Rudy Van Zijp - Austrian and New Classical Business Cycle Theories - A Comparative Study Through The Method of Rational Reconstruction-Edward Elgar Pub (1993)

AUSTRIAN AND NEW CLASSICAL
BUSINESS CYCLE THEORIES

VRIJE UNIVERSITEIT
AUSTRIAN AND NEW CLASSICAL

BUSINESS CYCLE THEORIES
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad van doctor aan

de Vrije Universiteit te Amsterdam,
op gezag van de rector magnificus
dr. C. Datema,
hoogleraar aan de faculteit der letteren,
in het openbaar te verdedigen
ten overstaan van de promotiecommissie
van de faculteit der economische wetenschappen en econometrie
op maandag 5 oktober 1992 te 13.30 uur
in het hoofdgebouw van de universiteit, De Boelelaan 1105
door
Rudolf Willem van Zijp
geboren te Arnhem
Thesis Publishers Amsterdam

1992
Promotores : prof.dr. H. Visser
prof.dr. N. de Marchi
Co-promotor: dr. W. Keizer
Referent prof.dr. E.E. Berns
To my parents
The book is nr. 33 of the Tinbergen Institute Research Series. This
series is established through cooperation between Thesis Publishers
and the Tinbergen Institute. A list of books which already appeared in
the series can be found in the back.
PREFACE
The present study could not have been written without the advice and
support of mal?-y people. I am particularly indebted to Hans Visser and
Willem Keizer. The former's comments on monetary theory have been
very useful, and his adherence to James Meade's 'common sense' preven-
ted me from running into either of the two extremes which are considered
in the present study. Willem Keizer has been very helpful as regards
Austrian economics. Our discussions about what we called the 'sniffling-
rat theorem' have allowed me to gain some insight into both Austrian and
'Keizerian' entrepreneurship. His corrections of my English are also
appreciated. Furthermore, I thank Neil de Marchi, whose stimulating and
challenging methodological remarks from afar caused considerable
distress on my part. The quality of this thesis would have been much
lower without his influence. I also wish to thank Guido Berns for his
willingness to read the final draft version of the entire text.
I highly appreciate the support of Bert Huussen, who was always
ready to comment upon my work, even though he was not formally invol-
ved in my research project. He has taught me a lot about philosophy of
science in general. I also thank Jack Birner (Rijksuniversiteit Limburg),
Roger Garrison (Auburn University, USA), Edmar Sprokholt (Harvard
University, USA), Ernie! Wubben (Erasmus University Rotterdam), and
the members of the Dutch research group on Austrian economics ('Club
Austria') for their help, cooperation, and comments. Furthermore, Hans
Vijlbrief (Free University) and Rob van de Wijngaert (Free University)
deserve praise for listening patiently to my stories about such idiosyncratic
issues as entrepreneurship and voluntary unemployment.
Finally, I express my gratitude to my family and friends. They
have provided an excellent social environment for doing research, taking
my research caprices for granted. This has been more important than they
may have expected.
LIST OF CONTENTS
1. INTRODUCTION 1
1.1. The aim of the study 1
1.2. The method of the study 2
1.3. The structure of the study 6
PART I. AUSTRIAN BUSINESS CYCLE THEORY
2. EARLY VERTICAL MALADJUSTMENT THEORIES 11

2.1. Introduction 11
2.2. Tugan-Baranowski 11
2.3. Spiethoff 13
2.4. Cassel 14
2.5. The structure of production 15
2.5. Wicksell 19
3. THE ONSET OF THE AUSTRIAN THEORY 23
3.1. Praxeology and economics 23
3.2. Monetary theory 27
3.3. Business cycle theory 30
3.4. Misesian methodology 36
3.4.1. Methodological individualism 36
3.4.2. Methodological singularism 38
3.4.3. Radical subjectivism 39
3.4.4. Methodological dualism 41
3.5. Conclusions 41
4. HAYEK'S YEARS OF HIGH THEORY 43
4.2. Intertemporal equilibrium 44
4.3. Hayek's early business cycle theory 45
4.4. The coordination problem 50
4.5. Capital, investment, and discoordination dynamics 53
4.6. The debate on the Ricardo effect 55
4.6.1. The Ricardo effect 56
4.62. The debate on the Ricardo effect 58
4.6.3. The re-opening of the debate: Hicks's
'The Hayek story' 61
4.7. The Hayek programme 64
ix
4.7.1. The Hayek Problem 65
4.7.3. Subjectivism 68
4.7.4. The compositive method 70
4.7.5. Summary 76
4.8. Competition, learning and the Hayek Problem 77
4.9. Conclusions 80
5. THE YEARS IN THE WILDERNESS 83
5.2. Innovation and the Austrian business cycle theory 84
5.3. Knowledge and expectations 86
5.4. Process analysis 91
5.5. Capital, expectations, and process analysis 92
5.6. Rothbard's historical study 96
5.7. Coordination and the role of entrepreneurs 98
5.8. Summary and conclusions 99
6. THE AUSTRIAN REVIVAL 101
6.2. The synthesis of competition and entrepreneurship 103
6.3. Two types of knowledge acquisition process 107
6.4. Creative entrepreneurship 109
6.5. The market as an evolutionary process 110
6.6. The different abilities of individuals 112
6.7. The revival of Hayekian business cycle theory 114
6.8. Austrian economics and mathematical formalization 118
6.9. Testing Hayekian business cycle theory 124
6.10. Conclusions 130
PART II. NEW CLASSICAL ECONOMICS
7. MONETARIST ROOTS OF NEW CLASSICISM 135

