans essay runs: Truly important and significant hypotheses will be found to have assumptions that
are wildly inaccurate descriptive representations of reality, and, in general,
the more significant the theory, the more unrealistic the assumptions (in
this sense) (14).2 Samuelson (1963, 2323) sees this as the extreme
version of Friedmans general proposition that [a] theory is vindicated
if (some of) its consequences are empirically valid to a useful degree of
approximation; the (empirical) realism of the theory itself, or of its
assumptions, is quite irrelevant to its validity and worth, which he
stigmatizes with the shorthand F-twist.3
The view that Friedman does not care about the truth of the assumptions faces a serious difficulty. After recounting the detailed evidence
on the cyclical behavior of money and the real economy, Friedman and
Schwartz (1963b, 21314) raise the question, how we can be sure about
the direction of influence?
It might be, so far as we know, that one could marshal a similar body of evidence
demonstrating that the production of dressmakers pins has displayed over the past
nine decades a regular cyclical pattern; that the pin pattern reaches a peak well
before the reference peak and a trough well before the reference trough; that its
amplitude is highly correlated with the amplitude of the movements in general
business. It might even be demonstrated that the simple correlation between the
production of pins and consumption is higher than the simple correlation between
autonomous expenditures and consumption; that the partial correlation between
pins and consumption holding autonomous expenditure constant is as high
2
3
Except where context would render it ambiguous, references to F53 will be referred to
hereinafter by page numbers only.
Replacing Friedman by his initial is an indication that Samuelson (1963, 232) is aware that
the F-twist may be a misinterpretation of [Friedmans] intention. While Samuelson rails
against Friedmans apparent abandonment of truth and realism, Frazier and Boland
(1983) turn it into a positive virtue. They interpret Friedman as an instrumentalist and
defend instrumentalism as a logically sound doctrine. My suggestion here is that they have
misinterpreted Friedman he is not an instrumentalist and my suggestion in Hoover
(1984) is that Frazer and Bolands version of instrumentalism is not sound doctrine.
Milton Friedmans stance: the methodology of causal realism
305
as the simple correlation; and that the correlation between consumption and
autonomous expenditures holding the production of pins constant is on the
average zero [B]ut even if [these statements] were demonstrated beyond a
shadow of a doubt, they would persuade neither us nor our readers to adopt a
pin theory of the cycle.
But why not? In Friedman and Schwartzs thought experiment, the pin
theory of the cycle has implications that are confirmed facts. Why should
the same variety of evidence that supports a monetary theory not equally
well support a pin theory? Primarily, the difference is that we have other
kinds of evidence (Friedman and Schwartz 1963b, 214): (i) pins are a
trifling element in the economy, and [w]e expect the effect to be in rough
proportion to the cause, but money is pervasive in economies that
experience business cycles; (ii) with money we can, and with pins we
cannot, conceive of channels through which large autonomous changes
in them might affect the economy; (iii) in contrast to a monetary theory,
no serious student of business cycles has ever seriously suggested a pin
theory.
It is not good enough for Friedman that a pin theory implies the facts.
His criticisms are based on evidence relevant to the truth of its assumptions. The Monetary History can be seen as a detailed marshalling, not only
of the facts implied by Friedmans monetary theory, but of the evidence
for the truth of its underlying mechanisms and initial conditions: its
assumptions. A partisan for the assumptions-dont-matter interpretation
of Friedmans essay might argue that there is a flat contradiction between
his rejection of the pin theory and his methodological position. That
would be too easy and too lazy. To say that Friedman does not care
about the truth of the assumptions is wrong. On the contrary,
Friedmans methodological stance in the essay is best described as causal
realism, which can be defined as the view that the object of scientific
inquiry is the discovery through empirical investigation of the true causal
mechanisms underlying observable phenomena.
One object of this chapter is to support this controversial claim. To do
so, I propose to read Friedmans essay charitably not as the work of an
amateur philosopher but as the work of an economist, one whose methodological reflections are consistent with his intellectual antecedents and
with his own empirical practice. Friedmans most important intellectual
antecedent is Alfred Marshall. Marshalls writings are Friedmans economic Bible. When, after a long and illustrious life, Friedman finally
passes on, he may well leave a scrap of paper pinned to the lining of his
coat on which will be written: The Methodology of Marshall, not the
Methodology of the philosophers.
306
Kevin D. Hoover
The absence of causal talk
If Friedman is a causal realist, there surely is a puzzle in his work: he

rarely talks about causes. In the 814 pages of Monetary History, a work
that I maintain is an exemplary piece of causal realist research, the words
cause or causal are used as far as I can tell only nine times, and
four of those nine are attributed to agents of the Federal Reserve (see
table 12.1).4 Similarly, Friedmans claim that the effect should be proportional to cause cited earlier is one of only two instances of causal talk
in Money and business cycles. What is more, Friedman is conscious
of his own unease with respect to causal talk. Tobin (1970, 301) asserts that
Friedman claims that changes in the money supply are the principal cause
(sometimes unique) of changes in money income. In reply, Friedman
(1970, 319) not only denies the accuracy of Tobins claim, but assails the
imprecision of Tobins causal talk. Friedman (1974, 101) writes: I myself
try to avoid use of the word cause it is a tricky and unsatisfactory word
(cited by Hammond 1992, 912, as part of interview with Friedman).
There is less here than meets the eye. Friedman routinely employs a large
variety of synonyms and circumlocutions, so that he clearly engages in causal
talk, even when explicit causal words are not in play. Instead of A causes B,
Friedman frequently writes: A produces (or influences, engenders, affects, brings
about) B; B reflects (or is a consequence of, is a result of, is an effect of) A; A is a
dependent variable, while B is an independent variable. Instead of causal chains
Friedman writes of chains of influence. Table 12.2 gives a number of examples
from the Monetary History, which plainly show that, while Friedman and
Schwartz have avoided the use of cause, causal, and etymologically
related terms, they use a variety of words that must be regarded as synonyms.
Friedmans avoidance of explicit causal talk in the face of clearly causal
intent might be dismissed as a mere quirk. I will argue subsequently that it
is of a piece with his overall Marshallian methodology. But first we must
examine that methodology somewhat more closely.
The F-twist in the light of Alfred Marshall
In describing Friedmans intellectual relationship to Marshall, the metaphor of the Bible was not chosen lightly. Methodologically Friedman is
like a fundamentalist Christian who constructs his theological position
by direct quotation from scripture. Once one has read Marshalls essay,
4
My count is as accurate as I could make it on a single reading of the text, but I cannot
warranty that I have caught every instance. Causal language would be exceedingly uncommon in the Monetary History even if I had shortened the count by half a dozen instances.
307
Table 12.1 Uses of causal language in Friedman and Schwartzs Monetary

History of the United States, 18671960
The proximate cause of the world price rise was clearly the tremendous outpouring of
gold after 1890 that resulted from the discoveries in South Africa, Alaska, and Colorado
and from the development of improved methods of mining and refining. [p. 137]
And, of course, the Board also took the position that the expansion in the stock of money
was a result, not a cause, of rising prices. [p. 223]
[The Federal Reserve Board] argued that changes in U.S. prices were effect rather than
cause, that the Reserve Board was powerless to do more than adapt to them, and that the
Boards policies had prevented financial panic at home and moderate the price changes.
[p. 237]
Doubtless there were effects both ways and common causes as well. [p. 288]
By relevant standards, the discount policy was abnormally tight, not easy. The System
regarded the lack of discounting as a reflection of the large accumulation of excess reserves
and hence as a lack of need for accommodation. That view no doubt had some validity, but
the causal chain ran the other way as well. [p. 514]
To his directors, he [Governor George L. Harrison of the Federal Reserve Bank of New York]
pointed out in December that most of the executives of the Reserve System do not believe
that monetary action would afford relief in the present business situation, regardless of
whether the causes of the recession are monetary or non-monetary but rather that it
was felt that improvement in the business situation would be more influenced by actions
of the Administration. [pp. 5289]
Precisely the same is true of a draft of a memorandum prepared at the New York Federal
Reserve Bank by John H. Williams in answer to the question whether the reserve
requirement changes caused the 193738 depression. [p. 544]
Consideration of the effects of monetary policy on the stock of money certainly strengthens
the case for attributing an important role to monetary changes as a factor that significantly
intensified the severity of the decline and also probably caused it to occur earlier than
otherwise. [p. 544]
Again, if the common movement of the stock of money and of money income and prices is
not coincidental or the consequence of a common cause, the direction of influence
must run from money to income. [p. 687]
Notes: emphasis in bold type added; page numbers in square brackets. The table attempts
to identify every use of explicitly causal language in the Monetary History. It was, however,
constructed from a single reading, so that completeness cannot be guaranteed.
The present position of economics (1885), the methodological antecedents of Friedmans essay and his other well-known methodological
work, The Marshallian demand curve (1949b), are obvious and fairly
complete. The word in Cambridge used to be Its all in Marshall. This
appears to have been true in Chicago as well.
I take the identity of Friedmans and Marshalls methodological positions as a postulate. In juxtaposing arguments from their two essays, I do
not aim to persuade anyone of that identity. Rather I explicate what I take
to be their commonly held views.
308
Kevin D. Hoover
Table 12.2 Examples of synonyms for causal language from Friedman and
Schwartzs Monetary History of the United States, 18671960
[The Monetary History] traces changes in the stock of money for nearly a century, from just
after the Civil War to 1960, examines the factors that accounted for changes, and analyzes
the reflex influence that the stock of money exerted on the course of events. [p. 3]
That outcome was widely regarded as, at least partly, a delayed reaction to the large
wartime increases in the stock of money. [p. 13]
As a matter of economics, there can be little doubt that [the contrast between a rise of
1.1 per cent per year in the money stock and a decline of 3.5 per cent per year in prices]
reflects primarily a rise in output. [p. 34]
On this interpretation the chain of influence ran from the attempted deflation to the
economic stagnation. [p. 41]
Not only did [the deflation of 50 percent in the decade and a half after 1865] not produce
stagnation; on the contrary it was accompanied and produced by a rapid rise in real
income. [p. 41]
the amount of high-powered money is a dependent rather than an independent
variable, and is not subject to governmental determination. [p. 52]
Such shifting expectations could affect the price of gold only as they affected the demand
for and supply of foreign exchange [p. 58]
the government did succeed in bringing about a minor reduction in the stock of highpowered money [p. 81]
The major channel of influence was from the stock of money to the level of money
income to the level of prices, and thence to the rate of exchange between the dollar and
other currencies, though undoubtedly some influences ran in the other direction. [p. 89]
These measures, in turn however, had offsetting effects, since debt redemption reduced
the amount and raised the price of bonds available to serve as backing for national bank
notes, and so led to a reduction in national bank notes from a peak of some $350 million
in 1882 to a trough of some $160 million in 1891. [p. 128]
there was no evidence on the length of the lag between action and effect. [p. 239]
the decline in the stock of money and the near-collapse of the banking system can be
regarded as a consequence of nonmonetary forces in the United States, and monetary
and nonmonetary forces in the rest of the world. [p. 300]
if it did initiate a worldwide disturbance, it would inevitably be affected in turn by reflex
influences from the rest of the world.
the rapid rise in the money stock certainly promoted and facilitated the concurrent
economic expansion. [p. 544]
Notes: emphasis in bold type added; page numbers in square brackets.
Van Fraassen (2002) refers to a stance as a kind of commitment to

a creed that is maintained by faith. Within the empirical stance, certain
kinds of empirical evidence have force, but the stance itself cannot be
justified by evidence of the same kind. Friedmans stance is Marshallian.
But Friedman sees himself to be in a kind of Babylonian exile, singing
Marshalls song in a foreign land, while other economists sacrifice at the altar
of strange gods. Baal in this parable is Walras. Friedman sees the Walrasian
309
stance as the requirement that, to know anything, one must know everything. A theory must articulate a structure that can accommodate every
economic actor in its full particularity. The Walrasian economist recognizes the impracticality of doing this completely. Instead, the Walrasian
offers the perfect general-equilibrium model as a transcendent ideal from
which one can criticize the compromised and inconsistent realities of
applied economics. Walrass is truly the Methodology of the philosophers.
The Walrasian approach suggests viewing the whole economy from an
Olympian height. But Friedman denies that there is any such standpoint
or that anyone could grasp the economy in its totality. Marshall argues
that [t]here is no use in waiting idly for [a unified social science]; we must
do what we can with our present resources. He goes on, common sense
does not deal with a complex problem as a whole. Its first step is to break
the problem into its several parts the human mind has no other method
of inquiry than this (Marshall [1885] 1925, 164).5 For Friedman knowledge is the product of sweaty labor among gritty facts. Marshall (p. 171; cf.
Friedman [1949] 1953b, 90) writes that the economist must stand by the
more laborious plan of interrogating the facts in order to learn the manner
of action of causes singly and in combination. Friedmans and Marshalls
object, then, is the acquisition of causal truth. The barrier is the complexity. Marshall (157) writes that economic causes often lie below the surface
and are likely to be overlooked by the ordinary observer. Similarly, in his
essay Friedman (33) writes:
A fundamental hypothesis of science is that appearances are deceptive and that
there is a way of looking at or interpreting or organizing the evidence that will
reveal superficially disconnected and diverse phenomena to be manifestations of a
more fundamental and relatively simple structure.
The Monetary History can be seen as an attempt to put that hypothesis to the
test. In the Summing up at the end of the book, Friedman and Schwartz
(1963a, 676) echo the Marshallian methodological conclusion: In monetary
matters, appearances are deceiving; the important relationships are often
precisely the reverse of those that strike the eye.
Opposed to a naive empiricism, Friedman and Marshall give theory a
special role. [F]acts by themselves, writes Marshall, are silent. Observation
discovers nothing directly of the action of causes, but only sequences
in time (166) [T]he most reckless and treacherous of all theorists is
he who professes to let the facts and figures speak for themselves (168;
5
Except where context would render it ambiguous, references to Marshall ([1885] 1925)
will be referred to hereinafter by page numbers without dates.
310
Kevin D. Hoover
cf. Friedman [1949] 1953b, 90). In the essay, Friedman concurs: A

theory is the way we perceive facts, and we cannot perceive facts without a theory (34).
What is a theory? Friedman ([1949] 1953b, 91) draws his interpretation
of theory directly from Marshall:
Economic theory, in this view, has two intermingled roles: to provide systematic
and organized methods of reasoning [Marshall, 164; cf. F53, 7] about economic
problems; to provide a body of substantive hypotheses, based on factual evidence,
about the manner of action of causes [Marshall, 171]
Theory, or what Marshall (164) calls the economic organon is not a

body of concrete truth, but an engine for the discovery of concrete truth,
similar to, say, the theory of mechanics (Marshall, 159; quoted in part by
Friedman [1949] 1953b, 90). The ideal theory for Marshall is universal,
saying nothing about particulars, and needing to be supplemented with
particular facts if it is to be useful. Economic theory, however, has not yet
reached universality so that it is necessary to sacrifice generality of form
to some extent (Marshall, 160).
In the essay, Friedman characterizes theory as, in part, a language
(7), an analytical filing system (11) constituted of tautologies that are
useful in organizing refractory empirical phenomena (12) and, [i]n
part, a body of substantive hypotheses designed to abstract essential
features of complex reality (7).
Both Friedman and Marshall conceive of theory as a purely deductive
system, whose claims are universal. Even the particulars that, on the one
hand, fill the lacunae of incomplete theory and, on the other hand, tie that
theory to empirical applications are strained and filtered to pick out what
is essential: the color of the wheat-traders hair or number of members
of his family are not relevant particulars (F53, 32). There is little difference here between Cartwrights (1999, ch. 2) treatment of theory as fables
in the sense of Lessing or some accounts of theories as idealizations
(Nowak 1980; Hoover 1994) views which are compatible with realism.
Friedman (F53, 36) himself refers to perfect competition and monopoly
as ideal types, the application of which to concrete cases requires judgment about their suitability and about the objects of the analysis.
Cartwright thinks that fables are true; Nowak thinks idealizations get
to the genuine heart of the matter. So what do we make of Friedmans
famous denial that the realism of the assumptions matters? The key is
to read Friedmans claim carefully. What he actually argues is that checking the realism of the assumptions does not provide a test in addition to
the empirical implications of a theory. He does not claim that those
assumptions are immune from indirect tests (F53, 14, 23). A theory for
311
Friedman is a deductive system in which some propositions are taken as

