Marchi The Popperian Legacy in Economics
Marchi The Popperian Legacy in Economics
Marchi The Popperian Legacy in Economics
10
economics
Edited by
NEIL DE MARCHI
Department of Economics
Duke University and
University of Amsterdam
oil manner
of books
was granted by
Henry
VIII
in
15J4.
158.f.
A catalogue record for this publication is available from the British Library
cm.
I. de Marchi, Neil.
HBI03.P66P66
-Views on Economics
1988
330. I-dcl988-15 I 9
ISBN 978-0-521-35576-6 hardback
ISBN 978-0-521-07023-2 paperback
To loop Klant
Contents
page ix
Preface
List of contributors
xi
Introduction
Neil de Marchi
Discussion
Part I:
17
65
87
Part II:
139
121
Part IV:
169
183
199
Contents
viii
8
Part V:
213
231
245
11
259
Economics as discourse
Arjo Klamer
Index
279
Preface
This volume had its origin in a desire on the part of Joop Klant's col
leagues at the University of Amsterdam to mark his retirement from the
Chair of History and Philosophy of Economics. That desire translated
into an attempt to focus scholarly effort on some of the themes that have
informed his work over the past quarter-century.
Those themes are clearly evident in his own chapter, "The Natural
Order," and are also spelled out in his contribution to the Discussion.
They have to do with the nature of economics and with certain implica
tions for being an honest practitioner in that discipline. First of all,
economics is not value-free. That means that the choices we make about
theories and policies in economics inevitably reflect our preferred no
tions of how the world is constituted. Furthermore, because we do not
have natural constants in the world of economics and because our theory
in economics is often so general (for instance, "agents optimize") that
only specific versions can be tested, this leaves the basic theories them
selves immune to test results. In our efforts at self-criticism, therefore,
we have to go beyond mere testing for falsifying instances. This does not
mean that striving after falsifiable theory is unimportant. It does mean
that we must acknowledge and identify, as far as possible, the role of
"vision" (perceived natural order) and art - the art of the good
practitioner- in our economic "science. "
Klant, it should be said, i s a novelist of some repute i n the Nether
lands and also an artist in his own right, and was for years an analyst and
adviser in a major commercial bank before taking up his chair at the
University of Amsterdam. His perspective on economics bears the
marks of one familiar with the various ways of engaging and persuading
others that go beyond the empirical core that he too would acknowledge
lies at the heart of his methodological writing.
The initiative for a conference came from Wim Driehuis, Professor of
Macroeconomics at the University of Amsterdam. We agreed that such
a gathering should pick up on the doubts that lay behind Klant's study
on "the logical structure of economic theories," which is the subtitle to
his The Rules of the Game. These were, as he wrote in the preface,
"doubts . . . about how we, economists, can prove our theories. How do
IX
Preface
Contributors
xu
Contributors
Introduction
NEIL D E MARCHI
de Marchi
Introduction
de Marchi
Quite apart from what he had to say viewed as matters of logical rela
tions and of the properties of statements, he represented an attitude-to
be critical. Neither fact nor theory is more than an element in the pro
cess of identifying error. This was liberating for economists in a special
way. Popper's balanced insistence on empirical content and on the epis
temological priority of theorizing might have been tailored to appeal to
practitioners in a discipline where experimentation is difficult and incon
clusive and theory seems more solid, yet where numbers are seen to be
essential to adapting theory to yield advice for policy making. His stress
on methodological conventions- rules of the game- was helpful to a
group of social scientists anxious to be useful and to explain themselves
to a somewhat reluctant public, yet conscious of the fallibility of their
pronouncements. In short, in contrast to much writing in the philosophy
of science, Popper's work was not only accessible to economists but
seemed relevant.
Economic methodologists and historians of economics took to Popper
with avidity. Some methodologists - Hutchison (1964, 1968, 1977, 1981 ,
in addition to his pioneering 1938) and Blaug (1975, 1980a, 1980b) in
particular, but also Klant (1979) and Boland (1982, 1985) - used Pop
per's methodological rules to criticize economic practice. Historians of
economics, encouraged by Popper's attention to historical episodes in
science and to "problem situations," began trying to elucidate episodes
in their own parent discipline with the aid of Popper's rules and perspec
tive (Blaug, 1968, ch. 16; de Marchi, 1970; Latsis, 1972) . Both groups,
however, quickly ran up against what might seem to be a fundamental
difficulty: Economists pay lip service to notions like empirical content
and falsifiability and deductive testing but do not actually put the Pop
perian rules for strenuous testing into practice (see esp. Blaug 1980b).
This caused no widespread consternation, however, both because it
had always been known in some sense and because the Kuhn, Lakatos,
and Feyerabend alternatives to Popper were being developed and made
available. Kuhn gave a place to actual practice and real history that Pop
per's and Lakatos's "rational reconstructions" could not. On the other
hand, Lakatos's methodology of scientific research programs, with its
uncriticizable hard core, offered a seductive way to rationalize the very
failure of economists to apply the Popperian conventions of good science.
In this sense it, too, came closer to actual economic practice. Feyera
bend's "anarchism," for its part, served to render inquiry into all kinds of
methods and all kinds of subgroups among economists free from the
judgments of "scientific" acceptability or its opposite, which seemed to be
part of every attempt to apply Popper's demarcation criterion.
Preliminary attempts to apply Kuhnian and Lakatosian schema quickly
Introduction
showed that they held problems of their own. How many paradigms are
there in the history of economics- one or many? Is there predictive force
to the notion that revolution is preceded by an accumulation of anoma
lies, or is Kuhn's framework primarily useful for writing history? Is it
damaging if we cannot reach precise agreement on the elements of the
hard core of a research program? What exactly is a novel fact? How
should one describe testing between noncomparable but supposedly com
peting paradigms or research programs? And so on. These sorts of ques
tions led historians of economics to wonder about the usefulness of the
grand visions of Kuhn and Lakatos. Both of these frameworks and more
traditional philosophy of science rested heavily on studies of physics and
astronomy and, in the case of Lakatos, mathematics. The difficulties that
arise in trying to transfer the results to economics have also led a few
philosophers to examine economics and its literature afresh as a distinct
area of inquiry (Rosenberg, 1976, on the nature of microeconomic laws;
Mackay, 1980, on Arrow's Theorem; Hausman, 1981, on capital theory;
Nelson, 1984, 1 987, on the reduction of macroeconomic phenomena to
microfoundations; Bicchieri, 1987, on rationality, information, and equi
librium). The growing skepticism about grand visions and the new-found
interest in economics as a subject of philosophical investigation have
given rise to increasingly detailed examinations of how economists actu
ally go about their work (Wong, 1978; Hamminga, 1983 ; Klamer, 1983;
Diesing, 1985 ; McCloskey, 1985 ; Weintraub, 1985 ; Hirsch and de
Marchi, 1986; also, for econometrics, though with a slightly different
motivation, Hendry, 1980, 1983, 1986).
Two further elements that have been moving things in this direction,
though at the same time away from traditional philosophical concerns,
are the work of philosophers such as Richard Rorty and of adherents of
the so-called Strong Program in the sociology of science at Edinburgh
University. The early response of Popper and Lakatos to Kuhn was that
he was replacing rational history of science with the history of science as
social psychology and sociology. Rorty (and others, in complementary
ways) have challenged this kind of criticism by pointing out that it con
founds explanation and justification, and presumes that the rational
account of knowledge is in some privileged position. An " 'account of
the nature of knowledge'," Rorty points out, "can be, at most, a descrip
tion of human behavior" (1979, p. 182). Aiding and abetting these phi
losophers have been literary critics, such as Stanley Fish (1980), who are
intent on dethroning structure and text as touchstones of truth (or at
least meaning). Klamer and McCloskey independently have embarked
upon research involving the rhetorical analysis of economic writing and
a close look at the discourse that takes place among economists and
de Marchi
between them and other audiences. This reinforces the efforts of a group
of philosophers and social thinkers linked to the Strong Program to
construct a social theory of knowledge (Bloor, 1983; see also Barnes and
Edge, 1982, and Knorr-Cetina and Mulkay, 1983) .
Historians of economics and economic methodologists might find
much in the new "rhetorical" and "social" approaches that is attractive.
These approaches at least claim to relieve them from having to engage in
demarcation disputes. Nor need they spend time rationalizing the dis
crepancy that exists between the lip service economists pay to
falsifiability and testing and their very different practices. They are free
to examine just what counts as supporting evidence without fear of
dismissal by anti-inductivists. And they can begin to attend to the actual
"conversations" that take place among economists, and between econo
mists and their audiences, without expecting to find debate being closed
by some criterion of truth or adequacy imported from outside.
2
Why Popper? Proximately because Klant's The Rules of the Game drew
inspiration from Popper's philosophy. More basically, however, be
cause, as should be clear by now, Popper is the one philosopher from the
years of dominance of the Positivists who, to many economic methodol
ogists, has seemed to convey a clear and simple set of guidelines for
exposing the self-indulgence of economists who cared little about the
empirical world. Simply put, Popper made it clear that this self
indulgence could not be tolerated if economics was to be counted among
the sciences. It is important to note that this reading of Popper is re
garded by some methodologists as a travesty. Lawrence Boland has
Introduction
consistently urged that Popper aims to be critical, not to offer rules for
providing a "sound" foundation for our knowledge (1982, esp. ch. 10) .
The fact remains, however-as Boland also acknowledges- that Pop
per's economics audience has tended to read him in the way suggested.
It would be a useful exercise, not undertaken here or elsewhere in this
volume, to trace the course of this divergence between the intended and
the received message. Popper's demarcation criterion was first brought
to the attention of economists in the 1930s by Terence Hutchison, who
used it to show just how much of what then passed for economic analysis
was tautology masquerading as substantive propositions. Popper himself
used the criterion and the notion that we cannot establish truth, but can
only identify error, in the fight against fascism. His ideas were put to
similar use by followers such as Hans Albert and Imre Lakatos in the
ideological struggles that wracked European universities in the 1960s.
Economic methodologists came to share the conviction that if their disci
pline is not value-free, it - and they - can at least strive after falsifiability.
That has been Klant's persistent plea.
The idea of positive (i.e. , nonnormative) economics, meanwhile, was
given a new lease on life by Milton Friedman - consciously influenced by
Popper, but no Popperian - when he linked it to the Popperian ideal of
science as conjecture and refutation. In Friedman's version the conjec
tures need not be realistic, provided only that they give rise to anticipa
tions that check out against the facts. This provided generations of uni
versity teachers with a handy rationalization for the apparent irrelevance
(unrealism) of much economic theory. More importantly the formula,
substantive hypothesis followed by empirical test, became the interna
tional standard for acceptable Ph.D. dissertations. By the 1970s this
bowdlerized message had become established in the minds of many
economists as the contribution of Sir Karl Popper.
In setting up the symposium, the organizers saw it as one of their tasks
to reexamine what Popper really did have to say to economists on mat
ters of methodological import. Further, in light of the debate of the last
twenty-five years in philosophy of science, it seemed appropriate to ask,
how much of the original message, in its English-language version just
three years older than Kuhn's book, is worth retaining?
4
There is much in Popper that need not be rejected along with rationalist
views on the way science develops, or should develop. Those elements
that can be retained stand out more clearly in the oral tradition surround
ing Popper's London School of Economics teaching than they do in his
de Marchi
writings, though they are there too. 2 The oral tradition stresses that he
was very concerned with imparting to students a sensitivity to evolving
problem situations, and to the creative element in formulating discrimi
nating hypotheses, and that he did not place too much emphasis on
falsifiability and testing, except in the sense of testing the real mettle of a
particular solution to a problem. That is to say, his instructional concern
was with the history of science - albeit a "rationally reconstructed his
tory" (Popper's phrase, before it was popularized by Lakatos) -as a
lead-up to criticizing current theories. Thus criticism (of proposed solu
tions to problems) was important, as were modeling and things like the
the rationality principle (or logic of the situation) ; but none of these
depended upon the strict dictates of logical, as distinct from conven
tional, falsificationism. As noted previously, they reflected an attitude
rather than faith in rules or methods.
If we can accept this, it actually puts Popper much more in line with
the limitations to which economists know they are subject. Moreover,
we see a Popper who at one level is by no means out of place alongside
Rorty, for whom, too, "there are no assertions which are immune from
revision" ( 1979, p. 181). Indeed, as Klant points out, although parame
ter inconstancy and the generic nature of our underlying theories in
economics make it pointless to try to apply a sharp demarcation line
based on falsifiability, the role of criticism is thereby actually enhanced.
Curiously, however, the symposiasts in Amsterdam were not taken with
the idea of seeing how much of Popper could be retained, but spent a
great deal of time discussing- inconclusively- the need for and possible
alternatives to Popper's demarcation criterion. This preference may re
flect the current confusion about standards noted already.
5
Four positions emerged in the course of the two days of discussion. First,
Klamer and McCloskey identified themselves from the outset as fanciers
of the- to most still somewhat exotic- Rorty-Fish agenda. The majority
response to this was that advocacy must be backed by evidence of superi
ority, and that accepting the "rhetorical" as one approach to understand
ing the way economics has evolved ought not preclude others. Some of
the promise of rhetoric, it should be said, has since been made actual in
a fascinating book by Don McCloskey, The Rhetoric of Economics
(1985) .
2 For a comprehensive introduction to Popper's philosophy, see O'Hear (1980). I am
Introduction
10
de Marchi
"right" practice, yet not deferred to in the actual doing. Why not? If
they are of little use, why even refer to Popper? If they are of genuine
value, why only pay lip service?
The first two papers, those by Hausman and Klant, ask, in effect,
what can we conclude if we take seriously both Popper and the limita
tions under which real economists do their work? Hausman concludes
that logical falsificationism is artificially narrow because so much more
than simple "basic statements" is taken as established background knowl
edge in actual practice. Worse , Popper's anti-inductivism distracts us
from examining how we produce the support for those things we do
accept. Klant reminds us that Popper himself backed away from apply
ing strict demands for falsification to economics. That is in accord with
the lack of numerical constants in the economic environment. Even
rational criticism, however, entails values and commitments, and is
strictly on the wrong side of Popper's demarcation criterion. Klant infers
that if we take Popper seriously we must acknowledge that falsifiability
is only an ideal and that economics is some mix of science (a la Popper) ,
"philosophy," or vision (a la Schumpeter) and art (as in J.M. Keynes's
famous dictum: "Economics is a science of thinking in terms of models
joined to the art of choosing models which are relevant to the contempo
rary world" (Collected Writings, vol. XIV, p. 295).
What really happens when economists take Popper seriously and
when they do not? Hands's study of how economists use notions of ad
hocness shows that economic theorists at least, in defining non-ad
hocness, tend to prefer Lakatos's idea of program continuity to Popper's
notion, which has to do with declining to adopt defensive stratagems. De
Marchi's paper examines the way Popper was taken up by a group of
young economists at the London School of Economics in the 1950s and
1960s. They found his views helpful in what was really (in part, anyway)
a struggle to accelerate the quantification of economics. That the encoun
ter was convenient and superficial, however, is suggested by the fact that
as soon as problems appeared in applying logical falsification - as in
qualitative comparative statics and probabilistic models -the economists
quickly abandoned Popper rather than explore his own efforts to deal
with the difficulties, some of which he had recognized.
If these two case studies and the preceding papers by Hausman and
Klant imply that the Popperian legacy in economics is insubstantial, or
survives only if we are prepared to modify substantially his logical falsifi
cationism, balance is provided by Terence Hutchison's restatement of
the case for falsifiability. His argument will be familiar to students of his
writing: He protests that hubris and mystification flourish where stan
dards are lax, and although falsifiability has its limitations in economics,
Introduction
11
12
de Marchi
Introduction
13
14
de Marchi
Fish, S.E. (1980). Is There a Text in This Class? Cambridge , Mass . : Harvard
University Press.
Hacking, I. (1983). Representing and Intervening. Cambridge: Cambridge Uni
versity Press.
Hamminga, B . ( 1983). Neoclassical Theory, Structure, and Theory Develop
ment: A n Empirical-Philosophical Case Study Concerning the Theory of
International Trade. New York: Springer-Verlag.
Hausman , D. (1981) . Capital, Profits and Prices: An Essay in the Philosophy of
Economics. New York: Columbia University Press.
(1984). The Philosophy of Economics: An Anthology. Cambridge: Cambridge
University Press.
Hendry, D . F . (1980). "Econometrics: Alchemy or Science?" Economica n.s.
47:387-406.
(1983). "Econometric Modelling: The Consumption Function in Retrospect."
Scottish Journal of Political Economy 30: 193-220.
(with N. Ericsson) (1986). "Assertion without Empirical Basis: An Economet
ric Appraisal of Monetary Trends in . . . The United Kingdom by Milton
Friedman and Anna Schwartz." International Finance Discussion Paper
No. 270, Federal Reserve Board of Governors, Washington, D.C.
Hirsch, A., and de Marchi, N. (1986). "Making a Case When Theory Is
Unfalsifiable: Friedman's Monetary History." Economics and Philosophy
2:1-2 1 .
Hutchison, T . W . (1938). The Significance and Basic Postulates of Economic
Theory. London: Macmillan.
(1964). 'Positive' Economics and Policy Objectives. Cambridge , Mass. : Har
vard University Press.
(1968). Economics and Economic Policy in Britain, 1946-1966. London: Al
len & Unwin.
(1977). Knowledge and Ignorance in Economics. Chicago: University of Chi
cago Press.
(1981). The Politics and Philosophy of Economics: Marxians, Keynesians, and
Austrians. New York: New York University Press.
Keynes, J.M. ( 1973). The Collected Writings ofJohn Maynard Keynes, Vol. xiv.
The General Theory and After. Part II: Defence and Development, ed. Don
ald Moggridge. Cambridge: Macmillan for the Royal Economic Society.
Klamer, A. ( 1983). Conversations with Economists. Totowa, N.J.: Rowman and
Allenheld.
Klant, J.J. (1979) . Spelregels Voor Econmen, 2nd ed. Antwerp: Stenfert Kroese.
(1984). The Rules of the Game: The Logical Structure of Economic Theories.
Trans. by Ina Swart. Cambridge: Cambridge University Press.
Knorr-Cetina, K.D., and Mulkay, M. (1983). Science Observed: Perspectives on
the Social Study of Science. London: Sage Publications.
Kuhn, T.S. ( 1970). The Structure of Scientific Revolutions, rev. ed. Chicago:
University of Chicago Press.
Introduction
15
Discussion
Hausman paper
Summary remarks
My thesis is that Popper's philosophy of science is a mess, and that
Popper is a very poor authority for economists interested in the philoso
phy of science to look to.
I am not concerned with all of Popper's philosophy of science. My
focus is on falsifiability, the idea that what distinguishes science from
nonscience is that scientific claims are falsifiable and that there are
certain rules of procedure which one can derive from the notion of
falsifiability, which characterizes how a scientist should go about his
business.
Popper employs two contradictory notions of falsifiability. Logical
falsifiability is logical inconsistency with some set of basic statements.
This gives rise to the asymmetry between verification and falsification
and to Popper's insistence that we can never have reason to believe that
theories are correct. All we can find out is that they are incorrect. In
other passages, however, Popper maintains that we can only decide
whether we are dealing with an empirical science by examining the
methods the supposed scientists employ.
A serious difficulty with logical falsifiability is that any test requires
background knowledge as well as basic statements. If we could rely on
nothing but relations between theories and basic statements, no science
would be possible.
Popper knows this and allows that we must combine our theories with
auxiliary hypotheses, and from these larger "test systems" derive predic
tions which involve basic statements. The problem with this is that we can
no longer falsify theories, only whole system amalgams. To insist that it be
possible to embed a theory in a logically falsifiable test system is, more
over, terribly weak. Virtually nothing is, in this sense, unfalsifiable.
So, logical falsifiability gets us nowhere. If we insist that theories be
logically falsifiable, there'll be no science, while if we insist merely on the
falsifiability of whole test systems, we are demanding almost nothing.
17
18
Discussion
Discussion
de Marchi: If Hausman is right, why did we ever take Popper seriously?
I suspect one part of the answer is that economists, like other scientists,
have found in Popper's rules of procedure both an encouragement to be
specific about what they are claiming and a feeling of freedom from an
obligation to be right. There is a challenge to be tough with oneself and a
sense of liberation.
But that aside, once we move away from logical falsification and in the
direction of "conventional" falsification, then there's not very much to
distinguish Hausman's position from the sorts of things McCloskey argues
for. "Shared background beliefs" puts one in mind of Kuhn. What then is
the unit of analysis? Whose beliefs? Who decides what is well established?
There are genuine difficulties here. We want a certain amount of clarity
about grounds for belief. The empirical study of the sociology of science
tends to abstract from this. Nor does McCloskey deal adequately with the
problem in telling us simply that we know when the conversation is
"good. " Dan Hausman encourages us to prospect in this old epistemologi
cal territory in a new way, but we desperately need a map.
To mention just one specific element that troubles me: It is not clear
that the alternative to logical falsification is conventional falsification.
"Plausibility" and "warranted belief" seem to be rich in content, and
both merit further exploration. Surprisingly, though, we seem to have
been caught up for several decades with matters of justification, and not
very much has been done on how economists actually explain. I would
like to see a shift in emphasis, but also to have Hausman tell us about
some of the next steps that should be taken.
McCloskey: What would Hausman make of the assertion that Popper's
arguments aren't so much arguments in epistemology as in ethics? If
Discussion
19
true, what is wrong with Popper is that he isn't "thick" in his ethical
argumentation . He doesn't use the long Western tradition of discussion
of ethical issues to help him understand why one might want to be a
good scientist.
20
Discussion
detailed studies of how theories are supported - and they are supported
in economics. That's not to say that, having examined economics more
carefully, we won't find that the sorts of arguments economists give are
very bad arguments. But to understand better what sorts of arguments
they do give is likely to be helpful. Philosophers really have a very, very
difficult task, to understand the nature of knowledge acquisition. There
are some complicated and not very well established systems which econo
mists would be crazy to rely upon in attempting to understand and judge
the work of other economists, but which may be useful in some cases. I
don't see why it should be simple to map out what the proper rules of
procedure are to understand the world. And it shouldn't be surprising
that some quite simple account like Popper's turns out to be terribly
inadequate.
Caldwell: But you do feel that we should look for rules of procedure?
Hausman: Implicitly . . . In any institution, and science is obviously an
institution, there are and always will be implicit accounts of how one
should do one's work. I don't see why they should be worse for being
made explicit.
Blaug: As a neo-Popperian, I find Hausman's position, while at first
sight seeming to be an onslaught on Popper, is, on second viewing, much
less controversial. What he says about Popper has been said by a lot of
Popperians. The strong asymmetry between verification and falsification
largely disappears once Popper in The Logic of Scientific Discovery
takes on board the Duhem-Quine thesis that we can never conclusively
falsify a theory, because any test involves not just the theory but also
auxiliary assumptions. So, one can never attribute the blame exclusively
to the theory. Popper is aware of this. That's why he introduces the set
of rules forbidding immunizing stratagems. That means he was perfectly
well aware that falsification is "conventional" falsification. It is also true
that he later goes on making statements that suggest that he believes that
the strong asymmetry is somehow maintained.
Now once you've appropriated the idea that falsification in Popper
involves a whole set of rules forbidding immunizing stratagems, and that
what he is laying down is a set of rules for good conduct, then you're left
at the end of the day asking, well, what then is Popper's contribution to
the philosophy of science? I'm inclined to say that what does survive is a
normative prescription, the idea that strong testing of theories implies
that the most desirable theories are those which involve a falsifiable
prediction. If you ask what states of the world does this theory really
Discussion
21
forbid, you get a much greater insight into the true discriminating empiri
cal strength of the theory than if you ask what states of the world will
corroborate it.
Summary remarks
I am interested in what economists do and in what they say they do, that
is, their pretensions.
One pretension that we economists have is that economics is objec
tive, or value free. Nobody, however, including Popper, has solved the
objectivity problem in economics.
There are similarities and differences between physics and economics.
Three are relevant to the objectivity problem.
22
Discussion
1.
2.
3.
Discussion
Kastelein: I want to challenge Klant's use of the term "plausibility" and
its meaning and function as a methodological concept. Klant's "plausibil
ity" has the same meaning as Popper's "assessment." Consequently,
Klant writes in terms of "justification," "acceptance and rejection," and
"choice. " In the end, however, economists are said to defend their
positions and to try "to convince each other of the plausibility of their
point of view."
Now everyone is free to use the words he wants; I am simply not very
Discussion
23
24
Discussion
Discussion
25
Summary remarks
I wanted to find a particular issue, with the aid of which I could relate
the responses of three groups of individuals: Popperian philosophers,
economic methodologists, and economic theorists. The issue I chose is
that of ad hocness, the modification of an economic theory in an "ad
hoc" manner. How does each of the three groups use this term?
In general, ad hoc modification of a theory means a patching up. The
patching up may be merely specifying exceptions so as to protect or
preserve the theory. Such modification may be acceptable, of course,
but on Popper's terms would be so only if it increased the falsifiability of
the theory or hypothesis. Otherwise it is to be classed among immuniz
ing stratagems.
Lakatos takes a somewhat different line. He was, in addition to the
above sense of ad hoc, concerned with the continuity of a research
program. His ad hocness has to do with some breach of this continuity.
A look at the literature turns up some surprises. Economic methodolo
gists seem to complain about ad hoc moves in Popper's sense. Economic
theorists, however, frequently use more the Lakatosian notion of ad
hocness. This emerges very clearly in the rational expectations litera
ture, though not only there.
The literature also contains some confusion between the problem of
the independent testability of hypotheses and the notion of novel facts.