7.2. The quantity theory of money and its critics 136
7.3. Friedman's restatement of the quantity theory 138
7.4. Interpretations of the Phillips Curve 140
7.5. Friedman and the Phillips Curve 143
7.6. The incorporation of expectations 145
7. 7. Summary and conclusions 147
X
8. THE RISE OF NEW CLASSICAL ECONOMICS 149
8.2. The formalization of the NRH 151
8.3. The Rational Expectations Hypothesis 156
8.4 ..From partial to general equilibrium 159
8.5. The Lucas critique 163
8.6. Lucas's partial test of the NRH 166
8.7. Sargent's tests of the NRH 168
8.7.1. Fisher's solution to the Gibson paradox 168
8.7.2. The implausible length of the transition
period 170
8.7.3. Muth-rationality as a criterion for
plausibility 172
8.7.4. The inclusion of omitted variables 173
8.7.5. Sargent's tests 176
9. PERSISTENCE, CAPITAL AND GLOBAL INFORMATION 183
9.2. The problem of persistence 184
9.3. Persistence and local capital markets 189
9.4. Current global information and the signal
extraction problem 193
9.5. The Lucas Programme 196
9.5.1. The Lucas Problem 196
9.5.3. Subjectivism 200
9.5.4. The method of rational model-
construction 200
10. THE EFFECTIVENESS OF MONETARY POLICY 207
10.2. The policy-ineffectiveness proposition 208
10.2.1. Minimizing the gap between actual and
capacity output 209
10.2.2. Minimizing the gap between actual and
full-information output 213
10.2.3. Rules versus discretion 217
10.2.4. The debate on the neutrality proposition 218
10.3. Testing for neutrality 221
xi
10.3.1. Sargent's tests 221
10.3.2. Barro's tests 226
10.3.3. Conclusions 228
10.4. Differential information and the effectiveness
monetary policy 229
10.5. Perceived versus unperceived money growth 232
10.6. The switch from monetary to real business cycle
iliwry m
10.7. Methodological assessment 237
PART Ill. A COMPARISON
11. AUSTRIAN VERSUS NEW CLASSICAL ECONOMICS 245

11.2. The Hayek Problem versus the Lucas Problem 247
11.3. Types of explanation 248
11.4. Methods of discovery 250
11.5. The subjective nature of the data 252
11.6. Risk and uncertainty 254
11.7. The homogeneity postulate 256
11.8. The 'degree of explanation' 258
12. SUMMARY, CONCLUSIONS AND EPILOGUE 263
12.1. Development of Austrian economics 263
12.2. Development of New Classical economics 266
12.3. Epilogue 269
BIBLIOGRAPHY 271
NAME INDEX 295
SUBJECT INDEX 299
DUTCH SUMMARY
xii
1. INTRODUCTION
Sisyphus, the founder and first king of Corinth, had twice tricked the god
Death, once by binding him and once by persuading him to let him return
to earth. As a result, Homer tells us, he was sentenced by the gods. His
punishment consisted in pushing a boulder uphill to the top of a moun-
tain. Everytime he was about to succeed, its weight would turn the rock
back, returning it to the plain, so that he had to start all over again. The
tragedy of this myth is that its hero is aware of the absurdity and inevita-
bility of his task. Nevertheless, as Camus (1975, p. 143) pointed out,
Sisyphus may be a happy man, because on his way downhill he realizes
that his torment results from his own desire, or choice, to live.
Although Homer does not elaborate on Sisyphus's mental pro-
cesses, it appears that the former king could not exert any influence on
the specific nature of his eternal task. Fortunately, historians of science
(think that they) differ from Sisyphus, in the sense that they may choose
their rock and the way in which they push it uphill. This chapter eluci-
dates these choices made for the present study. In the first section the
choice of the boulder is explained, and in the second the way in which it
will be pushed up. Section 1.3 gives the plan of the book.
1.1. THE AIM OF THE STUDY
In the last two decades economics has experienced a remarkable shift in