axioms (or postulates) from which others can be deduced (F53, 23). The
axioms are part of conceptual world or abstract model simpler than the
real world and tied to it through rules for a class of applications and
correspondences (F53, 24). The parts of a theory are interdeducible, so
that what may, from one point of view, be taken to be an axiom may, from
another point of view, be taken to be a theorem (F53, 267). How to
classify various theoretical propositions is mainly a matter of finding the
most economical mode of stating them (F53, 23). Economical axioms
need not themselves be superficially obvious nor tied individually directly
to reality.
Friedmans view of the status of theoretical axioms is virtually identical
to that of Bertrand Russell (1918, 1456):
When pure mathematics is organized as a deductive system i.e. as the set of
all those propositions that can be deduced from an assigned set of premises it
becomes obvious that, if we are to believe in the truth of pure mathematics,
it cannot be solely because we believe in the truth of the premises. Some of
the premises are much less obvious than some of their consequences, and are
believed chiefly because of their consequences. This will be found to be always
the case when a science is arranged as a deductive system. It is not the logically
simplest propositions of the system that are the most obvious or that provide
the chief part of our reasons for believing in the system Electro-dynamics,
for example, can be concentrated into Maxwells equations, but these equations are believed because of the observed truth of certain of their logical
consequences.
Friedman subscribes both to Russells claim that axioms are often not
obvious (appearances are deceiving) and that they are supported by
indirect evidence. In dismissing as futile Walrasian criticism of the
realism of assumptions, Friedman is attacking naive theoreticism the idea
that a theory should (or must) mirror casually parsed facts about the world.
Instead, Friedman argues that a theory needs to get to the essence of the
matter. Generally, that requires the creation of categories of entities that
are not directly observable, governed by rules that omit irrelevant details,
whose success is to be judged holistically.
In his monetary analysis, Friedman illustrates this strategy. The velocity
of circulation of money, for instance, is a theoretical construct that can
be observed only indirectly ( = nominal GDP/money or = 1/money holdings
expressed as a number of weeks income). Friedman treats velocity as a causally
significant, real category. But the evidence that he offers for its causal and
ontological significance (in the Monetary History or in Friedman 1959,
1966, or Friedman and Schwartz 1982) is the general empirical stability of
312
Kevin D. Hoover
the entire analytical schema of which it is but one part.6 What is more,
Friedman treats velocity as a variable that measured in one context can
be transferred to another context.
Permanent income provides another and perhaps more compelling
example. Like velocity, permanent income is not directly observable, but
is indirectly measured and validated in the context of the consumption
function. Permanent income is not only a causally significant category
in this context, but Friedman regards it as sufficiently freestanding and
independent of its original context, that he routinely uses the measured
quantity in other contexts. For example, Friedman takes permanent
income to be a causal determinant of the demand for money (i.e. permanent velocity = permanent income/money).
Critics who hold that Friedman does not care about the truth of
assumptions or even positively endorses their falsehood in some unqualified sense overlook Friedmans claims about the import of indirect
evidence. In a famous example (F53, 20), Friedman talks about leaves
of a tree arranging themselves as if they were maximizing received light.
Since trees do not maximize, this is often taken to mean that the truth
of the assumption does not matter. But Friedman goes on to say that
we should prefer an explanation in terms of the mechanisms that make
the tree behave as if (counterfactually) it were a conscious maximizer.
Finding and elaborating such mechanisms is useful. First, a theory that
encompasses those mechanisms will be more general and, therefore,
more fruitful in making predictions about a wider variety of phenomena. Second, evidence for the predictive power of such a theory in
other domains lends indirect evidence in favor of the prediction based
on the assumption that trees act as if they maximized received light.
The assumptions underlying such a theory may also be unrealistic in
Friedmans peculiar sense of not giving us a photographically accurate
description of reality, but that does not make them untrue. And, just as
for Russell, the more accessible implications of a theory lend support
to the logically more primitive, but inaccessible assumptions.
4
Friedmans anti-realism reconsidered
If Friedman is a causal realist after all, what then are we to make of

statements in which he denies the necessity of realistic assumptions and
even seems to promote the desirability of unrealistic assumptions?
The goal of science for Friedman (F53, 7) is to construct theories that
yield valid and meaningful predictions about phenomena not yet
6
As an economist, Friedman does not think or write explicitly in these metaphysical terms.
313
observed.7 Such theories are generalizations, which necessarily omit

many particular and, implicit in the theory, irrelevant details. This lack
of particularity or realisticness (to use Mkis apt coinage) is probably
the sense of realism in play when he attacks the need for a theory to
provide photographic descriptions of reality (Friedman [1949] 1953b,
91). Lack of realism for Friedman is just another (and for many philosophers, perhaps paradoxical) way of referring to the desirable property
that a theory captures the essence of a deep relationship.
Marshall (166), in arguing for a theoretical approach to empirical phenomena, notes that every event is the complex result of so many causes, so
closely interwoven that the past can never throw a simple and direct light on
the future. In much the same vein, Friedman (F53, 25) argues that a
model is a half-truth: although there is nothing new under the sun (that is,
generalization is the essence of the model), history never repeats itself
(all generalizations omit particularities). Yet, both Marshall and Friedman
see the immersion of the theorist in the particularities of factual inquiry as
the source of further generalizations. The more the theorist is successful at
eliminating necessary reference to particularities, the more general and
more powerful the theory becomes, and the more nearly it approaches
Marshalls ideal of a universal economic doctrine (Friedman 1953b, 159;
Marshall, 14). Paradoxically, it is unrealisticness that serves and underwrites Friedmans realism with respect to the deep causal mechanisms.
Marshall, of course, did not claim that universal economic doctrine was
ready to hand. Friedman (F53, 34) sees the complexity of economic
phenomena as itself an argument for the incompleteness of economic
theory. To accept the complexity of phenomena as an irreducible fact is
to deny the tentative state of knowledge that alone makes scientific activity
meaningful. By way of illustration, he heaps scorn on John Stuart Mills
claim that the laws of value (in 1848!) were complete and required no future
development. Economic theory is unrealistic in the sense that it is incomplete. But the only completeness that is valuable scientifically is that which
reduces its particularity and simplifies the complexity of phenomena.
The essence of Friedmans Marshallian stance is that the pursuit of
substantive knowledge cannot be stymied by the incompleteness and,
therefore, unrealism of theory. We cannot have a complete, realistic
theory first and apply it to concrete phenomena afterwards. Nor can we
start with atheoretical facts, for there are none; facts are theory-laden. All
we can do is to work back and forth between theory and facts, starting with
the primitive and highly tentative and working toward the sophisticated
7
Friedman (F53, 9) is clear that by prediction he means the inference of the unobserved
from the observed, and not only forecasts of the future from the past.
314
Kevin D. Hoover
and more secure. In this process, we can tolerate a large lack of realism in
another sense. Predictions based on theoretical assumptions that are only
approximate and, therefore, unrealistic may be as accurate as we can use
given our ability to measure or as accurate as we need for some practical
purposes (F53, 1517).
Lack of realism in all the senses just discussed is perfectly compatible
with the project of a constructive causal realism that is, with the project
of seeking to identify the true mechanisms underlying observed phenomena. Friedmans implicit commitment to such a project is clear in his
empirical practice as a monetary economist. Most obviously, what would
be the point of Friedman and Schwartzs massive historical project, not
only of the Monetary History, but of its sister volumes, Monetary Statistics of
the United States (1970) and Monetary Trends in the United States and the
United Kingdom 18671975 (1982), if somehow it were possible to be
unconcerned about the accuracy of the antecedents of economic inferences? One might think of history as just a giant testbed for Friedmanian
models with unrealistic assumptions, but that would miss the modus
operandi of Friedman and Schwartz as historians: far from merely collecting data on which to test predictions, they seek detailed accounts of the
institutions and even the sociological or psychological bases for the
actions of particular actors; in other words, they seek to describe the actual
mechanisms that generated the behavior of the economy accurately and
realistically.
In Friedmans Marshallian methodology, it is possible to have causal
knowledge without knowing all of the links, yet filling in the gaps is a
positive virtue. In his essay (42), Friedman argues that the static monetary
theory is satisfactory, but that we are largely ignorant of dynamic monetary theory. One of the promises of detailed empirical investigations is to
seek new generalizations to amplify theory. His point is not merely methodological. In Friedman and Schwartz (1963b, 222) it is elaborated:
We have little confidence in our knowledge of the transmission mechanism, except
in such broad and vague terms as to constitute little more than an impressionistic
blueprint. Indeed, this is the challenge our evidence poses: to pin down the transmission mechanism in specific enough detail that we can hope to make reasonably
accurate predictions of the course of a wide variety of economic variables on the
basis of information about monetary disturbances. (Cf. Friedman and Schwartz
1963a, 6789.)
The dual call for realistic elaboration of the assumptions of the theory and,
simultaneously, for securing breadth of generalization could not be clearer.
In principle, any theoretical propositions are at risk of being overthrown
in the face of predictive failure. Although the collapse of a bridge might be
315
taken as evidence by some of a failure of the theory of mechanics, Marshall

(15960) cautions that the theory ought to be regarded as secure in this
case and its application questioned.
Friedman maintains a similar position with respect to elementary
economic theory. He is well known to have attacked the utility of elaborating price theory with a theory of imperfect competition (F53, 349).
This is an empirical claim of the lack of fruitfulness of imperfect competition. It could not, consistent with Friedmans own principles, be upheld if
imperfect competition were discovered to underwrite better predictions.
It is important to understand that Friedman does not dismiss criticism
of analysis that divides the world into either perfect competition or monopoly, only because good predictions can be had by acting as if these
assumptions were true. Rather Friedman makes a conceptual claim that
the alternative assumption, imperfect competition, undermines the practical division of the economy into distinct industries and that, without this
division, practical testing of the hypothesis is impossible. Since every case
would be sui generis, imperfect competition could not support a general
rule needed for prediction. Friedman does not deny the desirability of true
assumptions in this case, instead he questions whether certain assumptions lend themselves to even indirect inquiry into their truth. For both
Friedman and Marshall facts are theory-laden: the choice of theoretical
assumptions helps to define what constitutes the facts. Friedman argues
for one set of assumptions and against another set of assumptions on the
basis of each sets effectiveness as a potential servant of the truth.
Friedmans deepest theoretical commitments are in principle, but not in
practice, at risk in theoretical inquiry. Auxiliary assumptions are, in contrast, very much at risk. One nice example is provided by Friedman and
Schwartzs comparison of the banking crises in the contractions of 19078
and 192933. The assumptions underlying their analysis of each recession
are adjusted to fit the facts of the other. Such adjustments provide the basis
for conjectural histories in which one can speculate how things might have
gone differently under different policies. Such conjectures are grounded in
common assumptions about the mechanisms governing different episodes.
Friedman and Schwartz (1963a, 168) conclude:
all truly analytical history, history that seeks to interpret and not simply to record
the past, is of this [conjectural] character, which is why history must be continuously rewritten in the light of new evidence as it unfolds.
Friedman and Schwartz practice what they preach. The Monetary History
provides numerous examples of counterfactual analysis, the most famous
example of which is their extended analysis of how the US economy might
not have collapsed so deeply in the early 1930s had Benjamin Strong,
316
Kevin D. Hoover
the influential (and, to Friedman and Schwartz, right-minded) President

of the Federal Reserve Bank of New York, not died in 1928 (Friedman
and Schwartz 1963a, ch. 7). Counterfactual experiments of this sort
make no sense without a commitment to the reality of a causal structure
that can be faced with alternative initial conditions (see Hoover 2001, esp.
chs 2 and 4).
5
Effects, not causes
We must now revisit the puzzle of why, if Friedman is a causal realist, he is

so averse to using causal language. In his interview with Hammond (1992,
92), Friedman said: The problem that bothers me about cause is that
it almost invariably leads into a problem of infinite regress. There is no
such thing as the cause of anything. Friedman sees every proximate
cause standing in a chain of influence. What is more, influences between
two variables can go in both directions. While much of Friedman and
Schwartzs work has aimed to document the dominant influence of money
over income and business activity, they have candidly acknowledged that
influence runs from income and business activity to money as well (e.g.
Friedman and Schwartz 1963a, 695; 1963b, 214; Friedman 1970, 321).
It is not only the existence of causal chains that concern Friedman; he
also worries that independent factors might influence any variable at one
time and that there are no unambiguous ways to count causes (Friedman
1970, 319).
Friedmans squeamishness with respect to causal language can be seen
as consistent with his Marshallian stance. The problem with talking about
causes is that, working backwards in time, the chain never ends and,
working contemporaneously, the array of causes is (perhaps infinitely)
wide. Causal talk could be regarded as the enemy of Marshalls strategy of
analyzing problems in manageable units: where do we draw the lines
delimiting the scope of our causal interests? To think in causal language
is to risk losing focus.8
As table 12.2 shows, Friedman frequently replaces talk of causes with
circumlocutions that concentrate on effects. One might argue that effects
are necessarily correlative to causes, so that this circumlocution is pointless. But that would miss the asymmetry of causation. Effects are defined
as a terminus. From the point of view of an effect, the whole world and all
of history spreads out around and behind us as potential causes. From the
point of view of the cause, the effects lie in relatively compact lines. This is
8
This interpretation is drawn from Hoover 2004, which concerns the use of causal language
in econometrics in the postwar period.
317
obvious in tracing genealogies: it is a relatively manageable task to trace

the descendents of a particular man; it is a completely open-ended task
to trace his ancestors. Friedman is by no means alone. In his account of
causal analysis, the statistician Paul Holland (1986) argues that we should
attempt to assess the effects of a cause and not the causes of an effect,
on the grounds that the first will produce stable knowledge, while the
second is necessarily always an incomplete task. Holland adds, on top of
Friedmans worries, that anytime we assign a proximate cause to an effect,
we place ourselves hostage to future research that may find a more elaborate causal chain lying between the (now obviously not proximate) cause
and the effect. In contrast, once the effect of a cause is established, further
knowledge elaborates the mechanism perhaps, but does not threaten the
truth status of the original claim. Friedman is clearly thinking along similar
lines when he claims that changes in nominal income are an effect of
money, while at the same time calling for investigation into the transmission
mechanism that would elaborate that channel as well as others.
There is, as he himself recognizes (Hammond 1992, 97) a large element
of semantic choice in Friedmans avoidance of explicit use of cause and
etymologically related terms. It is a semantic choice that reinforces, and is
reinforced by, his Marshallian stance. His usage is not, however, substantive in that it in no way undermines his general commitment to causal
realism, which is the commitment to the existence of structures governing
the causal influences among economic variables.
6
Milton Friedmans causal legacy
My reading of Methodology of positive economics turns large amounts

of Friedman scholarship upside down. Most methodologists and most
practicing economists have read Friedmans essay as offering a rationale
for being unconcerned for the truth status of the assumptions of a theory.
What matters is not whether markets are perfectly competitive or whether
people form rational expectations, but whether they act as if they do.9
Hutchison (1992, 2000) and Blaug (2002a, b) have taken Friedman to
have licensed the rise of formalism in economics on the basis of such an
interpretation of the essay. Hands (2003) argues that Friedman does no
such thing and that Hutchison and Blaug misinterpret his essay (see also
Mayer 2003). While Hands is likely right that Friedman should not be
regarded as licensing formalism, the average economist interpreted it along
the same lines as Hutchison and Blaug and probably regard it as a defense
9
See Hammond (1992, 93) for Friedmans own criticisms of the rational-expectations
hypothesis.
Kevin D. Hoover
90
85
25
Friedman's positivism in causal context
(right scale)
20
80
Trend
Percent
75
15
70
65
Trend
10
Percent
318
60
55
5
Causal language
(left scale)
50
45
1953
0
1958
1963
1968
1973
1978
1983
1988
1993
1998
Figure 12.1 The reception of Friedmans positivism and the fate of

causal language in econometrics
Notes: Each series based on counts of articles in the JSTOR archive
of economics journals 19532001. Causal language shows articles in
the econometric family and causal family (= cause, causes, causal,
causally, causality, or causation) as a fraction of the econometrics
family (= econometric(s), regression(s), structural model(s),
estimate, or estimation). Friedmans positivism in a causal context =
Friedman (and positive economics or as if) and causal family and
econometric family expressed as a fraction of articles using the
econometrics family and causal family. Heavy, smooth lines are sixthdegree polynomial trends.
against criticism of formalism. That widespread understanding of