One way to interpret independent testability is to say that a hypothesis
which is independently testable should predict something we have never
thought of before - a novel fact. It's not clear whether this is a sufficient
or only a necessary condition for independent testability. The issue is
important given that Lakatos's views on theoretical and empirical prog
ress make much use of "novel fact. "
I have been surprised too by the importance of the idea of heuristic
progress in the Lakatosian program, or non-ad hocness. Much of what
economists do that looks to be immunizing, however, is consistent with
what Lakatos calls "heuristic progress. " Weintraub's paper shows that
26
Discussion
you can find empirical progress in economics, but we all know that you
have to work hard to find it. On the other hand, heuristic progress, or
changes in the theory to avoid ad hocness, is something that happens all
the time.
Discussion
Nooteboom: It seems to me that the distinction between the two types
of ad hocness - the Popperian and the Lakatosian- is close to the old,
well-worn distinction between empiricism and rationalism. Seen in this
light, Professor Hands's observation that most (neoclassical) economists
are preoccupied with heuristic rather than empirical progress ties in with
our knowledge of a priorism as a traditional and still very strong ten
dency in economic thought.
What I miss in Hands's paper, however, is a clear normative judg
ment. In the actual conduct of neoclassical economists, program continu
ity has prime or even exclusive priority. But is this good or bad? To what
extent and when is priority of program continuity justifiable? I would
like to provide a basis for normative judgment.
We must recognize, firstly, the need for a wider basis for criticizing
both core assumptions and subsidiary assumptions. I propose that such a
widening of the basis for criticism is provided by considerations of plausi
bility. They are not by themselves sufficient, but we cannot do without
them.
Ad hocness, in the Lakatosian sense, is the acceptance of patched-up
explanations that do not follow from basic principles. Heuristic progres
siveness requires coherence with basic principles. But why stop at the
boundaries of one program? Why coherence only within a program and
not between programs? If it is ad hoc, within a given program of econom
ics, to accept explanations that do not follow from some established
conception of economic rationality, isn't it also ad hoc, in a wider per
spective, to accept economic conceptions of rationality that are in con
flict with common sense, direct observation, and psychology? I propose
that implausibility is also a form of ad hocness. It is lack of coherence, a
patching up, the acceptance of something which does not tie in with
some important part of knowledge. The concept of implausibility is thus
an extension of the concept of ad hocness. If ad hocness is lack of
coherence with the basic principles of a theory or program, implausibil
ity is lack of coherence with background knowledge. Should we perhaps
call this its "ad hoc4-ness"?
What normative judgments result from all this? To arrive at method
ological prescriptions that are realistic and useful, we should consider
Discussion
27
theories not just as finished products, but we should look also at their
life cycles and at the underlying processes of invention, development,
and use. A theory is a program in development.
The methodological rules we choose should reflect this fact of stage of
development. As an epistemological basis for this stage approach to
methodological rules, I favor using the "genetic epistemology" of Jean
Piaget. To illustrate, novel infant explanatory principles deserve protec
tion. The principles may be considered implausible for a considerable
time, and a benign neglect of falsifiers is justifiable.
As a theory develops, it is bound to accumulate problems and to
become more and more patched up. This is no cause for shame. The
anomalies are useful to indicate where and why explanatory principles
fail and in what direction alternatives should be sought. It is a cause for
shame to stick blindly to the program and pretend the problems are not
there. Program continuity is too easy when there are no clearly identifi
able forbidden events, when empirical anomalies can be circumvented
and implausibilities are not considered.
Methodological rules here are rather loose guidelines that cannot be
reduced to one simple, rigorous rule, and may be conflicting. But norma
tive order need not be logical order, and my claim is that the question,
Which rules deserve priority? is a matter of judgment. The judgment
must reflect, in part, the stage of development of a theory.
28
Discussion
Hands: We can give such examples, but there is a problem. The type of
novel fact discussed by Popper and Lakatos- physical novel facts, which
provoke a response of "Ah ha! - are not as likely to occur in economics.
That does not mean that we cannot discover a suitable sense of "novel
fact" for economics, but we do have a translation problem.
Hausman: A comment on ad hocness. For Lakatos a theory has to be
non-ad hoc in all three senses, not just in terms of preserving program
continuity. Let me give an example to show how difficult it can be to
apply Lakatos's requirement. The change from cardinal to ordinal utility
involved a large decrease in content. It wasn't ad hoc in the third
Lakatosian sense, but both Popper's and Lakatos's general account of
ad hocness would describe it as an ad hoc theory modification, even
though most economists regard it as an important piece of progress in
economics.
Hands: If you have a case where more than one type of ad hocness
surfaces at the same time, Lakatos is unclear as to which ought to be
addressed first.
de Marchi paper
Summary remarks
Why is Popper so highly regarded, at least among some methodologi
cally inclined economists, whereas most economists in their practice
seem to pay him little heed? There is evidence, in the recent book by
Paul Levinson [1982] , that a similar situation exists in philosophy. Pop
per's following is devout but quite small. On the other hand, there is also
a hint that practicing scientists find, or have found, Popper's views help
ful. Sir John Eccles, for instance, recounts in the Levinson volume that
Popper significantly shaped his own practice in two respects. Firstly, he
found it helpful to be encouraged to express his working hypothesis as
sharply as possible. This forced him to identify both what it ruled out
and how it differed from rivals. Secondly, Popper's emphasis on
falsifiability freed him from a felt obligation to speak only the Truth.
It is perhaps not surprising that scientists at this level of generality find
Popper's views exciting and directly applicable. I suspect that a survey of
practicing economists would yield similar responses. Indeed, some of
the young economists at the LSE in the late 1950s seem to have found in
Popper mainly a reinforcement of their desire to remake economics as a
rigorous, empirical science. But take Popper's philosophy of science as
Discussion
29
Discussion
Blaug: Did Lipsey and Archibald give up on Popper? That seems to be
hinted at in what you say, but it is surely false in the light of successive
editions of Lipsey's Introduction to Positive Economics. Lipsey became
a sophisticated falsificationist, but he always asks after he's presented a
theory if there's any good empirical evidence for this theory. The book is
entirely in a Popperian strain. Furthermore Lipsey has been working
more recently on the problem of world inflation, and in a recent paper
[1981 ] , where he lays out his research program, his emphasis is on set
ting down sharp predictions that will discriminate between a monetarist
30
Discussion
Discussion
31
32
Discussion
Discussion
33
de Marchi: But they were concerned about what they took to be Rob
bins's antiempiricism and by his perceived use of nonquantitative theory
to justify free market policies. Whatever the rights of the matter, Rob
bins was taken to be standing in the way of remaking economics as a
quantitative and essentially empirical science, and he became the target
to be attacked .
Klant: The mystery is, what was Popper thinking about economics? Did
the LSE economists ever have contact with Popper?
de Marchi: I agree that Popper's views on economics are mysterious. In
The Poverty of Historicism [p. 143], he made an exception of economics
because of parameter inconstancy and measurement problems- in the
following way. He removed - at least implicitly- the requirement of test
ability from falsifiability for economics. This looks as if he suspended the
rules of good science for economists without, however, abandoning the
demarcation criterion . Where that leaves economics is not clear, and
Popper doesn't clarify it any further for us.
As to contact between the LSE economists and Popper, there was
none, because they smoked and he had an aversion to smoke.
Hutchison paper
Summary remarks
I thought I would offer a few supplementary remarks rather than summa
rize the paper, remarks arising partly out of what's been said at the
sessions so far.
Over the years I've come to reject a good deal of Popper . . . not as
much as Daniel Hausman [laughter] . Popper was not all that interested
in economics. He didn't know much about it and it was indeed a brief
encounter, as Neil described it, that he had with economics. I'd only add
that Lakatos, similarly, was mainly interested in physics and mathemat
ics.
Popper's Poverty of Historicism I'd largely reject now; in particular,
his emphasis, which I at one time accepted, but feel much more reserved
about now, on the close resemblances between the natural and social
sciences. I think that on this a balanced view is necessary. One must
recognize the historical dimension in economics, which Popper does, but
doesn't really take into account.
I'd also reject, with Hausman, Popper's exclusively anticonfirmation
ist line. I think this is an attitude of somebody who takes physics as the
34
Discussion
Discussion
35
Discussion
Klamer: There is a great temptation to think in dualities . . . to divide
the world in twos, into black and white; subjective and objective; irratio
nal, rational; nonscience, science; values and fact. This dual thinking
concurs with binary logic and is very tempting. However, I suggest that
it is misleading and that thinking in threesomes may make more sense
when we come to try and understand what economists do.
Economists as scientists strive to know the truth about economic rela
tionships. We seek objective knowledge in the sense of a correspon
dence with the way things are. In principle, such knowledge should be
independent of our intentions and our interests. In this striving, then, we
turn against those who let intention and interest prevail. This striving is
expressed in the discourse in which we partake. It is expressed in the talk
about statements, the logic of statements, and the role of empirical tests.
It all serves the purpose of giving us certainty, or at least something to
hold on to. Logic and fact supersede intention and interest. But we've
heard many times today that the authority of logic and fact is highly
questionable. How do we react to that? One reaction is to think that if
economic knowledge is not objective, it must be subjective. It must all
be a matter of taste, class interest, ideology, or whatever. Anything
goes, apparently. These, however, are not the only two alternatives.
Both I and McCloskey are keen to suggest that this either-or thinking is
misleading.
There is a third possibility: We may talk about what economists do
in a way that allows a role for intentions and interests, as well as for the
constraints under which they operate. This may mean that we become
more sensitive to the intersubjectivity of what we say, and the context
of conversations and the discipline of these conversations. This implies
that not anything goes - there is a discipline; there are constraints on
conversations.
Professor Hutchison's writing seems to fit more into the first way of
talking- the "objective." But this paper actually brings him somewhat
closer to my suggested third alternative. He does not maintain that it is
logic and fact that rule our world. He does not grant explicitly that inten
tion and interest are involved, but he does admit that any significant case
for adopting or striving after a methodological principle must rest ulti
mately on some ethical, moral, or political argument. Both Hausman and
Hands focus on logic and fact, and to them Hutchison provides a sort of
counterweight. He stresses the desire of economists to go beyond form
and [theoretical] structure and to intervene. Many even seem to have an
36
Discussion
axe to grind. This opens the way to take seriously intentions and interests
and to study those things directly. That means, however, that rather than
judging the desire to be involved . . . to intervene, we should look at
motives very carefully. We should behave more like anthropologists,
trying to understand rather than to render judgments.
Discussion
37
Summary remarks
My paper is a case study of the methodological views of one of the most
famous of all living economists, Sir John Hicks. Since he rejects Pop-
38
Discussion
Discussion
39
Discussion
Cramer: Let me complement Blaug's paper with a few remarks about
this - may I say - desperate longing for empirical tests or falsifications of
economic theory. I should like to stress that this business of testing or
falsifying is a highly intricate, technical matter. It is very difficult to
decide here, on the basis of some examples, whether economic theories
are testable or not. I think this is insufficiently realized in the requests
we have heard for a simple - or any- demarcation criterion. Nor do I
think that it is altogether fortunate that we refer to "the barbarians at
the gates. " I do not think that macroeconomic theories, for instance, are
in the same category as Marxism or psychoanalysis, even though they
are difficult to falsify.
I want to illustrate my point with an example: the duration of unem
ployment. It is common knowledge that if you take a sample of people
who are unemployed, you will find that those who have been unem
ployed for a very long time stand little chance of finding a job. The
chance actually diminishes with the duration of unemployment. Why is
this so? One hypothesis is that unemployment itself is the reason; that
unemployment has a debilitating effect and that to be unemployed for a
very long time renders one unfit for the labor market. Thus, for the
individual, unemployment is the cause of further unemployment. An
alternative view is that unemployment is not the cause, but that it is a
symptom of ineptitude lodged in the individual ; that he is simply an
unsuitable individual from the point of view of the labor market. Such
people are, of course, overrepresented in a sample of people who have
been unemployed for a long time.
40
Discussion
Discussion
41
and not criticize. Yet he is very critical without having critical principles
to put in place of Popper's.
Summary remarks
My paper is a fairly detailed discussion of two emerging econometric
concerns in the 1920s and 1930s: finding an empirical model and the
notion of what constituted a "satisfactory" model.
42
Discussion
Discussion
Driehuis: I find myself comfortable with Mary Morgan's view that
econometricians have been concerned primarily with translating eco
nomic theories into empirical or measurable models, and not with trying
to prove theories incorrect or correct. Econometricians make economic
theories concrete. They try to operationalize general ideas for a specific
period and for a specific political, social, and technical context. The aim
of the econometrician is modest: [to answer the question] "How well
does the empirical model work with reference to the data?"
The central issue in Mary's paper is whether satisfactory empirical
models can be found. A crucial issue, of course, is what is meant by
"satisfactory. " In Section 2 of her paper, it is shown that early econome-
Discussion
43
tricians had clear ideas about this. First of all, there were requirements
for the individual model equations in terms of coefficients, their signs
and stability and statistical significance. These requirements can be
called "statistical requirements. " Secondly, there were requirements re
sulting from the purpose of the model. A forecasting model must yield
satisfactory forecasts; a policy model should give satisfactory informa
tion on alternative policy options; and so on.
These latter two sorts of requirements relate to macro-models. It
seems to me that we need little discussion on the statistical criteria; they
are well understood. It's also not difficult to decide on what a satisfac
tory forecast is. But much more discussion is possible - and maybe
needed - on whether a model is satisfactory from an economic-analytical
point of view. What are satisfactory effects of alternative policy simula
tions? Since these effects, by definition, are only dependent on the
structure of the model, the question becomes, What is a satisfactory
structure? The answer clearly is not that structure which fits the data
best. It is, in principle, possible that more than one model fits the data
very well, so it is not possible to prove in that way which model is to be
preferred.
I conclude, therefore , that a satisfactory model structure is one consid
ered most plausible by its designer, using "plausible" in Klant's sense.
So there is not one, but many, "satisfactory" model structures. This
situation is, as we know, considered disappointing by journalists, politi
cians, and others who accuse macroeconomists of quarreling all the
time, but it seems inevitable.
From a scientific point of view, what is important is that researchers
follow the rules of the game; that is, that they make clear what data they
use, how they estimate, why they choose what they choose, so that they
can constantly check each other's work. For me this is the enduring
message of Klant's book.
44
Discussion
tivity. A fourth interpretation: If you have markup pricing, this will also
lead to a significant coefficient.
Thus, because here we are not testing the basic theory of optimiza
tion, of production functions with substitution, etc . , we must a priori
select one interpretation. Now econometricians are modest. They say,
"Well, look, I haven't invented the background theory; all I am doing is
fitting it, and this is what emerges. " This opens the way to arbitrary or
ideological influences in the selection among interpretations, and if we
are not careful, we miss this and the discussion about interpretations
which should occur does not.
Discussion
45
about testing, they are not really testing in the sense of comparing
serious alternative theories. Even in Neyman-Pearson one has only a
straw alternative. They are "testing" in order to get a better and better
version of what they have, assuming that the underlying theory is right.
Summary remarks
My paper is an adumbration of a theme implicit in Professor Pen's last
comment: In the future, only econometricians will have the right to be
intelligent! My job is to begin where Popper might have taken us, but
did not. The argument follows very naturally on Neil de Marchi's paper
and on Mary Morgan's paper.
On the other hand, I disagree entirely with the view that Mark Blaug
just expressed, that there is a division between theory and testing or
empirical work; that there are theorists here and empirical workers
there, and that it is somebody's task to bring them together. I'd argue
that there is only one task, and that is to make sense of what's in the
middle. The paper tries to do something like that.
I take a series of three studies and try to show how theory and evi
dence intertwine, in ways that are associated in a Lakatosian research
program and are associated with Lakatosian progress. I used a Lakato
sian framework because it limited me to asking only a certain number of
questions, questions which interested me.
The sequence I examine is a series of papers in microeconomics.
These embody three theories in the Lakatosian belt of what I call the
"neo-Walrasian program. " Mr is a neoclassical theory of household be
havior; Mr+ i is Becker's theory of the household; and Mn 2 is a Nash
bargained theory of the household.
What is this Nash-bargained theory about? You can set up utility
46
Discussion
functions for a man and a woman - marriage partners. Each utility func
tion has as arguments some that are geared to the individual, to the
person, and some that are held in common. There are constraints for
each individual: income that is common and income that is specific.
These functions can be set up and one can work out an interdependent
optimization problem- a bargaining problem, in the sense of John Nash.
Using the Nash solution to the special bargaining problem, which in
volves the choice of a status quo point, one can come up with a solution
which is a division of utilities, and come up with a generalized demand
system which involves, in an intrinsic way, the interrelationship of two
individuals who happen to be married to one another and who in that
relationship hold certain kinds of income in common.
That's a theorist's problem. But what happens when you ask, exactly
how does this theory work? What data would be used to support it? How
could the theory be developed further? I describe in the paper how the
move from Mn1 to Mn 2 was accomplished. The creators of Mn 2 con
cluded that the Nash model provided, for the first time, an analytical
framework for examining the effect of the marriage tax on the joint labor
supply of husbands and wives. Her current married after-tax marginal
wage rate is an argument in the full family income constraint, whereas her
hypothetical unmarried after-tax marginal wage rate is an argument of
her threat point. This is just one of a whole class of applications where,
due to taxes and transfers, the prices and nonwage incomes faced by an
individual differ according to his or her marital status.
Thus we have a specific implication associated with the Nash model
which was not present either in the Becker household model or the
standard neoclassical model of two individual decision makers. In the
second paper, a data set was painstakingly developed to appraise the
excess content of the Nash demand system. This was a nontrivial part of
the overall task. It required much time (two years) and a great deal of
ingenious analysis. It is not sufficient to state, as some falsificationists
do, that in good science facts falsify theories and that since economists
don't look for falsifying facts, they are not doing good science. Getting
the "facts" is not trivial.
In the third paper, the bargaining model is applied to an entirely new
problem, that of household membership. Specifically, where some ear
lier work treated household membership as an exogenous dummy vari
able in studies of youth earnings, the third paper examines joint deci
sions on wages, consumption, and household membership.
What does all this mean? Whatever one believes about the value of
research programs, the methodological literature leaves one with the
impression that empirical economics is unimportant to the economics
enterprise. Yet most of Kuhnian normal economic science is incoherent
Discussion
47
Discussion
Birner: If this series of papers that Professor Weintraub discusses is part
of the same (neo-Walrasian) research program, and if they are indeed
empirically progressive, they provide one instance of an application of
general equilibrium theory's [GET] being progressive. I agree with Wein
traub that many more case studies of empirical work are needed in order
to appraise GET as a whole. But preliminary to this is an examination of
the nature of the neo-Walrasian program and the manner in which it is
supposed to operate. In my comment I shall concentrate on two items:
1.
2.
1.
2.
3.
Discussion
48
These are strong claims. I shall argue that all three claims cannot be
maintained consistently.
If one maintains that a program develops (progressively) by interpreta
tive shifts in the hard-core propositions, the question naturally arises of
whether the resulting theories are still part of the same program. Lakatos
would deny this, condemning the reinterpretation as being ad hoc; the
program would lack continuity, as it does not stick to its hard core. But
perhaps we may be less strict than Lakatos. In Becker's case, where
"agent" was redefined from "person" to "household," my intuition would
still allow me to say that the resulting theory was part of the same research
program. But replacing HC3, independent choice, by HC3 ' , interdepen
dent choice, and calling this a "reinterpretation" surely stretches the
meaning of that word. If in the development of a research program a hard
core proposition is replaced by its denial, we must conclude that we have
left the original program behind and are dealing with a new one.
Weintraub's very strong claim that the progressive shift (or, more
neutrally, the development) he describes could only have been made by
a neo-Walrasian general equilibrium theorist, and not by someone using
a Marshallian partial equilibrium approach, presupposes that the neo
Walrasian program exerts a very strong directing influence. (At the very
least, it presupposes that it provides constraints.) And this is indeed
what he claims for it; the positive heuristic of the program "forces atten
tion to the interdependent optimizing that occurs in households, and
predisposes the analyst to consider adopting a game theoretic model"
(p. 223 4; italics added) . But it does so by altering the content of HC3.
This raises severe problems.
Firstly, if the positive heuristic is derived from the hard-core proposi
tions, as Weintraub's examples suggest, how can it give directions for
altering the hard-core propositions themselves?
Secondly, I do not see the difference between taking the leap toward
an interpretative shift in one of the hard-core propositions and "a judi
cious use of the ceteris paribus clause" (p. 224). Indeed, there is nothing
in a partial equilibrium framework that prevents one from coming up
with an Mr+2-type bargaining model.
Thirdly, lacking any rules guiding interpretative shifts, I think Wein
traub's use of the idea that hard-core propositions are partially inter
preted makes him vulnerable to Alexander Rosenberg's criticism: "If the
partial interpretation claim is correct, any set of non-self-contradictory
propositions can become a perfectly adequate theory about any sort of
phenomenon antecedently described. All that is required is the appropri
ate correspondence [ = interpretation) rules" [1984).
To sum up, I do not think that Professor Weintraub has proved his
-
Discussion
49
50
Discussion
Birner: The fact that Chicago economists did not in fact agree with the
McElroy approach shows only that they did not act along those lines. It
does not show that they could not have developed their theory in that
direction.
Hands: I would like to talk about the possibility of violating some hard
core proposition in negotiating the transition from MT to MT+ i to MT+2
My question is not about interdependency- I think that's an unhelpful
pursuit - but about independent agents in the Becker model. There you
have a household production function and utility function. I find the
idea of household maximizing to be quite non-neo-Walrasian. You
might argue therefore that the second stage of theory development here
was not in the neo-Walrasian hard core. I'd answer my own question by
saying that consistency with the hard core is perhaps a matter of degree
there are degrees of ad hocness. A way to think about Becker's contribu
tion, therefore, might be to think of it as a gain in empirical content at
the cost of heuristic progress. The third stage, however- the game theo
retic approach - seems to involve both empirical and heuristic content.
There may be trade-offs of this sort.
Weintraub: I don't disagree with this. The Becker move involves a re
interpretation of "agent" as "household. " I'm not wholly comfortable
with appraising this move within the neo-Walrasian tradition, though
the next move can, I think, be appraised within that tradition. It could
not have been made, however, by Becker for very specific reasons.
Klamer: It is once more clear that there is a difference between a
spoken and a written presentation of an argument. The argument as
written [by Weintraub] is not entirely convincing. It depends critically
on the McElroy shift being within the neo-Walrasian program. In his
informal comments, however, Weintraub adds factual information
about how Chicago economists preferred to operate.
Nooteboom: Does not the whole debate turn on whether the word "in
dependent" belongs to the core? If so, this merely tells us that cases and
programs are not sharp-edged.
Weintraub: I like that. I have found, in trying to write down what I
mean by the hard core of the neo-Walrasian program, that the same
general propositions have satisfied me over some years and through
several attempts. But words like "independent" and "interdependent" I don't want these to be fixed. I don't regard my reconstruction in this
paper as fixed and definitive. As to the heuristics, these are capable of
Discussion
51
variation and extension almost indefinitely. They are meant to make the
conversation more productive.
Birner: I don't think we are having a quarrel about words only. The
claim is that the neo-Walrasian program exerts a strong, definite influ
ence on further research. The whole point of the hard core in a
Lakatosian program is that certain propositions have a privileged status;
but now we hear that they don't. So what's the point of using the
Lakatosian framework?
Weintraub: There's a marked difference between the hard core and
what's in the belt. The core is regulative; nonetheless, if one of our
axioms involves, say, something like the concept of a line, then in differ
ent geometries that will have different interpretations. There are some
things that a neo-Walrasian will not give up; that's a way to get at the
hard-core propositions.
Birner: But if we have the freedom to reinterpret hard-core proposi
tions, what really drives the program? It would be nice if you could
make a prediction as to what the next move would be in the program.
Weintraub: I would make the next move if I could make the prediction!
Caldwell paper
Summary remarks
My paper is concerned with the problem, How should we do methodol
ogy?
What is the purpose of methodological work? Answers I've run across
include the following:
1.
2.
3.
I have long felt that the philosophical literature did not argue for any
of these three alternatives very adequately. Indeed, I've come to be
more interested in problems that I've run into in looking at economic
methodology than in philosophical questions as such. Let me mention
four such problems.
52
Discussion
1.
2.
3.
4.
Now both B laug and Klant have a certain coherence in their positions,
imparted by their acceptance of some sophisticated variant of falsifica
tionism. I have gone a different route - a pluralistic route.
To come back to the question of how and why we should do methodol
ogy, one purpose is to better understand economic science (and, it is to
be hoped, improve it). We should also be evaluating research programs
to identify their strengths and weaknesses. Then, too, it seems appropri
ate to emphasize novelty.
This may not sound like a very tight set of goals. It is not. But it
contrasts sharply with some alternatives that I think we need to be rid of.
Firstly, there are no a priori grounds for excluding anything. I make no
attempt to solve the demarcation problem. Science is too rich and varied
to separate it from nonscience very successfully. We have not succeeded
up to this point. [Secondly,] but worse, trying to solve this problem has
created many other problems in methodology today. For the attempt
sets methodologists up as experts, arbiters. It is one reason for the gap
between scientists and methodologists - a gap that emerges clearly in
Mary Morgan's paper. Thirdly, demarcation criteria often seem to be
used dogmatically to exclude unwanted approaches.
Now an obvious response is, "Well, doesn't your position lead to anar
chy?" I defend myself with three arguments. Firstly, this fear of anarchy
misunderstands science as it is currently practiced. Science tends more
toward conservatism and dogmatism than toward anarchy. Secondly, it
misunderstands the role of methodology. It is not the role of the
methodologist to keep science "free. " Citizens in general must oppose
restrictions on discourse. Thirdly, it ignores that criticism is very impor
tant in my own position. Only via criticism are the strengths and weak
nesses of research programs brought to light. And the criticism may be
from a multiplicity of standpoints. I give more detail in the paper.