' focus. Keynesian macroeconomics, at least in its Hicksian IS/LM-version,
has been the ruling orthodoxy since World War II. Although it was occa-
sionally and sometimes fiercely challenged by monetarists, it retained its
dominant position until the 1970s. In that decade, however, monetarist
criticism received support from two other research traditions, the Austrian
School and New Classical Economics. These traditions stressed the alloca-
tive efficiency of markets. Their focus on business cycle theory led to a
revival of interest in this type of theory.
The emphasis of both research traditions on the efficiency of
markets has led to the idea that both share common roots. After all, both
criticize Keynesianism, stress the communicative role of prices, point to
the crucial role of expectations, and claim that business cycles are real
phenomena, caused by monetary disturbances. This 'common-roots idea'
was enhanced by Lucas's (1977, p. 216) claim that New Classical Econ-
omics (NCE) builds on Hayek's work of the 1920s and 1930s. Several
authors subscribed to this view. For instance, Kantor (1979) argued that
the New Classical Rational Expectations Hypothesis (REH) is not foreign
to Austrian economics. Colander and Guthrie (1980) claimed that this
hypothesis made New Classicals bed-fellows of the Austrians. Laidler
(1982) even proposes to relabel the New Classicals into Neo-Austrians.
And Scheide (1986) observed that in the post-war period the term 'equili-
brium' has changed, and that the New Classical concept of 'rational-
expectations equilibrium' is a more explicate and narrowed-down version
of Hayek's notion of 'disequilibrium'.
Butos (1986) criticized the 'common-roots claim', on the grounds
that it completely neglected Hayek's work on other issues than business
cycle theory. He concluded that the claim was correct though misleading.
Other analyses of the differences between Austrians and New Classicals
soon followed. Most informative in this regard are Kim (1988), Hoover
(1988), Garrison (1989), and Klausinger (1989a) and (1991). These contri-
butions provide excellent overviews on Austrian and New Classical
economics. Unfortunately, they consider both traditions in their fully-
fledged forms. This neglect of their respective historical developments
means that they cannot fully appreciate the problems which Austrians and
New Classicals faced, and can only analyze the theoretical aims of some
'representative' Austrian or New Classical theorist. The present study aims
to provide such histories, in order to increase our understanding of the
problem situations with which the theorists in both research traditions
were confronted. The aim of the present study, i.e. the Sisyphan boulder,
is thus to understand the respective Austrian and New Classical research
traditions by analyzing the way in which they emerged and developed, in
order to better define the differences and similarities. The analysis will be
restricted to the respective explanations of business cycles, although
developments in other fields are discussed as well if deemed relevant.
1.2. THE METHOD OF THE STUDY
Following Popper (1972, p. 164), a theory is a tentative solution to some