Friedmans methodology may have had implications for the history of
causality in economics.
Figure 12.1 uses the proportion of articles available through the JSTOR
journal archive that use a family of terms associated with causality and a
family of terms associated with econometrics as a proportion of all articles
that use the econometrics family.10 It shows that causal language fell
during the postwar period to reach a nadir in about 1980 and to revive
subsequently. The figure also shows the proportion of articles that use
Friedman and either positive economics or as if in the same article
as a proportion of all articles. The rise and fall of the proportion of such
references to Friedman strikingly mirror the fall and rise of causal language in econometrics.
10
Figure appears as figure 6 in Hoover (2004).
319
The relationship in Figure 12.1 is itself not necessarily causal. I find

it hard to believe that Friedmans own fastidiousness about the use of
causal language was either directly influential or even shared widely in the
economics profession. But I have argued at length elsewhere that the
fall and rise of causal usage is connected to the rise and fall of formalism
in econometrics, which is itself closely connected to the rise and fall of
formalism in economics generally (Hoover 2004). Thus, to the extent that
the common interpretation of Friedmans essay set the stage for overvaluing formalism in economics, we are left with an ironic conclusion:
The methodology of positive economics is best read as advocating
causal realism. At the same time, it was a contributing cause in the
suppression of causal language in economics.
References
Blaug, Mark (2002a). Ugly currents in modern economics. In Uskali Mki (ed.)
Fact and Fiction in Economics: Models, Realism, and Social Construction.
(2002b). Is there really progress in economics? In S. Boehm, C. Gehrke,
H. D. Kurz, and R. Sturn (eds.), Is There Progress in Economics? Cheltenham:
Edward Elgar
Cartwright, Nancy (1999). The Dappled World. Cambridge: Cambridge University
Press
Frazer, William J., Jr. and Lawrence A. Boland (1983). An essay on the foundations
of Friedmans methodology. American Economic Review, 73(1), 12944
Friedman, Milton (1949). The Marshallian demand curve. In Essays in Positive
Economics (1953b), 4799
(1953a). The methodology of positive economics. In Essays in Positive Economics
(1953b), 343
(1953b). Essays in Positive Economics. Chicago: Chicago University Press
(1959). The demand for money: some empirical and theoretical results. In
Milton Friedman, The Optimum Quantity of Money and Other Essays (1969).
Chicago: Aldine, 11140
(1966). Interest rates and the demand for money. In Milton Friedman,
The Optimum Quantity of Money and Other Essays (1969). Chicago: Aldine,
14156
(1970). Comment on Tobin. Quarterly Journal of Economics 82(2), 31827
Friedman, Milton and Anna J. Schwartz (1963a). A Monetary History of the United
States, 18671960. Princeton: Princeton University Press
(1963b). Money and business cycles. Review of Economics and Statistics 45(1,
part 2: supplement). Reprinted in Milton Friedman, The Optimum Quantity of
Money and Other Essays. Chicago: Aldine, 189236
(1970). Monetary Statistics of the United States: Estimates, Sources, Methods. New
York: Columbia University Press
320
Kevin D. Hoover
(1982). Monetary Trends in the United States and the United Kingdom: Their Relation
to Income, Prices and Interest Rates 18671975. Chicago: University of Chicago
Press
Hammond, J. Daniel (1992). An interview with Milton Friedman on methodology. In W. J. Samuels (ed.), Research in the History of Economic Thought and
Methodology, vol. 10, 91118. Greenwich, CT: JAI Press
Hands, D. Wade (2003). Did Milton Friedmans methodology license the formalist revolution. Journal of Economic Methodology, 10(4), 50720
Holland, Paul W. (1986). Statistics and causal inference [with discussion]. Journal
of the American Statistical Association 81(396), 94560
Hoover, Kevin D. (1994). Six queries about idealization in an empirical context.
Poznan Studies in the Philosophy of Science and the Humanities, 38, 4353
(2004). Lost causes. Journal of the History of Economic Thought, 26(2), 14964
Hutchison, Terence (1992). Changing Aims in Economics. Oxford: Blackwell
(2000). On the Methodology of Economics and the Formalist Revolution. Cheltenham:
Edward Elgar
Marshall, Alfred ([1885] 1925). The present position of economics. In
A. C. Pigou (ed.), Memorials of Alfred Marshall. London: Macmillan, 15274
Mayer, Thomas (1993). Truth versus Precision in Economics. Aldershot: Edward Elgar
(1995). Doing Economic Research: Essays on the Applied Methodology of Economics.
Aldershot: Edward Elgar
(2003). Fifty years of Friedmans The methodology of positive economics.
Journal of Economic Methodology 10(4), 4934
Nowak, L. (1980). The Structure of Idealization: Towards a Systematic Interpretation
of the Marxian Idea of Science. Dordrecht: Reidell
Russell, Bertrand ([1918] 1972). The Philosophy of Logical Atomism, ed. David Pears.
London: Fontana
Samuelson, Paul A. (1963). Discussion. American Economic Review, 53(2), 2316
Tobin, James (1970). Money and income: Post hoc ergo propter hoc? Quarterly
Journal of Economics, 82(2), 30117
Van Fraassen, Bas C. (2002). The Empirical Stance. New Haven: Yale University
Press
13
On the right side for the wrong

reason: Friedman on the
MarshallWalras divide
Michel De Vroey
Introduction
In 1949 Milton Friedman published an article on the Marshallian demand

curve (Friedman ([1949b] 1953) which is still frequently quoted, less for
its main argument than for its short final section entitled Alternative
conceptions of economic theory in which he launched a fierce attack on
Walrasian theory. Drawing a contrast between the Marshallian and the
Walrasian approaches, Friedman expressed his regret that the latter had
superseded the former. In his oft-quoted words, We curtsy to Marshall
but we walk with Walras (Friedman [1949b] 1953, 89).
Eminent Friedman scholars, such as Roger Backhouse (1997), Mark
Blaug (2003), Michael Bordo and Anna Schwartz (2004), Dan Hammond
(1996, 2003), Abraham Hirsch and Neil De Marchi (1990), Kevin Hoover
([1984] 1990), and Tom Mayer (1993, 2009), have endorsed Friedmans
claim about the MarshallWalras divide.1 The following quotation from
Hammond summarizes the gist of their argument:
The hallmark of Friedmans methodology is Marshallianism. With this label he
sets his view apart from the Walrasians. In the 1940s the differences in the two
methodologies were recognized, and as we have seen, they were not Friedmans
creation. As time passed, the Walrasian approach ascended to a position of
dominance in both micro and macroeconomics, with the virtues of the general
equilibrium approach and mathematical qualities of elegance and generality taken
for granted by most economists. (Hammond 1996, 43)
This work was supported by the Belgian French-speaking Community (Grant ARC 03/08302) and the Belgian Federal Governement (Grant PAI P5/10). The author is grateful to
Mark Blaug, Milton Friedman, Dan Hausman, Dan Hammond, Kevin Hoover, Robert
Leeson, Tom Mayer, Ivan Moscati, and Roy Weintraub as well as to the participants at the
Duke Workshop on the History of Political Economy for their comments on an earlier
version of this chapter.
Nonetheless the prevailing view, which can be traced back to Hicks, is that the Marshallian
and Walrasian approaches are complementary. Cf. De Vroey (2009).
321
322
Michel De Vroey
The aim of this chapter is to assess the validity of Friedmans claim, first
by studying the argument offered in section V of his 1949 article and, second,
by comparing it with his other contemporary methodological papers his
reviews of Robert Triffins Monopolistic Competition and General Equilibrium
(Friedman 1941) and Oscar Langes Price Flexibility and Employment
(Friedman ([1946] 1953), his comment on Christs assessment of
Lawrence Kleins econometric model (1951), his famous Methodology
of Positive Economics essay (F53) and finally his review of William Jaffs
translation of Walrass Elements dconomie pure (Friedman [1955] 1993).
Most of these papers pursued a critical aim. As will be seen, Friedman
had many foes at the time the concept of indifference curves, Hicksian
demand analysis, Henry Schultzs, work aiming at measuring interrelated
demands, macro-econometric models la Klein, monopolistic competition, Keynesianism la Lange. To Friedman, the common thread in all
these disliked models was their connection to Walrasian theory. Hence
his attack on the latter and the commentators difficulty in assessing
whether his target was Walrasian theory per se or the models he associated
with it. His positive hero was Marshall, whose insights, he felt, had been
unduly discarded, and his indictment of the Walrasian models rested
on their alleged departure from the Marshallian agenda.
The questions that I wish to address in this chapter ensue from these
observations. Was Friedman right in wanting to draw a contrast between
the Marshallian and the Walrasian approaches? Did he characterize it in
an apposite way? Should we follow him when he claims that the
Marshallian program was good and the Walrasian bad instead of simply
opposing them as alternatives? What did he actually understand by the
Walrasian modifier? Why was he so strongly anti-Walrasian in his 1949
article and less so in his 1955 piece?
A complete discussion of these issues must evolve at two levels, concerned respectively with the politics underlying Friedmans stance and
the contents of his arguments. The first is a tale of a group of Chicago
economists (Milton Friedman, George Stigler, W. Allen Wallis, and
Aaron Director) gradually taking the reins of the University of Chicago
community of economists, with the aim of implementing a defence of freemarket economic liberalisms agenda.2 A twofold offensive, external and
internal, was implied. The external fight opposed Chicago and Harvard
and concerned the relevance or otherwise of the monopolistic competition
theory put forward by Chamberlin.3 On the internal front, one enemy was
Walrasian theory as upheld within the Department first by Henry Schultz
2
Cf. Stigler (1988, 150).
See Leeson (2000).
On the right side for the wrong reason
323
(who died in 1938) and later by Oscar Lange. Adhesion to the Walrasian
approach was also the hallmark of some prominent economists (such as
Tjalling Koopmans, Jacob Marschak, and Lawrence Klein) on the Cowles
Commission (which had been relocated in Chicago in 1939). While they
were only residents in the Universitys Department of Economics, their
presence might be viewed as a threat to the laissez-faire agenda since, in the
wake of the debate on the possibility of socialism, Walrasian theory was
associated with collectivism, or at least with social democracy. So, when
studying what was going on in Chicago in the 1940s and 1950s at the level
of ideology and politics, a possible shorthand answer to the question What
did Friedman mean by Walrasian? is socialism.4
While, for all its bluntness, I may sympathize with this interpretation, it
does not provide a complete assessment of the issue. Whatever Friedmans
motivation for entering into the fray and attacking Walrasian theory, his
objections ended up being expressed as theoretical statements whose intrinsic validity needs to be examined. The shortcoming of Mirowski and Handss
work (Mirowski 2002, Mirowski and Hands 1998), which provides vivid
narratives of the contextual-historical dimension, is that they fail to discuss
this validity. The aim of this chapter is complementary to their work, in that
its exclusive concern is an assessment of Friedmans theoretical arguments.
The MarshallWalras divide in Friedmans

early writings
2.1
Friedmans 1941 review of Triffin
Triffins Monopolistic Competition and General Equilibrium (1940) was

based on his doctorate, written under the supervision of Chamberlin. As
far as can be judged from its introduction and first chapter, its aim was to
integrate monopolistic competition and Walrasian theory:
Monopolistic and imperfect competition theories have evolved in the United
States and in England alike along the lines of the theoretical tradition dominant
in both countries: the particular equilibrium of Alfred Marshall. What we might
well do now is to restate the whole problem in terms of the Walrasian, general
equilibrium system of economic theory, so much more influential in economic
thought on the continent of Europe. We shall find with happy surprise that
monopolistic competition begins to bridge the canyon which has for years separated two schools of theoretical thought. (Triffin 1940, 3)
4
As stated by Mirowski and Hands, In the minds of many at Chicago, Walrasian

mathematical theory became conflated with socialism, crude numerical empiricism, and
politically nave welfare economics (Mirowski and Hands 1998, 268).
324
Michel De Vroey
However, going beyond rhetoric, not much of this programme ended up

being realized. While Triffins book contains many scattered references
to Walrasian theory, they concern particular points rather than Walrass
general system. No model of a general equilibrium incorporating imperfect competition is present. The bulk of the book rather consists of a
comparison of different imperfect competition models. Still, one claim
that could not but attract Friedmans attention stands out, namely that as
soon as product differentiation is introduced, the notion of an industry
ceases to be relevant. Only the notions of the firm, on the one hand, and
the economy as a whole, on the other, remain relevant.
In his two-page review of Triffins book, Friedman (1941) did not evoke
the MarshallWalras opposition. Actually, Walrass name was not mentioned. Friedmans main complaint was that Marshallian tools have been
thrown away too light-heartedly by imperfect competition theorists
(1941, 390). He had in mind the notion of industry. While, like Triffin,
he considered the notions of monopolistic competition and industry
antagonistic, he drew the opposite conclusion for Friedman, the notion
of monopolistic competition had to be abandoned and that of industry
maintained. The underlying reason was his enduring concern for practical
issues. The latter, he believed, were to be grasped at the level of the industry
instead of that of the firm or the economy as a whole (Hammond 2003).
2.2
Friedmans 1946 review of Langes Price Flexibility

and Employment
The aim of Langes (1944) book, whose inspiration was Hickss Value and
Capital (1939; 2nd edition 1946), was to examine the Keynesian issue of
whether a decrease in the money wage could restore full employment in
the face of involuntary unemployment. Langes interest lay in the persistence of underemployment, rather than in the study of the logical existence of a state of equilibrium with involuntary unemployment. Not
surprisingly, his conclusion was that Keynes had been right:
Only under very special conditions does price flexibility result in the automatic
maintenance or restoration of equilibrium of demand for and supply of factors of
production. (Lange 1944, 83)
Langes book started with a rejection of partial equilibrium analysis on the

grounds of its narrow range of validity. Although Lange did not refer to the
industry notion, it is clear that he had no sympathy for it. This was certainly
not to please Friedman. Yet his criticism was not directly addressed to this
point. Rather Friedman wanted to indict Langes analysis, for consisting of
empty tautologies or being mere taxonomic theorizing ([1946] 1953,
325
277). He concluded that Lange was just enumerating theoretical possibilities, not describing the real world ([1946] 1953, 271) and that his work
lacked any empirical anchorage no fact-gathering at the initial stage of the
inquiry, no refutable predictions at its end stage.
Lange largely dispenses with the initial step a full and comprehensive set of
observed and related facts to be generalized and in the main reaches conclusions
no observed facts can contradict. His emphasis is on the formal structure of the
theory, the logical interrelations of the parts. He considers it largely unnecessary to
test the validity of his theoretical structure except for conformity to the canons of
formal logic. His categories are selected primarily to facilitate logical analysis, not
empirical application or test. (Friedman [1946] 1953, 283)
This was the very criticism that Friedman was later to address to the
Walrasian approach. Yet at this juncture, no reference was made to Walras.
A further, yet less emphasized, reason for Friedmans dissatisfaction
with Langes work concerned its policy conclusion. In Langes paper,
general equilibrium theory was used as a vehicle for making a Keynesian
case. That general equilibrium might serve the purpose of dismissing the
working of the invisible hand did not bode well for it in Friedmans eyes.
2.3
Friedmans 1949 paper on the Marshallian demand curve
Like the Triffin review, Friedmans 1949 paper on the Marshallian

demand curve was a complaint about the turn taken by economic theory,
a regret that Marshallian tools had been thrown away too airily.5 The
object of Friedmans dissatisfaction was the mutation that Marshallian
demand theory had undergone. He felt that the newly prevailing interpretation of the Marshallian demand curve was unsatisfactory and needed
to be replaced by an alternative more faithful to Marshalls interpretation.
The currently accepted interpretation can be read into Marshall only by a liberal
and, I think, strained reading of his remarks, and its acceptance implicitly
convicts him of logical inconsistency and mathematical error at the very foundation of his theory of demand. More important, the alternative interpretation of the
demand curve that is yielded by a literal reading of his remarks not only leaves his
original work on the theory of demand free from both logical inconsistency and
mathematical error but also is more useful for the analysis of most economic
problems. (Friedman [1949] 1953, 48)
While some readers may be at a loss to see who Friedman had in mind
when speaking of the current interpretation, his target was actually well
5
This theme is also visible in Friedmans lecture notes for his 1940s Economics courses at
the University of Chicago (Hammond 1996, 31).
326
Michel De Vroey
defined: the ideas developed by Hicks in his analysis of demand in part I of