But I suspect the more deep-seated objection is this: Can I, by mere
criticism, be confident that I will be led to true theories? No. I can't
establish that being open but critical will guarantee truth. I do claim,
Discussion
53
Discussion
Hamminga: Dr. Caldwell defends his pluralism in a pluralistic way. As
an economic methodologist, I do not think there is a professional reac
tion to pluralism.
Let me, then, try to supply an economist's reaction to pluralism. The
pluralist seems to be considerably more tolerant toward what I am doing
than the referees of journals and reviews whom I want to accept my
drafts (and he seems to be so on purpose). To me, as an economist,
reviews and j ournals are like huge dragons, and I want them to swallow
the food I prepare for them. Consequently, I would accept the assistance
of anyone who has clear-cut empirical information concerning the chemi
cal structure of the food the dragons refuse and of the chemical structure
of the food the dragons swallow. That is, I would certainly be willing to
listen to the results of those descriptive studies about science that are
welcomed by the pluralist.
The pluralist, however, calls these studies "nonmethodological. " That
means, I conjecture, that he does not consider them to be his own
specific affair. What, then, is the specific affair of the pluralist, apart
from welcoming all the different ideas and all the different ways to
criticize them? All in all, as an economist, I would not disagree with the
pluralist, but I do not see how the assistance of a pluralist expert in
argumentation could improve my performance in making these dragons
swallow my food. And that is what I want.
Now I could have done wrong in dressing as an economist. Pluralism
could be meant as assistance not to the individual economist, but to the
dragons: what they ought to refuse and what they ought to swallow.
Pluralism may be a recommendation for more variety in the dragons'
menu. Dressed as a dragon, I would answer pluralism as follows: My
stomach is of limited size. If I start accepting some kinds of papers that I
used to refuse , this implies that I have to refuse papers that I used to
accept. If I would follow the pluralist's recommendation to enlarge my
bouquet of standards for criticism, I thereby stop revealing stable and
predictable digestion behavior to those who present food to me. As a
result, submitting economists will soon start to look for other dragons to
feed. Whimsical evaluation behavior repels economists, especially those
who make the kind of food that they know to be desired by many
dragons. I would become ill-fed.
As you see, it is as if an invisible hand makes dragons and economists,
54
Discussion
Discussion
55
Klamer: I would like to suggest that rather than talk about pluralism,
we talk about argumentative strategies - theoretical, empirical, histori
cal. This does not imply any particular epistemological position.
Hutchison: About fifty years ago, I wanted a distinction between sci
ence and nonscience. I wouldn't use these words now, but distinction is
vital - as to how far a particular theory has been tested, and so on, as I
have said before. We don't necessarily want those who "fail" in some
sense to be cast into outer darkness, but we want the distinction.
Again, Bruce Caldwell went on about criticism, criticism, and Daniel
Hausman has urged empirical testing. But that's just a kind of falsifica
tionism; without some such principle, criticism is merely a matter of
likes and dislikes. Otherwise talk of criticism is empty.
Blaug: If you really take methodological pluralism seriously - that there
are no meta-methodological principles on the basis of which one might
compare different programs - then programs can only be criticized from
within. But if you take this position, why should logical consistency be a
desirable attribute? Such consistency is restricting; it creates anomalies
in the light of empirical evidence. Such a criticism must come from some
higher-order standard. Methodological pluralism is just a copout.
Birner: I've always considered pluralism to be harmless and uninterest
ing. If you want a true theory, you look for falsifiable statements. If you
want to be able to apply a theory, you may be happy to apply instrumen
talism. Typically, however, we look for theories which are true, applica
ble, and perhaps satisfy other criteria as well. How could Caldwell make
the trade-off?
Klant: I have the feeling that by "pluralism" Caldwell has in mind a
uniform theory, in the using of which, however, various methods can be
brought to bear. That is not methodology, but method.
As to the demarcation line, I would argue that economics is on both
sides of Popper's line. I am not interested in where economics falls along
this particular line. He does not describe economics. I claim that in
economics we are trying to follow plausibility strategies. And I agree
with Hamminga that we should find out exactly what strategies are being
followed.
I agree with Caldwell insofar as I think that there are many strategies,
some more, some less empirical. The only thing methodological about
this is the recognition that there are various plausibility strategies.
56
Discussion
Summary remarks
I think I should start by reminding you of George Stigler's conference
handbook, and in particular his comment that in assessing the work
under discussion it is always good form to say, "I can be very sympa
thetic with the young author; up until a few years ago, I was thinking
along similar lines" [1977, p. 442] . Up until a few years ago, I was a
logical empiricist. You don't see any former methodological anarchists
who have converted to logical empiricism, but you do see a great num
ber of the other kind. This is an argument for a move away from logical
empiricism. I claim that philosophical criticism of science looks thin
beside what actually happens in scientific conversation. Philosophy
might have proven to have been helpful in understanding science, but it
hasn't worked out that way. As Roy Weintraub says in his recent Gen
eral Equilibrium Analysis: Studies in Appraisal, a use of the Lakatosian
framework raises problems that rest uneasily on a Lakatosian bed. But
even if, with certain trimming of the feet, a piece of economics could be
made to fit easily in the bed, I ask, "So what?" What have we gained by
forcing science into a Popperian or Lakatosian mold?
I see evidence that philosophy, which has urged economists to impov
erish the economic argument, has not been of much help. We might use
philosophy to judge progress in economics, but I don't think Lakatos
claims that his framework is an accurate guide to history. Again, one
might think that economic theory leads to truth. But I don't think this
works either. If the philosophical criticism of science is not about what
we know, what is it about?
I think it is about morality. An ethical conversation cannot be taken
seriously, however, when the conversationalists themselves don't know
that they are dealing with ethical questions. To the extent that they
understand but are not prepared to defend their conversations as ethi
cal, the same holds. They are unprofessional conversations. We are
invited to believe that certain conversations have political effects- the
barbarians again - but these effects are undemonstrated.
If the philosophers of science will reason together with us, they will
find that the values they want to protect are the same as those Klamer
Discussion
57
and I want to protect. The sprachethik is the same for us all: Don't
shout; pay attention; don't lie; don't use violence. A good scientific
society is the ideal speech situation. The logical empiricists and the
rhetoricians both have in mind the preservation of the same values, and
these are the values of good conversation. It is a conversational model of
science.
This raises the question of standards. But we know when the conversa
tion in our own field is going well. Conversations overlap and the
sprachethik is transmitted through the scientific society in this way. A
value in a good conversation, for example, is that we are careful to read
what has gone before. Not only would the ideal methodology have this
character of studying conversations, but that is what historians of eco
nomic thought already actually do.
The close study of conversation is a natural - indeed, empirically
progressive - research program in methodology and history of thought.
The only close competitor is a fine program in the sociology of science
the Edinburgh or Strong Program. What is unique about that program,
and what is interesting about history of science generally, is its rhetorical
dimension. The sociology of science in fact reduces to a rhetorical study:
how arguments of all kinds are actually made. But with an explicit use of
the long tradition of rhetoric, you can go much further than conventional
sociology of science or history of thought. You can analyze the audience
for economics, for example, and show how the texts of economics make
appeals to certain audiences. You can show how genres such as the scien
tific paper developed. You can show how metaphors, theorems, and
statistical appeals to authority are employed. In other words, the conver
sational approach to economics has virtue for both the positive and the
normative parts of methodology and for the history of economic thought.
Discussion
Pen: The organizers of this conference thought of me, apparently, as
one who would be able to characterize . . . shall we call it the "thick"
school [laughter] . . . and discuss it in an objective way. Until recently,
that would have been correct. I am a structuralist and an economic
journalist, but I've always tried to keep the two separate. But this clear
distinction started to crumble some years ago, a critical point being my
reading of Don McCloskey's article in the Journal of Economic Litera
ture. I'm still not a certified member . . .
McCloskey: Certifiable, perhaps [laughter]
58
Discussion
Summary remarks
My paper presents a philosophical argument in favor of a practice differ
ent from the one normally adopted with respect to understanding what
economists do.
I am interested in the problems experienced by novices in economics.
I maintain four things. Firstly, it is by paying attention to the problem
of communication, and not to the problem of knowledge, that we will
make economic conversations once more interesting. The problem of
knowledge is concerned with how we manage the relationship between
our thoughts and objects. This is the relationship between Popper's first
and third worlds. We don't get very far by proceeding this way. Proceed
ing analytically, we ignore the sender, the audience, and most of the
signals, and concentrate on a series of statements. That is how we sim
plify the communication process. Form and the structure of statements
is all we see handled by methodologists. Thereby we ignore the ethos of
the speaker- can we trust him? What is his reputation? - and the interpre
tation of signals at the other end. Economists get their message across by
Discussion
59
means that go far beyond mere statements. I submit that we will learn
much more about what is going on if we pay attention to discursive
practice.
Point number two. If this is true, we must become like anthropolo
gists, studying and interpreting economists as they communicate.
Thirdly, to think of economics as discourse is better than thinking of it
as rhetoric. It helps us to think beyond an individual facing others and
trying to persuade them. Speech communities, as distinct from the 1them model, are more complex and richer.
Fourthly, a focus on problems of communication makes sense only if
we recognize differences. The prevalent emphasis on form, structure,
criteria of justification stresses identity, consensus, and the basis of
agreement. But there are many conversations, different ways of world
making. We can reach an understanding of similarities when we have
recognized differences. But seeking identity will not help us.
Discussion
Weintraub: I would have liked to open up an additional line of communi
cation by dancing. I can't dance. What I want to do, then, is to describe
a ballet- one of my own invention. The prima ballerina, l'economie, is
fought over by the black prince, Le Popper, and the white prince, Le
Rorty, or their armies, Les Positivistes et Les Pragmatistes. The dances
of the former are heavy and plodding, martial in step. The latter's
dances are opportunistic, free. In the end, Les Positivistes die of self
infticted exhaustion, and sprightly Rorty carries off l'economie.
Short of being able to dance this- all parts - myself, I'd like to extend
an invitation to Don McCloskey and Arjo Klamer to have a conversa
tion, and not just to talk about conversations [applause] . My recent
book on general equilibrium theory is an extended conversation with
Mark Blaug (the reader over my shoulder) . I used many styles and
contexts to come at points in different ways. We need more studies, of
the sort that Klamer gave us of the New Classical discourse, in his
unpublished dissertation [1980] , and of the sort that McCloskey gives us
of Fogel and of Purchasing Power Parity, in his new book [1985, chs. 7,
9] . We can do with less talk. As Eliza Doolittle said:
Words, words, words ;
I'm so sick of words.
I hear words all day through
first from him now from you.
Is that all you blighters can do?
[Laughter. Hear, hear.]
60
Discussion
Discussion
61
sort, gaining demarcation order yet not being able to define a demarca
tion line?
Blaug: Of all the different ways one can analyze what economists do,
why should one choose to invest exclusively in rhetorical analysis? I
think there are many more interesting ways to look at what is going on in
the history of economics. I'm not unsympathetic, however.
McCloskey: I understand that you are not unsympathetic, and I have no
answer in the abstract. It would be absurd for me to stand up here and say
that I have a philosophical theory that rhetorical analysis is the way to do
history of economic thought. That would be self-contradictory. So I ac
cept that I have to show you. I have analyzed particular papers, con
structed partial grammars of economics, made this an empirical question.
Caldwell: Both novelty and criticism are important to me. I welcome
your program because it is novel, and when it has a bit more content, I'll
be happy to criticize it.
McCloskey: That's fair comment.
Klamer: In reaction to Mark Blaug, I am working on discourse analysis,
and part of that deals with what he would call the "sociology of econom
ics." That's essential if you want to understand the audience. If the
economics audience prefers formal over historical arguments, this can
be understood in part by looking at how the discipline is organized. I do
not exclude that.
References
Archibald, G.C. ( 1965). "The Qualitative Content of Maximizing Models,"
Journal of Political Economy; reprinted in M. Morishima et al. (1973),
Theory of Demand. Oxford: Oxford University Press.
Hamminga, B . (1982) . "Neoclassical Theory Structure and Theory Develop
ment," in W. Stegmi.iller, W. Balzer, and W. Spohn, eds. Philosophy of
Economics. Berlin: Springer-Verlag.
Hicks, J. (1969). A Theory of Economic History. Oxford: Clarendon Press.
(1979). Causality in Economics. Oxford: Basil Blackwell.
(1983). "A Discipline not a Science ," in Hicks, Classics and Moderns. Col
lected Essays on Economic Theory, Vol. III. Oxford: Basil Blackwell.
Klamer, A. (1980). "A Methodological Appraisal of Rational Expectations Eco
nomics. " Ph.D. dissertation, Duke University. Mimeo.
(1984). Conversations with Economists. Totowa, N.J . : Rowman and Al
lenheld.
62
Discussion
Klant, J.J. (1984). The Rules of the Game. The Logical Structure of Economic
Theories. Cambridge: Cambridge University Press.
Levinson , P. , ed. (1982) . In Pursuit of Truth. Essays on the Philosophy of Karl
Popper on the Occasion of his 80th Birthday. Atlantic Highlands, N.J.:
Humanities Press.
Lipsey , R.G. (1981). "The Understanding and Control of Inflation: Is There a
Crisis in Macro-Economics?" Canadian Journal of Economics 14: , 545-76.
McCloskey, D.N. (1985). The Rhetoric of Economics. Madison: University of
Wisconsin Press.
Popper, K. (1957). The Poverty of Historicism. London: Routledge & Kegan
Paul.
Rosenberg, A. (1984). "Partial Interpretation in Microeconomics," in W.
Leinfellner and E. Kohler, eds., Developments in the Methodology of Social
Science. Dordrecht: Reidel.
Stigler, G.J. (1977). "The Conference Handbook," Journal of Political Econ
omy 85:441-43.
PART I
C H A PT E R 1
Many of the points made in this chapter are not new. The critique of Popper's views
on induction follows, to a considerable extent, criticisms made by Levison (1974),
Lieberson (1982a, 1982b), Putnam (1974), Grunbaum (1976), and Salmon (1981),
although only Lieberson was a major influence. Isaac Levi's work (1967, 1980) has also
influenced me, and in my argument for the possibility of verifications, I was assisted by
Nisbett and Thagard's ( 1982) discussion of induction. Lakatos's critique and his plea
for a "whiff of inductivism" is much weaker, since his own treatment of induction
differs so little from Popper's (Lakatos, 1974, pp. 241-73). David Miller's defense of
Popper's account of induction (1982) does not touch the simple and, in my view,
decisive arguments given in this chapter. I would like to thank Ernie Alleva, Clark
Glymour, Wade Hands, Kevin Kelly, Jonathan Lieberson, Michael McPherson, Jona
than Pressler, Teddy Seidenfeld, and Wilfrid Sieg for comments on earlier versions of
this chapter.
65
66
Hausman
1.
Logical falsifiability
67
separation of what he calls the "logical" from the "pragmatic" problem of induction.
Although I do not know precisely what the identity criteria for problems are, it seems to
me that there is basically only one problem here. Thus Popper manages to be extremely
misleading when he makes the entirely correct (within his set of definitions) claim, "For
anybody who adopts an instrumentalist view, the (logical) problem of induction [as
formulated by Popper] disappears" (1983, p. 1 17). But, as Popper himself notes (1983,
p. 120), the point is an entirely trivial one, since the instrumentalist faces an entirely
analogous problem of deciding which statements to rely on. The point would hardly be
worth making but for the fact that Popper's formulation has perhaps misled Lawrence
Boland, who argues mistakenly that instrumentalism is a response to the problem of
induction (1979, p. 508).
68
Hausman
argument" one means "valid deductive argument" (and I shall not dis
cuss any alternative interpretations here) , than there are no good argu
ments with only basic statements as premises and universal statements as
conclusions. But one can provide valid deductive arguments against uni
versal statements. One can thus (albeit fallibly, since basic statements
are not themselves infallible) find out that theories are wrong. By the
elimination of error our knowledge can, in this Pickwickian sense, grow.
Notice that even if Popper's insistence that scientific theories be logi
cally falsifiable were not plagued by the difficulties to be discussed soon,
this so-called solution to the problem of induction would be profoundly
unsatisfactory. Only the fallacy of elimination enables one to find theory
T meritorious merely because an alternative theory, T' , has been re
futed. The logical conclusion, although clothed in soothing words about
how rational the procedure is (for example, 1972, pp. 22, 27, 58, 81 , 95) ,
is a skepticism that is as extreme as Hume's, for Popper explicitly denies
that there is any room for argument in support of any theory or law. He
writes, for example, "that in spite of the 'rationality' of choosing the
best-tested theory as a basis of action, this choice is not 'rational' in the
sense that it is based upon good reasons for expecting that it will in
practice be a successful choice; there can be no good reasons in this
sense, and this is precisely Hume's result," (1972, p. 22) . We have no
better reason to expect that the predictions of well-tested theories will
be correct than to expect that completely untested theories will predict
correctly.
But this purported solution to the problem of induction falsely presup
poses that individual scientific statements, or at least individual scientific
theories, are falsifiable. Statistical and probabilistic claims are obviously
not logically falsifiable. Even flipping a million heads in a row does not
logically falsify the claim that a particular coin is unbiased. Even more
seriously, all claims that cannot be tested individually are not logically
falsifiable. For if a sentence is not, by itself, inconsistent with some finite
set of basic statements, then it is not logically falsifiable. And the fact is
that virtually no scientific claims of any interest are by themselves incon
sistent with basic statements. To falsify even an utterly simple scientific
claim, such as Galileo's law of falling bodies, requires not only basic
statements, but also some nonbasic statements concerning whether
nongravitational forces are present. If individual statements can be re
garded as scientific only if they are logically falsifiable, it will turn out
that almost all interesting science is not science after all.
Although Popper talks constantly of the falsifiability of scientific theo
ries, he nevertheless knows all this; and he has discussed at length not only
the role of background knowledge in testing, but also the sort of "conven-
69
Watkins (1984, p. 326) also notices Popper's blunder here. Popper says that the observa
tion report would falsify the theory if we ignore the possibility of immunizing strata
gems. But it is not just the observation report that falsifies the theory. We also need the
knowledge that the other explanations of the predictive failure are in fact merely immu
nizing stratagems.
6 Zahar's proposed phenomenological reconstrual of basic statements (1983, pp. 156-61)
and of logical falsifiability would only underscore how small a role logical falsification
plays in the process of theory revision. Popper disputes this claim, which was empha
sized by Duhem (1954, p. 187), by arguing that H2 might entail - P, whereas Hl entails
P when Hl and H2 are conjoined to a common set of additional premises, S. But even in
such a case, H2 does not, of course, follow deductively from our acceptance of P as a
true basic statement. We still need to decide (for whatever reasons) to regard the
members of S as true. See Grunbaum (1976, pp. 248-50) and footnote 15.
-
70
Hausman
"But what kind of clinical responses would refute to the satisfaction of the analyst not
merely a particular analytic diagnosis but psycho-analysis itself?" (1963, p. 38n)
71
the same goes for science as a whole. And I know of no economic theories
that cannot enter into logically falsifiable test systems (which would, of
course, include specifications of statistical techniques and ceteris paribus
clauses in addition to theories from economics and from other domains) .
Popper's claims about the asymmetry of logical falsifiability and logical
verifiability (of whole test systems) are as correct as they are irrelevant.
They provide no solution to the problem of induction.
2.
72
Hausman
excuse that can itself be readily tested. A scientist might, for example,
question whether a particular instrument was operating properly. There
is no simple algorithm stating when it is unreasonable to make further
reasonable excuses, but provided that one is seeking to falsify rather
than to hold on to theories, no algorithm is needed. s
3. If a theory has not yet been falsified, scientists should accept it
tentatively. Theories that have survived harsh testing are not better
supported or more worthy of belief than are theories that have not been
tested at all. They are merely particularly difficult to falsify and worth
testing further. "The theoretician's choice," in Popper's view, "is the
hypothesis most worthy of further critical discussion (rather than accep
tance)" (1963, p. 218n). Although Popper argues that there may be
evidence that one theory has more true and fewer false consequences
than another ( 1972, pp. 58, 81-2; 265), there is never, in his view, any
good evidence that a law or a theory is correct or even close to correct.
What we do-or should do - is to hold on, for the time being, to the most
improbable of the surviving theories or, more precisely, to the one that can be
most severely tested. We tentatively "accept" this theory-but only in the sense
that we select it as worthy to be subjected to further criticism, and to the severest
tests we can design.9 (1968, p. 419)
Are these good rules of procedure for scientists to follow? The answer
presumably depends on what the objectives of scientists should be. And
therein lies a long and tangled story, which we had better avoid here. In
order to keep this chapter short and focused on Popper's views, I shall
not argue that these are in fact bad rules. Instead I shall merely show
that Popper has not provided any good arguments in their defense.
Given their implausibility, this demonstration of their lack of justifica
tion should be a sufficient criticism.
One can, at least in outline, grant the first rule, that proposed scien8
This requirement, like the last, is not so much a requirement of each individual scientist
as it is of the institution of science as a whole ( 1969, p. 1 12; but see 1972, p. 266).
Provided that there is open and free communication, the institution as a whole may be
critical, even though each individual scientist attempts to protect his on her own theory
from criticism.
Popper chides Lakatos ( 1974, p. 1003) for, in effect, ignoring the sentence that follows:
"On the positive side, we may be entitled to add that the surviving theory is the best
theory- and the best tested theory - of which we know." But this last sentence changes
nothing concerning what Popper takes corroboration and "acceptance" of scientific
theories to be. Popper has, of course, a theory of verisimilitude that met with an
unfortunate formal demise (Tichy, 1974). But regardless of formal vicissitudes, Popper
has insisted that corroboration is no evidence of verisimilitude. In very special and
limited circumstances we can justifiably maintain that one theory has greater verisimili
tude than another, but there is never any justification for maintaining that a theory has
high verisimilitude (1972, pp. 47ff) .
73
tific theories be testable, that science seeks theories with much content
and theories that can be tested harshly and in many ways. But apart
from the important insistence on content, there is no news in any of
these methodological rules. Inductivists have been saying these things
since at least the seventeenth century (Grunbaum, 1976, pp. 217ff).
The second and third rules are less plausible. To claim that scientists
should seek only to falsify theories, never to support them, and that they
should never regard theories as anything more than conjectures that
may be worthy of criticism is to make two apparently outrageous sugges
tions. 10 Why should anyone accept them? Popper offers four reasons.
First, he maintains that confirming evidence is worthless, since "it is easy
to obtain confirmations, or verifications, for nearly every theory- if we
look for confirmations" (1963, p . 36). So it is a mistake to look for
confirmations or to regard what one finds as really supporting or estab
lishing any theory. But Popper is mistaken. Cheap confirmation is
readily available; merely locating another instance of a generalization
may be as easy as it is worthless. Good supporting evidence is, on the
other hand, hard to obtain. Much of it comes from just the sort of harsh
testing that Popper insists on (see Grunbaum, 1976, pp. 215-29) .
Second, Popper argues that to seek confirmation or to believe that
one has found quite a lot is to show a dogmatic attitude rather than the
sort of open and critical attitude shown by those who seek falsifications
(1963, pp. 49ff ) . But this is just name calling. Even someone devoted to
giving evidence in support of a favorite theory need not be credulous,
closed-minded, or dogmatic in doing so. And it is certainly not the case
that a person seeking the solution to a problem, who is thus concerned
both with confirming and with disconfirming evidence, automatically
qualifies as a dogmatic, unscientific sort.
Third, Popper suggests that to seek supporting evidence or to regard
scientific theories as sometimes well established rests on a view of scien
tific knowledge as infallible. Popper argues that the overthrow of Newto
nian physics renders such a view completely untenable. But one can, of
course, concede that human knowledge is fallible without denying that
particular knowledge claims may be well supported. 11
Popper's fourth argument for the injunction to seek to falsify theories
and never to seek to support them rests on the thesis that it is impossible
10 And indeed, Popper is unable to follow his own advice consistently, for he writes:
What we believe (rightly or wrongly) is not that Newton's theory of Einstein's theory
is true, but that they are good approximations to the truth, though capable of being
superseded by a better one.
But this belief, I assert, is rational. (1983, p. 57).
11 And Isaac Levi (1980) has even argued that one can regard one's knowledge as infalli
ble, without regarding it as as incorrigible.
74
Hausman
There are, however, some verbal difficulties here, since Popper offers an explicit analy
sis of what it is for evidence to support a theory (see esp. 1983, pp. 236ff). This is not a
contradiction, however, for in the Popperian sense in which evidence "supports" or
"corroborates" theories, it gives one no reason to believe that the theory is true or close
to the truth and no reason to believe that the theory will pass any future tests. Evidence
that supports or corroborates theories in Popper's sense does not support theories.
75
depend on basic statements and deductive logic but that require deci
sions to regard portions of a test system as background knowledge "con
ventional falsifications," as opposed to logical falsifications that depend
only on logic and basic statements. 13 Perhaps Popper's solution to the
problem of induction (and the methodological injunctions that depend
on this purported solution) can be saved if we interpret Popper as argu
ing that there is an asymmetry of conventional falsification that enables
us to provide good arguments against scientific theories even though no
good arguments can be provided for them. The premises in such argu
ments would not, of course, be restricted to basic statements. Back
ground knowledge must be allowed in as well.
But there is no asymmetry of conventional falsification and verifica
tion. There are two ways to establish this claim. Others have argued that
if we cannot verify statements, we do not have enough premises to get
beyond mere logical falsification of whole test system conglomerates - if
we can get even that far (Lieberson, 1982b). In answer to this criticism,
Popperians maintain that decisions will do in place of verifications (Wat
kins, 1984). In response, one can point out (correctly, in my view) that
arbitrary decisions will not do. We need to consider how well confirmed
different statements are. Popperians implausibly (in my view) dispute
this claim. 14 But I shall not pursue this line of argument. Regardless of
13
These labels may be somewhat misleading. Remember that our tentative acceptance of
some basic statements as true is just as conventional as our decision to regard nonbasic
statements as background knowledge.