problem. Understanding the theory means that it is known (1) what
problem it was intended to solve, and (2) how satisfactory it is as a
solution to that problem. The present study attempts to answer these
2
question by using Popper's version of the method of Verstehen (under-
standing), the so-called method of rational reconstruction~ This method
aims to reconstruct the theorist's problem situation, which consists of the
theoretical aims and the situational constraints. According to Wong (1978,
p. 11), the former "... generate the main question(s) to which the new
theory is directed", and the latter form the constraints or limitations on
the answers to these questions. The problem of the theorist is thus to
devise a theory which satisfies the theoretical aims as well as the situa-
tional constraints. 'Understanding' a theory means to conjecture this prob-
lem situation, and to explain why the solution was considered satisfactory,
or otherwise significant, to the theorist. As Popper (1972, p. 189) pointed
out, the historian of science thus faces the task of reconstructing the prob-
lem situation as perceived by the agent, in such a way that the latter's
actions can be regarded as being adequate to the situation. This conjec-
tured situation must be logically sufficient to have brought about the
theorist's solution (i.e. the theory). Since the theorist then acted according
to the logic of the situation as he perceived it, the method of rational
reconstruction is also called the method of situational logic.
The method of rational reconstruction does not imply that each
and every theory provides a satisfactory solution to the theoretical aims
for which it is designed. Instead, the theorist need not have perceived his
problem situation correctly. As Popper (1972, p. 179) stated, the historian
of science must therefore "... distinguish between the [problem] situation
as the agent saw it, and the situation as it was (both, of course, conjec-
tured)." The differences between the perceived and reconstructed situ-
ations may explain why a theory fails to solve the problem for which it
was intended. The differences may result from (1) the theorist's misper-
ceptions of his problem situation (particularly his situational constraints),
and/or (2) the historian's misinterpretations and 'misreconstructions'.
However, since the historian will consider his interpretation of the
theorist's problem situation the best conjectures available, the latter
possibility can be ruled out. This means that any difference between the
theorist's and the reconstructed problem situ~tion is explained as the
result of the former's misperceptions. These misperceptions are most
likely to concern the situational constraints, and not the theoretical aims.
Wong (1978, p. 12) classified these constraints as follows:
(1) an appraisal of the theories relevant to the pursuit of the theoretical
aims;
(2) the general theory or theoretical framework of which the theory under
3
study is an integral part;
(3) the epistemological theory of the theorist;
(4) the methodological theory of the theorist; and
(5) the metaphysical 'doctrines' of the theorist.
The present study will be mainly concerned with the first, second, and
fourth classes. The epistemological theories and metaphysical doctrines,
however, fall outside the realm of the present study. 1 Furthermore, follo-
wing Wong (1978, p. 19), problems intrinsic in logics and mathematics are
neglected as well, even though these disciplines form part of the theorist's
problem situation.
To what kind of history does the method of rational reconstruc-
tion lead? In Popper's (1972, p. 164) view, theorists start with a problem
(P) which they try to solve by conjecturing some theory or tentative
solution (17). The solution is subsequently scrutinized in order to elimin-
ate any errors. This phase of error-elimination (EE) leads to a new
problem (P'), for which another theory is conjectured, and so on. The
process of scientific research can thus be formulated as follows:
P --> IT--> EE --> P' --> IT'--> ...
This representation of the process of scientific research, and the underly-
ing method of rational reconstruction thus leads to an interpretation of
the history of scientific thought as a sequence of problem situations. Or,
in Popper's (1972, p. 177) words, "... the history of science should be
treated not as a history of theories, but as a history of problem situations
and their modifications ... through the intervention of attempts to solve
the problems." As a corollary, the resulting history stresses the relation-
ships between problem situations through time.
Popper (1963, p. 215) argued that 'science' needs to grow. In his
1 This is not to say that they do not play an important role in the development of both
research traditions. The point merely is that their inclusion would overstretch the domain
of this analysis. Furthermore, the metaphysical doctrines are also excluded because they
cannot be unambiguously specified. Even if it would be possible to fully explicate one's
ideological and ontological views (which it presumably is not), economists do not give such
an explication, so that the historian of thought must derive them from the their methodo-
logical practice. Since there does not exist a one-to-one correspondence between meta-
physics and methodology, the former cannot be derived unambiguously from the latter. For
instance, van Zijp (1991a, pp. 17 - 18) shows that 'methodological individualism' may be
based on either 'ontological individualism' or 'ontological collectivism'. These doctrines are
incompatible, in the sense that the latter holds that in reality collective (social) entities
exist, whereas the former denies this existence. Ontological collectivism can nevertheless
lead to methodological individualism on the grounds that the existence of collective entities
is discernible only through the actions of individuals. This suggests that conjectures about
the underlying metaphysical doctrines cannot be based on methodological practice alone.
4
view, continued growth is essential to the rational and empirical
character of scientific knowledge; that if science ceases to grow it must
lose that character." Obviously, the method of rational reconstruction
should then allow for an appraisal whether scientific knowledge has 'pro-
gressed'. Popper (1963, p. 219) suggested that the notion of verisimilitude
('truthlikeness', 'nearness to truth') can be used as a criterion of progress.
He defined this notion in terms of the (empirical) content of a theory,
which consists of the class of non-tautological statements logically entailed
by that theory (1972, p. 48). Since the growth of scientific knowledge
presupposes an increase in the empirical content of theories, Popper
(1972, p. 53) advised theorists to strive after 'deep' theories, which have
greater content, larger classes of allowed statements, and hence are more
easily falsified. 2 The method of rational reconstruction is to be used to
appraise such progress. It should be noted, however, that the present
study merely aims to understand the Austrian and New Classical research
traditions, and hence does not constitute an attempt to appraise their
progressiveness. This may lead to the criticism that it uses a method for a
different purpose than for which it was originally intended. This criticism
may be answered by pointing out that Austrians do not subscribe to the
prescription that economics should try to provide ever 'deeper' theories.
Instead, as will be shown in Chapters 4 and 11, they hold that social
phenomena cannot be explained in full because of the subjectivity and
dispersion of knowledge. As a result, they argue that the explanatory
precision of social theories is rather limited. The alleged complexity of
social phenomena thus hampers the 'deepening' of these theories, so that
Popperian progress cannot be attained in the social sciences. The Austrian
research tradition thus does not strive after such progressiveness. This
does not mean, of course, that its progressiveness cannot be appraised,
but it should be recognized that such an appraisal disregards the reasons
which Austrians give for refusing to 'deepen' their theories. Since these
reasons form an important part of the Austrian tradition, their elimination
distorts our understanding of Austrian economics. Therefore, the present
study does not attempt to appraise the respective research traditions.
Instead, it uses the method of rational reconstruction as a method of
understanding only. It yields reconstructions of both Austrian and New
Classical business cycle theories, which are compared as regards their
methodological features.
2 This has led followers of Lakatos to engage in a quest for novel facts. _
5
Some final remarks are now in order. Firstly, the present study
concentrates on the internal history of economics as a science, so that
considerations from outside economics are mostly disregarded. This
reflects the belief that scientific developments have an internal dynamic
(which, again, need not constitute 'progress'). Secondly, the analysis of
Austrian economics takes a rather unusual route, in the sense that it is
mainly concerned with the Austrian business cycle theory, to the neglect
of the socialist-calculation debate as held in the 1920s and 1930s. How-
ever, this debate enters in a disguised form when attention is paid to the
Austrian view of the role of competition as a process. It will be argued
that this view has some implications for understanding Austrian business
cycle theory. Finally, there is a considerable difference in opinion between
Austrians and New Classicals as regards the appropriate language which
economics should use. New Classicals opine that mathematics is the
appropriate language whereas Austrians are highly sceptical about math-
ematical formalization because they consider it to enlarge the danger of
the mechanistic interpretation of social relationships. A comparison of
these research traditions presupposes, though, that a common language
exists. This means that either the Austrian theories should be translated in
mathematical terms, or New Classical views should be expressed in verbal
form. The present study follows the latter strategy, because the former
could only be pursued by eliminating one of the central tenets of Austrian
thought, namely its subjectivism and its emphasis on fundamental (Knight-
ean) uncertainty and the complexity of economic relationships. At the
same time it should be noted that in many cases mathematical, econo-
metric or statistical terminology can be used as a short-hand for verbal
descriptions. In the case of the NCE such short-hand will be used to
describe the most important test procedures, and to explain Sargent's
(1976b) 'observational equivalence'.
1.3. THE STRUCTURE OF THE STUDY
The study is organized as follows. Part I consists of five chapters, which