Value and Capital and expanding on Hicks and Allen (1934), on the one
hand, and Henry Schultzs work (his 1935 article his book, The Theory and
Measurement of Demand, 1938), on the other. Both Hicks and Allen, and
Schultz had independently rediscovered views that had been put forward
by Slutsky in a 1915 Giornali degli Economisti article.6 Hickss views soon
gained wide recognition. Although Hicks and Schultz shared the same
insight about building demand theory on the apparatus of indifference
curves, they pursued different purposes. While Hicks was interested
in pure theory without aiming at measurement, Schultzs aim, in contrast, was to make the new demand theory amenable to empirical testing.
Friedman wrote his masters dissertation at the University of Chicago
under Schultzs supervision. He was also Schultzs research assistant for
a year. Yet at the end of the day he barely adhered to Schultzs views.7 He
was not alone in this respect. Critiques of Hickss conception had already
been delivered by Stigler (1939) and Knight (1944),8 and in the 1942
Festschrift edited by Lange et al. in memory of Schultz, Friedman himself
had already written an article (jointly with Wallis) criticizing the indifference curve notion (Wallis and Friedman, 1942).
Hickss aim was to rebuild the Marshallian theory of demand while
disposing of the idea of diminishing marginal utility. Beginning with the
case of two goods, Hicks studied the effect of the variation in the price of X
on the demand for Y, the nominal income of the agent under study being
fixed. The renowned distinction between the substitution effect and the
income effect ensued. The analysis was subsequently extended to the case
of a great number of goods.
The contrast drawn by Friedman between the Hicksian reconstruction of
Marshallian theory and what he believed to be its proper interpretation is as
follows. In Hickss model the money income and the price of every other
commodity is fixed, on top, of course, of tastes and preferences. The real
income is changing due to the eponymous effect. In contrast, Friedman
argues that the real income should be considered fixed as well as the price of
all close substitutes for, and complements of, the particular commodity.
6
7
8
For a study of the relationship between these authors, see Chipman and Lenfant (2002)
and Moscati (2004).
Hammond (1992, 223) gives a vivid account of Friedmans opinion of Schultz. See also
Stigler (1994, 1200).
Knight introduced his paper as follows: The objective of this paper is largely negative to
criticize certain recent innovations in the treatment of demand which have been generally
hailed as representing an advance but which, in the writers opinion, constitute a movement in a backward direction. The particular reference is to the treatment of demand and
utility by J. R. Hicks, pioneered by E. Slutsky, and also followed, and more or less
independently worked out, by Henry Schultz and many others (1944, 289).
327
Whenever a change in the price of the good considered occurs, the average
level of prices of unrelated goods needs to be changed in such a way as to
freeze the purchasing power of money. Adopting Friedmans stance amounts
to eliminating the income effect, thus leaving only the substitution effect to be
active. Therefore, the Marshallian demand curve la Friedman is necessarily
negatively sloped, which is untrue for the Hicksian demand curve.
Friedman was critical of the indifference curve apparatus in general, but
his precise target was its use by Hicks, i.e. employing the substitution effect
while keeping utility constant that is, consumers are compensated by
just enough money to bring them back to their initial indifference curve.
Friedman dismissed this view on the grounds that indifference curves were
not an observable phenomenon. In contrast, he endorsed Slutskys formulation, where the substitution effect gives the agent enough money to get
back to his or her initial level of consumption. Slutsky was thereby tilted
towards Marshalls side.9
In defence of his interpretation, Friedman claimed that it was more
modest (because it does not try to take account of all prices in the system,
not even by keeping them constant (Yeager 1960, 54)) and more useful than its rival when it came to practical problems.
A demand function containing as separate variables the prices of a rigidly defined and
exhaustive list of commodities, all on the same footing, seems largely foreign to this
[Marshallian] approach. It may be a useful expository device to bring home the mutual
interdependence of economic phenomena; it cannot form part of Marshalls engine
for the discovery of concrete truth. The analyst who attacks a concrete problem can
take explicit account of only a limited number of factors; he will inevitably separate
commodities that are closely related to the one immediately under study from
commodities that are more distantly related. (Friedman [1949] 1953, 57)10
Friedman went to great lengths to convince his readers that his interpretation was more faithful to Marshall than the Hicksian one. In effect, most
of the paper is devoted to this task. While I doubt that Friedman was able
to make the point, I will refrain from entering into this issue, and turn at
once to the papers penultimate section.11
10
11
On my interpretation, Marshalls demand curve is identical with the constructions

introduced by Slutsky in his famous paper on the theory of choice, namely, the reaction
of quantity demanded to a compensated variation of price, that is, to a variation in price
accompanied by a compensating change in money income. Slutsky expressed the compensating change in money income in terms of observable phenomena, taking it as equal
to a change in price times the quantity demanded at the initial price (1949] 1953, 523).
See also Friedman ([1949] 1953, 85, 91).
Let me nonetheless give a quotation from Aldrich that is rather dismissive: Friedmans
essay did not generate new economic theory but it established Marshalls demand theory
328
Michel De Vroey
For all its brevity, section V develops several distinct themes. Instead of
trying to paraphrase what Friedman wrote, I find it preferable to quote at
length from this section.
A difference in purpose and method between Marshall and Walras

The important distinction between the conceptions of economic theory implicit in
Marshall and Walras lies in the purpose for which the theory is constructed and
used. To Marshall to repeat an expression quoted earlier economic theory is an
engine for the discovery of concrete truth Economic theory, in this view, has two
intermingled roles; to provide systematic and organized methods of reasoning
about economic problems; to provide a body of substantive hypotheses, based on
factual evidence, about the manner of action of causes. In both roles the test of the
theory is its value in explaining facts, in predicting the consequences of changes in
the economic environment. Abstractness, generality, mathematical elegance these
are all secondary, themselves to be judged by the test of application. The counting of
equations and unknowns is a check on the completeness of reasoning, the beginning
of the analysis, not an end in itself. Doubtless most modern economic theorists
would accept these general statements of the objective of economic theory. But our
work belies our profession. (Friedman [1949] 1953, 90)
Friedmans indictment of the Walrasian approach

According to Friedman, the Walrasian approach must be indicted on
several grounds:
(a) Means i.e. abstractness, generality, and mathematical elegance
have become ends in themselves (Friedman [1949] 1953, 91).
(b) The Walrasian approach is flawed because it gives precedence to the
realism of assumptions over their predictive ability:
Facts are to be described, not explained. Theory is to be tested by the accuracy
of its assumptions as photographic descriptions of reality, not by the correctness of the predictions that can be derived from it. (Friedman [1949] 1953, 91)
(c) It eliminates the industry notion and takes the monopolistic competition line, a blind alley:
From the Walrasian viewpoint, to take another example from recent
developments in economic theory, it is a gain to eliminate the concept of an
industry, to take the individual firm as the unit of analysis, to treat each firm
as a monopoly, to confine all analysis to either the economics of the individual
firm or to the general equilibrium analysis of the economy as a whole. From
as a topic for historians of economic thought. Its bold thesis and its detailed attention to
what Marshall really meant were provocative. It set a pattern: the subject text is tortured
until a confession is produced that can be corroborated by modern theory (1996, 212).
329
the Marshallian viewpoint this logical terminus of monopolistic competition

analysis is a blind alley. Its categories are rigid, determined not by the problem
at hand but by mathematical considerations. It yields no predictions, summarizes no empirical generalizations, provides no useful framework of analysis. (Friedman [1949] 1953, 912)
The difference between the two approaches is not a matter

of partial versus general equilibrium
The distinction commonly drawn between Marshall and Walras is that Marshall
dealt with partial equilibrium, Walras with general equilibrium. This distinction is, I believe, false and unimportant. Marshall and Walras alike dealt with
general equilibrium; partial equilibrium as usually conceived is but a special kind of
general equilibrium analysis unless, indeed, partial equilibrium analysis is taken to
mean erroneous general equilibrium analysis. Marshall wrote to J. B. Clark in 1908:
My whole life has been and will be given to presenting in realistic form as much as
I can of my Note XXI. Note XXI, essentially unchanged from the first edition
of the Principles to the last, presents a system of equations of general equilibrium.
(Friedman [1949] 1953, 90)
Praising Keynes
Section V ends with Friedman praising Keynes on the ground that he was
a real Marshallian:
Of course, it would be an overstatement to characterize all modern economic
theory as Walrasian in this sense. For example, Keyness theory of employment,
whatever its merits or demerits on other grounds, is Marshallian in method. It is a
general equilibrium theory containing important empirical content and constructed to facilitate meaningful prediction. (Friedman [1949] 1953, 92)
2.4
Friedmans 1951 comments on Christs article
In a paper presented at a NBER Conference on Business Cycles, Christ

(Christ 1951) set himself the task of testing Kleins econometric model.
The conclusions he reached were rather negative.
Friedmans comment on it is interesting for my purposes, because he
took the opportunity to plead against using simultaneous equation models
aiming at representing an entire economy to make short-term predictions.
In Friedmans eyes, Christ has shown that Kleins model was a failure.
This was music to his ears. He then raised the question of what to do next.
Does it not then follow that despite the unsatisfactory results to date,
the appropriate procedure is to continue trying one after another of such
systems [of general equilibrium] until one that works is discovered?
(Friedman 1951, 112). No, was his answer, for the probability that such
330
Michel De Vroey
a process [of constructing a model for the economy as a whole] will yield a
meaningful result seems to me almost negligible (Friedman 1951, 113).
Instead, his proposed line was to return to the study of industries:
The direction of work that seems to me to offer most hope for laying a foundation
for a workable theory of change is the analysis of parts of the economy in the hope
that we can find bits of order here and there and gradually combine these bits into a
systematic picture of the whole. In the language of model builders, I believe our
chief hope is to study the sections covered by individual structural equations
separately and independently of the economy. (Friedman 1951, 114)
Friedman might aptly have evoked the names of Walras and Marshall in
this discussion. In effect, studying the individual structural equations
separately was nothing other than doing Marshallian industry analysis,
while Kleins attempt at grasping the economy as a whole suggested a
Walrasian affiliation. Yet he did not do so.
2.5
The 1953 Methodology of positive economics essay
While the MarshallWalras opposition was not mentioned in Friedmans

celebrated 1953 paper (F53), two related themes were nonetheless
broached. First, Friedman engaged in a criticism of empty tautologies,
taking up the theme developed in his Lange paper ([1946], 1953).
Second, as in the Triffin review (1941), he crossed swords with imperfect
competition theory.12 It is to the latter that he resorted when criticizing the
realism of assumptions view:
The relevant question to ask about the assumptions of a theory is not whether
they are descriptively realistic for they never are, but whether they are sufficiently good approximations for the purpose in hand. And this question can be
answered only by seeing whether the theory works, which means whether it yields
sufficiently accurate predictions The theory of monopolistic and imperfect
competition is one example of the neglect in economic theory of these propositions. The development of this analysis was explicitly motivated, and its wide
acceptance and approval largely explained, by the belief that that the assumptions
of perfect competition or perfect monopoly said to underlie neoclassical
economic theory are a false image of reality. And this belief was itself based almost
entirely on the directly perceived descriptive inaccuracy of the assumption rather
than on any recognized contradiction of predictions derived from neoclassical
economic theory. (F53, 15)
12
According to Leeson, his methodology of positive economics was formulated as a

reaction to Robinson and Chamberlin (2000, 62). On this, see also Hammond
(2003, 13 ff.).
2.6
331
Friedmans 1955 review of Jaffs translation of Walrass Elements
Friedmans tone regarding Walrasian theory was more positive here than in
his Marshallian demand curve article, as witnessed by the following statement: The Elements is a great work which marked an important step
forward in the development of economics as a science (Friedman [1955]
1993, 19). His review started with the recognition that Walras was pursuing
an aim different from Marshalls to give a birds eye-view of the economic system as a whole instead of trying to solve concrete issues yet of
no lesser importance.
Walras solved a different, though no less important, problem. He emptied
Cournots problem of its empirical content and produced a complete and
rigorous solution in principle, making no pretence that it could be used
directly in numerical calculations. His problem is a problem of form, not of
content: of displaying an idealised picture of the economic system, not of
constructing an engine for analysing concrete problems. His achievement
cannot but impress the reader with its beauty, its grandeur, its architectonic
structure; it would verge on the ludicrous to describe it as a demonstration of how
to calculate the numerical solution to a numerically specified set of equations.
([1955] 1993, 234)13
Walras was thus to be credited with having given economists a framework for organizing their ideas. Nonetheless, it was time, Friedman
claimed, to return to the more serious business of meaningful theory
([1955] 1993, 27).
Substantive hypotheses about economic phenomena of the kind that were the goal
of Cournot are an essential ingredient of a fruitful and meaningful economic
theory. Walras has little to contribute in this direction; for this we must turn to
other economists, notably, of course, Alfred Marshall All these [factors] have
combined to favor the Walrasian emphasis on form, to make it seem not only an
essential part of a full-blown economic theory, but that economic theory itself.
This conception or misconception of economic theory has helped to produce
an economics that is far better equipped in respect of form than substance. In
consequence, the major work that needs now to be done is Marshallian rather
than Walrasian in character itself a tribute to Walrass impact. (Friedman [1955]
1993, 27)
In writing the above, Friedman fell back on Stiglers opinion as expressed

in Production and Distribution Theories:
13
The same more balanced view is to be found in his Debate with his Critics (Friedman 1974,
1456).
332
Michel De Vroey
Indeed the general equilibrium theory has contributed little to economic analysis
beyond an emphasis on mutual dependence of economic phenomena: the problems are far too complicated to be grasped in toto. Yet this particular theory
describing the nature of general equilibrium was essential: such an idea had to
appear before rigorous study could proceed. It was Walrass greatest contribution
one of the few times in the history of post-Smithian economics that a fundamentally new idea has emerged. (Stigler 1941, 242)
Note finally that, in spite of his praise to Walras, Friedman still stuck to his
empty tautologies indictment, as the following excerpt makes clear:
In his final sentence, Jaff speaks like a true Walrasian in methodology. One first
constructs a pure theory, somehow on purely formal considerations without
introducing any empirical content; one then turns to the real world, fills up
the empty boxes, assigning numerical values to constants, and neglects secondorder effects at this stage. As I have argued extensively elsewhere, this seems to
me a basically false view. ([1955] 1993, 29 note 6)
Friedmans subsequent views
Two main claims were made in Friedmans 1949 article, first that there
existed a better alternative to Hickss account of demand theory, second,
that the Marshallian and the Walrasian approaches were poles apart. Since
Friedman continued to contribute to economic theory for almost a century, it is worth raising the question as to whether he stuck to his earlier
views. The answer is yes.
As to demand theory, it suffices to turn to the first chapter of his Price
Theory essay (1976, second edition; first edition 1962). It starts with the
exposition of the old Marshallian conception of demand and includes
the different concepts which Hicks wished to ban. As in the 1949
article, the ordinary demand function is criticized because changes
in real income are not rigorously excluded and contrasted with a
favored alternative in which the real income, in the sense of money
income divided by the purchasing power of income, is held constant
(1976, 29).
A concession to orthodoxy, Friedman broaches the notion of indifference curves an apparatus fiercely criticized in Wallis and Friedman
(1942) on the grounds of its separating preferences from prices and
income. Yet when it comes to deriving demand functions from indifference curves, he falls back on the wedge between Slutsky and Hicks, with
the former being the good and the latter the bad guy! Thereby, the
earlier contrast between a conception of demand that is somewhat loose
yet lends itself to empirical work and another, which for all its possible
higher rigor fails to do so, turns up again:
333
The advantage of the Slutsky measure, even though in one sense it is an approximation which the Hicks measure is not, is that it can be computed directly from
observable market phenomena and behavior, namely prices and quantities purchased. The Hicks measure cannot; it requires knowledge of the indifference
curves. (1976, 50)
Likewise, there is little doubt that Friedman remained faithful to his

earlier views as to the existence of a divide between Marshall and
Walras. Two testimonies suffice to support this view. The first comes
from Friedmans interview with Snowdon and Vane, one of several
interviews in which he was invited to develop his current views on
economics,
Question: Kevin Hoover has drawn a methodological distinction between your
work as Marshallian and that of Robert Lucas as Walrasian. Is that
distinction valid?
Answer: There is a great deal to that. On the whole I believe that it is probably
true. I have always distinguished between the Marshallian approach
and the Walrasian approach. I have always been personally a
Marshallian. That doesnt mean that the Walrasian approach is not a
useful or appropriate approach. Peoples temperament and attitudes
are different I guess. (Snowdon and Vane 1997, 202)14
14
Unfortunately, when asked to delve into the reasons why he endorsed the Marshall
Walras divide, Friedmans testimony has been of little help, as the following extract from
Hammonds interview makes clear:
Hammond: In The Marshallian demand curve you echo Marshalls description of
economic theory as an engine for the discovery of concrete truth. You
compare the Marshallian conception of economic theory with the Walrasian
conception of what theory should be and should do Do you have a sense of
how you first came to make this distinction and how or why you saw it as
important?
Friedman: I dont really have the sense of how I first came to make the distinction or why
I said it was important. I havent thought about the question, and offhand, in
thinking about it, I really dont know. It is a distinction I made fairly early
on
Hammond: You made it early, and Ive come to think that it may be one of the keys to
your methodology and perhaps
Friedman: I suspect that came from Burns. Thats my guess; but I really couldnt
document it because he was so imbued with Marshall. You see, he was
very much a disciple of Marshall on the one hand, and Wesley Mitchell
on the other. And Wesley Mitchell would have impelled in him aversion
to the pure abstract Walrasian, while Marshall would have impelled in
him his problem-seeking approach. I suspect that thats where it comes
from, but I really cant say. Thats just pure rationalization. (Hammond
1992, 226)
334
Michel De Vroey
The second testimony is the well-known discussion between Friedman

and his critics organized by Gordon (Friedman 1974). Section 3 of
Friedmans Debate with his Critics is entitled What explains the difficulty
of communication? (between Tobin and him). His answer to this question was that the difficulty is a different approach to the use of economic
theory the difference between what I termed a Marshallian approach and
a Walrasian approach in an article I wrote many years ago (1974, 145).
Friedman went on quoting extensively from his 1949 article. Later on,
when discussing Patinkin he returned to the same theme of the contrast
between Marshall and Walras:
Patinkin, even more than Tobin, is Walrasian, concerned with abstract completeness, rather than Marshallian, concerned with the construction of special
tools for special problems. (1974, 159)
A criticism of Friedmans section V argument
Friedmans 1949 paper section V raises a host of questions. Had

Friedmans claim that Walrasian theory dominated the profession any
plausibility at the end of the 1940s? Did Friedman mean that the
Marshallian and the Walrasian approaches differed in purpose or in
their implementation of a common purpose? Was Friedman right when
stating that the divide was not a matter of partial versus general equilibrium? Can his portrait of the Walrasian approach be considered fair and
well documented? Finally, how does his 1949 article relate to his other
pieces?
The alleged dominance of the Walrasian approach
Considering the time at which he was writing (the end of the 1940s),
Friedmans fierce attack on the Walrasian approach is surprising.
We curtsy to Marshall but we walk with Walras is definitely a
powerful image, but is it a correct account of what was going on? I
doubt it.
At the time, Walrass Elements had not been translated into English.
Arrow and Debreu had not started their collaboration. Only a few
sources (e.g. Stigler 1941) and the works of Cassel and Wicksell were
available to economists who wished to become acquainted with
Walrasian theory while being unable to read French. Admittedly, a
revival of Walrasian theory, triggered by Hickss Value and Capital, was
effectively under way. A few works of mathematical economics had seen
the light of day, led by authors such as Moore, Hotelling, and Schultz.
335
Their aim, which could not but have looked odd to Walras, was to render
the Walrasian conceptual apparatus statistically operational. Thus, their
perspective was poles apart from the line pursued by Arrow, Debreu, and
Mackenzie, which I for one view as being more faithful to Walrass own
methodological precepts. It is also true that Friedman was in direct
contact with some of the few Walrasian American economists and
enjoyed many opportunities to confront them with his views when
attending seminars at the Cowles Commission.
Yet, all in all, these economists were a definite minority in the profession. Therefore, Friedmans image is totally inappropriate. At the time,
the majority of economists were not walking with Walras.
A fair portrait of the Walrasian approach?