14 In actual scientific practice, the bulk of the premises needed to derive the refuted
implication are regardless as well established and either true or good approximations to
the truth. This judgment may change, of course, but it is crucial to the interpretation of
experimental failures. If, for example, someone reports experimental results showing
that particular objects fell toward the earth with decreasing acceleration, physicists
would, of course, conclude that there was some experimental error or that other forces
besides gravity were involved. This conclusion, obviously (unless one is a Popperian), is
based on the judgment that Galileo's law of falling bodies is a good approximation to the
truth. If this judgment is unsupportable, as Popper alleges, science proceeds irrationally
and should be replaced by a rational Popperian science.
It is, perhaps, possible to describe a completely noninductive enterprise (see Watkins
1984). In deciding what to do in light of the failure of a prediction of a whole theoretical
system, one might be guided entirely by consideration of which revisions are maximally
content increasing, least ad hoc, etc. Questions about how well supported the various
constituents of the system are might conceivably play no role. But such an enterprise is
radically unlike science, and, indeed, Popper is hesitant in presenting it. Zahar (1983, p.
168) quotes the following passages from Popper (1979) (written in 1930-31), which
illustrate vividly Popper's early hesitance.
We unquestionably believe in the probability of hypotheses. And what is more
significant: our belief that many a hypothesis is more probable than others is moti
vated by reasons which undeniably possess an objective character (Grunde, denen ein
objecktiver Zug nicht abgesprochen werden kann). (1979, p. 145)
The subjective belief in the probability (of hypotheses) can be based on their cor
roboration, but it goes beyond what corroboration can effectively do. This belief
76
Hausman
the basis upon which the decisions are made, only decisions to take
other statements as premises in refuting arguments make it possible to
falsify particular scientific theories and claims. 1s
What I shall argue instead is that if it is permissible to enlarge the set of
permissible premises to include background knowledge in order to make
conventional falsifications possible, one also makes conventional verifica
tions possible. Given the background knowledge that scientists employ,
there may be no more difficulty in verifying universal statements than
there is in falsifying them.
Consider, for example, a scientist attempting to determine the spec
trum of a newly discovered metallic element (Nisbett and Thagard,
1982). The scientist already knows that the spectrum of an element is
invariant in particular ways from pure sample to sample. Given (1) this
background knowledge, (2) the report of a particular Bunsen burner's
flame turning orange, and (3) the claim that the particular sample was
pure, the scientist can deduce that all pure samples of the element will
turn a Bunsen burner's flame orange. So conventional verifications are
no more impossible than are conventional falsifications. Straightforward
probabilistic arguments can, of course, also be made. Given background
knowledge in addition to basic statements, one can thus provide good
15
77
78
Hausman
79
Hausman 1981b) . The economist takes the axioms as already well estab
lished and explores their implications deductively. In doing so, he or she
provides positive reasons to accept the derived conclusions. Testing the
deductive conclusions is important to determine whether one has left out
some relevant factor, not to determine whether the axioms or their
deductive implications are correct.
So, for example, the strongest argument for the hypothesis of rational
expectations is surely not that it survives hard tests, but that it seems to
follow from the axioms of neoclassical theory, once one accepts the
claim that knowledge is, from an economic perspective, a commodity
like any other. Consider the famous argument that John Muth offered:
. . . I should like to suggest that expectations, since they are informed predic
tions of future events, are essentially the same as the predictions of the relevant
economic theory. . . .
If the prediction of the theory were substantially better than the expectations
of the firms, then there would be opportunities for the "insider" to profit from
the knowledge - by inventory speculation if possible, by operating a firm, or by
selling a price forecasting service to the firms. The profit opportunities would no
longer exist if the aggregate expectation of the firms is the same as the prediction
of the theory. (196 1 , pp. 316, 3 18)
80
Hausman
Even if Popper's rules for scientific procedure were free of the difficul
ties discussed previously, they would be of little or no value to econo
mists. For within Popper's philosophy of science, there are no interest
ing questions to be asked concerning the falsifiability of economic theory
(or of any other theory) . Are economic theories logically falsifiable by
themselves? No, of course not, but neither are any interesting theories
in science. Can economic theories be incorporated into logically falsifi
able test systems? Yes, of course they can, but the same goes for theo
ries of practically all disciplines, no matter how patently unscientific they
may appear to be. Can one take the other statements in such test sys
tems to be background knowledge and regard economic theories as
conventionally falsifiable? It all depends on whether the other state
ments have been falsified and thus, ultimately, on whether one can
decide to take still other statements to be background knowledge. Such
decisions are, in Popper's view, more or less arbitrary. If one wants to,
there is little difficulty in taking any statement to be conventionally
falsifiable.
So, if one is concerned about whether economic theory is falsifiable,
as many economists have been, what question is one asking? Once one
leaves Popper's frame of reference, the answer is simple. Economists
have wanted to know whether one could come, as the result of experi
ment or observation, to have good reason to believe that economic
81
82
Hausman
83
84
Hausman
Duhem, P. (1954) . The Aim and Structure of Physical Theory. Princeton, N.J.:
Princeton U niversity Press.
Friedman, M. ( 1953). "The Methodology of Positive Economics" in Essays in
Positive Economics. Chicago: University of Chicago Press, pp. 3-43.
Glymour, C. (1980). Theory and Evidence. Princeton N.J . : Princeton University
Press.
Grunbaum, A. (1976). "Is Falsifiability the Touchstone of Scientific Rational
ity? Karl Popper versus Inductivism" in R. Cohen et al. , eds. , Essays in
Memory of Imre Lakatos. Dordrecht: Reidel, pp. 213-52.
Hands, D. (1985). "Karl Popper and Economic Methodology." Economics and
Philosophy 1 : 83-100.
Hausman , D. ( 1981a). Capital, Profits and Prices: An Essay in the Philosophy of
Economics. New York: Columbia University Press.
(1981b) . "John Stuart Mill's Philosophy of Economics. " Philosophy of Science
48:363-85.
(1985). "Is Falsificationism Unpractised or Unpractisable?" Philosophy of the
Social Sciences 15:313-19.
(1988). "Arbitrage Arguments." Erkenntnis.
Hutchison, T. ( 1977). Knowledge and Ignorance in Economics. Chicago: Univer
sity of Chicago Press.
(1978). On Revolutions and Progress and Economic Knowledge. Cambridge:
Cambridge University Press.
Klant, J. (1984). The Rules of the Game. Cambridge: Cambridge University
Press.
Kuhn, T. (1970). The Structure of Scientific Revolutions, 2nd ed. Chicago: Uni
versity of Chicago Press.
Lakatos, I. (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, pp. 91-96.
(1974). "Popper on Demarcation and Induction," in Schilpp (1974), pp. 24173.
Laudan, L. (1976). Progess and Its Problems. Berkeley: University of California
Press.
Levi, I. (1967). Gambling with Truth. Cambridge, Mass.: MIT Press.
(1980). The Enterprise of Knowledge. Cambridge, Mass . : MIT Press.
Levison, A. ( 1974). "Popper, Hume, and the Traditional Problem of Induc
tion ," in Schilpp (1974), pp. 322-31 .
Lieberson , J . (1982a). "Karl Popper," Social Research 49:68-115.
(1982b) "The Romantic Rationalist," New York Review of Books 29 (Decem
ber 2, 1982).
McCloskey, D. ( 1986) . The Rhetoric of Economics. Madison: University of
Wisconsin Press.
Miller, D. (1982). "Conjectural Knowledge: Popper's Solution to the Problem
of Induction," in P. Levinson , ed. , In Pursuit of Truth: Essays on the
Philosophy of Karl Popper on the Occasion of his 80th Birthday. Atlantic
Highlands, NJ . : Humanities Press, pp. 17-49.
85
Muth, J. (1961). "Rational Expectations and the Theory of Price Movements. "
Econometrica 29:315-35.
Nisbett, R . , and Thagard, P. (1982). "Variability and Confirmation," Philosophi
cal Studies 42:379-94.
Popper, K. ( 1957). The Poverty of Historicism. New York: Harper & Row.
(1963). Conjectures and Refutations; The Growth of Scientific Knowledge.
London: Routledge & Kegan Paul.
(1968) . The Logic of Scientific Discovery, rev. ed. London: Hutchinson & Co.
(1969). "Die Logik der Sozialwissenschaften," in Adorno (1969), pp. 103-23.
(1972). Objective Knowledge; An Evolutionary Approach. Oxford: Clarendon
Press.
( 1974) . "Replies to My Critics," in Schilpp (1974), pp. 961-1200.
(1979). Die Beiden Grundprobleme der Erkenntnistheorie. Tubingen: Mohr
Siebeck.
(1983) . Realism and the Aim of Science; From the Postscript to the Logic of
Scientific Discovery, ed. W. Bartley III. Totowa, N.J . : Rowman and Little
field.
Putnam, H. (1974). "The 'Corroboration' of Theories," in Schilpp (1974), pp.
221-40.
Salmon , W. ( 1981 ) . "Rational Prediction. " British Journal for the Philosophy of
Science 32: 1 15-25.
Settle, T. ( 1974) . "Induction and Probability Unfused," in Schilpp (1974), pp.
697-749.
Tichy, P. (1974) "On Popper's Definition of Verisimilitude," British Journal for
the Philosophy ofScience 25: 155-60.
Watkins, J. ( 1984) . Science and Scepticism. Princeton, N.J . : Princeton Univer
sity Press.
Williams, M. (1977). Groundless Belief. New Haven, Conn. : Yale University
Press.
Zahar, E. (1983) . "The Popper-Lakatos Controversy in the Light of 'Die
Beiden Grundprobleme der Erkenntnistheorie. ' " British Journal for the
Philosophy of Science 34: 149-74.
C H A PT E R 2
87
88
Klant
89
adhere to these rules may be run over. His advice in that case belongs to
art, but the theory he applies regarding the rules that are followed also
proves not to be free of the a priorism that gives shape to an order of
which the evident naturalness must be recognized rationally. Despite his
sincere endeavor to base his theory on facts, it proves to be derived from
a number of principles that require no proof because everyone knows
them (1972, p. 16) .
Malthus too was not lacking in meddlesomeness, the basis of art. In
the introduction to his Principles of Political Economy he states:
One of the specific objects of the present work is to prepare some of the most
important rules ofpolitical economy for practical application [my emphasis], by a
frequent reference to experience , and by endeavouring to take a comprehensive
view of all the causes that concur in the production of particular phenomena.
(195 1 , p. 16)
Ricardo described the laws of wages, rent, and profits by the logical
analysis of a number of principles and did not require his theory to be
tested. After all- as he wrote to John McCulloch when the latter had
queried the truth of his revised theory regarding unemployment as a
result of mechanization - that theory was "as demonstrable as any of the
truths of geometry" (1973, p. 390). However, the result of his treatise
90
Klant
91
In its dealings with science art keeps a respectful distance, but in his
definition of the science of political economy, Mill nevertheless reintro
duces art.
For the "operations of mankind" spoken of in the definition relate to a
"pursuit of objects." Economists postulate laws of behavior by assuming
that individuals act in accordance with rules as they pursue their objec
tives. But this paradox too can be solved, like other ones, by making a
distinction between object language and metalanguage. According to
Mill, political economy is not an art but an investigation of the (possible)
consequences of behavior in accordance with a certain art. In economics
the agent acts as someone who aims at maximum net returns in order to
satisfy wants, as a homo economicus, a protector of personal interests,
and a hunter of profit. Since 1871 the agent has been a rational decider
who weighs utility or neatly ranks preferences. Economics is not an art
but examines art and, like every empirical science, yields pronounce
ments from which those who wish to practice art may learn how the
objectives set by them can - or cannot- be attained.
The natural order, the ideal that inspires deeds, was, however, not yet
banished therewith. Schumpeter has called the fundamental assumption
on which neoclassical theory bases the description of the behavior of
deciders, as sociologists also do, "methodological individualism" (1908,
pp. 88-98) . The explanation of economic events is based on the deci
sions and corresponding actions of separate individuals. It has nothing to
92
Klant
That Locke's natural order did not disappear from economics with ei
ther the physiocrats or the classical economists is shown by the work of
Leon Walras, who laid the foundations of contemporary neoclassical
general equilibrium theory. He distinguishes between three forms of
political economy. The first is that of pure political economy. To him
this consists of a static theory in which transactions are assumed to be
subject to an unspecified process of tiitonnement that leads to a general
equilibrium, as represented by a number of equations relating to de-
93
mand, production, and supply of goods and services. Prices are formed
through perfect competition on interdependent markets of finished
goods, productive services of labor, land, and capital, and of financial
assets. The participants are fully informed about the prices that have
come about and act in accordance with utility functions that are invari
ant during the process of tatonnement.
Walras was aware that economic equilibrium does not occur in reality
and that in the latter the conditions of his model are insufficiently satis
fied: "the Walrasian attempt to explain the market system takes place in
a cognitive, motivational and institutional vacuum" (Albert, 1979, p.
1 19) . His proof of existence - at least the attempt to do so- is purely
mathematical, namely, of a unique solution of his system of equations.
But, like Quesnay and Smith, he wanted to reform the world in terms of
his ideal. He investigated not out of pure curiosity, but in order to be
able to use science (Boson, 1950, p. 106). In applied political economy
he endeavored to examine how the ideal could be realized as much as
possible. Disturbances of his desired system ought to be avoided and
countered: "la science etant definie /'idealisation de la realite; et l'art etant
definie la realisation de /'ideal" (1936, p. 21 ; my emphasis). Science
brings about an idealization, and art realizes the ideal. Idealization func
tions in W alras too as ideal.
To that end, the necessary inroads had to be made into the principle
of laisser-faire. Walras advocated government intervention, such as con
trol of money creation , protection of consumers and supervision of ad
vertising, nationalization of businesses with increasing returns to scale,
supervision of railways and freight rates, a ban on speculation on the
stock exchange by other than well-informed specialists, and interna
tional labor legislation. He believed that it was not possible to leave all
enterprises free. On the other hand, he considered it quite feasible to
expropriate them all, for that would not be at variance with liberty,
equality, order, and justice, if only the free market mechanism were to
continue to exist. The operation thereof thus proves, in his opinion, to
conform to the values of the French Revolution and his natural order
allows of variants. Market socialism is a form of Walrasian nature. No
wonder, therefore, that his general equilibrium theory could be applied
in the theory of planned production and that Oskar Lange managed to
combine Marxism with neoclassical microeconomics.
The reason Walras wanted to see the world organized in accordance
with his pure theory was that, he believed, thanks to the perfect market
mechanism, a maximum of utility is produced within certain limits, or,
as he could have put it today, that, given the circumstances, an optimum
situation is realized for each individual participant in the system. Every-
94
Klant
one aims at maximum satisfaction of wants and achieves the best posi
tion compatible with that of the others. Because of some obiter dicta of
Vilfredo Pareto on maximum collective ophelimity, we today call that
imaginary state of affairs "Pareto-optimal" and a system of making its
achievement possible "Pareto-efficient . " The welfare of no single indi
vidual can be increased without at least one other individual suffering a
setback as a result.
Walras was convinced that by demonstrating the efficiency of his po
tential ideal, he had supplied no proof of its justness. The Pareto
optimum defines an ideal called "efficiency," which is worth pursuing,
but for the rest, it has no ethical effect. It depends on a number of
boundary conditions, such as the distribution of property, that are open
to change and then result in a different income distribution. Each distri
bution has a different optimum.
Walras wished to see the question of a just distribution posed in
"social economics. " His own contribution to this consisted principally in
a plea for nationalization of land; in emulation of his father, Auguste
Walras, and of Hermann Heinrich Gossen and John Stuart Mill (1926,
pp. 267-80) ; and in agreement with the demands of the Communist
Manifesto. He was an advocate of consumer and producer cooperatives,
by which everyone would be a capitalist and a worker at the same time,
firms would be democratized, and democracy would be activated. He
called himself a "scientific socialist" in contradistinction to the Marxists,
who took no account of the laws of supply and demand (1936, pp. 22933). Perhaps for that reason he will one day be given a statue in Yugosla
via, Hungary, or China. His socialism was responsible for the rules he
devised, taking into account the laws of the market mechanism, for
realization of the ideals of efficiency and justice.
According to William Jaffe, Walras had much more drastic moralistic
intentions with his pure theory ( 1977). Donald A. Walker has rightly
disputed this interpretation (1984) , but Walras's statement on the rela
tion between science and reality is not affected by his criticism. Walker
is of the opinion that Walras used the word ideal for "idealization"
(1984, p. 452) . However, that opinion is not supported in the cited
statement. Walras considered art necessary to realize the ideal, and he
therefore gave the word a normative connotation. As Jaffe mentions,
Pareto was in any case not very taken with Walras's idealism. With
reference to it, he wrote to Pantaleoni that he did not agree that pure
theory demonstrates what the facts ought to be. He considered it inad
missible to study what ought to be instead of what is. In his publications
too he evinced such empirical scientific intentions. In his opinion, pure
science describes but does not prescribe.
95
96
Klant
97
researcher's view of the world cannot be an empty one. Der Mann ohne
Eigenschaften does not exist. "Do your research without bias" is there
fore a piece of advice that will not lightly be given to natural scientists
unless the intention is that, if they are to be able to maintain their theory
successfully, they must make allowance for criticism. "To attain objectiv
ity we cannot rely on the empty mind," writes Karl Popper; "objectivity
rests on criticism, on critical discussion, and on the critical examination
of experiments" (1975 , p. 79) . The objectivity of theories consists in the
fact that they can be tested intersubjectively (1959, pp. 37-44) . As a
result, the suspicion arises that Brentano's requirement could proceed
from the belief that it is possible for research scientists to resist all
criticism if they stick to their ideals. The disputes between professors
would then be impossible to resolve.
The value judgment conflict
98
Klant
99
100
Klant
101
102
Klant
103
free," because the theories that have come about in that way do not
contain any explicit prescriptions. If, however, as Mill wanted, "art
hands its end over to science,'' it receives in return a nonobjective theo
rem that is dependent on value judgments that, in Weber's view, are
irrelevant in science to the decision of whether to accept or reject.
In accordance with Hume's law, "is judgments" cannot imply prescrip
tions. Ought does not follow from is. Prescriptions in which norms are set
are justified by recourse to values. They can, it is true, be derived from a
conjunction of prescriptions and is judgments. However, nonobjective
ideological pronouncements, as meant by Weber, are not pure is judg
ments. They are based in part on value judgments. The facts are exam
ined in the light of ideas about a natural, that is, attractive order. If a
nonobjective theory of this kind is to be called "value free," Schmoller's
ethical theory of just distribution will have to be taken as value free. After
all, that theory too can be reproduced without explicit prescriptions. "A
value judgment may be found without being uttered in an evaluation
sentence" (Taylor, 196 1 , p. 3) . It is possible to describe evaluatively and
to evaluate descriptively. According to R. M. Hare, the distinction be
tween is judgments and value judgments is not exclusive (1952, pp. 1 1 126; 1972, pp. 55-75). If art makes an appeal to Schmoller's "positive"
science, it promptly receives an ethical reply, but not from Schmoller
alone.
Just as ought does not follow from is, is does not follow from ought.
What should be objective or value free, in accordance with the method
ological norm, is not so if it is not satisfied because of whatever weak
ness. For clarity's sake, it would also be better if we never spoke again of
value freedom but simply of objectivity, for, as Weber himself clearly
realized in his 1904 article, that norm forms the problem proper.
If philosophies of life always interfere with research, it will not be
possible anyway to keep them away in the case of testable idealizations
either. For, because we require a set of abstract laws to explain a con
crete situation and to interpret our observations of that situation, hy
potheses cannot be independently tested. According to Pierre Duhem,
critical experiments therefore do not exist, and Willard Van Orman
Quine has rightly established that "our statements about the external
world face the tribunal of sense experience not individually but only as a
corporate body" (1964, p. xii). The one theory depends on the other.
However, Duhem-Quine's holistic thesis does not exclude the possibil
ity that researchers, by making alternating attempts with different combi
nations of hypotheses, may arrive by testing at a decision as to what they
will choose and what they will reject. But the conditions of the static
metatheory of naive falsificationism (Klant, 1984, pp. 40-6) , which com-
104
Klant
105
cation show what imposing structures can be erected right in the swamp
of nature.
In the natural sciences, central ideas can be used to continue to see the
world in a certain way. In the social sciences, in which the consequences
of art are investigated, they are, however, also used to change the world.
The ideas then become ideals, that is to say, values. For the social world
is a changing system that we make ourselves. It is inevitable that the
view of the structure or of the process is at the same time an idea about
what can be done with it. Every social theory is a train that is ready to be
boarded to take us to a- somewhat - better country. In his faith in the
possibility of "causal imputation" without speculation, Weber has not
shown us how the mix of idealization, idea, and ideal desired by Othmar
Spann can be avoided in order to realize the ideal of a rightly value-free
social science.
The value j udgment conflict yielded a large number of publications,
especially in Germany, by both the Wertiker and the Wertfreien (Weber,
1952). The former succeeded in involving in the conflict the philosophy
of the day, such as that of Edmund Husserl. However, they made little
impression on the mainstream economists. Political economy had in the
meantime become economics or, better still, economic science. The prin
ciples of value freedom that had been proclaimed by the verstehende
sociologist Max Weber were well received by the neoclassical econo
mists. However, his related philosophy of the heuristic ideal types did
not take root. The metatheory of value freedom, purged of hermeneu
tics, was taken for granted, the problems of provability, which were
sensed but not solved by Weber, being overlooked. In probably the
most widespread view among economists, the objectivity of research is
guaranteed by the "analysis," which takes place without value judg
ments and values (except the methodological ones) being involved.
According to Schumpeter, analysis consists of taking note of facts,
forming concepts, positing hypotheses, and applying logic. However, he
distinguished two phases. One starts with a prescientific vision of pre
sumptive relations that can be ideologically influenced. Unlike Bren
tano, he believed that prejudice can be a condition for inspiration. How
ever, next analysis takes place, through which the vision, if it does not
prove tenable, is discarded or converted into a theory from which the
ideological element has been eliminated (1949, pp. 345-59).
Schumpeter acknowledges that "in economics, and still more in other
social sciences , [the] sphere of the strictly provable is limited in that
there are always fringe ends of things, that are matters of personal
experience and impression from which it is practically impossible to
drive ideology, or for that matter conscious dishonesty, completely"
106
Klant
The life expectancy of a metatheory of the heuristic ideal types was not
great in an environment in which economists increasingly felt that they
had to behave like natural scientists. They concentrated on making for
malized theories and testing them by statistical analysis. Their preten
sions are often called "positivistic," which is not of benefit to the clarity of
the philosophical discussion. In 1944-5 they had already received critical
rationalistic recognition by Karl Popper, when he published what became
The Poverty ofHistoricism as a series of articles in Economica.
In Popper's view the method of the social sciences ought to be a
"technological approach," imposing "a discipline on our speculative incli
nations" and forcing us "to submit our theories to definite standards,
such as standards of clarity and practical testability. " Perhaps we owe it
to the intercession of Friedrich von Hayek, but Popper showed great
respect for economics. He recommended social scientists not to look
further than their Galileo or their Pasteur, but added in a footnote: "It
must be admitted, however, that mathematical economics shows that
one social science at least has gone through its Newtonian revolution"
(1957, pp. 59-60). The pretensions of the economists were indeed New
tonian, but were their deeds too? Popper did not go into that question.
In the same work he drew attention to what he regarded as "a really
fundamental similarity between the natural and the social sciences" by
pointing to "the existence of sociological laws or hypotheses which are
analogous to the laws or hypotheses of the natural sciences" (1957, p.
62) . Strangely enough, in so doing he brought forward only the similar
formulation of hypotheses that, in both cases, render possible logical
transformations into corresponding "technological corollaries" : q is
107
108
Klant
109
110
Klant
111
1 12
Klant
world with which they are confronted, if they try to solve economic
problems couched in causal terms, happens to be subject to much more
rapid historical change than nature. Humans, who learn and create,
change their world constantly and evoke unforeseeable situations one
after the other. The verbal interpretations, which can be much more
subtle but less exact than the econometric ones, also fail in the process of
give-and-take converging toward the truth that Schumpeter thought he
saw happening.
The cultural sciences are indeed different from the natural sciences.
The absence of universal numerical constants, as a result of which syn
thetic combination and fictitious isolation do not yield testable results,
means that the world of the economists appears much more complex
than that of the physicists. Their basic hypotheses are not "conclusively
decidable. " They are heuristic systems, decided on in a rational critical
discussion to which the inconclusive recourse to experience also belongs.
Economics is philosophy, science, and art. John Maynard Keynes was
rightly of the opinion that economics is a moral science that cannot be
practiced in the same way as the natural sciences (Klant, 1985).
Economists who were to decide on acceptance or rejection of their
theories in accordance with the system of values of their colleagues in
the natural sciences would have to be desperate and, in any case, unde
cided and highly uncertain researchers. However, the opposite is the
case. In economics it is almost only metatheoreticians and historians,
who write not about the world but about theories, who permit them
selves skepticism and neutrality. The real theoreticians, when they pres
ent their speculative solutions, are as a rule very sure of themselves.
Even the word "probably" does not pass their lips. They have more
about them of prophets or mathematicians than of careful, empirical,
scientific researchers. Even if they remain much further removed from
the realization of the empirical scientific ideal than physicists, they are
usually more firmly convinced of the correctness of their theory.
However much economists may evoke their purity, they want to
change the world. They want to contribute to the solution of urgent
practical problems. The theories of Quesnay, Smith, Ricardo, Mill,
Marx, Walras, Marshall, Pareto, Wicksell, Veblen, Keynes, Schum
peter, Hayek, Tinbergen, Hicks, Friedman, and Samuelson contain mes
sages. Perhaps the tame cattle, peacefully grazing in the manner of
Kuhn's normal scientists in an area enclosed by definitions of Robbins,
will succeed in simply enjoying their own digestion. The creative
economists- and they are much more numerous than the "extraordi
nary" ones, according to Kuhn, and the seventeen I have listed
entertain more or less veiled but recognizable ideas about a natural
1 13
order and a "meaning" of history, ideas that are not scientifically prov
able but do help to determine the content of their theories. They try to
make the latter acceptable by invoking plausibility.