describe the development of Austrian business cycle theory. Chapter 2
gives a short pre-history of this theory, and contains the Bohm-Bawerkian
framework of the structure of production and the Wicksellian distinction
between the natural and the market rate of interest. Chapter 3 discusses
Mises's contribution to Austrian methodology and economics. It concen-
6
trates on his monetary theory and his explanation of the business cycle.
Chapters 2 and 3 depict the Austrian situational constraints of the types 1
and 2 above, as faced by Hayek. Chapter 4 treats the latter's transform-
ation of the Misesian general-equilibrium construct, which led to the
incorporation of knowledge and time. In this manner the relevant equilib-
rium construct became an intertemporal general equilibrium. By distin-
guishing between individual and general equilibrium Hayek could address
the so-called 'coordination problem', which forms the central problem of
what will be called the Hayek Programme, as outlined at the end of
Chapter 4. This outline also entails Hayek's situational constraints of type
4, his methodological constraints. Chapter 5 describes Austrian economics
in the Keynesian era, during which Lachmann's work on the role of
expectations constitutes its main development. This work is interpreted as
an attempt to elaborate on Hayek's views. Chapter 6 contains the revival
of Austrian economics, which took place in the 1970s and 1980s. This
revival started with Hicks's re-opening of the debate on the Ricardo
effect. Its main focus, however, was Kirzner's analysis of the role of
entrepreneurship. The dis~ussion shows that one of Hayek's methodologi-
cal constraints changed over time.
Part II consists of the description of the developments in the New
Classical research tradition. Chapter 7 gives some of its pre-history, in
particular concentrating on the Friedman-Phelps explanation of the
Phillips curve. Analogous to Chapters 2 and 3 of the Austrians, it dis-
cusses the New Classical situational constraints of type 1 and 2. Chapter 8
describes the rise of New Classical Economics (NCE), particularly its
central propositions and modelling strategies. Furthermore, it identifies
some defensive strategies, which New Classicals use to immunize their
economic model from empirical discorroboration and refutation. Chapter
9 is concerned with an apparent anomaly of New Classical models, namely
the problem how to generate persisting cyclical movements of the econ-
omic system as a result of random shocks in a general-equilibrium frame-
work. This problem of persistence was solved by adopting the Slutzky-
Frisch hypothesis. Furthermore, it includes a reconstruction of Lucas's
problem situation in the shape of the 'Lucas Programme'. Chapter 10 dis-
cusses the problem shift which the NCE experienced in the mid-1970s. It
increasingly became concerned with the theory of economic policy, and
more in particular, with the 'neutrality proposition'. This proposition holds
that anticipated monetary policy does not have real effects. The debate on
this proposition contained a debate on the optimal form of monetary
7
policy, the so-called 'rules-versus-discretion' debate. Two approaches
which tried to settle the debate on empirical grounds are distinguished.
These approaches led to opposite results, one favouring the joint NR/RE
hypothesis and the other discorroborating it. The latter generated a
problem switch from monetary to real business cycle theory, which is also
discussed in Chapter 10, albeit briefly.
The descriptions of the respective developments in Austrian and
New Classical business cycle theory allow for a comparison between them.
This comparison is the subject of Chapter 11. It will explain the differ-
ences between the two traditions in terms of the different problems which
they try to solve. The NCE is shown to abstract from those features which
Austrians regard crucial. This comparison thus makes some caveats to
Lucas's claim that the NCE builds on Hayek's work. Finally, Chapter 12
contains a summary, some conclusions, and an epilogue.
8
PART I. AUSTRIAN BUSINESS CYCLE THEORY
2. EARLY VERTICAL MALADJUSTMENT
THEORY
2.1. INTRODUCTION
Austrian business cycle theory evolved from nineteenth- and early twenti-
eth-century business cycle theories. Its most important predecessors were
the theories of Michail Tugan-Baranowsky, Arthur Spiethoff and Gustav
Cassel, which stressed the importance of the rate of interest in the indivi-
duals' savings and investment decisions, as does the Austrian theory. The
latter thereby adopts Menger's and Bohm-Bawerk's views on capital as a
set of heterogeneous physical goods. These goods can be classified in so-
called stages of production. In fact, the theories referred to above can be
characterized by the term vertical maladjustment theories (Haberler (1937,
p. 30) ), because they stress that the business cycle is characterized by
maladjustments between the stages of production. Obviously, the idea of
maladjustments presupposes some notion of adjustment. Such a situation
of perfect adjustment can be viewed as an equilibrium situation. When
referring to an equilibrium position, we must make clear how such a
position is defined. In doing so we base our analysis on Wicksell's distinc-
tion between the natural and the money (or market) rate of interest.
This chapter aims at providing a short overview of the 'pre-Austri-
an' business cycle theories. It is organized as follows. Firstly, some pre-
Austrian 'vertical adjustment' business cycle theories are discussed,
namely those of Tugan-Baranowski, Spiethoff, and Cassel respectively.
Section 2.5 will provide a brief introduction into Austrian capital theory.
This introduction will be confined to the Mengerian structure of producti-
on. Finally, Wicksell's distinction between the natural and the market rate
of interest is addressed.
2.2. TUGAN-BARANOWSKI
One of the earliest 'vertical-maladjustment business cycle' theorists was