The Lausanne economist would hardly have recognized his brainchild
in Friedmans sketchy portrait. Take the statement that to Walras
facts are to be described, not explained. Assuming its meaning is
clear, which is questionable, I do not see how it relates to Walrass
methodology. Likewise, having Walras ranked as a defender of the
realism of assumptions looks odd. Ascribing to him the view that a
theory is to be tested by the accuracy of its assumptions as photographic descriptions of reality, not by the correctness of the predictions that can be derived from it (Friedman [1949] 1953, 91)
indicates a total ignorance of his methodological principles. It is
true that, unlike Friedman, Walras did not believe that economic
theory needed to be gauged by its predictive capacity. Yet, he never
viewed the realism of its assumptions as the criterion by which to
assess it. Finally, another indication of Friedmans ignorance of
Walrasian theory is his claim that Walras adhered to monopolistic
competition and treated firms as monopolies. In fact, nothing was
more alien to Walras than such a view.
Furthermore, on reading section V it is unclear whether Friedman had
in mind that the Marshallian and the Walrasian approaches were pursuing
different purposes or whether he indicted the Walrasian approach for
having lapsed from some purpose that would have been common to
both of them. He started by asserting that they pursued different purposes.
Yet, when it came to documenting this difference, he shifted toward
denouncing abstractness, mathematical elegance, and means becoming
ends, all of which suggests that Walrasian theory had derailed from the
supposed common purpose of economic theory.
336
Michel De Vroey
Even this derailment claim is poorly supported. Friedmans opposition

between mathematical elegance and relevance has little grounding. While
mathematical elegance has indeed been a distinctive feature of neoWalrasian theory which, remember, had not seen the light of day at
this time Walrass own work can hardly be characterized in this way. If
anything, Walrass mathematics was, as noted by Harrod (1956), clumsy
and tortuous.
Friedmans claim that the MarshallWalras divide is not

a matter of partial versus general equilibrium
Can Friedman be taken at his word when he claims that Marshallian
analysis is a fully fledged general equilibrium theory? I do not think so.
The most plausible definition of general equilibrium analysis consists of stating that a model can be considered as general equilibrium as soon as its object of study is an entire economy rather than
a fraction of it. The immediate implication is that all models that fail
to meet this criterion must be considered as partial equilibrium
models. The joint study of two markets (for example, the market
for a given good and the markets for its inputs) constitutes partial
equilibrium even if it is concerned with an interdependency phenomenon. As soon as this definition is adhered to, the matter is settled:
Marshallian analysis la Friedman, centered on the study of an
industry, is partial equilibrium.
Unfortunately, Friedman refused this characterization. The underlying
reason, I believe, is a matter of territorial claims. The term general
equilibrium has positive connotations of completeness, while the term
partial equilibrium suggests an admission of incompleteness. Hence the
struggle to stop the Walrasian camp having a monopoly over the positively
connoted field.
As soon as it is accepted that general equilibrium must be equated with
the analysis of an economy as a whole, Friedmans stance can be ascertained unequivocally: he is against it. It suffices to return to his comment
on Christs assessment of the Klein model, where he argued that industry
analysis has advantages of its own that are lost when the economy as a
whole becomes the object of study. Therefore, Friedmans rationale for
calling his preferred approach general equilibrium analysis is hard to
understand. Does it mean that admitting that one is not doing general
equilibrium analysis amounts to an unbearable stigma? Be that as it may,
nothing but confusion is to be gained by such a semantic move.
337
As to Friedmans claim that Marshalls Mathematical Note XXI

generalized his partial equilibrium into a general equilibrium analysis,
it breaks down as soon as the line between general equilibrium and
non-general equilibrium is drawn along the above lines. Marshalls
note is concerned with the phenomenon of joined and composite
supply and demand. Clearly an element of interdependency is
involved. Yet widening the scope of the analysis from a single market
to a group of related markets cannot be viewed as tantamount to
analyzing the entire economy. Likewise, the fact that Marshall wrote
to John Bates Clark that all his life had been devoted to giving flesh
to this note should not be taken for more than it was, a declaration of
intention. Authors must be judged on their theories rather than on
their metatheoretical comments.
Three main lessons can be drawn from my survey. First, it is
surprising to see how numerous Friedmans foes were at the time.
The concepts or works that he disliked and felt the need to attack
ranged widely: indifferences curves, Hicksian demand theory,
Schultzs statistical demand curves, econometric models aiming at
representing the economy as a whole, monopolistic competition,
and Langes Keynesianism. All this conveys the image of Friedman
as an outcast, or at least as somebody fighting against the tide.
Positively, he had one hero, Marshall, and to him this meant three
associated features: (a) the search for concrete truth and an understanding of the ordinary business of life; (b) the need to focus on the
study of particular industries; and (c) the need to resort to concepts
that were directly operational, i.e. measurable. The second lesson is
that Friedman seems to have believed that all the models he disliked
were Walrasian. The third lesson is the difference in tone between
Friedmans criticism of Walrasian theory in his 1949 and 1955
articles. In the latter, he is still dismissive of the Walrasian approach,
but in a far less vehement way.
Two tasks ensue from these observations. The first is to assess
whether there are good grounds for characterizing all the above models
as Walrasian. That is, what might the Walrasian modifier have
meant for Friedman, and does his understanding of it stand up retrospectively? The second is to elucidate his change in tone between 1949
and 1955.
5
Friedmans understanding of the Walrasian modifier
As I see things, Friedmans characterization of Walrasian theory can be

summarized in the following five points:
338
Michel De Vroey
(1) Walrasian theory aims at discovering fundamental principles governing abstract economies.
(2) Walrasian models comprise theoretical concepts that have no direct
measurable counterparts.
(3) Empirical measurement and prediction-making and testing are no
part of the research program.
(4) Walrasian models are concerned with the economy as a whole rather
than with a part of it, such as an industry.
(5) The Walrasian approach is formalized; whenever necessary, relevance
will be sacrificed to rigour.15
On all these points, the Marshallian approach is supposed to stand in
sharp contrast to the Walrasian. Moreover, to Friedman, bringing out a
contrast between the Marshallian and the Walrasian approaches was
tantamount to pleading for the superiority of the former over the latter.
So all the above features were considered by Friedman as flaws and
reasons to condemn any model which adopted them.
The question must be asked as to whether such a characterization stands
up to a retrospective test would a present-day defender of the Marshall
Walras divide agree with Friedmans way of putting the issue? As expected,
the answer is not clear cut. Positively, Friedmans characterization of the
scope and aim of Walrasian theory is acceptable. Negatively, three critical
observations come to mind. First, Friedman insists too much on broad
methodological traits such as those above, failing to realize that the contrast
between the two approaches is also a matter of narrower methodological
choices, the effect of which is to put the theoretical construction on divergent
tracks. These bear on the notion of equilibrium that is adopted, the role of
money, the trade organization and information assumptions, the treatment
of time, etc.16 Second, the fifth criterion, on which Friedman insisted, has
failed to pass the test of time. Accepting that game-theoretical and industrial
organization models can be considered neo-Marshallian, it has turned
out that the Walrasian approach no longer has a monopoly over formalization. Finally, Friedmans manicheist stance must also be questioned. In
my opinion, no one approach should be considered superior to another.
They are rather alternative research programs, each having its pros and cons.
In order to qualify as Walrasian, a given model must abide by the above
criteria.17 Let me then ask whether there are good reasons to consider
15
16
17
Some further traits mentioned in section V of Friedman ([1949] 1953), such as the
monopolistic nature of Walrasian theory, have been dropped because they are patently
mistaken.
On this, see De Vroey (1999a, 1999b, 2003).
Henceforth, I will consider only the first four conditions, eliminating the fifth one.
339
the models criticized by Friedman as Walrasian according to his own

definition.
Hickss demand theory
Unlike Friedman (and myself), Hicks saw no fundamental reason for
opposing Marshall and Walras. To him, these two authors held complementary approaches (Hicks ([1934] 1983). He viewed the study of the
Marshallian market as a first step in the analysis of the economy as a
whole.18 Be that as it may, separating the elements of Hickss Value and
Capital (1946) that are Marshallian from those that are Walrasian is quite
easy. The very fact that part II of the book is entitled General equilibrium
analysis indicates that part I, entitled The theory of subjective value,
where Hicks develops his demand analysis, is not a general equilibrium
theory. In effect, Hickss recurrent reference in the chapters composing
part I of Value and Capital is Marshall. So Hickss demand theory analysis
belongs to the Marshallian part of the book rather than to its Walrasian
part.
It may be argued that the fact that every commodity composing the
economy is individually listed in the Hicksian demand theory makes
it Walrasian.19 Yet this is not a sufficient condition. For example, the
category of income as understood by Hicks in these chapters a quantity
of money that agents have at their disposal before entering the goods
markets and which accrued to them from their antecedent sale of inputs
has no place in a Walrasian framework.
Finally, comparing Hickss demand theory with the set of conditions
above, it turns out that it abides by conditions (1), (2), and (3), but
definitely not by condition (4).
To conclude, it is dubious whether the Hicksian demand theory can be
characterized as specifically Walrasian. Friedmans dissatisfaction with it
may well rest on a trait that he associates with the Walrasian approach
basing the analysis on concepts having no observable counterparts yet
this association is far from compelling. The dispute between Hicks and
Friedman is one that takes place within the Marshallian partial equilibrium framework rather than opposing the Walrasian and Marshallian
approaches. As a result, Friedmans transition from the discussion of
18
19
If a model of the whole economy is to be securely based, it must be grounded in an

intelligent account of how a single market functions (Hicks 1965, 78).
This is the step taken by Hirsch and De Marchi: In comparing two demand curves, one
of which he calls Walrasian (the conventional one) and of which he refers to as Marshallian
(constant real income) Friedman gives us an example of how he applies the Marshallian
Walrasian distinction in a specific case (1990, 23).
340
Michel De Vroey
demand theory in sections I to IV of the 1949 article, where he discusses

demand theory, to section V, devoted to the broader MarshallWalras
contrast, is far from straightforward. While there is a link between these
two parts of the article, they are hardly part and parcel of each other. In
fact, they could be split without harming the argument in either.
Schultz
Schultz was an author who claimed to be part of the WalrasPareto lineage
while pursuing the aim of making Walrasian concepts statistically measurable, a project that Walras would have refused to consider as his. Be
that as it may, Schultz cannot be considered Walrasian according to
Friedmans criteria. His agenda fails to abide by conditions (1) and (3),
and possibly also condition (2). Friedman might have been expected to
credit Schultz with having breached the Walrasian reluctance to enter into
empirical work, but this was not the case. In Friedmans eyes, something
more than merely engaging in empirical work was needed, namely that the
empirical results exerted a feedback on the theoretical structure.20
Moreover, Schultzs work shows that there is many a slip twixt cup and
lip. To achieve his aim of measuring Walrasian concepts, Schultz had to
go through many twists as, to give just one example, his transmogrification
of the Walrasian notion of endowment into the Marshallian notion of
monetary income witnesses. While the theoretical and empirical parts of
Schultzs paper and article are excellent contributions on their own, the
link between them is virtually non-existent.21
To conclude, there is more to Friedmans dislike of Schultzs work than
its Walrasian character. Moreover, Schultz should not be considered a
fully fledged Walrasian economist.
20
21
This is evidenced in a letter that Friedman wrote to Edwin Wilson in 1946 in which he
criticized Schultz in the following terms: Schultz took the theory as fixed and given, and
tried to measure what he thought were essential functions in the theory. He imposed
extremely high standards of care and thoroughness on the measurement process but he
nowhere attempted what seems to me the fundamental important task of reformulating
the theory so that it would really generalize the observable data; he also tried to wrench the
data into a pre-existing theoretical scheme, no matter how much of a wrench was
required (Stigler 1994, 1200).
This has led Mirowski and Hands to conclude that it is seldom recognized that
[Schultzs] book is essentially a swan song for Walrasian economics: Schultz bravely
reported the empirical debacle in detail and then produced a litany of excuses why things
had not worked out as hoped (1998, 264).
341
Lange
Price Flexibility and Employment was an attempt to combine Walras and
Keynes, two authors to whom Friedman was strongly opposed. Was there
then some domino process at work, Friedmans dislike of Keynes spilling
over to Walras? Actually, the matter is more complicated than that. Recall
Friedmans hailing of Keynes as a genuine Marshallian in section V of the
1949 article. In other words, while Friedman was opposed to Walrasian
theory because of its methodological principles, his opposition to Keynes
had no such groundings. Keynesian theory was to be rejected, not, like
Walrasian theory, because of its faulty methodological basis, but rather
because it had been contradicted by empirical evidence.22 So, Friedmans
target in his criticism of Langes book is less the Keynesian than the
Walrasian element of the proposed KeynesWalras alliance.
The irony here is that, while Walrass name does not appear in
Friedmans review of Lange, of all the authors attacked by Friedman he
was the only one who really deserved the Walrasian label. And, in effect,
his theory abides by the criteria set up to this effect.
Triffin
Two reasons may explain Friedmans hostility to Triffins project. First, it
amounted to shelving the study of industries. The second objection has a
more ideological nature. Triffin claimed that a deep affinity linked imperfect competition and Walrasian general equilibrium, to the effect that a
synthesis between them was within reach. This could not but displease
Friedman on the ground of a bedfellows argument any friend of any
enemy must also be an enemy. Friedman did not question Triffins cooptation of Walrasian theory. Saddled with such a bedfellow, Walrasian
22
The following extract from Friedmans Theoretical Framework for Monetary Analysis makes
this clear: Keynes was no Walrasian seeking, like Patinkin and to a lesser extent Tobin,
a general and abstract system of all-embracing simultaneous equations. He was a
Marshallian, an empirical scientist seeking a simple, fruitful hypothesis. And his was a
new, bold, and imaginative hypothesis, whose virtue was precisely how much it could say
about major problems on the basis of so little. Of course, his assumptions were not in
literal correspondence with reality. If they had been, he would have been condemned to
pedestrian description; his whole theory would have lost its power. Of course, he could be
wrong. There is no point to any scientific theory that cannot be. The greater the range of
evidence that, if observed, would contradict a theory, the more precise are its predictions
and the better a theory is provided it is not, in fact, contradicted. I believe that Keynes
theory is the right kind of theory in its simplicity, its concentration on a few key magnitudes, its potential fruitfulness. I have been led to reject it, not on these grounds, but
because I believe that it has been contradicted by evidence: its predictions have not been
confirmed by experience. The failure suggests that it has not isolated what are really the
key factors in short-run economic change (Friedman 1974, 134).
342
Michel De Vroey
theory could hardly look congenial to him. However, what he failed to see
was that the synthesis between Chamberlin and Walras evoked by Triffin
was just wishful thinking. Triffins thesis hardly took the first step toward
constructing such a synthesis. If the MarshallWalras divide makes sense,
it would be better to put monopolistic competition on Marshalls side
(as argued in De Vroey 2004).
The general lesson to be drawn from this quick assessment is that, with
the exception of Lange, the reasons for Friedmans dissatisfaction with the
models he rejects actually have little to do with their alleged Walrasian
character.
Friedmans 1949 account of the MarshallWalras