Of course , they also pursue the consistency of the theories they make,
for he who contradicts himself proves nothing. However, it testifies to
excessive indolence in furnishing proof to do as Walter Eucken did, and
take consistent theories for true ones (1947, p. 269). It equally testifies
to dogmatic blindness, and thus also to indolence, to regard a consistent
theory that one considers plausible as having been conclusively demon
strated. Discussions in economics often lead to strengthening of both
opposite points of view and not to "objective" knowledge, that is, estab
lished in accordance with the rules of the game, but provisional and
uncertain.
If we look at things from the aspect of the logical structure of their
theories, it can hardly be otherwise. Whatever Popper may have thought
of economics, it can be demonstrated with his own means: A door is
open to extrascientific values because basic economic theories are not
falsifiable. Even if they were, ideals are given an opportunity to pene
trate the hard core of ideas that is maintained as long as possible. The
fact is that ideas about a world that is lived in easily become ideals if they
relate to action. Welfare theory, for example, which is the result of a
formal analysis of rational action, shows us the properties of an ideal
ized, timeless system in which the attainment of joint ends is in equilib
rium. Of course, the implications can be shown without the introduction
of value judgments other than those already contained in the premises,
but because it is an idealized social decision system, it is a potential
ideal. If welfare theory is applied, it functions as an ideal.
Sometimes the values of the economists are professed in a wide circle,
such as laissez-faire in the nineteenth century or methodological indi
vidualism or Pareto efficiency today. Sometimes too the supporters of
differing values are irreconcilably opposed, such as the Keynesian inter
ventionists and the monetarist neoliberals, or the "bourgeois" and Marx
ist economists. However, there is little reason to regard the one ideal as
legitimate, formal, indifferent, or neutral and the other as a sign of lack
of respect for value freedom and objectivity. Economic theories cannot
be objective . What matters is to explicate the values by self-examination
and by unmasking others; to continue the discussion on each relevant
aspect, without paying attention to what connoisseurs of the subject
matter of economics forbid; to extend empirical research and the study
of history; and to abandon the hypocrisy of the so-called value freedom,
about which Weber said untenable things more than seventy years ago.
This does not mean to say that in economic research we can waive the
1 14
Klant
ideal of objectivity and give free play to what our desires suggest to us.
The empirical scientific ideal cannot be abandoned by any serious partici
pant in the joint discussion on problems in terms of cause and effect.
"All science, and all philosophy, are enlightened common sense" (Pop
per, 1983, p. 34). In our attempts to convince one another, we are again
and again confronted with decisions that cannot be postponed, because
it is not only a matter of acquiring knowledge of what there is, but also
of what has to be done. In that case, our methodological value judg
ments are mingled with other value judgments if we are lacking in proof.
In our attempts to refute the empirically irrefutable, we seek consistent
and plausible solutions to problems and an association of minds on the
basis of common value, not only -with due respect to Marx - to under
stand our world, but also to change it. Economics is political economy.
References
Albert, H. (1979). "The Economic Tradition ," in K. Brunner, ed. , Economics
and Social Institutions. Boston: Martinus Nijhoff.
Blaug, M. (1980). The Methodology of Economics, Cambridge: Cambridge Uni
versity Press.
Boson, M. (1950) . Leon Walras, fondateur de la politique economique scientific.
Rennes: Imprimeries Reunies.
Brentano, L. (1911). "Uber Werturteile in der Volkswirtschaftslehre ," Archiv
fur Sozialwissenschaft und Sozialpolitik 33:695-714.
Cohen, M . R . , and Nagel, E. (1963). An Introduction to Logic and Scientific
Method (1934) . London: Routledge & Kegan Paul.
Entreves, A. P. d' (1970). Natural Law (1951). London: Hutchison.
Eucken, W. (1947). Die Grundlagen der Nationalokonomie (1939). Godesberg:
Kiipper.
Frondizi, R. (1971). What Is Value? La Salle, Ill . : Open Court.
Gehrig, H. (1914). Die Begrundung des Prinzips der Sozialreform. Jena:
Fischer.
Hare, R. M. (1952) . The Language of Morals. Oxford: Oxford University Press.
(1972). Essays on the Moral Concepts. London: Macmillan.
Hayek, F.A. (1 964 [1955]). The Counter-Revolution of Science. London: Free
Press.
Helvetius, C.A. (1973 [1758]). De !'esprit, ed. F. Chatelet. Verviers: Gerard.
Hutchison, T.W. ( 1964). "Positive" Economics and Policy Objectives, London:
Allen & Unwin.
(1978). On Revolutions and Progress in Economic Knowledge. Cambridge:
Cambridge University Press.
Jaffe, W. (1977). "The Normative Bias of the Walrasian Model," Quarterly
Journal of Economics 91:371-87.
Kant, I. [1781] Kritik der reinen Vernunft, Werke, III . Wiesbaden: Suhrkamp.
115
Keynes, J.M . ( 1972 [1933]). Essays in Biography, The Collected Writings ofJ.M.
Keynes, Vol. X. London: Macmillan .
Keynes, J.N. ( 1963 [ 1880]) . The Scope and Method of Political Economy. New
York: August M. Kelley.
Klant, J.J. ( 1982). "ldealisatie: idee en ideaal," in B. de Gaay Fortman, ed. ,
Economie en waarde. Alphen aan de Rijn: Samson.
(1984). The Rules of the Game. Cambridge: Cambridge University Press.
(1985). "The Slippery Transition," in T. Lawson and H. Pesaran, eds., Keynes'
Economics. London: Croom Helm.
Latsis, S.J. (1976a). "A Research Programme in Economics," in S.J. Latsis, ed. ,
Method and Appraisal in Economics. Cambridge: Cambridge University
Press, pp. 1-41.
(1976b). "Situational Determination in Economics," British Journal for the
Philosophy of Science 27:207-45.
Laudan, L. (1981). Science and Hypothesis. Dordrecht: D. Reidel.
Losee, J. (1980 [1972]). A Historical Introduction to the Philosophy of Science.
Oxford: Oxford University Press.
Mach, E. (1906 [1905]). Erkenntnis und Irrtum. Leipzig: Barth.
Machlup, F. ( 1969). "Positive and Normative Economics," in R.L. Heilbronner,
ed. , Economic Means and Social Ends. Englewood Cliffs, N.J. : Prentice
Hall.
Malthus, T. R . (1951 [1823]). Principles of Political Economy. Oxford: Black
well.
Mannheim, K. (1979 [1928]). Ideology and Utopia. London: Routledge & Ke
gan Paul.
Mill, J.S. (1843). A System of Logic, various editions.
(1974 [1844]). Essays on Some Unsettled Questions of Political Economy.
Clifton: August M. Kelley.
Myrdal, G. ( 1970) . Objectivity in Social Research. London: Dackworth.
North, D. (1954 [ 1691]). Discourse upon Trade, in J.R. McCulloch, Early English Tracts (1856). Cambridge: Cambridge University Press.
Papandreou, A . G . (1958) . Economics as a Science. Chicago: Lippincott.
Plato, 1976. Protagoras. Trans. C.C.W. Taylor. Oxford: Clarendon.
Pope, A. (1924). Collected Poems, ed. B. Oobree. London: Dent.
Popper, K.R. (1956 [ 1945]). The Open Society and Its Enemies. London: Routledge & Kegan Paul.
(1957). The Poverty of Historicism. London : Routledge & Kegan Paul.
(1959). The Logic of Scientific Discovery. London: Hutchison.
(1965 [1963]) . Conjectures and Refutations. London: Routledge & Kegan
Paul.
(1983 [1972]) . Objective Knowledge. Oxford: Clarendon.
(1972). "Die Logik der Sozialwissenschaften," in T.W. Adorno, ed. , Der
Positivismusstreit in der deutschen Soziologie. Darmstadt: Luchterand.
(1975). "The Rationality of Scientific Revolutions," in R. Harre, ed. , Prob
lems of Scientific Revolution. Oxford: Oxford University Press.
116
Klant
Quine, W.V.O. (1964 (1950]). Methods of Logic. New York: Holt, Rinehart &
Winston.
Reisman, D. ( 1976) . Adam Smith's Sociological Economics. London: Croom
Helm.
Rescher, N. (1969) . Introduction to Value Theory. Englewood Cliffs, N.J . :
Prentice-Hall.
Ricardo, D. (1971). Speeches and Evidences, in Works, Vol. V. Cambridge:
Cambridge University Press.
(1973). Letters 1819-June 1821, Works, Vol. VIII. Cambridge: Cambridge
University Press.
Robbins, L. (1938). "Live and Dead Issues in the Methodology of Economics ,"
Economica n.s. 5 :342-52.
(1946 [1932]). An Essay on the Nature and Significance of Economic Science.
London: Macmillan.
Runciman, W.G. (1972). A Critique of Max Weber's Philosophy of Social Sci
ence. Cambridge: Cambridge University Press.
Say, J.B. ( 1972 [ 1803]). Traite d'economie politique. Paris: Calman-Levy.
Schmoller, G. (1873) . "Sendschreiben," Jahrbiicher fur Nationa!Okonomie und
Statistik 20:47-70.
(191 1). "Volkswirtschaft, Volkswirtschaftslehre und methode," Handworter
buch der Staatswissenschaft, Vol. VIII. Jena: Fischer.
Schumpeter, J . A . (1908). Das Wesen und der Hauptinhalt der theoretischen
Nationa!Okonomie, Leipzig: Daneker & Humblot.
(1949). "Science and Ideology," American Economic Review 39:349-59.
(1954). History of Economic Analysis. London: Allen & Unwin.
Senior, N.W. (1966a [1826]). An Introductory Lecture on Political Economy, in
N.W. Senior, ed. , Selected Writings on Economics. New York: August M.
Kelley.
(1966b [ 1852]) . Four Introductory Lectures on Political Economy, in N.W.
Senior, ed . , Selected Writings on Economics. New York: August M. Kelley.
Simmel, G. (1907 [1900]). Die Philosophie des Ge/des. Leipzig: Duncker &
Humblot .
Smith, A. (1976a ( 1750]). The Theory of Moral Sentiments, ed. D. D. Raphael
and A.L. Madie. Oxford: Clarendon.
(1976b [ 1776]) . An Inquiry into the Nature and Causes of the Wealth of Na
tions, ed. R.H. Campbell, A.S. Skinner, and W.B. Toell. Oxford: Claren
don.
(1980 [1795]). Essays on Philosophical Subjects, ed. W.P.D. Wrightman and
J.C. Pryce. Oxford: Clarendon.
Stegmiiller, W. (1979). The Structuralist View of Theories. Berlin: Springer
Verlag.
Taylor, P.W. ( 1961). Normative Discourse. Englewood Cliffs, N.J.: Prentice
Hall.
Toulmin, S. (1960 [1953]) . The Philosophy of Science. New York: Harper &
Row.
117
PART II
CHAPTER 3
Ad hocness: Popper
121
122
Hands
ad hoc. Although the term "ad hoc" has no universally accepted defini
tion within the philosophy of science, it is usually reserved for a particu
lar type of face-saving/falsification-avoiding theoretical adjustment. For
instance, Laudan's (1977, p. 1 15) popular definition states, "a theory is
ad hoc if it is believed to figure essentially in the solution of all and only
those empirical problems which were solved by, or refuting instances
for, an earlier theory . "
A s a n example o f the a d hocness problem, consider the following
(hypothetical) case from economic theory. Suppose a macroeconomic
theory Tl predicts that for any country whose money supply changes by
the amount m, the price level in the following period will change by the
amount p. Suppose further that after this result has been successfully
corroborated for several countries, a contrary result is obtained for coun
try A , that is, Tl is refuted by the evidence from country A. How can Tl
be saved? How can Tl be adjusted so that the potential refutation is
avoided? One approach is to simply modify Tl in an ad hoc way to avoid
the refutation. For instance, we could modify Tl so that it now says, "for
all countries except country A, m implies p. " This new theory-call it
TI- is confirmed by all the countries that confirmed Tl , and in addition,
it is confirmed in country A, where Tl failed. Thus, it could be argued
that since Tl was refuted and T2 is more confirmed (or, in Popper's
terms, more "corroborated"), the move from Tl and T2 constituted
scientific progress.
It seems obvious that this type of theory modification should not
count as progress. Tl was refuted, but T2 is not really any better. The
change from Tl to T2 was patently ad hoc; it was contrived solely to
protect the theory from an observed anomaly, and served no other
purpose. Not only does this amount to "cheap success" (Worrall, 1978,
p. 49) , it also makes "progress" trivially easy. 2 By this technique a
successful theory can always be constructed out of one with a dismal
track record. The history of human speculation abounds with such (ulti
mately unsuccessful) face-saving moves. 34
Within the Popperian tradition, the ad hocness problem is even more
2
For this reason, Watkins (1984, p. 166) refers to the non-ad hocness requirement as an
"antitrivialisation principle. " The principle states that "any philosophical account of
scientific progress must be inadequate if it has the (no doubt unintended) implication
that it is always trivially easy to make theoretical progress in science."
3 Grunbaum (1976, pp. 329-30) cites historical examples.
4 It should be noted that although the undesirability of ad hoc theories is being presented
as obvious, not all philosophers of science would agree. Laudan (1977, pp. 1 14-18), for
instance, finds no difficulty with this type of ad hoc modification, claiming that to
condemn such adjustments puts "an epistemic premium on theories which work the first
time around" (p. 1 17) . Grunbaum (1976, pp. 358-61) advocates a similar position.
123
124
Hands
125
Ad hocness: Lakatos
126
Hands
15 Based on what Lakatos actually wrote, this is not literally correct. He says that a change
is theoretically progressive if it predicts novel facts- he does not say only if it predicts
novel facts. Thus the door is (formally) left open for other things that would be sufficient
for theoretical progress. Although this is an interpretation that can be defended on the
basis of what Lakatos actually wrote, it is certainly not the standard interpretation. The
standard interpretation is (as above) that progress requires novel facts, that is, "if and
only if" is read in rather than "if." The standard interpretation is probably more consis
tent with the overall spirit of the MSRP.
16 For example, he says "produces novel facts (that is, it is 'independently testable')"
(1970, p. 126) and "theories which had no excess content over their predecessors (or
competitors) that is, which did not predict any novel facts" (1970, p. 175, note 2).
17 This follows his presentation in (1978a), a paper originally published in 1968.
18 Actually, of course, Popper requires some confirmation of this excess content (see note
8) and Lakatos requires at least intermittent confirmation.
19 This equivalence is implicit in Lakatos (1968, 1970); it becomes explicit in later
"Lakatosian" works such as Zahar (1973).
20 "The problem of when a hypothesis-replacement is 'ad hoc', i.e., irrational, degenerat
ing, bad, has never been discussed with more attention and detail than by Popper and
myself" (Lakatos, 1978b, pp. 221-2).
21 Actually, this identity requires a slight modification of Lakatos's definition; see Zahar
(1973, p. 101, note 1).
127
22
Lakatos (1968, pp. 170-3; 1970, pp. 134-8) and Worrall ( 1978, p. 59). Worrall (1978, p.
69, note 35) provides a partial list of the things that the positive heuristic might include.
23 As an economic example, in Weintraub's (1985a) specification of the neo-Walrasian
research program, the positive heuristic consists of propositions like "go forth and
construct theories in which economic agents optimize" and "construct theories that
make predictions about changes in equilibrium states. "
24 Zahar i s quoted here simply because his statement i s more straightforward than
Lakatos's, which must be extracted from his discussion on pp. 174-7 and 182-7 of his
1970 paper.
128
Hands
This third type of ad hocness is clearly different from the type that
concerned Popper (and was discussed previously) . 25 It is apparent that
even an empirically progressive (non-ad hoc2) adjustment might still be
ad hoc3 For instance, it may be that T'2, proposed in the previous
economic example, is ad hoc3 . Unless we know the macro research
program the theory is embedded in, and can show that something in the
hard core or positive heuristic of that program suggests that deficits
should matter, T'2 could be accused of being ad hoc3 Lakatos clearly
considered program continuity to be extremely important and some
thing that was not guaranteed by Popper's attempts to exclude ad hoc
adjustments. For Lakatos, real progress occurs when a research pro
gram achieves theoretical/empirical progress "while sticking to its hard
core" ( 1970, p. 187) .
Thus there are two separate notions of ad hocness within the Pop
perian tradition. Ad hoc1 _2 , or Popperian ad hocness, is the traditional
notion of modifying a theory for the sole purpose of avoiding falsifica
tion. It can be prevented by requiring independent testability (or test
ing) , that is, the prediction (or confirmation) of novel facts. The second
notion, due to Lakatos, is ad hoc3 , and it pertains to the continuity of the
program in which the theory is embedded. Ad hoc3 adjustments can be
avoided by requiring that any change be consistent with the core and
positive heuristic of the program. Armed with these two concepts of ad
hocness, let us now turn to the way the term is used in economics.
III.
Ad hocness in economics
Though Popper did require a new theory to be "deeper," to have "a certain coherence
or compactness" (1972, p. 197) or "simplicity" (1963, p. 241). Although these concepts
are related to Lakatos's non-ad hoc3 , Popper was pessimistic about formalizing such
requirements and never provided more than an intuitive discussion. Nor did Popper
ever directly relate this coherence and simplicity to his notion of ad hocness. Recently,
though, such concepts have been analyzed more rigorously (and related to ad hocness)
by neo-Popperians, especially Watkins (1984) .
129
Not only economic theorists but natural scientists as well "use the term ad hoc to cover a
multitude of sins. A theory may be called ad hoc because it is unaesthetic and clumsy,
because it is arbitrary and uninteresting, or because it is wildly implausible" (Koertge,
1978, p. 267) . Grunbaum (1976, p. 361) cites a quite extensive list of scientific uses of ad
hocness (compiled by Holton).
27 Similar criticisms of Marxian economics are made by Blaug ( 1980b).
28 Although the ad hocness that concerns economic methodologists always seems to be of
the Popperian type, it is interesting that Popperian methodologists such as Blaug,
Hutchison, and Klant explicitly discuss the "disease" itself, that is, ad hocness, whereas
Lakatosian methodologists are more likely to focus exclusively on the "cure," that is,
"excess content" and "novel facts. " In either case, there is seldom any recognition that
the fundamental issue is the same.
29 From T. Sargent and N. Wallace, "Rational Expectations and the Theory of Economic
Policy," Journal of Monetary Economics 2, April 1976; quoted by Maddock (1984, p.
295).
130
Hands
131
needed to generate the main Keynesian results, yet it is not derived from
or supported by any analysis grounded in the motivational assumptions
economists have found to be generally applicable" (Olson, 1984, p.
299) . Even Solow's disequilibrium dynamics are potentially ad hoc be
cause they are "certainly not the solution of some vast intertemporal
optimization problem" (Klamer, 1983, p. 139) . These statements do not
indicate that nonrational expectations models are ad hoc because they
lack empirical novelty or independent testability (though they might) ;
rather, they are ad hoc because they are not derived from individual
optimizing behavior. Nonrational expectations models are accused of
being ad hoc3 , not ad hoc1_2; according to their critics, the assumptions
they require do not follow from the core or positive heuristic of the
neoclassical/individual optimization research program, and that is the
source of their ad hocness.
What about criticism going the other way, that is, what about theorists
who accuse the rational expectations models of being ad hoc? Are they
concerned with ad hoc3 as well? Certainly this is the case for Hahn. The
basis of Hahn's ad hocness charge is that Lucas-type models assume
equilibrium but fail to explain the presence of that equilibrium on the
basis of rational optimizing behavior. According to Hahn, "much impor
tance is attached to rationality until it comes to price changes: then
anything goes. For the Lucasians, prices change to keep Walrasian mar
kets cleared by a mechanism that is entirely secret in the Lucasian mind"
(1983, p. 54) . 32 The accusation here, as in the case of nonrational expec
tations models, is that the theory in question fails to follow from the
positive heuristic of the neoclassical research program, not that it is ad
hoc1 -2 33
What about the situation outside of the rational expectations litera
ture? Is the concern over ad hocness only a rational expectations phe
nomenon, or is it an economic theory phenomenon? Certainly if the
32 Similarly, "I am in Lucas's methodological spirit when, instead, I propose that prices are
flexible when there are no obstacles to price change when it is to someone's advantage to
do so. More formally, prices in a given theory are flexible when their formation is
endogenous to the theory . . . . Now as a matter of fact, prices in the Lucasian world are
not properly endogenous to the fundamental theory, because there is no theory of the
actions of agents that explains how prices come to be such as to clear Walrasian mar
kets" (Hahn, 1983, p. 49).
33 Following a time-honored Lakatosian tradition (Lakatos, 197 1 , p. 107) , theorists who
have "misbehaved" relative to the preceding reconstruction are confined to footnotes.
One such case is Tobin (1980). In his discussion of Lucas's model, Tobin accuses (p. 42)
it of being ad hoc in the traditional (ad hoc1 _2 ) sense. In response, Lucas fully admits that
it is. "If ever there was a model rigged frankly and unapologetically, to fit a limited set of
facts, it is this one" (1981 , p. 563). The reader is reminded that the preceding argument
tries to offer only an empirically progressive generalization about the way economic
theorists use the term "ad hoc," not a claim that it is not used in any other way.
132
Hands
133
134
Hands
"I define a research programme as degenerating even if it anticipates novel facts but
does so in a patched-up development rather than by a coherent, pre-planned positive
heuristic" (Lakatos, 1971, p. 125, note 36).
37 A partial list is provided in Hands (1985), but many more works have appreared since
that paper was written (for instance, Maddock, 1984, and Weintraub, 1985a).
38 For instance, Zahar (1973) argues that Einstein's program superseded Lorentz's pro
gram because it was heuristically more progressive.
39 Clearly, the best example is Weintraub (1985a).
40 Although the absence of novel facts is a theme in Hands (1985), in all fairness it should
be noted that "paucity" does not imply "nonexistence. " Maddock (1984) and Weintraub
(1985b), for instance, both do a very good job of finding novel facts in the economic
programs they discuss.
It may be just a humorous coincidence, but whenever philosophers in the Popperian
tradition discuss authors who are critical of novel facts, that is, those who think that facts
135
136
Hands
References
Begg, D . K.H. ( 1980) . "Rational Expectations and the Non-Neutrality of System
atic Monetary Policy," Review ofEconomic Studies 47:293-303.
(1982). The Rational Expectations Revolution in Macroeconomics. Baltimore:
Johns Hopkins University Press.
Blaug M. (1980a). The Methodology of Economics. Cambridge: Cambridge
Univesity Press.
(1980b) , A Methodological Appraisal of Marxian Economics. Amsterdam:
North-Holland.
Debreu, G. (1983). Mathematical Economics: Twenty Papers of Gerard Debreu.
Cambridge: Cambridge University Press.
Fisher, F.M. (1983). Disequilibrium Foundations of Equilibrium Economics.
Cambridge: Cambridge University Press.
Gardner, M . R. ( 1982). "Predicting Novel Facts," British Journalfor the Philoso
phy ofScience 33: 1-5 .
Grunbaum, A. (1976). "Ad Hoc Auxiliary Hypotheses and Falsificationism,"
British Journal for the Philosophy of Science 27:329-62.
Hahn, F. (1983). Money and Inflation. Cambridge, Mass.: MIT Press.
Hands, D.W. (1985) . "Second Thoughts on Lakatos," History of Political Econ
omy 17:1-16.
Hutchison, T.W. (1981). The Politics and Philosophy of Economics. New York:
New York University Press.
Klamer, A. (1983) . Conversations with Economists. Totowa, N.J.: Rowman and
Allenheld.
(1984). "Levels of Discourse in New Classical Economics," History of Political
Economy 16:263-90.
Klant, J.J. (1984). The Rules of the Game. Cambridge: Cambridge University
Press.
Koertge, N. (1978). "Towards a Theory of Scientific Inquiry," in G. Radnitzky
and G. Anderson, eds . , Progress and Rationality in Science. Dordrecht:
Reidel, pp. 253-78.
Lakatos, I. (1968). "Criticism and the Methodology of Scientific Research Pro
grammes," Proceedings ofthe Aristotelian Society 69: 149-86.
(1970). "Falsification and the Methodology of Scientific Research Pro
grammes," in I. Lakatos and A. Musgrave, eds. , Criticism and the Growth
of Knowledge. Cambridge: Cambridge University Press, pp. 91-196.
(1971). "History of Science and Its Rational Reconstructions," in R.C. Buck
and R.S. Cohen, eds . , Boston Studies in the Philosophy of Science, Vol. 8.
Dordrecht: Reidel, pp. 91-136.
(1978a). "Changes in the Problem of Inductive Logic," in Mathematics, Sci
ence and Epistemology, Philosophical Papers, Vol. 2. Cambridge: Cam
bridge University Press, pp. 128-200.
(1978b). "Anomalies versus 'Crucial Experiments' ," in Mathematics, Science
and Epistemology. Cambridge: Cambridge University Press, pp. 211-23.
137
C H A PT E R 4
139
140
de Marchi
141
142
de Marchi
143
II
Lipsey entered the LSE as a research student in 1953 and two years later
was appointed an assistant lecturer. He had attended Robbins's seminar
as a student and continued to do so as a staff member. Robbins was
Professor of Economics, Analytical and Descriptive, and his seminar
was the focus of theoretical endeavor within the School. The style of the
seminar was the then common one of analytical dissection. Theory was
regarded as "a method of classifying the universe of possible cases"
(Johnson, 195 1 , p. 826; cf. Lange, 1944). Models were examined for the
realism of their assumptions and for internal consistency. 3 Results might
be tested for robustness (to assumptions change) but were rarely sub
jected to quantification using actual numbers. Overall assessment was
made by an economic-linguistic test- is it economically meaningful? - or
according to a largely undefined notion of relevance: Has light been
shed? Insight gained?