Michail Tugan-Baranowski, a Russian economist who lived during the
turn of the century. Tugan-Baranowski did not resort to any exogenous
11
influences, like Jevons or Moore. 1 Instead, he referred to investment and
savings behaviour. He argued that there are two types of capital, free
(loanable) capital and fixed capital (e.g., machinery). These types fluctu-
ate differently over time. Free capital, which may be interpreted as
savings, is accumulated continuously, so that its accumulation not only
takes place during booms but also during depressions. This latter result is
brought about by Tugan-Baranowski's assumption that in the latter
periods real incomes of 'the salaried and creditor classes' will increase for
quite some time. Hence the amount of free capital grows during all
phases of the cycle. By contrast, fixed capital ('machinery') will only be
increased during the boom. The transformation of free capital into fixed
capital therefore takes place by 'spurts'. When it is not yet transformed, it
forms latent purchasing power. As soon as some use for these accumula-
ted funds is found (e.g., a new market is opened up), the latent purchasing
power becomes effective. 2 Demand for capital goods is then increased,
prices of these goods rise, and after some time output of the capital goods
industries will begin to rise. When the new equipment is finished and all
free capital is transformed into fixed capital, the capital goods industries
are confronted with a fall in demand, leading (among other things) to a
decrease in the quantity of raw materials demanded by these industries. It
then turns out that they have been overexpanded as compared to the
consumer goods industries. That is, vertical disproportionalities emerge.
The capital goods industries are then confronted with excess productive
capacity. Tugan-Baranowski argued that entrepreneurs will nevertheless
produce at full capacity, because inactive capital 'cannot pay interest'.
Entrepreneurs in the capital goods industries will therefore produce at
more or less full capacity, meanwhile gradually reducing excess capacity.
General overproduction follows, leading to falling prices of machinery.
This deflationary process will continue as long as overproduction occurs.
It will only come to a halt when entrepreneurs have eliminated their
1
Jevons's business cycle theory considered the effect of sunspot on harvest cycles,
which in tum were supposed to cause industrial fluctuations. Moore explained these
fluctuations by raincycles, which in tum were seen to depend on the movements of the
planet Venus. For an overview of both cycle theories, see Morgan (1990, pp. 18- 39).
2
Tugan-Baranowski used the effect of steam in a steam engine as a metaphor of the
effects here described. When the pressure of the steam (accumulated free capital) reaches
a certain force, the piston (industry) is forced into motion (is forced to invest in fixed
capital) and is pushed to the end of the cylinder (is bound to expand productive capacity as
far as possible). The steam (free capital) escapes, and the piston (industry) returns to its
former position.
12
excess capacity. However, during the depression fixed capital is trans-
formed into free capital because, as was already stated, people continue
to save, that is, to accumulate free capital. As soon as the latent pur-
chasing power is invested, a new boom starts.
2.3. SPIETHOFF
Arthur Spiethoff was strongly influenced by Tugan-Baranowski. His