divide and his review of Jaffs translation of
Walrass Elements
Finally, let me compare the 1949 article with the 1955 review. While the
former is virulent, the latter, although still dismissive, is milder in tone. In
particular, the Jaff review improves on the 1949 article by bringing out
the difference in purpose between Marshall and Walras more satisfactorily. Walras is no longer assessed through a Marshallian lens. The specificity of his project is recognized.
Two factors may be evoked to explain the difference between the 1949
and the 1955 articles. A first one is the bad company explanation. Yet,
as seen, it played only for the imperfect competition / Walrasian theory
connection. In 1949, Friedman might still have believed that Triffin was
right in arguing that Walrasian theory and imperfect competition were
compatible.23 At the time, he had only an indirect knowledge of Walrasian
theory, while the works against which he felt it necessary to react were
claiming a Walrasian affiliation. Professor Blaugs testimony can be called
in support of this view, as he has written to me that Friedman told him that
the first time he read Walras was when he reviewed Jaffes translation.24
Once Walrass Elements became available in English, Friedman realized
that Walrasian theory did not carry the associations which Triffin and
Lange had wanted to impose upon it. It was still not his cup of tea, but at
least he recognized its specificity and was ready to give it due credit.
23
24
An extract from Hammond (2003) is worth quoting in this respect for it reveals that
Friedmans mistake is still being made: So the impetus of Friedmans discussion of
methodology with George Stigler was the challenge posed by Walrasian monopolistic
competition to Marshallian industry analysis (2003, 16; my emphasis). Treading in
Friedmans footsteps, Hammond wrongly takes it for granted that Walrasian monopolistic competition existed and raised a challenge to Marshallians.
This point was confirmed to me by Professor Friedman in private correspondence.
343
The second factor relates to the political dimension evoked in the introduction. On the one hand, the Cowles Commission had left Chicago. On
the other hand, the socialist debate had petered out, and on reading
Walrass Elements, Friedman must quite rightly have discovered that no
plea in favor of socialism was to be found in it.
7
Concluding remarks
My aim in this chapter has been to assess the validity of Friedmans claim,
made in his 1949 article, as to the existence of a MarshallWalras divide.
I have shown that Friedmans arguments in favor of it stand up to scrutiny
only partially. I have also put forward the view that when he criticized
specific models for their Walrasian flaws he might have had in mind other
targets for which the Walrasian label served as a proxy. The change in tone,
from a vehement to a softer anti-Walrasian stance, between Friedmans
1949 article and his 1955 review of Jaffs translation of Walrass Elements of
Pure Economics has also been brought out. While keeping his Marshallian
affiliation, he followed Stigler to a position combining praise for Walrass
achievement with the conviction that it was not the right line to take. Thus,
to my mind, the 1955 article is a better embodiment of Friedmans definitive view as to the relationship between Marshall and Walras than the 1949
article.
My judgment that Friedmans justification of the MarshallWalras
divide was unsatisfactory should not be taken as meaning that he was
wrong in believing that the two approaches were poles apart. On the
contrary, he should be credited for having blazed the trail for the recognition of this divide, which in my opinion constitutes a clue for understanding the development of economic theory over the last century. He
was right in perceiving the need to draw the MarshallWalras divide yet
unable to give a solid argumentation as to its grounding. The task ahead is
to fill in this lacuna.
References
Aldrich, J. (1996). The course of Marshalls theorizing about demand, History of
Backhouse, Roger E. (1997). The rhetoric and methodology of modern macroeconomics. In Brian Snowdon and Howard Vane (eds.), Reflections on the
Development of Modern Macroeconomics. Cheltenham: Edward Elgar, 3154
Baranzini, R. and P. Bridel (2003). Echange et utilit: Walras, versus Pareto.
Communication au X Colloque de lAssociation Charles Gide pour ltude de la
Pense Economique, Grenoble, September 2003
344
Michel De Vroey
Blaug, Mark (2003). The formalist revolution of the 1950s. Journal of the History of
Bordo, M. and A. Schwartz (2004). IS-LM and monetarism. In M. De Vroey and
K. Hoover (eds.), The IS-LM Model: Its Rise, Fall and Strange Persistence,
annual supplement of History of Political Economy, 36, 21739
Buchanan, James (1958). Ceteris paribus: some Notes on Methodology. Southern
Economic Journal, 24, 25970
Chipman, J. and J.-S. Lenfant (2002). Slutskys 1915 article: how it came to be
found and interpreted. History of Political Economy, 34, 55497
Christ, C. (1951). A test of an econometric model for the United States,
19211947. Paper presented at the 1949 NBER Conference on Business
Cycles, 35105
Clower, R. and A. Leijonhufvud ([1975] 1984). The coordination of economic activities: a Keynesian perspective. In D. Walker (ed.), Money and
Markets: Essays by Robert Clower. Cambridge: Cambridge University
Press, 20917
Colander, David (1996) (ed.). Beyond Micro-Foundations: Post-Walrasian Macroeconomics. Cambridge: Cambridge University Press
De Vroey, Michel (1941). Review of Monopolistic Competition and General
Equilibrium Theory by Robert Triffin. Journal of Farm Economics, 23, 38991
([1946] 1953). M. Lange on price flexibility and employment: a methodological
criticism. In Essays in Positive Economics. Chicago: University of Chicago Press,
277300
([1949] 1953). The Marshallian demand curve. In Essays in Positive Economics.
Chicago: University of Chicago Press, 4799
(1951). Comments on Christ, Conference on Business Cycles. New York: National
Bureau of Economic Research, 10713
(1953). The methodology of positive economics. In Essays in Positive Economics.
Chicago: University Chicago Press, 343
([1955] 1993). Lon Walras and his economic system. In J. Cunningham Wood
(ed.), Lon Walras, Critical Assessments. London: Routledge, vol. I, 1929
(1968). The role of monetary policy. American Economic Review, 58, 117
(1999a). The Marshallian market and the Walrasian economy: two incompatible bedfellows. The Scottish Journal of Political Economy, 46(3), 31938
(1999b). Equilibrium and disequilibrium in economic theory: a comparison of
the classical, Marshallian and Walras-Hicksian conceptions. Economics and
Philosophy, 15, 16185
(2002). Equilibrium and disequilibrium in Walrasian and neo-Walrasian
economics. Journal of the History of Economic Thought, 24, 40526
(2003). Perfect information la Walras versus perfect information la Marshall.
Journal of Economic Methodology, 10, 46592
(2004). The history of macroeconomics viewed against the background of the
MarshallWalras divide. In M. De Vroey and K. Hoover (eds.), The IS-LM
Model: Its Rise, Fall and Strange Persistence, annual supplement of History of
(2009). Marshallian and Walrasian theory: complementary or alternative
approaches? History of Political Economy, forthcoming
345
Friedman, Milton (1974). Milton Friedmans Monetary Framework: A Debate with

his Critics, ed. R. J. Gordon. Chicago: University of Chicago Press
(1976) Price Theory. Chicago: Aldine
Hall, R. E. (1979). A theory of the natural unemployment rate and the duration of
employment. Journal of Monetary Economics, 5, 15369
Hammond, J. Daniel (1992). An interview with Milton Friedman on methodology.
In Bruce J. Caldwell (ed.), The Philosophy and Methodology of Economics, vol. I.
Cheltenham: Edward Elgar, 21638 (interview took place on May 24, 1988)
(2003). Predictions as the goal of economic analysis. Paper presented at the
Conference Milton Friedmans Essay at 50, Erasmus Institute for Philosophy
and Economics, December 2003
Harrod, Roy F. (1956). Walras: a re-appraisal. Economic Journal, 46, 30716
Hausman, Daniel (1990). Supply and demand explanations and their ceteris
paribus clauses. Review of Political Economy, 2, 16886
Hicks, J. R. ([1934] 1983). Lon Walras. In Classics and Moderns: Collected Essays
on Economic Theory, vol. III. Oxford: Basil Blackwell, 8595
(1946). Value and Capital, 2nd edition. Oxford: Clarendon Press
(1965). Capital and Growth. Oxford: Clarendon Press
Hicks, J. R. and Allen, R. G. D. (1934). A reconsideration of the theory of value.
Economica, 5276, 196219
Hirsch, Abraham and Neil De Marchi (1990). Milton Friedman, Economics in
Theory and Practice. New York: Harvester Wheatsheaf
Hoover, Kevin ([1984] 1990). Two types of monetarism. In J. Cunningham Wood
and R. N. Woods (eds.), Milton Friedman: Critical Assessments, vol. III.
London: Routledge, 52751
Jaff, William ([1980] 1983). Walrass economics as others see it. In Donald
A. Walker (ed.), William Jaffs Essays on Walras. Cambridge: Cambridge
Knight, Frank (1944). Realism and relevance in the theory of demand. Journal of
Lange, Oskar (1944). Price Flexibility and Employment. Bloomington, IN: Principia
Press
Leeson, Robert (2000). The Eclipse of Keynesianism: The Political Economy of the
Chicago Counter-Revolution. Basingstoke: Palgrave
Mayer, Thomas (1993). Truth versus Precision in Economics. Aldershot: Edward
Elgar
(2009). The influence of Friedmans methodological essay. This volume
Mirowski, Philip (2002) Machine Dreams. Economics Becomes a Cyborg Science.
Mirowski, Philip and D. Wade Hands (1998). A paradox of budgets: the postwar
stabilization of American neoclassical demand theory. In M. Morgan and
M. Rutherford (eds.), From Interwar Pluralism to Postwar Neoclassicism, annual
supplement to History of Political Economy, 30. Durham, NC: Duke
346
Michel De Vroey
Moscati, I. (2004). History of consumer demand theory 18711971: a neoKantian rational reconstruction. European Journal of the History of Economic
Thought, 14, 11956
Schultz, H. (1935). Interrelations of demand, price and income. Journal of Political
Economy, 43, 43381
(1938). The Theory and Measurement of Demand. Chicago: University of
Chicago Press
Slutsky, E. (1915). Sulla Teoria del bilancio del Consumatore. Giornale degli
Economisti, 3, 121
Snowdon, Brian and Howard Vane (1997). Modern macroeconomics and its
evolution from a monetarist perspective: an interview with Professor Milton
Friedman. Journal of Economic Studies, 24(4), 19222
Stigler, George (1939). The limitations of statistical demand curves. Journal of the
American Statistical Association, 34, 46981
(1941). Production and Distribution Theories: The Formative Years. New York:
Macmillan
(1988). Memoirs of an Unregulated Economist. New York: Basic Books
(1994). Correspondence on methodology between Milton Friedman and
Edwin Wilson; November-December 1946. Journal of Economic Literature,
32, 11971203
Triffin Robert (1940). Monopolistic Competition and General Equilibrium.
Wallis, W. A. and M. Friedman (1942). The empirical derivation of indifference
functions. In O. Lange, F. MacIntyre and T. Yntema (eds.), Studies in
Mathematical Economics and Econometrics. In Memory of Henry Schultz.
Yeager, L. (1960). Methodenstreit over demand curves. Journal of Political Economy,
68, 5364
Part 5
Concluding perspectives
14

Mark Blaug
The methodology of positive economics (F53) is a classic text in

economic methodology and like any classic text it can be read in or
out of context. Its original context, as Roger Backhouse reminds us in
chapter 8 of this volume, was that of antimarginalist attacks on profit
maximization by heterodox economists using survey data to demonstrate that businessmen do not calculate marginal costs and benefits.
Much of the essay is easier to understand when we keep that context
in mind. But as time passed, that particular context became irrelevant
and the essay was increasingly read in the light of Friedmans subsequent career as a monetarist, leading opponent of Keynesian demand
management and outspoken advocate of free-market fundamentalism.
After, say 1970, few economists could read the essay as simply a
methodological defense of patently unrealistic assumptions in economic theories because the rest of Friedmans message was always in
the back of their minds. In short, in examining interpretations of what
F53 really meant, it is important to take account of the year the
interpretation is advanced. To underline that point, let me briefly
reminisce.
I first read F53 for a 1955 Yale University seminar organized by Tjalling
Koopmans when he was writing the second of his Three Essays on the State
of Economic Science (1957). Koopmans was quite skeptical of the essay,
saying it apologized too much for the lack of realisticness of standard price
theory, but I myself, knowing no better, thought it was a remarkably
effective defense of standard theory and I was totally convinced by the
idea that the assumptions of a theory did not really matter provided the
theory predicted accurately. The now familiar proposition that there is
something wrong with a black-box theory that predicts accurately but
provides no mechanism to account for those predictions did not occur
to me at the time. Besides, I was impressed by Friedmans fall-back
argument that the beliefs of businessmen do not matter anyway, because
competition is a Darwinian selection mechanism for weeding out firms
that fail to maximize profits in accordance with the marginal-cost-pricing
349
350
Mark Blaug
rule of orthodoxy. In other words, standard theory predicts accurately

because it captures the essential consequences of a survival mechanism
that produces results that are indistinguishable from those of perfect
competition.
It was some years later that I read Sydney Winters PhD thesis at Yale,
which argued that standard price theory is a precisely stated static theory,
whereas the Darwinian selection mechanism is a vaguely stated dynamic
theory. That is to say that Friedman employed a radically different evolutionary explanation to justify his original contention about the accurate
predictions of comparative static reasoning. It seems perfectly obvious
now but, with tremendous efforts, I can still recall my confusion about all
these issues fifty years ago. Fifty years ago? Was it really that long ago?
Why, that is longer than the years that separate World War I from World
War II, or the publication of Marshalls Principles from Keyness General
Theory.
When I came to take another look at F53 in the late 1970s in preparation for my Methodology of Economics (1980), it was easy to read
Friedman as yet another weak-kneed Popperian who would not practise
what he preached. After all the fuss about judging theories by their
predictions and not their assumptions, he finally reached the disappointing conclusion that An even more important body of evidence for the
maximization-of-returns hypothesis is experience from countless applications of the hypothesis to specific problems This evidence is
extremely hard to document; it is scattered in numerous memorandums,
articles, and monographs concerned primarily with specific concrete
problems rather than with submitting the hypothesis to test (F53, 223).
Really? Not even one concrete prediction? For example, that advertising
will never be observed in a perfectly competitive market and therefore
that the more nearly the market structure of an industry approaches that
of perfect competition, the less is the quantity of advertising we will
observe; or, that a rise in the money wages of a particular category of
labor will reduce the volume of employment of that type of labor, if not in
the long run then certainly in the short run. Whatever the evidence for
both propositions, is there really nothing we can do to marshal that
evidence, and if there really is no such evidence, does that not tell us
something very significant about the theoretical beliefs of the economics
profession?
By 1980, Friedmans monetarism had arrived but was not yet gone
and again it was all too easy at that time to agree with Lawrence Boland
(1979) that Friedman was an instrumentalist who cared little about the
truth-value of theories but more about their utility in devising solutions to
practical problems. Besides, even Friedman himself endorsed Bolands
351
interpretation (Boland 1997, 4151; 2003).1 But ignoring the context in

which F53 was written and even the heated macroeconomic debates
surrounding Friedmans views on money in the 1970s, it is difficult to
see any explicit endorsement in F53 of the old MachPoincar philosophy
of instrumentalism. Uskali Mki (1986) is quite right, it seems to me, to
reject Bolands claim that when F53 is read carefully it will be found to be
both logically sound and unambiguously based on a coherent philosophy
of science-instrumentalism. Actually, it is at once wonderfully ambiguous and incoherent. One can find in it echoes, and sometimes much more
than echoes, of Popper, Kuhn, Quine, Toulmin, Laudan, and even
Feyerabend and that is precisely what made it one of the most influential
texts in the methodological literature of twentieth-century economics.
One of the most memorable ideas in F53 is the notion that competition
is an evolutionary selection mechanism weeding out businessmen who fail
to maximize returns. Jack Vromen (this volume) dismisses the importance
of this argument in F53 because Friedman expounds rather vaguely what I
once called the Alchian thesis2 and follows it almost immediately by the
countless applications paragraph cited earlier. However, without something like a Darwinian selection mechanism, Friedmans frequent appeal
to as-if reasoning lacks any grounding in a commonsense realist interpretation of economic behavior. This is a crucial point in the essay at which it
does matter whether we read it as an exercise in the philosophy of realism
or in the philosophy of instrumentalism because, as Uskali Mki rightly
observes in chapter 3 of this volume, we may argue that businessmen act as
if they only maximize profits (but of course they do many other things) or
that they act as if they maximize profits (but that they really dont). I side
with Mki and against Boland in this because Bolands reading implies an
uncommonsensical philosophical sophistication on Friedmans part that
rings false.
Besides, if we examine Capitalism and Freedom (1962) and Free to Choose
(1980), the principal statements by Milton and Rose Friedman of the
normative case for free markets, what we find is an informal dynamic
argument about competition between independent self-interested individuals doing good as an unintended by-product of doing well; not for
them the arid formalism of Arrows First and Second Fundamental
1
Boland modestly announced that he was the first to realize that Friedman was an instrumentalist while everyone else had pictured him as a sort of positivist. He also claimed to be
one of the few methodologists who understood that Popper was no falsificationist but
only a critical rationalist, that is, one who believes that everything can be criticized
including reason itself (Boland 2003, 268, 26878).
Vromen shows that Friedmans own presentation of the argument was made independently of Arman Alchian but it was credited to Alchian in F53.
352
Mark Blaug
Theorems of welfare economics (Arrow 1983).3 In other words, they lean

on something like a Darwinian selection mechanism.
This relates to one of the mysteries of F53, which is Friedmans (and
Stiglers) deep-seated antipathy to the theory of monopolistic competition
and the insistence that to understand the phenomena of the business
world, all we need is a theory of perfect competition and a theory of
monopoly, applying them alternatively as circumstances require. The
trouble with the theory of monopolistic competition according to
Friedman (F53, 349) is not the usual objection that the Chamberlin
case of product differentiation without oligopolistic interdependence
between firms is a quantitatively unimportant market structure in modern
economies (Blaug 1997a, 3769), but rather that it destroys the concept of
an industry which is the bedrock of the Marshallian analysis of firm
behavior. It is not always appreciated that Chicago-style economics
regards real-world competition as so nearly resembling the textbook
model of perfect competition as to be just about the same thing. Melvin
Reder (1982) in his famous exegesis of Chicago economics labeled this
the good approximation assumption, which was a principal element in
the tight prior equilibrium theory of Chicago economics. The state of
perfect competition may be regarded by other economists (for textbook
references, see Blaug 1997b) as an ideal, an unobservable analytical
fiction, which nevertheless serves as a benchmark against which to compare real-world markets, but that is not how it is regarded by Friedman
and company. To label Friedman an instrumentalist is to deny the realworld ontological status of his conception of what Marshall called free
competition to distinguish it from a theoretical notion of perfect competition. This is a subtle distinction that is easily lost sight of if we
reinterpret Marshall and Friedman in the light of the dominant
Walrasian end-state conception of competition.
Now, economic theory, orthodox or otherwise, provides no measure of
what would be a close or distant approximation to a perfectly competitive
economy in which no firm has a price policy and all firms are grouped
together with other firms producing the same homogeneous product. In
other words, tight prior equilibrium theory involves a judgment about
the real world that is incapable of being reduced to quantifiable elements.
Ah, this argument is beginning to sound very similar to the judgment
involved in appraising the predictive accuracy of the maximization-ofreturns hypothesis. The idea of conclusions which are not quasimathematical proofs but fuzzy dynamic qualitative judgments is in fact a
3
Although Arrows Second Fundamental Theorem, permitting a divorce between equity

and efficiency, turns up in the Friedmans as a proposal for a negative income tax.
353
principal feature of what Friedman calls the Marshallian, in contrast to the