Lipsey and Archibald's famous 1958 paper, "Monetary and Value
Theory: A Critique of Lange and Patinkin," is in this genre. Arising
out of discussions of Patinkin's Money, Interest, and Prices in Rob
bins's seminar, their paper was a thorough examination of out-of
equilibrium processes versus equilibrium conditions in models with a
real-balance effect. Through a refinement of the stock-flow distinction,
the authors discover that "full" classical "equilibrium" conditions do
not require the presence of a real-balance effect and that models em
bodying this effect cannot "invalidate" Keynesian claims about less
than full employment "equilibrium . " Notice all the terms that require
quotation marks. Was this merely a terminological dispute- a "sorting
out," as Lipsey has referred to it (notes to the author, 7 June 1985)?
Was that how economics advanced? Lipsey felt at the time the implied
objection: " I was never quite sure if this exercise was of purely scholas
tic interest" (ibid. ) . What Lipsey and Archibald did conclude with
confidence was that "one model cannot be used to invalidate another"
(1958, in Thorn, ed. , 1966, p. 320) . This made it a question of circum
stances or problem context as to which model is preferable. And that
question, as they correctly pointed out, "is one for empirical rather
than a priori investigation" (ibid.).
Robbins, though, had little truck with empirical investigation. Read3 Cf. Harry Johnson's remark about Cambridge in the 1950s: "the examination of the
realism or unrealism of analytical assumptions as a test of the validity of a theory . . .
provided a basic technique of British theoretical discourse in the 1930s and on well into
the 1950s" ( 1978, p. 158) .
144
de Marchi
There is a striking similarity between Robbins's convictions on this point and Keynes's.
(See Keynes's letter to Harrod, 16 July 1938, in Keynes's Collected Writings, Vol. XIV,
p. 299).
This is the interpretation that was placed on Robbins's basically negative judgment on
setting aggregate demand policy by a commitment to full employment; see his Money,
Trade and International Relations (1971a, pp. 2-3 and ch. 11) and his Autobiography of
an Economist ( 197lb, pp. 230-1 ) . His enthusiasm for Patinkin's work, too, seemed to
be linked to the view that the logical consistency of the full employment equilibrium had
145
146
de Marchi
147
148
de Marchi
but they were eager to be satisfied. Although they were not wholly
gullible, then, it was only a matter of time before appropriate weaponry
would be found.
Enter Popper. He had been in the Philosophy Department at the LSE
since returning to Europe from wartime teaching posts in New Zealand
in 1946. His ideas were not a matter of daily conversation among the
economists, but neither were they entirely unknown to them. In the late
1950s there was an optional logic and scientific method course available
to Part One B . Sc. (Econ. ) students at the LSE, and some good under
graduates who took it- David Laidler and Maurice Peston among
them - brought to their economics teachers in the final year of specializa
tion bits of what Popper was saying.
In the spring or early summer of 1957 Lipsey gathered the group of his
concerned young colleagues into an informal seminar, outside of the
Robbins seminar. They met to discuss how to get out from under the
Robbins incubus. As it turned out, they soon found themselves receiving
instruction in Popper's ideas. For Klappholz had got to know Joseph
Agassi, a graduate student of Popper's, who later became his research
assistant. Agassi was keen to proselytize. He became mentor to the
economists and, as Lipsey recalls it, over about half a year they
"learned, and came to accept, much of Popper's views on methodology"
(letter to the author, 17 May 1985).
This almost suggests mass conversion, and the group may well have
regarded themselves as disciples at first. Certainly, after their initial
instruction in Popper's views, they felt charged up and eager to go out
and apply them to economics. At this point they constituted themselves
formally as the LSE Staff Seminar on Methodology, Measurement, and
Testing, with Lipsey as chairman. The title of the seminar reflects their
confidence that at last they knew how to advance beyond Robbins.
What was it that Popper gave the group? Three things stand out. First,
he imparted an impetus to formulate theory in testable form. This was
picked up by Archibald, as I have already indicated, but he brought
Lipsey and others along. Notice that in supplying this impetus, Popper
legitimized the theoretical enterprise but made theorists accountable.
Neither Hutchison nor Friedman could have given quite this feeling of
comfort to a group whose upbringing as economists, we must remember,
was essentially in a pure theoretical tradition.
Second, Popper did not just urge testable theory; he actually issued
some guidelines on how to pursue testing. As Steuer has put it (in
conversation, 1 July 1985), with Popper "you had a tight procedure to
guide you through the maze of the fact-theory mess." It is often
forgotten how central to Popper's thought is the notion of method-
149
ological conventions. Here are several of his obiter dicta that bear on
testing:8
A refutable theory is one that denies something.
More pointed predictions deny more and are therefore more
refutable.
Testing should be "severe. "
Stratagems for saving the hypothesis when evidence is unfavor
able are to be avoided.
Theoretical progress involves accounting for what we know and
explaining new facts.
These pieces of advice add up to a set of test criteria. They too go
beyond anything to be found in Hutchison or Friedman.
Third, Popper exalted criticism above the idea that we can ever
know with certainty. What could be better for a band of would-be
revolutionaries?
Despite their eager embracing of Popper, the LSE economists were
never undiscriminating in their commitment, and the ways in which they
drew on his work became more selective as their early broad concerns
gave way to more precisely formulated problems.
It is time to consider this next phase in more detail, which we shall do
by looking at some of the research spawned by the M2T seminar.
III
An overlapping but longer list, with references, is to be found in Mark Blaug's The
Methodology of Economics (1980), p. 19.
9 Items in my list of references that should be regarded as M2T-related work are given an
asterisk.
150
de Marchi
that the theory being considered did not yield a large number of testable
predictions (e .g. , pp. 221 , 239, 326) ; or that testing it seemed to be
extraordinarily difficult (p. 326) ; or simply that it yielded few implica
tions about questions of interest (pp. 265-6, 325 n. 2, and 326) . But even
conclusions like these were discoveries. Nobody before had undertaken
this exploration.
It is not clear whether, at the time, Lipsey was disturbed that economic
theory seemed to be so little "positive. " 10 At any rate, his zeal was undi
minished. The essential message remained: "economic theory is unable to
produce something out of nothing" (p. 239). If we know no facts or cannot
or do not measure parameters and relations and judge their stability
quantitatively, "we cannot use economic theory to make useful predic
tions about the real world" (p. 161). Methodologically speaking, that is to
say, there was no excuse for Robbins's attitude that we needn't bother to
measure just because so many things are happening and the numbers are
bound to alter anyway ( 1952 [1932] , pp. 107ff.) , and certainly no basis for
leaping to policy from pure theory. Furthermore, since a main reason why
so few predictions seemed to emerge from accepted theories was that the
theories themselves were so poorly informed - they assumed almost no
knowledge of the real world - this very poverty of quantitative informa
tion was a kind of self-condemnation. These critical points are well taken.
It remains true, though, that Lipsey's constructive harvest was scanty and
he could only express a plausible optimism: "Generally, the more things
we do know . . . the more likely it is that we shall be able to deduce
interesting and possibly unsuspected consequences from these facts" (In
troduction, 1963, p. 239) .
In being concerned with getting quantitative knowledge, Lipsey ran a
risk of being accused of pursuing measurement without theory. His
major empirical preoccupation was a detailed reconsideration of Phil
lips's work on the relation between unemployment and the rate of
change of money wage rates (Lipsey, 1960) . This work occupied the
seminar for a full six months. Lipsey thought it important "to quantify
Phillips' results," 11 discovering, for example, just how much of the vari
ance in nominal wages is explained by unemployment and its rate of
change. He also wanted to test Phillips's subsidiary hypotheses in a
systematic way. And he thought that alternatives to Phillips's explana10 We do know that when he set out in 1959-60 to learn some econometrics, he was naive
about the amount of existing quantitative work. He recounts that he set an assistant to
work cataloguing past work, and expected that this might take a month or two at most
(notes to the author, 7 June 1985).
11 Here too there was a political undercurrent. Supporters of Labour among M2T members
took Phillips's results to be a useful antidote to Robbins 's fear of cost-push inflation.
151
tion should be tested too ( 1960, p. 1). This sounds admirably Popperian
in its repeated stress on testing. In fact, however, Lipsey's stated ideal
order of proceeding was somewhat inductive and reminds one more of
Friedman: The researcher first discovers what phenomena need explain
ing and then builds a model to rationalize these data. Finally, further
implications or out-of-sample implications of the model are tested.
Lipsey's penchant (not only in this instance: see Brechling and Lipsey,
1963) was to put in most of the work at stage one, since to build up a
clear picture of the explicanda "a rather elaborate treatment [of the
data] is required" (ibid. ) .
There was also a theoretical modeling stage in this Phillips Curve
work. A model was inserted at Archibald's insistence (Lipsey, 1960, p.
12, n. 1). He was the analytical alter ego to the LSE group. Lipsey, it
need hardly be added, thoroughly welcomed his interventions. As he
himself explained the desirability of having a model (I paraphrase very
liberally from ibid. , pp. 12, 21 n. 2, 23):
1.
2.
3.
Lipsey here comes over as being not only utterly antipathetic to Rob
bins's style but also very commonsensical in his approach to method
ological issues. In this respect he seems as close to Friedman as to
Popper, and one is reminded that it was Archibald who stressed testable
theory, whereas Lipsey pushed ways to measure and to capture what is
in the data. The contrast between the essentially empirical and the essen
tially theoretical emerges very clearly in two subsequent and separate
papers generated by these two extending the early Phillips Curve work:
Lipsey and Parkin (1970) and Archibald, Kemmis, and Perkins (1974).
Archibald, for his part, worked much on the theory of the firm and on
market structure. At least some of his determination to write out fully
specified and testable models arose out of a disappointment with Fried
man and Chicago on these subjects - not with Friedman's message, note,
but with what seemed to Archibald to be a complacent attitude, possibly
152
de Marchi
stemming from too lightly assuming that orthodox price theory had been
tested and was firmly established (Archibald, 1959a, pp. 61, 63; 1961,
pp. 3, 5).
There is an unmistakable tendency in Friedman to slip into a priori
defenses of propositions such as profit maximization (Archibald, 196 1 ,
p. 3; Hirsch and d e Marchi, 1985) . A s noted previously, there i s as well a
distinctly un-Popperian theme in Friedman's essay, and one strikingly at
odds with his apparent acceptance of falsification. He speaks repeatedly
of our "confidence" in and evidence for a hypothesis, even of "our
confidence in [its] validity," and of testing as if its function is to verify ( ! )
validity (1953, pp. 9, 12, 22, 23, 28, 40) . Whereas Popper tests for truth
"by eliminating falsehood" - a negative construction that fits his critical
view of knowledge (Popper, 1963, p. 81) - Friedman seeks to strengthen
confidence by accumulating positive instances. Friedman, for all his
verbal deference to the asymmetry thesis, is not a critical rationalist.
Archibald picked up the point. In "Chamberlin versus Chicago" (1961)
he accused "Chicago" of having asked "what do we know?" without
having asked "the inquiring, scientific [Popperian] question: What don't
we know?" (p. 5).
We noted earlier Archibald's complaint that Friedman had failed to
explain how we test and what is meant by a good test (1959a, p. 61) . In
the context of the theory of the firm, the complaint could be made fairly
specific. How do we test a purely static theory at all? We cannot in
general observe equilibrium conditions; and in the absence of detailed
quantitative information about functions, which mostly we do not have,
we apply qualitative comparative statics. But to do this successfully
requires, at a minimum, that we specify in advance which things are to
be regarded as constants in the hypothesis under test, and that we ensure
that they are observables (1959a, p. 61 ; 1961 , p. 9) . 1 2 All of this presup
poses an explicit model, "loose" enough to allow change away from
equilibrium but "tight" enough to yield qualitative predictions that are
refutable. Archibald attempted in a series of papers on aspects of the
theory of the firm (1961 , 1963a, 1964, 1965) , and also in isolated at
tempts to reformulate the implications of welfare economics and the
marginal productivity theory of distribution (1959b, 1960) , to state mod
els with testable consequences. He accused Friedman, in not having
bothered to do likewise, of having actually discouraged "that sceptical
reexamination of the allegedly obvious that is the prerequisite of prog12 Archibald felt attracted by Papandreou's stress on structure in his approach to testing,
but despaired of the very abstract way in which he was content to talk about the
environment ("social space") within which a test might be specified (Papandreou, 1958,
1963; Archibald, 1963b).
153
Maximizing Models" (1965, in Morishima, 1973); see esp. p. 76 and n. 7. See also
Section IV, this chapter.
154
de Marchi
155
were at pains to show that making oneself vulnerable to criticism is all there is. There is
no guarantee that good methodology will issue in good economics, although one has no
alternative but to be critical. The others in the group were shocked by this at first.
156
de Marchi
157
The probem may be put this way. Frequently we would like to be able
to conduct an experiment in which we alter one parameter (call it a) of a
system in equilibrium and emerge with a clear prediction about the
direction of change in variables of interest to us (x1 ,
, xn). It turns
out that we can never sign more than one of the n expressions dx/da
unless the cross-partials among the variables are either zero or we have
specific information about them. Moreover, we cannot even sign that
one element if parameter a appears in more than one equation (Archi
bald, 1965, in Morishima, 1973 , pp. 71-4). This second condition results
in a form of the "parametric paradox."
Think of a simple case of "second best,'' an example Archibald him
self used. 1 7 We wish to alter a tax rate kg and observe the impact on a
utility function. Knowing the sign of dU/dkg will tell us whether the tax
rate should be raised or lowered.
Now
dU
U;
dkg '
dkg L..
1 1
but we have two problems: the sign of
u.
' dkg
is unpredictable unless interfirm or interindustry effects of the change in
kg are zero or known in detail; and not even dx/dkg can be signed if the
parameter kg appears in more than one equation. We know that it will,
however, in all cases involving some degree of monopoly control over
price. Let there be a wedge, k;, between price and marginal cost in the
ith industry- this is the departure from Pareto optimum conditions that
puts us in a second-best situation to start with. It will not do, as a
welfare-improving policy change, simply to impose the same k in all
other industries. For if in the ith industry there is monopoly power, the
profit-maximizing k; will be a function of all the other k's (pp. 77-9).
Samuelson, in introducing qualitative comparative statics in its mathe
matical form, had argued that there is no serious loss of generality
because of the conditions necessary for unambiguous signing (Founda
tions, p. 33; cited in Archibald, 1965, in Morishima, 1973, p. 71). Archi
bald's explorations showed that Samuelson was mistaken, and in a way
that had very serious consequences for deriving testable predictions by
this method. There are almost no unambiguous qualitative predictions,
or predictions "in general,'' when one is trying to assess the effect of a
parameter change, if the change also alters other parameters. 1 8
17
18
158
de Marchi
159
ventions are inevitably somewhat arbitrary; but more than that, they
represent a marked departure from refutability in the sense of a relation
between a universal and a singular statement (1966, pp. 280, 287, 190) .
However, he held that it is not economic inquiry that should be jetti
soned as unscientific, but Popper's demarcation rule that should give
way. "My own judgment is that many of the irrefutable hypotheses in
economics are important, that they are incurably irrefutable for good
and fundamental reasons [here Lipsey's kinds of reasons] , and that the
activity of comparing them with observation is useful (too practically
useful to be acceptably called metaphysics)" (p. 279).
Archibald advocated demarcation by comparison. We should com
pare the predictive success of extant rival theories. If we exhaust these
critical experiments, naive hypotheses rivaling the more successful
theory can be constructed, subject to sensible restrictions such as that
they have no fewer degrees of freedom than the serious rival (p. 291 ) .
Archibald's proposed demarcation rule was the following: "that we call
a statement - or hypothesis - scientific if we may, at least in principle,
compare its probable truth or falsity with that of another statement by
appeal to observation (reference to facts)" (p. 293) . Having empirical
content thus meant "being potentially comparable" (with preexisting
rivals, with constructed rivals, or with the null hypothesis and, in every
case, with facts).
This new rule carried serious consequences. For Popper, "there are
sea-serpents" is confirmable but not refutable (we may always find one if
we maintain the search) , and hence metaphysical. For Archibald, "there
are sea-serpents" is a comparable hypothesis (pp. 293, 294) . Similarly,
"inductive confirmation," a phrase that would horrify a Popperian (cf.
Boland, 1982, ch. 1 ) , seems, on Archibald's new view, a perfectly sensi
ble way to convey a sense of the comparison procedures actually fol
lowed by scientists and inherent in statistics (pp. 294-5).20
It is not clear that Archibald's new rule represents quite the advance
upon Popper that he intended (or that Popper would seriously ob
ject!21 ) . It embraces all the conventions of statistical hypothesis testing
20 Archibald pointed out that even the familiar R2 is a comparative measure: "it tells us
21
how much better our fitted relationship predicts the independent variable than does its
own mean" (p. 282).
As Blaug notes, Archibald was in fact "knocking against an open door" (1980, p. 122, n.
30). It seems almost as if Archibald was looking for a reason to break with Popper. For it
is clear that Popper himself was busy with ways to reconcile falsifiability and probability.
He writes: "How is itpossible that probability statements- which are not falsifiable - can
be used as falsifiable statements? (The fact that they can be so used is not in doubt: the
physicist knows well enough when to regard a probability assumption as falsified.)"
(1959, p. 204). In particular, he was concerned to argue that "accepted basic statements
may agree more or less well with some proposed probability estimate; they may repre-
160
de Marchi
161
quite uncritically" (p . 293). Lipsey's way of saying the same thing (in
notes introducing the second edition of his textbook) was more blunt:
"I have abandoned the Popperian notion of refutation" (Introduction,
2nd ed. , 1966, p. xx) . A little further on he writes: "The choice is not
one between theory and observation but between better or worse theo
ries to explain our observations" (p. 14) .
v
How close did members of the seminar get to resolving the concerns of
their pre-Popperian youth? And what does their experience reveal
about why Popper so captivated them initially, but not for long? A brief
recapitulation of developments will help us toward clear answers.
1.
2.
3.
4.
162
de Marchi
5.
The logical next step for the LSE group, having substituted compari
son of theories for confrontation between theory and observation,
would have been to adopt Lakatos's notion of comparisons between
connected sequences of theories. Lakatos was at the LSE from 1959 on,
and some of the economists knew of his remarkable Proofs and Refuta
tions from seminar presentations prior to its publication (in article form)
in 1963-4. Lipsey, however, was preoccupied with his textbook and had
got involved in governmental advising. Archibald was trying to absorb
the discovery that economic theory does not lend itself to Popper's
refutability. In short, the moment was not right for grasping the signifi
cance of Lakatos's startling tale of how progress occurs in mathematics.
Thus Popper gave the LSE group a procedure, but it led nowhere.
There were in some cases logical barriers to falsifiability in economics.
There turned out to be a need for detailed information of a sort whose
availability was not guaranteed in cases where the qualitative calculus
was subjected to enough restrictions to make it yield predictions. Over
and above these problems, Popper's refutability quickly came to seem
too rigid for statistical propositions.
That Popper's ideas flourished at all among the LSE economists is
thus somewhat surprising and may be attributed in large measure to
circumstances that were quite time- and space-specific. Firstly, a revolt
23
163
164
de Marchi
165
166
de Marchi
PART III
CHAPTER 5
I
The preceding title and subtitle, as formulated by the authors of the
draft program, seem to correspond reasonably well with what I have
discovered that I want to say. But a little further clarification may not,
perhaps, be superfluous.
Any significant case for adopting, or "striving after," a methodologi
cal principle must rest ultimately on some ethical, moral, or political
argument (as Popper emphasized in the debate with Habermas and
elsewhere). Insofar as methodological principles are critical principles,
then, the criticism they contain, or imply, must rest on some kinds of
normative standards or presuppositions. Moreover, I would like to em
phasize, at the start, how closely the real questions of economics, much
more than is the case with any other subject, scientific or otherwise, are
constantly and intimately involved with controversial political, indeed
party-political, issues, an involvement that gives rise to persistent and
perennial problems in applying economic knowledge to policy.
Anyhow, it is with such ultimate ethical or political meta-arguments
for the (or a) falsification and falsifiability principle that this chapter is
concerned.
II
I want to start from, and take as far as it will go, Professor Johannes
Klant's analogy or comparison between the methodological principles of
economics and the "rules of the game. " Personally, however, I would
not want to emphasize at all appreciably the two definite articles. It may
seem that the main problems arise from the fact that a number of differ
ent games, or types of gamesmanship, are going on at the same time
(some more like chess, and some more like a game of academic
oneupmanship) . Anyhow, a game, including the "Scientific Economics
Game" or the "Economic Theory Game," is not given, in a particular
form, so that its rules and codes can be discovered and formulated if we
search and analyze far enough.
169
170
Hutchison
So, before one can usefully discuss the rules of the game, one must
start by asking about the aim of the game (or of playing the game) or
about why the players are playing it (beyond, that is, simply trying to
win).
The aim of a game, as a purely amateur pastime, is, presumably,
simply the enjoyment of the players. For many or most players, for a
game to be enjoyable, a code of more or less agreed-upon rules is
necessary. To employ Mark Blaug's analogy of playing tennis with the
net down (and possibly with no lines to demarcate the court) , this is
widely held to be less enjoyable than playing with a net and court in
accordance with the standard dimensions and demarcations, and to be
more enjoyable than with no net or with a net, say, twenty feet high.
Many or most players of games find that "anything goes" does not,
generally, make for enjoyment.
When, however, a game becomes a professionalized spectator sport,
and the players are paid to play, the immediate enjoyment of the players
may not be the sole or most important factor, or justification, for what
ever rules are adopted. The rules of the game will have to be formulated
with regard to the enjoyment of the paying customers, spectators, or
televiewers, for whose enjoyment considerable modifications in the
rules may be required - if not in the basic code, at any rate in what may
be called the "by-laws" for particular matches. Such modifications in the
rules regarding, for example, the timing and length of matches, designed
to increase the enjoyment of the paying public, might well be quite
unacceptable to unpaid amateurs.
III
171
172
Hutchison
173
174
Hutchison
then arises as to what this implies regarding any rules of the game, or
any code or demarcations, for whatever kind of activity economists are
engaged in. Politicians, journalists, and sometimes ordinary citizens are
concerned with economic policies and sometimes present their views at
great length , orally or on paper. But politicians, journalists, and ordi
nary citizens, when pronouncing on economic policy, do not have to
submit to rules or demarcations other than those laid down in the laws of
their country regarding libel, national security, and so on. So why should
there be any different or additional rules for economists beyond those
obtaining for politicians, journalists, and ordinary citizens? Why, on the
subject of economic policies and processes, shouldn't "anything go" for
economists j ust as for anyone else?
The answer to such a question obviously depends on whether econo
mists have made, or do make, any claims that their conclusions on
economic processes and policies generally differ, in some significant
qualitative respect, from those of noneconomists. If economists re
nounce any such claims regarding their pronouncements on economic
policy, as compared with those of politicians and journalists, then obvi
ously no additional rules or code need be applicable.
I would, however, maintain that over the last hundred and fifty years
or so, since, for example, the time of James Mill, many economists have
put forward considerable claims for what they have had to say on eco
nomic policies, in particular, as possessing a certain measure of author
ity. So what can be the source of such authority? It would not seem that
it can justifiably be regarded as deriving simply from the much larger
amount of time and effort that they have spent on the specialized study
of economic processes and policies. A kind of labor theory of value
hardly seems to work. The proof of the puddings cooked up by econo
mists cannot lie simply in the number of socially necessary hours of labor
applied by the cooks.
Moreover, as already maintained, economists do not only make
claims for the authority of their pronouncements; they also apply for,
and obtain, financial support for their research. The further question
follows, therefore, as to what kind of claims for their subject economists'
applications for research funding are based on, or what kind of rules,
discipline, or demarcations are claimed to obtain. It is possible that
economists' applications for financial support might not have been as
successful as they have been if it had been evident that the kinds of
claims for the subject on which these applications were based resembled
those of physicists and natural scientists much less than they did the
kinds of claims relevant for applications for sponsorship from artists,
175
176
Hutchison
177
It is not, however, for one moment being suggested that any legal
enforcement or sanctions, such as obtain with regard to the marketing of
drugs, are possible or desirable with regard to the testing of economic
theories and predictions. Needless to add, the falsification and falsifi
ability principle is not comparable with legislation, but must be put
forward simply as a voluntary code, upheld, as far as it can be, by
seriously critical opinion.
x
178
Hutchison
179
XI
180
Hutchison
181
C H A PT E R 6
Sir John Hicks's career in economics spans more than fifty years, in the
course of which he has published fourteen books, including six collec
tions of almost one hundred essays. In some of these, particularly Value
and Capital ( 1939) and A Revision of Demand Theory (1956), there are
hints of Hicks's general attitude to the nature of economics, but it is only
recently that he has become more explicit about his views on the method
ology of economics. A 1976 essay on ""Revolutions" in Economics"
voiced doubts about the applicability of Lakatos's philosophy of science
to economics; the opening and closing chapters of Causality in Econom
ics ( 1979) threw up similar doubts about the wider question of empirical
testing in economics; and, finally, an essay written in 1983 with the
pointed title of "A Discipline Not a Science" decisively parted company
with all varieties of empiricism, Popperianism, falsifiability, or call it
what you will, in economics.
1.
After observing that economic theories can offer no more than "weak
explanations" for economic events because they are always subject to a
ceteris paribus clause - a feature that he appears to believe is unique to
economics- Hicks concludes:
it becomes clear that they cannot be verified (or falsified) by confrontation with
fact. We have been told that "when theory and fact come into conflict, it is
theory, not fact, that must give way" [a quote from R.G. Lipsey] . It is very
doubtful how far that dictum applies to economics. Our theories, as has been
shown, are not that sort of a theory ; but it is also true that our facts are not that
sort of fact. ( 1983, pp. 371-2)
Economic facts, he goes on to say, are not the data produced by rep
licable, controlled experiments but the testimony of observers in histori
cal time. For that reason alone, economic facts are frequently defective.