analysis also allows for general overproduction. However, he did not
adopt Tugan-Baranowski's explanation of the cyclical pattern in terms of
investment and savings behaviour. In the 1903 edition of the Jahrbuch fUr
Gesetzgebung, Verwaltung und Volkswirtschaft (p. 696) he wondered what
the volume of savings had to do with the cyclical accumulation of capital.
That is, why are huge masses of free capital first piled up without being
invested? Tugan-Baranowski's analysis could not provide answers for these
questions. Spiethoff therefore turned to a different explanation of cyclical
fluctuations, and identified two causes. Firstly, the boom is brought about
by innovations, or the discovery of overseas markets, either of which could
raise profitability in some particular sector, thus starting the upswing.
However, this cause explains by no means the recu"ence of industrial
fluctuations. This led Spiethoff to resort to a second explanation, which
explains the crisis in terms of limited opportunities for investment. During
the boom, output of consumer goods will lag behind investment, so the
prices of these goods stay high, keeping profits high. At some point in
time, though, the newly produced investment goods will be installed and
will be starting to produce consumer goods. Production of these goods will
rise, thereby lowering their price. This affects profits in the consumer
goods industries. Investment opportunities now become limited. Entre-
preneurs in these industries will then curtail their investments, and the
crisis spreads to the producer goods inrlustries. It turns out that the wrong
goods have been produced. Too much of investment goods have been
produced and the resulting excess capacity must be eliminated. This
process of elimination takes place during the depression. Pessimism and
reluctance to invest arise, hampering the recovery, although the severity of
these phenomena will depend on the abruptness with which the boom
ended. During the depression monetary funds are accumulated again,
since the savings behaviour as presupposed by Spiethoff is similar to that
in Tugan-Baranowski's theory. Eventually the revival will emerge through
13
an increased profitability of hitherto unprofitable investment projects,
caused by a lowering of the cost of capital construction (due · to the
reduction in wages, the fall in the price of raw materials, the reduction of
interest charges, the adoption of improved methods of production, etc.; cf.
Haberler (1937, pp. 80 - 81)). Entrepreneurs will undertake these new
investments, thereby drawing on the monetary funds, which people have
accumulated during the depression phase. The cycle then repeats itself.
2.4. CASSEL
Cassel (1918 (1923)) built on Spiethoffs over-investment theory. He