Walrasian, approach to economics.
So, how can anyone say, as I am supposed to have said according to
Wade Hands, that Friedman licensed the formalist revolution? It is simply
that those who read F53 remembered the fundamental thesis that the
realisticness of assumptions does not matter and all that does matter for
the validity of economic theories is their predictive accuracy and ignored
everything else. Also, unlike Wade Hands (or Tom Mayer) they paid no
attention to how Friedman went on to validate his favorite economic
theory, the quantity theory of money. After all, Friedman gave no prescriptive advice in F53 on how best to validate economic theories. Should
we use econometric evidence or historical evidence? Should it be a thick
case-by-case examination of the evidence or a thin verification by a
perfectly general demonstration? Friedman may have won some methodological battles with F53 but he lost the methodological war because he
went on to write A Monetary History of the United States, 18671960 (1963)
with Anna Schwartz a thick examination of the historical evidence for the
theory that money causes prices and not the other way round while the
rest of the economics profession put its faith in thin corroborations by
means of econometric regressions. So, Hands has a case when he argues
that Blaug and Hutchison are examples of methodology-as-viewed-fromphilosophy, while Mayer gives us methodology from the standpoint of
the applied economist but, as usual, he overstates a good case. It is not
Blaug and Hutchison but the profession as a whole that took from F53 the
methodological message that suited them and simply ignored Friedmans
subsequent practice.
References
Arrow, Kenneth J. (1983). An extension of the basic theorems of welfare economics (1951). In Collected Papers of Kenneth J. Arrow. Oxford: Blackwell, ch. 2
Blaug, Mark (1997a). Economic Theory in Retrospect, 5th edition. Cambridge:
(1997b). Competition as an end-state and competition as a process. In Not Only
an Economist. Cheltenham: Edward Elgar
(1997). Critical Economic Methodology: A Personal Odyssey. London: Routledge
(2003). Methodological criticism as ideology and hypocrisy. Journal of Economic
Literature, 10(4), 151226
Friedman, Milton (1953). Essays in Positive Economics. Chicago: University of
Chicago Press
354
Mark Blaug
(1962). Capitalism and Freedom. Chicago: University of Chicago Press

Friedman, Milton and Rose D. Friedman (1980). Free to Choose: A Personal
Statement. New York: Harcourt Brace Jovanovich
Milton Friedmans methodology. Research in the History of Economic Thought
and Methodology. 4, 12943, Greenwich, CT: JAI Press
Reder, Melvin W. (1982). Chicago economics: permanence and change. Journal
of Economic Literature, 15(1), 138
15
Final word
Milton Friedman
I have somewhat mixed feelings about the splendid conference that Uskali
Mki organized on my ancient methodology article. On the one hand, it is
a source of great satisfaction that an article I wrote more than fifty years
ago should still be regarded as worth extensive scholarly discussion. On
the other hand, that very fact is a severe condemnation of the essay.
Surely, if the essay had been really lucid, scholars should not today still
be having different opinions about what it says.
I have myself added to the confusion by early on adopting a policy of not
replying to critiques of the article. I decided that I had a choice: I could
spend my time discussing how economics should be done a worthy
cause; or I could spend my time doing economics in my opinion, if not a
more worthy, a more attractive cause. That act of self-denial has quite
unintentionally been a plus for the discussion of methodology. It has left
the field open for all comers, and they have all come and produced a broad
stream of commentary. The articles at this conference are a good example.
I have read them all and I am impressed with their high quality and
intellectual seriousness. I would not have thought that my modest essay
would have relevance to so many issues. I feel like a proud father who has a
large brood of bright children all of them right, all of them wrong, and all
entitled to his or her own views. Yet I am also impressed that there is a
common thread that runs through them.
I would like to close by offering my personal thanks to the authors of the
chapters in this collection, for how responsibly you have taken your task of
commenting on my essay. I have learned much from your essays about my
own as well as about other ideas on methodology. I am sure other readers
of the book will have the same experience.
Milton Friedman, February 19, 2004
355
Index
abstract models 247, 34, 35, 42, 80, 82

Agassi, J. 160
Akerlof, George A. 178, 181, 182, 184, 186
Alchian, Armen A. 79, 258, 270, 279
Alchian thesis 351
Alexander, Sydney 82
Allais paradox 293, 294, 298, 299
Allen, Clark 129
Allen, R. G. D. 326
American Economic Review (AER) 15, 70,
77, 82, 85
analytical relevance 334, 73, 82
Andrews, Philip 2289, 238
arbitrariness 1089
Arrow, Kenneth 132, 1478, 1568, 160,
245, 24950, 335, 351
as-if methodology 1046
early drafts of F53 702, 79, 84, 85, 87,
1046
F53 and the marginalist controversy 231
F53 as a realist statement 90, 1046
Friedman and the theory of the firm 244
Friedmans causal legacy 318
Friedmans risky methodology 2923
Friedmans selection argument 2601,
263, 270
map of multiple perspectives, 47, 48, 54,
55, 56, 65
two kinds of as-if 1047
assumptions of theory
as idealizations 99101
early drafts of F53 6971, 73, 77,
7881, 87
in F53 1415, 1620, 2330, 313,
401, 42
F53 as a realist statement 928, 99104, 113
Friedmans positive methodology and the
formalist revolution 1501
influence of F53 1202, 128, 1301
map of multiple perspectives 47, 48, 535, 59
methodology of causal realism 304, 310
356
politics of positivism 207

roles of assumptions 99101
asymmetric information 130, 140, 209
Aumann, Robert 2668, 271
Backhouse, Roger 49, 56, 321
Bain, Joe S. 224, 229
Barnard, Chester 252
Baumol, William 128, 247
Becker, Gary S. 184, 259
behavior theories 2478
Berle, Adolphe A. 247
Bernouilli, Daniel 286
Bewley, Truman 177, 1824, 186
billiards player example 212, 71, 73, 78,
79, 84, 87, 88, 244, 258, 2613, 279
Binmore, Ken 2646, 267, 268, 271
Blaug, Mark 144, 14551, 152, 1547, 158,
159, 161, 162, 317, 321, 342, 353
Bloor, David 199
Blume, Lawrence 2701
Boland, Lawrence 61, 62, 350, 351
Bordo, Michael 321
Brain Trust 192
Brinkley, Alan 193
Brunner, Karl 60, 137
Buchanan, James 250
Burns, Arthur 75, 80, 81, 87
business cycle theories 33, 69, 82
Cambridge approach 42, 2213, 224, 227,
237, 307
Cartwright, Nancy 310
casual empiricism 123, 126, 127, 131, 132,
136, 137, 138, 140
causal realism, methodology of 30319
chalk mark example 256, 36, 80, 82
Chamberlin, Edward 38, 39, 72, 87, 93,
220, 2234, 226, 229, 231, 236, 244,
322, 323, 342, 352
Chicago school 68, 125, 127, 140, 151, 158,
246, 307, 3223, 352
Index
choice behavior 1734
Christ, C. 322, 32930
cigarette tax example 367, 72, 82
Clapham, John 222
Clark, John Bates 329, 337
Clark, John Maurice 223
Clark medals 127, 128
Coase, Ronald H. 172, 249
collectivism 323
Commons, John R. 249
consilience 1726, 186
conspiracy in restraint of trade 278
consumption function 134, 137,
140, 193
contingency tables 198
controlled experiments 1011, 76
cost controversy 222
Cournot, A. A. 223, 331
Cowles Commission 323, 335, 343
cross-elasticities 36, 38, 72
Cubitt, R. 296
Dawkins, Richard 265
De Marchi, Neil 52, 56, 159, 321
Debreu, Gerard 132, 1478, 1569, 335
demand curves 24, 356, 39, 123, 3259
demand theory 78, 15960, 196, 197
Descartes, Ren 303
description invariance 2945, 298
descriptive accuracy 334, 82
descriptive validity 73
Descriptive validity vs. analytical relevance
in economic theory (Friedman) 69
De Vroey, Michel 56, 64
Dewey, John 56
Director, Aaron 192, 322
discrimination 29, 81
distance formula see law of falling bodies
Dixit, Avinash 247
dominance heuristic 299
Duesenberry, James 130
Durkheim, E. 199
Easley, David 2701
econometrics 119, 130, 1356, 318
economic methodology 16587; see also F53
and the theory of the firm; F53 as a
realist statement; Friedmans
positive methodology and the
formalist revolution
economic policy 1545, 189
economic research 1267, 153, 18990, 208
economics 1734
early drafts of F53 68, 70, 76, 87
in F53 35, 18, 2931, 3940
357
F53 and the marginalist controversy 219
formalist revolution 1445, 14650,
153, 155, 156, 1589, 161
influence of F53 119, 125, 126, 130,
1312, 139
map of multiple perspectives 47, 48, 53,
55, 58, 60, 63
politics of positivism 193, 202, 20410
economies of scale 6, 222
efficiency wages 182
Eiteman, Wilfred 228
elasticity of demand 356, 218, 223
Enke, S. 279
epistemic equivalence 1089
Euclidean geometry 256, 36, 80
EUT see expected utility theory
evidence 49, 16587, 28995
early drafts of F53 76, 77
in F53 1011, 1214, 23, 289, 31, 31, 33
formalist revolution 1489, 151, 159
underdetermination of theory 1079
evolutionary economics 104, 248, 257,
25968, 26981, 2812, 351
expected utility theory (EUT) 56, 2645,
285, 2868, 28995, 2959, 301
experiments 1011, 49, 76, 174
explanatory unification 1067
F-mix 90, 112
F-twist 90, 304, 30612
F53 (The methodology of positive
economics, Friedman)
appraisal of evidence in economic
methodology 1656, 168, 1713,
1756, 1867
as a realist statement 110, 11213
early drafts 6888
external contexts 56
F53 and the marginalist controversy
21738
F53 in and out of context 34953
Friedman and the theory of the firm
1712, 241, 24254
Friedman on the MarshallWalras
divide 330
Friedmans original essay 37, 343, 3,
716, 1623, 2330, 309, 3943
358
Index
F53 (The methodology of positive

economics, Friedman) (cont.)
Friedmans risky methodology 285,
2923, 295, 297, 3001
Friedmans selection argument 25782
influence of F53 49, 11940
Friedmans themes 1204
macroeconomics 1204
monetarism 1367
Nobel Prize lectures 1312
new classical theory 138
reviews 12830
some difficulties in estimating the
influence 1248
sources of influence 13940
map of multiple perspectives 4766
evaluating and explaining F53 5961
examining and using F53 513
external contexts of F53 559
extra-economic and intra-economic
labels 616
internal characteristics of F53 535
methodology of causal realism 3034,
305, 309, 310, 31214, 31719
politics of positivism 18995, 196203,
204, 2056, 2078
rereading and rewriting F53 as a realist
statement 90115
assumptions and predictions, and the
non-predictivist torso 924, 967
assumptions and their roles: vacuum
and maximization 99101
how to rewrite F53 11214
indirectly testing approximation to the
truth 956
manipulation of truth-values by
paraphrasing assumptions 1024
from reception to the authors ideas
11415
subjective and social dimensions of
theory appraisal 10912
testing by assumptions and testing by
predictions 967
theoretical isolation and explanatory
unification 1067
truth of assumptions and truth of theory
1012
two kinds of as-if 1047
underdetermination and theoretical
values 1079
unrealisticness as irrelevant and as a
virtue 946
the whole truth and nothing but the
truth 979
fallibilism 91, 113
falsificationism 62, 63, 109, 110, 112, 148,

158, 160
Ferejohn, John 2634, 267, 268
firms 21, 357, 712, 82, 83, 16772
perfect competition 50
see also F53 and the marginalist
controversy; Friedman and the
theory of the firm; Friedmans
selection argument
Fishburn, P. C. 299
Fisher, Irving 42, 126
Fisher, Ronald 191, 198, 200
formalism see F53
free competition 352
Friedman, Milton
appraisal of evidence in economic
methodology 165, 169, 171, 174,
1856
21738
theories of the firm (plural) 24554
Friedman on the MarshallWalras divide
64, 32143
a criticism of Friedmans Section V
argument 3347
F53 330
Friedmans 1941 review of Triffin 3234
Friedmans 1946 review of Lange
3245
Friedmans 1949 paper on the
Marshallian demand curve 3259
Friedmans 1951 comments on
Christs article 32930
Friedmans 1955 review of Jaffs
Walras translation 321, 3312
Friedmans subsequent views 3324
Friedmans understanding of the
Walrasian modifier 33742
relation between 1949 article and 1955
review 3423
F53 licensing the formalist revolution
57, 94, 1527
on the formalist revolution 14550
Friedmanic license 1502
development of expected utility theory
2868
EUT as an empirical hypothesis 28995
market selection and the FS-twist 2957
normative notions in descriptive
dialogues 297300
Index
evolutionary economics 257, 27581
evolutionary game theory 26975
evolutionary reinterpretations of
economic theory 2608
methodology of causal realism 30319
absence of causal talk 306
effects, not causes 31617
F-twist in the light of Marshall 30612
Friedmans anti-realism reconsidered
31216
Friedmans causal legacy 31719
the God of Abraham; the methodology
of Marshall 3035
politics of positivism
Friedman rewritten? 20410
how to disagree on predictions 196200
making the contract explicit 2013, 204
politics in F53 18990
predicting for the policy makers 1905
see also F53 and other individual works
Friedman, Rose D. 200, 351
fruitfulness 10, 77, 108, 109, 241
FS-twist 289, 291, 2957, 301
FS48 (The utility analysis of choices
involving risk) (Friedman and
Savage) 285, 286, 289, 2923, 300
FS52 (Friedman and Savage) 285, 28992,
2978, 3001
full-cost pricing 217, 2256, 228, 230,
2356, 237, 258, 277
full employment 1945
359
Hands, Wade 57, 63, 161, 303, 317, 323, 353
Harrod, Roy 225, 226, 227, 228, 25960,
277, 281
Harsanyi, John 132
Harvard approach 2212, 2234, 2268,
230, 237, 246, 322
Hausman, Daniel 143, 237
Hayek, Friedrich 222, 252
Heflebower, Richard 22930
Hicks, John R. 222, 227, 324, 3267,
332, 334, 33940
Hirsch, Abraham 52, 56, 159, 321
Hirshleifer, Jack 259
Hitch, C. J. 97, 2256, 228, 258
Hitler, Adolf 193
Holland, Paul 317
homo economicus 265
Hoover, Kevin 54, 63, 92, 321, 333
Hotelling, Harold 191, 196, 198, 334
Hume, David 42
Hutchison, Terence 7, 915, 1623, 234,
2630, 31, 335, 401, 423, 702,
747, 7881, 846, 93, 108, 11011,
112, 1201, 127, 129, 144, 1459,
1501, 152, 1547, 158, 159, 161,
162, 175, 2078, 317, 353
imperfect competition, 15, 34, 38, 70, 167,
169, 220, 223, 315, 330
implications 268, 69, 77, 81, 92, 121
indifference curves 3323
indirect evidence, 2830, 40, 85, 104, 111,
209, 289
industries 356, 38, 39, 72, 82, 87, 16671,
224, 236, 280, 324, 328, 336
inflation 11, 76, 193
institutionalism 130, 139, 152, 166, 193,
2212, 2246, 227
instrumentalism 47, 623, 90, 91, 96, 101,
103, 105, 109, 114, 350, 352
insurance 289, 292
introspection 123, 126, 127, 131, 137,
173, 207
investment 135
invisible hand 57, 150, 156, 157, 165,
1669, 171, 176, 1801, 1857, 325
gambling 289, 292

game theory 73, 83, 130, 172, 266, 26975
general equilibrium theory 86, 120, 1478,
149, 1567, 158, 270, 325, 329, 332,
3367
General Theory (Keynes) 175, 1778, 179, 180
Georgescu-Roegen, Nicholas 253
God 3034
good approximation assumption 352
Gordon, R. A. 734, 85, 87
governance 249
government intervention 186, 191, 194
Granger, C. 125
Great Depression 191, 192, 224
growth models 136
Guthrie, G. E. 228
Jaff, William 322, 3312, 342