A lack of consilience between theory and fact, in economics (when that cannot
be ascribed, or readily ascribed, to the weakness of the theory . . . ) is most
commonly due to a lack of correspondence between the terms in which the
183
184
Blaug
theory runs, and the terms in which the fact is described . . . . When that clash
occurs, it may be that theory should be improved, so as to run more closely in
the terms in which the relevant facts are commonly described but it may also be
that the description of the facts should be improved, so that we may think about
them more clearly. I believe it is this last . . . which is the special function of
economic theory. Though the concepts of economics (most of the basic con
cepts) are taken from business practice, it is only when they have been clarified,
and criticized , by theory, that they can be made into reliable means of communi
cation. Now once one recognises that this is what economic theory very largely is
doing, one sees that the use of models, which are themselves quite unrealistic,
may be extremely defensible . . . . I might indeed go on to maintain that the
Value and Capital model, of General Equilibrium under Perfect Competition,
can be defended in much the same way . . . it is a laboratory, in which ideas can
be tested.
185
Keynes never did tell us how one would know that a conclusion drawn
by an economist was correct- and neither does Hicks.
2.
186
Blaug
187
of some of the great economists of the past. Thus, the classical concep
tions of flexible wages and fixed coefficients of production are defended
as realistic in the circumstances of the day (Hicks, 1967, p. 147; 1979, p.
47), and the decline of Ricardian economics in the second half of the
nineteenth century is attributed to the waning importance of the scarcity
of land (Hicks, 1983, p.38) . In the same way, Keynes's assumption of an
exogenous money wage is deemed to be justified by the facts of the
1930s (Hicks, 1977, p. 81), but changes in methods of collective bargain
ing and in monetary institutions have rendered Keynes's analysis increas
ingly obsolete (Hicks, 1983, p. 38) . Because the "facts" of economics are
subject to continual, nonrepeatable change, economic theories are for
ever doomed to be valid only for the historical circumstances in which
they are born. Therefore, if there are "revolutions" in economics, they
are merely changes of attention because what Kuhn called "loss of con
tent" in successive "paradigms" is considerable. It would seem, there
fore, that every "well-chosen" economic theory is true for its time
(Hicks, 1981 , p. 233), and the job of the historian of economic thought is
to make sense of these well-chosen ideas in the light of their historical
context. Such "relativism" is, of course, perfectly defensible if only we
were given some help in distinguishing the well-chosen from the ill
chosen theories of the past. If Ricardo was correct for his time, was
Malthus also correct for the same time even when he argued in diametri
cal opposition to Ricardo?
3.
Hicks has always upheld the distinction between positive and normative
economics, and of course, a great deal of his output has been concerned
with normative economics of the cost-benefit variety in which problems
of true or false take second place to problems of good and bad. Neverthe
less, his writings on questions of positive economics seem to me to suffer
from a continuous unwillingness to face up to the question of how it is
that we ever discover whether a piece of positive economics is true or
false. On the one hand, we are asked not to regard economics as a
science but only as a discipline, a form of applied logic without any
empirical content. Without worrying too much about the honorific label
of "science," Hicks's object in calling economics a discipline rather than
a science is, as we have seen, to discourage attempts to knock out
economic theories by empirical testing. At the same time, we seem to
come equipped, according to Hicks, with considerable background
knowledge about the workings of economic systems that enable us to
know when assumptions about economic behavior or the operation of
188
Blaug
189
Causality in economics
190
Blaug
191
192
Blaug
As Hicks said himself in later years: "They gave me a Nobel prize [in 1972] for my work
on "general equilibrium and welfare economics" . . . work which has become part of the
standard literature. But it was done a long time ago, and it is with mixed feelings that I
found myself honoured for that work, which I myself felt myself to have outgrown"
(1977, p. v).
193
By way of conclusion
As Lindbeck (1985, p. 42) notes: "it is indeed remarkable how strong the intellectual
influences of these four theorists (Hicks, Samuelson, Arrow, and Debreu] has been on
our profession - by influencing not only the choice of issues and methods but also the
'style' of analysis and exposition."
194
Blaug
Yes, indeed, some would say, and this only goe to demonstrate the
sterility of prescriptive Methodology with a capital M and the impor
tance of descriptive methodology with a lowercase m: We need to study
what economists actually do and not what they should do (McCloskey,
1983). It is, however, impossible rigidly to divorce the two, and even
those who profess to be totally open-minded about the best practice in
economics find themselves driven to criticize actual practice (Caldwell
and Coats, 1984).6 Besides, I have been preoccupied in this chapter less
with what Hicks does than with what he says he does. What Hicks says is
that economic theories cannot be tested for their truth value, and yet he
does not deny, or at least he does not appear to deny, that economic
theories have truth value: They are not merely conventional instruments
for organizing our background knowledge but genuine causal explana
tions of what does happen and why it so happens. Moreover, some
economic theories are said to be better than others even though we are
somehow precluded from ever comparing them in terms of their power
to account for empirical evidence. There is nothing wrong with such a
position, but it does lead inevitably to the conclusion that economic
theory is a subject precisely like pure mathematics, an intellectual game
to be justified in its own terms. And that position is incompatible with
the notion of economics as the handmaiden of economic policy, which
Hicks has consistently upheld. I conclude, therefore, that it is impossible
to extract any coherent methodology of economics from the writings of
Hicks.
References
Addison, J.T., Burton, J . , and Torrance, T.S. (1984). "Causation, Social Sci
ence and Sir John Hicks," Oxford Economic Papers 36(1):1-11.
Blaug, M. (1980). The Methodology of Economics. London: Cambridge Univer
sity Press.
Caldwell, B .J . , and Coats, A. W. (1984) . "The Rhetoric of Economics: A Com
ment on McCloskey ," Journal ofEconomic Literature 22(2):575-8.
Coddington, A. (1979). "Hicks's Contribution to Keynesian Economics," Jour
nal of Economic Literature 18(3): 970-88.
Elster, J. (1983). Explaining Technical Change. Studies in Rationality and Social
Change. Cambridge: Cambridge University Press.
Helm, D. (1984). "Predictions and Causes: A Comparison of Friedman and
6 Both in his 1983 paper and in his 1985 book, McCloskey is much concerned with the bad
rhetoric of significance tests in econometrics, arguing for Bayesian in preference to
Neyman-Pearson methods. Actually, if the methodology of falsificationism, or what he
calls "modernism," is as silly as he makes out, it is difficult to see why he is so concerned
about econometric practice.
195
PART IV
C H A PT E R 7
My thanks go to the discussant and participants at the Klan! conference and to Margaret
Schabas for their helpful comments on the ideas expressed in this chapter. Much of the
historical material in the chapter is drawn from an extended study of the history of
econometrics, initially funded by a grant from the ESRC (Grant 6727) and reported in
detail in Morgan (1984).
199
200
Morgan
Empirical models
While the methodological point of view is given, for example, by Keynes (1891),
Jevons's work gives the prime practical example of the separation of the two tools.
2 In discussing this aspect of empirical model building, I have been much influenced by
Nancy Cartwright's provocative book How the Laws of Physics Lie (1983), in which she
argues that the fundamental theoretical laws of physics are true in some fundamental
sense but that for any given real circumstance, they are false.
201
ing Boyle's law, the chemist has to rely on a set of empirically derived
approximations for each substance, and these have been gathered over
the years by applied chemists. Similarly, engineers have developed the
relevant approximations for physical theories to be applied, for exam
ple, in building a bridge.
In economics, finding an empirical model to match the theory in
volved making the theory operational in several ways:
1.
2.
3.
202
Morgan
At the same time in the late 1920s and the early 1930s, there was a
discussion within econometrics about how to deal with the time dimen
sion of the theoretical relationships in these models. Econometricians
adopted different ways of introducing dynamic aspects into empirical
models. One method involved introducing time lags of various lengths
into the model to capture the delays between causes and effects. An
other path was to add new forms of the theoretical terms into the model
to capture a more complex behavior than simple response to current
prices. These were not necessarily independent approaches; their use
would often depend on whether the starting point was mathematical or
statistical work.
The cobweb model provides a good illustration of the first path to
dynamic models. In the cobweb model, price responds to this year's
quantity demanded, while supply is determined by the previous year's
price. This model was first used by Moore in 1925 to model what he
called a "moving equilibrium" of demand and supply. Moore gave no
particular reason for this choice of model, although he might have ratio
nalized it on the basis of knowledge of the particular market. The model
formulation was analyzed and applied again in the early 1930s in three
different publications by Schultz, Leontief, and Tinbergen (Ezekiel,
1938). Tinbergen (1930) , for example, expressed reservations about the
model because he saw that under a certain range of parameter values the
model implied a path diverging away from an equilibrium solution, giv
ing an exploding cobweb pattern. The suggestion that such a divergent
path could form the basis of a crisis model was made by Evans in 193 1 ;
and i n 1938 Ezekiel proposed it as an alternative explanation of the
malaise of the 1930s economies: that is, the economies had adjustment
processes that led away from full employment. The lagged adjustments
or responses in crucial relationships that characterized the cobweb
model could also, in certain circumstances, produce cycles in the vari
ables. This version of the model entered econometric business cycle
literature in the early 1930s- for example, in Frisch's propagation and
impulse model of 1933 - and soon became an essential element in mathe
matical theories of the business cycle.
In this story, the cobweb model started off in the mid-1920s as an
approximation for the time relationships of an empirical demand model,
but by the mid-1930s it had become an essential element of empirical
cycle models and thence of cycle theory. This rapid assimilation , from an
empirical approximation to an essential element in theory, was no doubt
due to the integrated nature of econometrics during this early period,
when those who were undertaking empirical work were also responsible
203
204
Morgan
make the model relevant to the particular real circumstances. The latter
involved several possibilities. It could mean introducing elements to
match some general observed features of data, such as cycles, or more
specific observed features such as those seen in the market for a particu
lar commodity. Above all, an empirical model must be one that is mea
surable. This translation of theories into empirical models that can be
fitted to statistical data was described earlier as making theories "opera
tional. " This term was used deliberately, for Percy Bridgman was the
only philosopher of science whose work was referenced by those in
volved in the econometric work of the period. Both Roos (1934) and
Schultz (1938), important figures in econometrics, expressed admiration
for Bridgman's ideas and used their books to try to introduce opera
tionalism into economics. As Klant (1984) points out, Bridgman's ideas
on operationalism had considerable influence on economists in the
1930s. 3
These early econometricians were open and direct about their desire
to translate theory into empirical models. Within the demand field, for
example, there was some discussion about the relative roles of theorists
(meaning nonmathematical economists) and of econometricians in devel
oping such models. The econometricians blamed the theorists for their
failure to develop adequate models to cope with the problem of time,
and the consequent necessity for their own development of such models
(Gilboy, 1930; Stigler, 1939). Some economists clearly resented what
they saw as statisticians interfering in economic theory. For example,
Keynes's famous critique (1939) of Tinbergen's work accused him of
plucking lag lengths out of midair. Such criticisms were unfair, for as any
econometrician of the period well knew, theorists just did not supply
such information in their theories.
II.
Satisfactory models
205
Their aims and criteria ranged over the following set of ideas:
1.
2.
3.
4.
5.
206
Morgan
periods and their standard errors; and (4) the degree of statistical expla
nation achieved by the model (corrected R2) . He described the empirical
models that were most satisfactory according to these criteria as "fairly
good demand equations. "
Whitman's comparison o f alternative empirical models did not involve
choosing between different fundamental theories but rather between
different empirical models of the same theory involving different dy
namic processes. He was not seeking to prove one of these models as the
true one, but to find the most satisfactory of the different versions. His
comparison used a relatively sophisticated assessment process for the
period, for it tested the models' consistency with both theory and statisti
cal data in a coherent way. Those that performed better on the tests
were better models, but for Whitman they still held the interim status of
"hypotheses" pending further econometric investigation.
Tinbergen was a product of the more mixed European tradition, also
pragmatic in its way. In his macroeconometric modeling of the late 1930s
(1937 and 1939) Tinbergen was, as always, modest and cautious in his
work, though he did claim to "test" theories. He stated his belief on
testing clearly: that statistical work cannot prove a theory correct,
though it can go some way toward proving a hypothesis incorrect. Be
cause of the difficulties involved in building and estimating a large
macro-model, Tinbergen used graphic means to help choose the vari
ables in the equations and assemble the empirical model. This method
seemed to involve ad hoc adjustments and manipulations, but the work
gained rather than lost from Tinbergen's willingness to be open about
the problems involved. One of his aims was to use the final macro-model
to simulate different policy options and to assess which policy would be
most useful in reducing unemployment. He found that he could indeed
use hi empirical models for this task.
Tinbergen was one of the few econometricians to develop a well
defined testing program, using every possible criterion that had been
developed in econometrics by the late 1930s to assess his empirical mod
els. First, he considered the consistency of his results with economic
theories in respect to the individual relationships and then in terms of
the implied dynamic process of the system. He used statistical criteria to
assess how well his model characterized the data set and to test whether
assumptions about the technical properties required by the statistical
method held true. (For example, he tested for constancy of the equa
tions' parameters over different subperiods. ) He also used historical
criteria to see whether his empirical model explained certain peculiar
historical features of the period such as the 1929 Great Crash. The
information from these testing criteria was used to help him to reformu-
207
208
Morgan
III.
Testing
209
the recent listing of dynamic model formulations by Hendry and Richard (1982, Section
3).
See, for example, Hendry (1983) for a very recent, relatively nontechnical discussion of
empirical models and their evaluation that is close to the ideas discussed here. An earlier
example of a similar approach is given by Carl Christ's text (1966).
210
Morgan
211
bilistic Revolution, Vol. II: Ideas in the Sciences. Cambridge, Mass: MIT
Press.
Roos, C.F. (1934). Dynamic Economics. Cowles Commission Monograph 1 .
Bloomington, Ind.: Principia Press.
Samuelson, P.A. (1947). Foundations of Economic Analysis. Cambridge, Mass.:
Harvard University Press.
Schultz, H. (1928). Statistical Laws of Demand and Supply with Special Applica
tion to Sugar. Chicago: University of Chicago Press.
(1933). "A Comparison of Elasticities of Demand Obtained by Different
Methods," Econometrica 1 : 274-308.
(1938). The Theory and Measurement of Demand. Chicago: University of
Chicago Press.
Smith, B . B . (1925) . "The Error in Eliminating Secular Trend and Seasonal
Variation before Correlating Time Series," Journal of the American Statisti
cal Association 20:543-45.
Stigler, G.J. (1939) . "The Limitations of Statistical Demand Curves," Journal of
'
the American Statistical Association 34 :469-81 .
Tinbergen , J. ( 1930) . "Bestimmung und Deutung von Angebotskurven," Zeit
schriftfiir NationalOkonomie 1:669-79. (English summary, 798-99.)
(1933). "The Notion of Horizon and Expectancy in Dynamic Economics,"
Econometrica 1 : 247-64.
(1937). An Econometric Approach to Business Cycle Problems. Paris: Her
mann.
(1939). Statistical Testing of Business Cycle Theories, Vols. I and II. Geneva:
League of Nations.
Whitman, R.H. (1936). "The Statistical Law of Demand for a Producer's Good
as Illustrated by the Demand for Steel," Econometrica 4:138-52.
C H A PT E R 8
213
214
Weintraub
215
216
Weintraub
(1981a, 1981b), and the third paper was written by McElroy (1985)
alone. 1
Consider the allocation of time spent by the female in the labor force,
the amount of money spent on household goods, and the consumption
patterns of households. There had been two distinct themes in the pre1980 literature; first, demand was associated with an agent called a
"person," so that demand for an automobile was the demand by a per
son for an automobile. Alternatively, one could specify that demand is
generated by households as agents (Becker, 1981). In the former case,
one has to perform aggregation to mesh joint demands by married cou
ples, whereas in the latter case one must assume that household prefer
ences are well defined. Both approaches are standard, though each has
its difficulties, and, of course, each is only partially corroborated.
The models that each approach generates for testing are demand
models, or demand systems, which,are based on the agents, their prefer
ences, their constraints, and the assumption of optimization. Each pro
duces functions of the form X; f;(p;, w, I) , where X; is the quantity of the
ith good demanded, P; is its price, and w and I are wage and nonwage
income, respectively.
Suppose we are interested in the hours worked by married female
nonheads of households, and let X; be leisure for such a female. Whose
wage is w and whose nonwage income is I? If f is a household demand
function, so that both male and female wage incomes are combined,
whose preferences are maximized?
Paper 1 sets up a model of individual preferences of males and females
who choose x (x0, x1 , x2, x3 , x4) with prices p;, i 0, . . . 4. Here x0 is a
pure household good, a public good to the household, x1 is the husband's
market good, x2 is the wife's, x3 is the husband's leisure, and x4 is the
wife's. Thus the wife is concerned with x1 (x0, x2, x4), for instance, and
she is constrained by I1 + p4 (T x4) PoXot + px2, where T is total time
available.
The issue is how the preferences and constraints of the husband and
wife are interrelated in any household demand system that has testable
implications. McElroy and Horney specify a Nash-bargaining theory in
which the game theoretic structure of bargaining is used to model the
interdependent optimization problem. They produce, in paper 1 , a de
mand system of the form X;
h;(p, Im, I1) with i
0, 1 , 2, 3, 4. The
important point is that the functions h; are restricted not by the usual
=
1 I should declare my interest here. McElroy is a friend and colleague, and I served on
Horney's doctoral examination committee. My reason for using these papers is that I am
familiar with them and their origins, and I have had a chance to check my understanding
of them with their senior author.
217
218
Weintraub
219
seems to bring out the worst tendencies in economists and philosophers alike. There is,
for example, the appalling view that the facts can be used to test theories for truth or
falsity.
3 This move is associated with Gary Becker and will, one suspects, result in Becker's
being awarded the Nobel Memorial Prize some day.
220
Weintraub
221
theories (Popper, 1959, 1972) . Those who hold this view are skeptical of
applied economics. This skepticism is misguided. The issue of theoreti
cal progress requires the development of excess content in theories. It is
the role of applied empirical work to assess the nature of that content.
Because it is, in contrast to the Popperian notion, so easy to falsify
excess content, applied workers are legitimately wary of tests that theo
ries do not pass. What is more difficult is to develop appropriate tests,
using imperfect data, for appraisal of the theory's theoretical progress;
or, put another way, the role of empirical work is to appraise the move
from M(t) to M(t + 1) to M(t + 2). Attempts to falsify the excess content
of successor theories are but one thread of such appraisals. They must be
joined with attempts to develop data sets on which the excess content
can be identified, to present corroborations of some of that excess con
tent, and to perform all such tasks from the point of view of one who
does not believe that th excess content can be corroborated. That is,
the empirical worker is bound by a belief that the excess content is false,
and thus falsifiable, so that the claims of progress are false.
In the second McElroy- Horney paper, it is clear that months of work
went into the job of finding and refining a data set adequate to frame an
appraisal of the excess content of the Nash demand system. The NLS
has an immense data file, and the authors had to restrict the set to
married households with matched demographic characteristics. Their
belief that there might be separate marriage gains for whites and non
whites led to the development of separate white and nonwhite subsam
ples. They ended up with a white sample of 485 persons and a nonwhite
sample of 186. Their Table A . 1 presents descriptive statistics for the
nineteen variables that are associated with the variables in M(t + 2) .
Some of the variables could not be observed from the sample data, like
the wife's nonwage income, and so that variable required the authors to
impute its values from the available data set. In that particular case,
McElroy and Horney used a different NLS subsample of 144 divorced
women to impute a share of the family nonwage income to the wife,
since that income is a variable important to the appraisal of excess
content. Thus the stage of analysis associated with the construction of
the data set is nontrivial; it is not sufficient to state, as some falsifica
tionists do, that in "good science" facts falsify theories and economists
do not look for falsifying facts, so economics is not good science.
McElroy and Horney develop the theory M(t + 2), and ft1, in such a
way that it can be confronted with the data. Error and the stochastic
nature of the theory require that there be a stochastic specification of
M(t + 2). Statistical tests require further simplifications, which the ana
lysts make, about the distributions of the stochastic elements of the
222
Weintraub
model. When this is done, the model is estimated and the hypotheses
about the estimated parameters are tested. Note that stochastic specifica
tion, estimation, and testing are all done prior to falsification and cor
roboration, which is one reason why so many applied economists are
bemused by philosophers who focus on such a limited point. The idea
that facts can falsify theories, and that the role of applied work is to
produce the facts thatfalsify the theories that the theorists create, is simulta
neously to misunderstand facts, theories, tests, and falsification. The activ
ity of applied economic analysis is appropriately characterized as "devel
oping evidence to appraise excess content. "
McElroy and Horney, in paper 2, do not falsify the theory M(t + 2).
Neither do they corroborate it. They instead indicate that there is some
evidence that partially corroborates some of the excess content of M(t +
2). This result itself is presented with extreme caution, however, since
many modifications were required in the model to allow the available
evidence to confront the excess content of M(t + 2) . The authors' conclu
sion to paper 2 is an exemplar of the best Popperian science, a point of
view that is skeptical about claims of progress: "This paper . . . repre
sents a start in evaluating the empirical payoff to a bargaining ap
proach . . . (but] alternative data sources, bargaining models, systems of
demand equations, and estimation techniques need to be examined be
fore the empirical payoff on bargaining models can be judged" (p. 17).
In paper 3 a further step is taken in evaluating the bargaining ap
proach. That is, the bargaining model is applied to a new problem, that
of household membership. Specifically, where some earlier work had
treated household membership as an exogenous dummy variable in
studies of youth earnings, paper 3 examines the joint decision on
wages, consumption, and household membership. "The estimated
model agrees with the predictions of the Nash model and shows a
strong interaction between household membership and the work behav
ior of youths. In contrast, at least for these data, if either household
membership or work status is treated as exogenous one is led to the
false conclusion that household membership and work behavior of
youth are unrelated" (p. 294).
In short, the Nash model as M(t + 2) has excess content compared with
M(t) and M(t + 1 ) , content associated with the household membership
choice for young people; the first part of paper 3 explores that excess
content, and the second part of that paper attempts to corroborate it by
the use of maximum likelihood estimates of the implied model (a trino
mial probit model) using matched samples of parents and sons from three
NLSs. The results of that empirical analysis show that "at a sufficiently
low offered wage the youth lives with his parents and does not work.
223
When the offered wage rises above his asking wage, the youth remains in
his parents' household and works in the market. Finally, at a sufficiently
high offered wage, the youth works in the market and also separates from
his parents' household" (p. 294). From our perspective, the Nash model
has allowed a regularity of the economy to be uncovered.
The heuristics in action
What has been the role of the heuristics in this progressive problem
shift? How have the heuristics shaped the move to M(t + 2)? What is the
relationship of the hard core of the neo-Walrasian program to this par
ticular sequence of theories in the protective belt?
Let us examine each of the hard core propositions to see how they
entered in the models in the sequence:
HCl "Agents" were single individuals in M(t) , households in
M(t + 1 ) , and household members in M(t + 2).
HC2 Outcomes involve the disjunction between endogenous
and exogenous variables. Exogenous variables for M(t) in
cluded spouse income, household goods, and household
membership; for M(t + 1) spouse income and household
goods were endogenous, but their interrelationships and
household membership were exogenous. In M(t + 2) all the
previous variables are treated as endogenous.
HC3 In M(t) , constraints were placed on an individual's
choice by that individual's income and wealth. In M(t + 1),
the household faces constraints on joint income and wealth,
but in M(t + 2) the constraints are faced by the individuals in
a household and are combinations of constraints associated
with their individual resources and those resources held in
common.
HC4-HC6 There is no interpretive shift in these propositions
as the sequence proceeds from M(t) to M(t + 1) to M(t + 2) .
224
Weintraub
225
some
References
Becker, G.S. (1981) . A Treatise on the Family. Cambridge, Mass. : Harvard
University Press.
Blaug, M. (1980) . The Methodology of Economics. New York: Cambridge Uni
versity Press.
Coddington, A. (1975). "The Rationale of General Equilibrium," Economic
Inquiry 13:539-58.
4 As de Marchi notes in his own chapter in this volume (chapter 4), Popper was never
really understood by economists; what practical economists took from Popper was the
position that Lakatos called "naive falsificationism" (Lakatos, 1970) . Thus it is a bit
misleading to put the blame, if there be blame, on Popper.
226
Weintraub
De Marchi, N . (1985) . "Popper and the LSE Economists. " Chapter 4, this
volume.
Hahn, F.H. (1973). On the Notion of Equilibrium in Economics. Cambridge:
Cambridge University Press.
Handler, E.W. (1980). "The Logical Structure of Modern Neoclassical Static
Microeconomic Equilibrium Theory," Erkenntnis 15:33-53.
Hands, D.W. ( 1984a). "The Role of Crucial Counterexamples in the Growth of
Economic Knowledge: Two Case Studies in the Recent History of Eco
nomic Thought," History ofPolitical Economy 16:59-67.
( 1984b) . "What Economics Is Not: An Economist's Response to Rosenberg."
Unpublished working paper, Department of Economics, University of
Puget Sound, Tacoma, Wash.
(1985). "Second Thoughts on Lakatos," History of Political Economy 17:116.
Hausman, D . M . ( 1981). "Are General Equilibrium Theories Explanatory?" in
J . C. Pitt, ed. , Philosophy in Economics. Dordrecht: Reidel, pp. 17-32.
Klant, J. (1984) . The Rules of the Game. New York: Cambridge University
Press.
Kotter, R. (1983). "General Equilibrium Theory: An Empirical Theory?" in W.
Stegmuller, W . Balzar, and W . Spohn, eds., Philosophy of Economics.