Jorgenson, Dale 135
Hahn, Frank 136, 137, 156, 158, 160

Hall, R. J. 97, 134, 2256, 228, 258
Hamilton, Walton 225, 226
Hammond, Daniel 52, 56, 61, 63, 121, 123,
316, 321
Kahneman, Daniel 1745, 178, 295, 299

Kandori, M. 274
Keynes, John Maynard 42, 125, 134, 137,
149, 175, 1778, 202, 324, 329, 341
Keynes, John Neville 3, 4, 75, 76, 88, 124
360
Index
Keynesianism 84, 1346, 1378, 140, 148,

1512, 1589, 166, 176, 179, 1802,
186, 192
Klappholz, Kurt 160
Klein, Lawrence 135, 322, 323, 32930
Kneeland, Hildegarde 191
Knight, Frank 192, 193, 223, 233,
235, 326
Koopmans, Tjalling 54, 104, 2445, 279,
323, 349
Kuhn, Thomas 157, 243, 300, 351
Kuznets, Simon 196, 197, 199
labor market 1769, 1805, 186,
21719, 223
Laffont, Jean-Jacques 252
Lakatos, I 300
Lange, Oscar 122, 123, 128, 158, 198, 322,
323, 3245, 326, 341, 342
law of falling bodies 1619, 245, 70,
789, 82, 87, 95, 99100, 1012,
1023, 232
Lawson, Tony 146
leaves around a tree example 1920, 24, 79,
84, 88, 105, 107, 261, 2623, 312
lens of contract approach 24853
Lerner, Abba 124
Lester, Richard 97, 21720, 222, 228, 231,
2345, 258
life cycle hypothesis (LCH) 134, 140
linear regression 198
logic 11, 24, 74
Loomes, G. 297
Lucas, Robert 1789, 184, 333
Machina, M. 299
Machlup, Fritz 49, 127, 21820, 222,
2278, 229, 231, 233, 235, 238
macroeconomics 11920, 13940, 148,
1512, 159, 217
Keynesian program 1346
monetarism 1368
new classical theory 138
Mailath, George J. 2712, 275
Mki, Uskali 55, 57, 61, 63, 174, 351, 355
managerial theories 2478
marginal analysis 15, 22, 32, 70, 71, 73, 74,
77, 87, 93, 104, 171, 185, 223; see
also F53
marginalist controversy see under F53
Marris, Robin 247
Marschak, Jacob 323
Marshall, Alfred 345, 36, 378, 41, 56, 61,
72, 76, 82, 84, 86, 87, 149, 1567,
165, 167, 2204, 2368, 244, 305,
30610, 31315, 316, 3214, 3259,

330, 331, 3327, 33843, 352
Marshallian economics 634, 103, 120,
1223, 1257, 131, 132, 134, 135,
136, 138, 140, 149, 155, 190, 198
Martimort, David 252
Mason, E. 224, 2267
mathematics 11, 12, 24, 76, 122, 125,
1457, 149, 197, 2068, 221, 245
Matthews, Robbin 136
Mauss, M. 199
maximization of returns see profit
maximization
Mayer, Thomas 52, 57, 63, 52, 144, 1457,
14850, 1513, 156, 1589, 1612,
303, 353
Mayr, Ernst 267
McCloskey, Deirdre 146
Means, Gardner 2245, 226, 247
memes 266, 268
Menard, M. 20410
methodology 689, 867, 107, 11213,
11415, 119, 150, 152, 154, 1558,
1612, 231
see also economic methodology; F53
microeconomics 119, 13940, 147, 217
microfoundations 165, 178
Mill, John Stuart 34, 157, 313
minimum wage legislation 56, 76, 21819,
234
Mirowski, Philip 158, 323
Mitchell, Wesley 56, 83, 130, 152
Modigliani, Franco 134
monetarism 1368, 140, 1512, 159, 350
A Monetary History of the United States
(Friedman and Schwartz) 119, 124,
303, 305, 306, 309, 314, 315, 353
monetary policy 192, 225
monetary theory 42, 84, 124, 135, 314
money illusion 175, 178
money pump arguments 296
money supply 11, 194, 306
Mongin, Philippe 237
monopolistic competition 15, 346, 389,
49, 59, 70, 72, 77, 82, 87, 93, 94, 167,
169, 172, 223, 226, 2278, 2367,
244, 3234, 335, 352
monopoly 27, 29, 34, 35, 87, 93, 171,
2223, 227, 246, 310
monotonicity 298, 299
moral hazard 201, 210
Morgenstern, Oskar 73, 83, 245, 286
Musgrave, Alan 55, 102
Index
Nagel, Ernest 62
Nash equilibrium 2668, 2723, 2745
National Recovery Act (1933) 225
National Resources Committee (NRC)
191, 201
natural selection 22, 71, 79, 84, 88, 104,
244, 25782, 295, 298, 34950, 351,
352 see also evolutionary economics;
F53: Friedmans selection argument
Nelson, Richard R. 260, 276, 27881
neoclassical theory 15, 69, 93, 121, 130,
135, 1656, 167, 169, 1746, 1778,
181, 185, 187, 217, 244, 2467
new classical theory 138, 140, 148, 151,
1656, 178, 180, 184
New Deal 1902
Newman, Peter 128
Neyman, Jerzy 191
Nobel Prizes 120, 127, 1312, 174, 175,
178, 181
normative economics 35, 67, 40, 57,
756, 88, 111, 1234, 137, 187, 192,
195, 2046, 210, 242, 246, 297300
Nowak, L. 310
Noyes, Carl 200
NRC see National Resources Committee
objectivity 25, 30, 40, 801, 110; see also
F53: map of multiple perspectives
observational equivalence 1089
oligopoly 27, 224, 226, 227
Oliver, H. M. 81, 219, 231
ontological unification 1067, 109, 113
organization theory 248
Oxford approach 2212, 2246
Oxford Economists Research Group
(OERG) 225, 228
Pareto optimality 270, 2723, 281, 340
Pascal, Blaise 3034
path dependence 273
Patinkin, Don 56, 2305, 334
perfect competition 15, 348, 69, 72, 82, 87,
93, 220, 222, 224, 2367, 352
perfect monopoly 15
permanent income theory (PIT) 134, 140, 312
philosophy 63, 3034
philosophy of science 55, 59, 61, 64, 68,
155, 158, 160, 162, 199, 203
Pigou, A. C. 222, 223, 249
pin theory 3045
PIT see permanent income theory
pluralism 253
policy-making 127, 1905, 205, 206
Popper, Karl 130, 148, 158, 160, 280, 351
361
portfolio theories 135
positive economics 37, 716, 41, 42, 57,
756, 86, 88, 110, 111, 1234, 137,
186, 189, 195, 2046, 210, 2412,
3001, 318
positivism 62, 63, 91, 109, 112, 130, 152
politics of positivism 18995, 196203,
20410
postmodernism 124, 131, 139
predictions 2412, 253, 260, 31213
in F53 910, 12, 1415, 17, 18, 201, 33,
40, 41, 77
in testing 93, 957, 110
neutrality of 18990, 195
policy relevance of 1945
see also F53: influence of; map of multiple
perspectives; politics of positivism;
rereading and rewriting F53 as a
realist statement
predictivism 656, 93, 967, 110
price policy 2257
price theory 217, 222, 315
Principles of Economics (Marshall) 72, 82, 84,
87, 165, 167, 220, 223
products 35, 38, 39
professions 1967, 199
profit maximization 213, 27, 31, 49, 54, 59,
712, 73, 79, 846, 87, 95, 97, 99101,
104, 1667, 171, 1834, 21921,
2256, 228, 232, 2445, 247, 258,
2613, 269, 2768,
296, 351
psychology 30, 60, 1734
public choice theory 140, 153, 158
q theory 135
quantity theory of money 42, 84
Quiggin, J. 299
Rapping, Leonard A. 1789, 184
rational choice theory 2634, 267
rational expectations theory 138, 140,
148, 151
realism
evolutionary economics 281
in F53 1415, 1623, 313, 41, 42, 74,
77, 82, 87
see also under F53: Friedman and the
theory of the firm; influence of F53;
map of multiple perspectives;
methodology of causal realism;
politics of positivism; rereading and
rewriting F53 as a realist statement
470; invisible hand and validity of
theory
362
Index
reception methodology 578, 65, 92, 94,

11415
Reder, Melvin 523, 567, 244, 254, 352
reductionism 119, 138
regression curves 198
relative price theory 41, 42, 84
The relevance of economic analysis to
prediction and policy (Friedman)
7586
revealed preference theory 15960
risk see Friedmans risky methodology
Robbins, Lionel 131, 222
Robinson, Joan 38, 72, 148, 220, 223,
226, 227
Robson, Arthur J. 268
Roosevelt, Franklin D. 191, 192, 193,
2012, 225
routines as genes 27880
Russell, Bertrand 311
Samuelson, Paul 132, 15960, 217, 245,
287, 303, 304
Satz, Debra 2634, 267, 268
Savage, Leonard 21, 56, 71, 79, 262, 265,
285, 2889, 292, 294, 297, 300
Schneider, Erich 129
Schultz, Henry 322, 326, 334, 340
Schumpeter, Joseph 129, 165
Schwartz, Anna 119, 159, 3045, 306, 309,
31416, 321, 353
science 47, 55, 589, 60, 110, 1489,
1556, 1578, 161, 203, 2067,
2323, 312
in F53 25, 30, 33, 80
selection argument see natural selection
Selten, Reinhard 266
Sherman Antitrust Law 27
Shiller, Robert 179
Simon, Herbert 245, 253
simplicity 10, 77, 108, 109, 241
Slutsky, E. 326, 327, 3323
Smith, Adam 155, 1567
social constructivism 62, 63, 91, 112, 114
social epistemology 11314
social sciences 10, 14, 19, 40, 59, 158, 162,
2323, 245
social scientific contract 2023
socialism 323
sociology 2930, 110
Solow, Robert 2523
Soper, C. S. 129
Sraffa, Piero 148, 222
Starmer, C. 534, 56, 297, 299
static monetary theory 42, 314
statistics 1912, 193, 194, 196, 197, 202
Stigler, G. J. 80, 87, 139, 218, 219, 222, 229,

2335, 238, 265, 322, 326, 331,
343, 352
Stiglitz, Joseph 178, 181, 186
Strong, Benjamin 315
Study of Consumer Purchases (SCP) 191
subjective judgments 110, 111, 113
Sugden, R. 296, 297
supply and demand 78
tautology 1112, 74, 76, 122, 231, 233, 234
tax 9, 37, 72, 77, 108, 193, 248
Teira Serrano, David 556, 58
theory 186
early drafts of F53 6970, 724, 767,
801, 84, 87
in F53 78, 1115, 17, 1819, 20, 2330,
302, 34, 38, 402
F53 as a realist statement 923, 945,
969, 1012, 10612, 113
influence of F53 122, 124
map of multiple perspectives 54, 62
methodology of causal realism 310,
31214
theory-choice
normative criteria 285301
social shaping 10912
underdetermination 1079
Tobin, James 135, 306, 334
tort problem 249
transaction cost economics 252
Triffin, Robert 73, 87, 236, 322, 3234,
3412
truth
approximate 956
manipulations of truth-values 1024
of assumptions and of theories 73, 1012,
266, 309
the whole and nothing but 979
Tugwell, Rexford 192
Tversky, A. 174, 178, 295, 299
unemployment 17680, 1812, 186
unification 1067
utilitarianism 165
validity 166, 173, 175
early drafts of F53 69, 71, 73, 77
in F53 89, 1314, 20, 28, 402
F53 as a realist statement 93
influence of F53 122
van Damme, Eric 266
Van Fraassen, Bas 303, 308
Veblen, Thorstein 81
Index
velocity of circulation of money 31112
vertical integration 250, 252
viability analysis 279
Vickrey auctions 297
Vickrey, William 128
von Neumann, John 73, 83, 245, 286
Vromen, Jack 534, 351
wages 175, 186
wage behavior and involuntary
unemployment 17685
21719, 223, 234
Wallis, W. Allen 197, 322, 326, 332
Walrasian economics 1201
and F53 64, 86, 109, 353
formalist revolution 147, 149, 155,
158, 159, 160
influence of F53, 1203, 125, 1267, 131,
132, 134, 135, 138, 139
363
map of multiple perspectives 56,
59, 64
politics of positivism 190, 1979
see also Friedman on the MarshallWalras
divide
Weber, Max 129
Weintraub, Roy 145
welfare economics 156, 223, 245
wheat market example 37, 73, 82, 978
Williamson, O. E. 41, 53, 60, 66, 2478
Wilson, E. B. 196
Wilson, E. O. 173, 175
Winter, Sidney G. 54, 260, 270,
27681, 350
WTA (willingness to accept) 2967
WTP (willingness to pay) 2967
Yale school 135, 137
Young, Allyn 223
Zamora Bonilla, Jess 556, 58

LIBRO. Uskali Mäki (2009), The Methodology of Positive Economics

Uploaded by

Copyright:

Available Formats

LIBRO. Uskali Mäki (2009), The Methodology of Positive Economics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

LIBRO. Uskali Mäki (2009), The Methodology of Positive Economics

Uploaded by

Copyright:

Available Formats

This page intentionally left blank

The Methodology of Positive Economics

The Methodology of Positive

CAMBRIDGE UNIVERSITY PRESS

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, So Paulo

Cambridge University Press has no responsibility for the persistence or accuracy

Part 1 The classical essay in twentieth-century economic

The methodology of positive economics (1953)

Part 2 Reading and writing a classic

1 Reading the methodological essay in twentieth-century

2 Early drafts of Friedmans methodology essay

Unrealistic assumptions and unnecessary confusions:

Part 3 Models, assumptions, predictions, evidence

The influence of Friedmans methodological essay

Did Milton Friedmans positive methodology license

Appraisal of evidence in economic methodology

The politics of positivism: disinterested predictions from

Theoretical context: firm, money, expected utility,

Friedmans 1953 essay and the marginalist controversy

Friedman (1953) and the theory of the firm

Friedmans selection argument revisited

Expected utility and Friedmans risky methodology

Milton Friedmans stance: the methodology

On the right side for the wrong reason: Friedman on the

The debate over F53 after fifty years

9.1 The sciences of choice and contract

4.1 Effect of changes in economics on the acceptance

r o g e r e . b a c k h o us e is Professor of the History and Philosophy of

also the editor of Macroeconometrics: Developments, Tensions, and

of New Institutional and Austrian economics; in journals such as the

c h r is s t ar m e r is Professor of Experimental Economics at the

o l i v e r p . w il l i a m s o n is Professor of the Graduate School and Edgar

Milton Friedman is known for a variety of contributions. Professional

The classical essay in twentieth-century

The Methodology of Positive Economics*

N HIS admirable book on The Scope and Method of Political

Confusion between positive and normative economics is to

The Methodology of Positive Economics

Essays in Positive Economics

minimum wage without any counterbalancing increase in the

The Methodology of Positive Economics

The ultimate goal of a positive science is the development of

Essays in Positive Economics

ceding list of analogical questions. Viewed as elements of the

The Methodology of Positive Economics

comparison of its predictions with experience. The hypothesis

Essays in Positive Economics

The Methodology of Positive Economics

analysis and involved chains of reasoning, which seldom carry

Essays in Positive Economics

ogies if it is to be able to predict and not merely describe the

The Methodology of Positive Economics

consistent with the evidence at hand, its further testing involves

Essays in Positive Economics

hypotheses or revision of old ones. So the two methodologically

The Methodology of Positive Economics

Essays in Positive Economics

The Methodology of Positive Economics

dropped instead of a compact ball. The formula then gives wildly