Berlin: Springer-Verlag, pp. 103-17.
Lakatos, I. (1970) . "Falsification and the Methodology of Scientific Research
Programmes," in I . Lakatos and A. Musgrave, eds . , Criticism and the
Growth ofKnowledge. Cambridge: Cambridge University Press, pp. 91-196.
(1978a). "History of Science and Its Rational Reconstructions," in I . Lakatos,
ed. , The Methodology of Scientific Research Programmes: Philosophical
Papers, Vol. 1 , ed. J . Worrall and G. Currie. New York: Cambridge Univer
sity Press, pp. 102-38.
(1978b). The Methodology of Scientific Research Programmes: Philosophical
Papers, Vol. 1 , ed. J . Worrall and G. Currie. New York: Cambridge Univer
sity Press.
Leijonhufvud, A. (1976). "Schools, 'Revolutions', and Research Programmes in
Economic Theory," in S. Latsis (ed.), Method and Appraisal in Economics.
Cambridge: Cambridge University Press, pp. 65-108.
McElroy, M . B . (1985). "The Joint Determination of Household Membership
and Market Work: The Case of Young Men," Journal of Labor Economics
3:293-316.
and Horney, M.J. (1981a). "Nash-Bargained Household Decisions: Toward a
Generalization of the Theory of Demand. " International Economic Review
22:333-49.
(198lb) . "The Household Allocation Problem: Results from a Bargaining
Model. " Discussion Paper, Department of Economics, Duke University,
Durham, N . C.
Popper, K. (1965 [1959]). The Logic of Scientific Discovery. New York: Harper
Torchbooks.
227
PART Y
Non-Popperian perspectives on
economics
C H A PT E R 9
232
Caldwell
233
234
Caldwell
On the most general level, the pluralist believes that the primary pur
pose of methodological work in economics is to enhance our understand
ing of what economic science is all about and, with luck, by so doing, to
improve it. I employ the term "understanding" in its everyday, common
sense usage, and I will speak more about the term "improve" later. To
accomplish these goals, the pluralist undertakes critical evaluations of
the strengths and limitations of various research programs in economics
and economic methodology. In addition, our understanding of econom
ics may be enhanced by various descriptive studies, for example, histori
cal studies of the development of ideas, analyses of the sociological
milieu in which a research program or discipline develops, and studies of
the rhetoric of economics. Whether one wishes to consider such studies
235
236
Caldwell
237
238
Caldwell
239
240
Caldwell
This important question deserves special attention. Note first that plural
ism is a meta-methodological position. It offers no specific methodologi
cal advice to economists. Indeed, what economists do is taken as given
by the pluralist.
What economists do, whether they are part of the mainstream or part
of the heterodoxy, is to work on a specific research topic. Such research
takes place within a specific research tradition (e.g. , neoclassical, Marx
ist, Austrian) and, as such, presupposes a methodology. Further refine
ments within traditions are also frequently encountered: A priorism has
been challenged by an interpretive, hermeneutical turn in the Austrian
241
242
Caldwell
forces him to claim that we can never know if we have reached the truth,
we can know that we have eliminated error. His theory of truth makes
not one whit of difference operationally, of course, but it does solve the
problem philosophically.
I took my cues from Feyerabend, another pluralist who in his early
work (e.g. , 1962, 1970) tried to show how pluralism could lead to an
expansion of true knowledge. He argued that because all facts are
theory laden, the multiplication of theories multiplies empirical content,
or knowledge. He was crucified for the initial premise of his argument,
the strong theory dependence thesis, and as a result he turned to anar
chism, and ultimately to Dadaism (Feyerabend, 1975) . As a philosopher
he had little choice.
Luckily, I am not a philosopher; I have choices. But even so, I recog
nize this as a very serious criticism of my position. I can think of a
number of responses to this criticism, but I doubt that any will convince
my critics.
I could try to use a pluralist argument to defend pluralism by arguing
that a commitment to a particular theory of truth narrows one's ap
proach to questions of methodology. But such an argument is clearly
circular.
I could offer pragmatic arguments. For example, I could argue that
my concern is to make methodological discourse more useful, and as
such, I am not concerned with the philosophical question of truth. Or I
could simply claim that my approach should be measured by its effects,
and that I think the effects would be good. This seems sensible enough
to me, but I fear my critics would see little sense in it.
I could point out that others have supplied very helpful analyses while
still falling into the same trap as me. Kuhn, for example, is forced to use
the weasel word "evolution" rather than "progress" in the final pages of
The Structure of Scientific Revolutions for the same reasons: One cannot
talk of progress without some conception of what we are progressing
toward. Larry Laudan attempts to define progress without having re
course to the notion of truth in Progress and Its Problems. Though both
Kuhn and Laudan have helped us to understand science better, both
have been criticized on this point.
I could take the offensive and cite the antifoundationalist philoso
phers who play so prominent a role in McCloskey's "rhetoric of econom
ics" approach. Thus I too could capitalize Truth in order to ridicule it.
But I must say, this seems to me to be the least satisfactory approach.
Look what it's done for McCloskey. Five years after publishing his
important article in the Journal of Economic Literature, in which the
rhetoric approach was trumpeted, he's still trying to figure out the argu-
243
244
Caldwell
Boland, L.A. (1982). The Foundations of Economic Method. London: Allen &
Unwin.
Caldwell, B .J . (1982). Beyond Positivism: Economic Methodology in the Twenti
eth Century. London: Allen & Unwin.
(1984). "Praxeology and Its Critics: An Appraisal," History of Political Econ
omy 16(3):363-79.
Feyerabend, P.K. (1962). "Explanation, Reduction and Empiricism ," in H.
Feig!, G. Maxwell, and M. Scriven, eds . , Minnesota Studies in the Philoso
phy of Science, Vol. III. Minneapolis: University of Minnesota Press, pp.
28-97.
(1970). "How to Be a Good Empiricist -A Plea for Tolerance in Matters
Epistemological, " in B . Brody, ed. , Readings in the Philosophy of Science.
Englewood Cliffs, N.J . : Prentice-Hall, pp. 319-42.
(1975). Against Method: Outline of an Anarchistic Theory of Knowledge. Lon
don: New Left Books, 1975.
Hutchison, T.W. (1981) . The Politics and Philosophy of Economics. New York:
New York University Press.
Klamer, A. (1983) . Conversations with Economists. Totowa, N.J.: Rowman and
Allanheld.
Klant, J.J. ( 1984 [1979]). The Rules of the Game: The Logical Structure of
Economic Theories. Trans. I . Swart. Cambridge: Cambridge University
Press.
Kuhn, T.S. ( 1970) . The Structure of Scientific Revolutions, 2nd enlarged ed.
Chicago: University of Chicago Press.
Lakatos, I. ( 1970) . "Falsification and the Methodology of Scientific Research
Programmes," in I . Lakatos and A. Musgrave, eds . , Criticism and the
Growth ofKnowledge, Cambridge: Cambridge University Press, pp. 91-196.
(1974). "Popper on Demarcation and Induction," in Paul Schilpp, ed. , The
Philosophy of Karl Popper, Vol. XIV, Book I , The Library of Living
Philosophers. La Salle, Ill. : Open Court, pp. 241-73.
Laudan, L. ( 1977). Progress and Its Problems: Towards a Theory of Scientific
Growth. Berkeley: University of California Press.
McCloskey, D . (1983). "The Rhetoric of Economics," Journal of Economic
Literature 21(2):481-517.
Popper, K. (1972). "Conjectural Knowledge: My Solution to the Problem of
Induction," in Objective Knowledge: An Evolutionary Approach. Oxford:
Clarendon Press, pp. 1-3 1 .
Toulmin , S . ( 1970) . "Does the Distinction between Normal and Revolutionary
Science Hold Water?" in I. Lakatos and A. Musgrave , eds., Criticism and
the Growth of Knowledge, Cambridge: Cambridge University Press, pp.
39-47.
Weintraub, E . R. (1984). "Appraising General Equilibrium Analysis," Econom
ics and Philosophy 1(1) :23-37.
C H A PT E R 1 0
246
l\lcCloskey
Yet even those of us who, from time to time, make use of philosophy
of science complain about its thinness. Roy Weintraub, for example,
complains rightly that Popper reduces the rich conversation of empirical
work down to a falsifying "fact. " Lakatos's philosophical work was an
extended complaint about the lack of thickness in Popper's work, as
Popper's was a complaint about earlier and still thinner philosophies of
science. None of it works. Mark Blaug, J.J. Klant, and Lawrence
Boland accept Lakatos's program, the "rational (which is to say, philo
sophical) reconstruction of research programs," as what methodology
should do, but strain at its limits when applying his program to real work
in economics. Boland, for one, skirts the edge of an economic literary
criticism (e.g. , 1982, pp. 1 16-17). And Weintraub notes in General
Equilibrium Analysis: Studies in Appraisal (1985, p. 142) that his case
study "raises several other problems that rest uneasily in a Lakatosian
bed. "
The thinness of the philosophy comes from the thinness of the ques
tion it asks. The question in a rational reconstruction of a piece of
science is: Does the discourse fit, say, a Lakatosian model? What is the
hard core, the protective belt, a typical negative heuristic? Can it be
made to lie down on the bed, with suitable trimming at head and feet?
The question will strike the outsider as odd. A study that verifies or
falsifies the fit of such a simple notion as sophisticated falsificationism to
a part of economics does not ask very many questions. The one question
it does ask would not strike a working scientist as interesting. At the end
of the day, you are led to ask what has been accomplished.
Consider again Roy Weintraub's recent work, the brilliant imitation
of Lakatos just mentioned and his elegant paper for this conference,
"The Neo-Walrasian Research Program Is Empirically Progressive. " All
right, suppose that by the Lakatosian definition the neo-Walrasian pro
gram is empirically progressive. (Weintraub certainly persuades on the
point: His work exhibits precision and candor well beyond the call of
duty in intellectual history.) But what follows? What at the end of the
day has been accomplished? We are now persuaded (set aside the prob
lem of induction in talking about the problem of induction) that neoclas
sical economics can be rationally reconstructed to correspond with a
pattern adumbrated by a certain philosopher. Well, so what?
The question is pragmatic, but not in a vulgar sense. It will be per
fectly satisfactory if the cash value of the Lakatosian categories shows up
merely in their value for further thinking: for setting economics in con
text, for making economists more self-aware, for telling persuasive sto
ries about the history of economics, for understanding why economists
go on as they do. Among professional intellectuals these should count as
10
247
248
McCloskey
10
249
One can agree with a purpose here of attacking Nazism without agreeing
that some method of Science will achieve it. One can argue in fact the
other way around. After all, the Nazis were gloriously Scientific in their
experiments. Victorian and even British Science, with its elaborate cere
mony of testability (most skillfully practiced by the psychologist Sir Cyril
Burt), was a major source of racist theories (cf. Gould, 1981). And on
the other side it is not easy to blame, say, Jewish numerology or Gypsy
legerdemain for the rise of Nazism.
250
McCioskey
10
251
"[B)oys were seen quarreling all the time, but not once was a game
terminated because of a quarrel" - and explains that "it seemed that the
boys enjoyed the legal debates as much as they did the game itself, and
even marginal players of lesser size or skill participated equally in these
recurrent squabbles. In contrast, the eruption of disputes among girls
tended to end the game" (p. 9) . The parallel with methodological dis
putes is suggestive. Gilligan reports on Piaget's observation of "boys
becoming through childhood increasingly fascinated with the legal elabo
ration of rules . . . , a fascination that, he notes, does not hold for girls"
(p. 10) . The girls stressed community, conversation, solidarity, and the
other nonrule values of those known affectionately as the "new fuzzies"
(Rorty, 1987). Arjo Klamer and I can be viewed therefore as presuming
to bring a feminine perception to the matter.
I suggest, with Klamer, that the good that lies behind methodological
thinking is the goodness of community, solidarity, openness to ideas,
educated public opinion, and a better conversation of humanity. By
their moral fervor the methodologists reveal their values. Their values
are fine, and not much different from those of the terrible fuzzies they
fear.
The word is sprachethik, speech morality, the ethics of conversation.
That the word comes from a hive of Marxist fuzzies in Frankfurt-am
Main should not be alarming, for it is liberalism incarnate: Don't lie; pay
attention; don't sneer; cooperate; don't shout; let other people talk; be
open-minded; explain yourself when asked; don't resort to violence or
conspiracy in aid of your ideas. These are the rules adopted by the act of
joining a good conversation. Socratic dialogue - flowing first from a pen
devoted to finishing conversation- is the model for Western intellectual
life. An American philosopher put the point well. What is crucial, writes
Amelie Oksen berg Rorty, is "our ability to engage in continuous conver
sation , testing one another, discovering our hidden presuppositions,
changing our minds because we have listened to the voices of our fel
lows. Lunatics also change their minds, but their minds change with the
tides of the moon and not because they have listened, really listened, to
their friends' questions and objections" (1983, p. 562) . Good science is
not good method but good conversation.
We know when conversations are going well among our own intellec
tual friends. Most economists would agree, for example, that the conver
sation about international trade since 1950 has been through a bad
stretch, relieved only temporarily by a burst of creativity fifteen years
ago on the financial side. They would agree, too, that economic history
improved radically after 1958 and has flourished ever since. Working
economists do not need the advice of a philosopher- least of all an
252
McCloskey
10
253
254
McCioskey
10
255
are bound only by their close, if differing, interests in the controversy's outcome.
(1985, p. 142)
Collins treats the debate among physicists about gravity waves, for
example, as just that: as a debate, showing how one or another rhetorical
move led to the result. At its turning point, for instance, the chief propo
nent of gravity waves, Joseph Weber, "in accepting . . . electrostatic
calibration . . . accepted constraint on his freedom to interpret results"
(p. 105) . Collins notes that Weber did not have to accept the calibration
(which itself, by the way, is a rhetorical turn common to many fields: the
selection of a quantitative standard) . It was a rhetorical choice. But hav
ing made it, Weber was constrained by rules of debate, rules that can
themselves be studied and partially understood, and have in fact been
studied and partially understood since the time of the Greeks.
The varied rules of human debate, not godlike tests, decide the out
come. "It is control on interpretation which breaks the circle of the
experimenters' regress [Collins's phrase for Duhem's dilemma] , not the
'test of a test' itself" (p. 106). That is to say, it is rhetorical consider
ations, the workings of a human conversation, not mechanical applica
tions of rules within a closed system, that end a scientific debate. Scien
tists do not commit a crime when they argue beyond the constricted
realm of formal logic. "Scientists do not act dishonourably when they
engage in the debates . . . ; there is nothing else for them to do if a
debate is ever to be settled and if new knowledge is ever to emerge from
the dispute" (p. 143). It is not the logic of inquiry that allows scientific
progress, but the rhetoric of inquiry (cf. p. 153, note 5).
I am asserting that Collins and other observers of scientific contro
versy contribute unawares to the rhetorical tradition. Rhetorical criti
cism is the thickest approach. It draws on an immensely long tradition
from the Sicilian sophists to the present, running parallel to philosophy,
though spurned by philosophy in every age. At present the tradition
lives in law schools, in the literary world (Booth, 1974; Fish, 1980) , and
in writings on argument emanating from specialists in rhetoric (Scott,
1967). Occasionally it can be seen half-conscious in a philosopher gone
wrong (Toulmin, 1958; Steiner, 1975 ; Rosen, 1980; Rorty, 1982; Wal
ton, 1985 ) .
I t i s not "mere" rhetoric, and not an ornament to be distinguished
from the substance of argument. It is rhetoric in the ancient and honor
able sense:
the art of probing what men believe they ought to believe, rather than proving
what is true according to abstract methods . . . , of discovering good reasons,
256
l\1cCloskey
10
257
CHAPTER 1 1
Economics as discourse
ARJO KLAMER
259
260
Klamer
hasten to say that their work played a critical part in the development of
my argument. It is with deference to their insights that I disagree.
The following discussion presents philosophical reasons for changing
our talk about economics. It could be interpreted as an indication of
what happens when we replace Karl Raimund Popper as the central
character in our conversations. I shall have little to say about economics
per se; that has been and will be my concern elsewhere. 2
Let the argument begin with a story.
A story
11 Economics as discourse
261
262
Klamer
11 Economics as discourse
263
of explicit and implicit rules, just like a baseball game. The implicit rules
deserve emphasis: Participation in the economics game requires not only
knowing the methodological rules, but also how to talk with economists,
what questions to ask, which names to know.
For our study of economic discourse we can choose to confer with
semioticians, rhetoricians, communications theorists, literary critics, lin
guists, philosophers of science, cognitive psychologists, ethnographers,
sociologists, anthropologists, or information theorists. The possibilities
are too many, so we are compelled to choose an entry point.
The following simple model of communication is a useful start.
vehicle
sender
receiver
signal
The study of rhetoric is one way of delineating a specific domain for the
model of communication. Here Aristotle's Rhetoric points the way.
Aristotle defines rhetoric as "the faculty of discovering in the particu
lar case what are the available means of persuasion." Rhetoric, to him,
is the art of public speaking and is distinct from dialectic, or the art of
logical discussion. Accordingly, Aristotle's rhetoric is a form of commu
nication that takes place in the courtroom (deliberative oratory), in the
political arena (forensic oratory), and at celebrations (epideictic ora
tory). Translated in terms of his rhetorical triad, the communication
model reads: speaker- discourse (or logos) - audience.
264
Klamer
However, Aristotle carved out too small a domain for the application
of rhetoric. Besides the courtroom and the political arena, the novel, the
street, and scholarly seminars and writings can be considered the do
main of rhetoric. Aristotle's rhetorical triad has already turned out to be
useful for students of literature, colloquial discourse, and academic disci
plines such as mathematics, psychology, history, and economics.
Aristotle's discussion of the rhetorical process has psychological over
tones. He invites us to think of the emotions of the audience (pathos)
and points at the ethos or moral character of the speaker as significant
factors in the process. Translated into contemporary terms, the rhetori
cal model compels us to consider the authority of the speaker (in
economic discourse, the name of a Robert Solow or Robert Lucas on
an article is likely to have a stronger rhetorical effect than the name of
any other intelligent but little-known economist) and the role of the
reader(s) . The point is that the signs of the rhetor are never self
evident or complete and require decoding or interpretation. The art of
persuasion is therefore the art of being able to induce the intended
interpretation.
Characteristically rhetorical arguments are, according to Aristotle,
the enthymeme, which is a deduction that omits a premise or contains
premises that are merely plausible. An example of the enthymeme is the
economic statement: If the price of a good falls, the quantity demanded
will increase. The statement has the form of a syllogism but clearly
leaves out various assumptions, such as the assumptions that individuals
maximize their utility, that indifference curves are convex, that the sub
stitution effect is greater than the income effect, and that preferences
are constant. The user of the enthymeme presumes a critical contribution
of the audience, as the missing links have to be filled in. Decoding of the
statement, then, requires the application of information and knowledge
that is not contained in the statement itself.
The use of so-called topoi, or lines of argument, is a critical factor in
bringing about the decoding that the speaker intends. Topoi are assump
tions or arguments that are not necessarily true, but are reasonable and
generally accepted by the audience. For example, an audience of neo
classical economists is likely to accept the topoi that individuals are
rational and their preferences constant. They understand what the
speaker means when he or she uses these topoi, that is, they can imagine
the appropriate diagrams and are familiar with the literature in which
these topoi are developed.
Another characteristically rhetorical device is the example. When
economists, for instance, argue that government deficits do not necessar
ily inhibit economic growth, they may offer the case of West Germany in
11 Economics as discourse
265
266
Klamer
11 Economics as discourse
267
thinks; to translate what one knows, and something other than to play with the
structures of language (langue) ; to show that to add a statement to a pre-existing
series of statements is to perform a complicated and costly gesture, which in
volves conditions (and not only a situation, a context, and motives), and rules
(not the logical and linguistic rules of construction) ; to show that a change in the
order of discourse does not presuppose "new ideas," a little invention and
creativity, a different mentality, but transformations in a practice, perhaps also
in neighbouring practices, and in their common articulation. (p. 209)
268
Klamer
One and many: a reading of a discussion
1 1 Economics as discourse
269
Likewise, accepting the importance of logical studies, as I am quite willing to do, does
not imply that my version will be like Popper's or that of any other conventional
philosopher.
270
Klamer
11 Economics as discourse
271
272
Klamer
I have in mind the work of David Gordon, Thomas Weisskopf, and Samuel Bowles.
See, for example, Beyond the Wasteland (New York: Doubleday, 1984).
10 Like Bruce Caldwell.
1 1 They adhere to the analytical style but reject the major tenets of positive philosophy.
11 Economics as discourse
273
The finding and articulation of this "more" begs for interpretation and
thus consideration of the meaning of the signs economists and economic
methodologists communicate.
274
Klamer
Further differences: on interpretation and politics
The issue of meaning is taboo to Popper, who "avoids, like the plague,
discussing the meaning of words" (1979, p. 309). He does, however,
discuss the problems of understanding, in particular the understanding
of problems. Using Galileo's theory as an example, he argues that under
standing of problems is "a matter of handling third-world structural
units, [of getting] familiar with these units and their logical interrela
tions" (1979, p. 182) . He thus disentangles our understanding of the
phenomenon "understanding" from the common association with a men
tal activity (his world 2).
This perspective makes sense but only in a limited way, as the contrast
with Wittgenstein's discussion of understanding and meaning may indi
cate. Wittgenstein, like Popper, rejects the reference to mental activi
ties, but he creates a much larger discursive domain for interpretation
and meaning. The meaning of a word is, according to Wittgenstein, its
use in the context of what he calls a "language game" (or "discourse," or
"conversation"). The meaning of "money," "conflict," and other eco
nomic terms, therefore, is to be understood through the use in conversa
tion. From this perspective, Popper's discussion is restrictive for two
major reasons. Firstly, Popper's interpretation of world 3 represses the
possibility that one term has many incommensurable interpretations, a
possibility that is invoked through the metaphorical use of the "game"
and "conversation . " Secondly, whereas Popper wants us to believe that
understanding is a matter of recognizing the problem at hand- and not
the meaning of words- Wittgenstein urges us to consider the appropri
ate cultural context, or the language game, when we want to understand
the meaning of words.
Popper's restrictions on interpretive discourse are an immediate corol
lary of his three-worlds analogy. He constructs his world 3 to be the set
of human products. The "meaning" of world 3 is indicated by the use of
the expression "objective knowledge" : It makes us think of the compo
nents of world 3 as being independent of knowing subjects, or his world
2. For further clarificiation he presents a story- a "thought experiment,"
he calls it:
All our machines and tools are destroyed, and all our subjective learning, includ
ing our subjective knowledge of machines and tools, and how to use them. But
libraries and our capacity to learn from them survive. Clearly, after much suffer
ing, our world may get going again . (1979, p. 108)
11 Economics as discourse
275
what happens if the "we" in his experiment are illiterate people who
come out of the bushes after all literate people have been eliminated in a
series of neutron bomb explosions? For all they know, the meaning of
library books is that they burn well. 1 2
The problem with Popper's analogy is that world 3 is like world 1 . Both
are a complex whole of things and relationships among things that are to
be known; in the postmodernist manner of speaking, both are a collection
of texts to be read. Any of its elements, whether a particle or an article,
becomes real only through conversation, that is, by virtue of an interpre
tive community. Before the bush people can "read" the particle or article
the way any of "us" does, they need to learn how to participate in the
appropriate conversations. In that sense, their experience would be simi
lar to the experiences of novices in the games of baseball and economics.
Accordingly, they need to learn not only the proper language, the neces
sary jargon, but also the reasons why it is good to be interested in nuclear
physics and text analysis. Like my learning about baseball, their learning
involves a cultural experience that will change them.
Let me give one final argument. Popper tells us that understanding a
theory is a matter of understanding the problem that it attempts to solve.
But do we understand his epistemological theory when we recognize
that it attempts to solve the problem of induction? If so, how do we
understand the tenacity with which Popper clings to his theory in spite of
logical and historical difficulties? How do we understand the differences
between Rorty and Popper in spite of the similarities? These questions
inevitably expand the interpretive domain.
A significant extension is the political connection. Both Rorty and
Popper make the connection explicit in their writings. Popper writes in
his Logic of Scientific Discovery: "I hope that my proposals may be
acceptable to those who value not only logical rigour but also freedom
from dogmatism" (p. 38) . The dogmatists he has in mind are the Marx
ists, fascists, and Freudians, as he makes explicit in his other writings.
He favors the open society in which people are willing at all times to
submit their ideas (conjectures) to the strongest tests available. Rorty
shares Popper's liking for a democratic society that allows individual
expression, but his assessment of the most important features is differ
ent. Whereas Popper stresses criticism, Rorty talks about solidarity. In
Popper's world we seek peaceful coexistence through respect for objec
tivity; in Rorty's world the sense of community is crucial and is threat12
The movie "The Gods Must Be Crazy" is a humorous portrayal of such a learning
process. The story is about a bushman who tries to return to the gods a coke bottle that
dropped from heaven (an airplane) and is confronted with a Western culture for the first
time in his life.
276
Klamer
11 Economics as discourse
277
278
Klamer
Index
Haavelmo, T. , 207-8
Hahn, Frank, 38, 130, 131
Hicks, J. R., 34, 37-9, 170 ff. , ch. 6 pas
sim
Hutchison, T. W., 4, 7, 35, 81, 129, 141 ,
145-6, 236, 259
advocacy of falsifiability, 10- 1 1 , 249
empirical a priorism
and J. S. Mill, 78-9, 95
and Lionel Robbins, 3, 78
empirical model, 41, 42, 44, ch. 7 passim
empirical progress, 247
vs heuristic progress, 26, 122-3, 125,
127
"ethical" argument
vs epistemology, 18, 56
and goals of science, 19, 57, 169, 249
excess content (independent testability),
1 1 , 47, 123, 135, 219, 220, 221-2, 225
see also novel fact
279
280
Index