argued that at the beginning of the boom profits will be high relative to
wages. Banks will then be inclined to lend at too low a market rate of
interest, because there is a large supply of loanable funds. High profits
and low market rates of interest will stimulate investment. Free (monet-
ary) capital is transformed into fixed (real) capital. As the boom proceeds,
more production is devoted to capital formation. However, the propensity
to save does not rise, so that eventually and inevitably a shortage of free
capital arises, raising market interest rates. Additionally, labour becomes
more scarce, which pushes up the wage rate. Both effects will lower the
demand for investment goods. In .the subsequent depression these effects
are reversed, which means that eventually investment in fixed capital will
become profitable again.
According to Backhouse (1987, p. 184), Cassel identified several
lags which prevented the economy from being in equilibrium continuously.
Particularly, these lags concerned "... the response of investment to
changes in interest rates; the reaction of interest rates to changes in
investment; and the time taken between the start of an investment project
and its completion." As in Spiethoffs theory, fluctuations in the produc-
tion of capital goods spread over the economy, causing other industries to
experience such fluctuations as well. Where Cassel differed from Spiethoff
was in seeing a greater role for monetary factors. Spiethoff had argued
that the shortage of capital at the crisis was a shortage of real capital
goods. Cassel attributed the crisis to a shortage of monetary savings, thus
linking the concepts of capital and money. It is in this regard that his
theory is similar to the Austrian malinvestment theory. Ludwig von Mises
placed even more emphasis on the role of monetary factors. He linked his
monetary theory, including its nonneutrality of money, with Menger's
14
structure of production and Wicksell's analysis of interest rates. 3 Before
addressing Mises's theory, we must first discuss these latter two elements.
2.5. THE STRUCTURE OF PRODUCTION4
The early business cycle theories under consideration distinguished

between 'free' ('loanable') and 'fixed' capital, or between 'monetary' and
'real' capital. This latter type of capital consists of heterogeneous, physical
investment goods. Menger (1871 (1968), pp. 8 - 10) tried to bring some
sort of order into this heterogeneous set by classifying the goods according
to the functions they perform in the process of production, using their
remoteness from consumption as a criterion. In Menger's terminology,
consumer goods are 'goods of the first order'. Capital goods which are
used in the production of these first-order goods are called second-order
goods; capital goods which produce second-order goods are called third-
order goods, and so forth. It is then possible to distinguish stages of
production which can be categorized in a manner similar to that of capital
goods, and which use capital goods of the same order. Taken together,
these stages form a vertical structure of production, or capital structure,
which can be seen as the reflection of all past investment decisions.
Bohm-Bawerk (1889 (1921)) reformulated Menger's capital theory
by adopting three concepts, namely the absolute and average period of
production and the average waiting time. 5 The absolute period of production
(Produktionsperiode) is defined as the time interval elapsing between the
first application of primary factors and the emergence of the final (con-
3 Lachmann (1943) even called the Austrian business cycle theory the 'Austro-
Wicksellian theory'.
4 The purpose of this section is not to give a complete account of controversies in

Austrian capital theory. It is merely intended to give the reader an idea of the hete-
rogeneous and subjectivistic nature of the Austrian ·notion of 'capital'. For a more detailed
analysis of Austrian capital theory, see e.g. Garrison (1981) and (1985}, Endres (1987}, and
Zuidema (1989).
5 It should be noted that Bohm-Bawerk (1889 (1921}, pp. 39 - 43) distinguished

between three types of capital: (1) the subsistence-fund ('Subsistenzfonds'), which contains
all goods in the economy under consideration (except its landownership); (2) social capital
('Produktivkapital' or better 'Sozialkapital'), which consists of all goods which are used for
production purposes; and (3) private capital ('Privatkapital' or 'Erwerbskapital'}, which
renders its owners some rent. In our analysis, the differences and similarities in these
notions will be disregarded.
15
sumer) product (p. 117). This concept is important insofar as it enables
the determination of the average period of production (durchschnittliche
Produktionsperiode ). The latter is defined as the time which on average
elapses between the first application of primary factors and the final
emergence of consumer goods. This average period equals the absolute
period only if the inputs uniformly ('gleichmat3ig') enter the production
process. That is, the absolute and the average period of production are
identical only if all periods of production are identical. Bohm-Bawerk's
third concept is the average waiting time (durchschnittliche Wartezeit). It ex-
presses the average time that elapses between the moment at which a
specific factor input is applied in production, and the moment at which
the output becomes available for consumption.6 The average waiting time
is half the average period of production.7 Bohm-Bawerk argued that both
average periods could be used as a measure for the 'degree of capitalism',
that is, for the 'roundaboutness' of production. By contrast, Menger did
not consider Bohm-Bawerk's introduction of these concepts an improve-
ment. 8 In the Mengerian-subjectivist view, the structure of production
should reflect the use which is made of the heterogeneous capital goods.
By contrast, the concepts 'average period of production' and 'average
waiting time' are intended to give a one-dimensional measure for this
structure. Their use implies that such a measure is at all possible. In turn,
this means that the heterogeneous capital goods can be aggregated into
an unambiguous, homogeneous concept 'capital'.9 According to Streissler
and Weber (1974, p. 231), Menger (1888) forestalled Bohm-Bawerk's
publication by emphasizing the heterogeneity of capital goods. Bohm-Ba-
werk's student Mises and Wieser's pupil Hayek shared his criticisms, and
6 Blaug (1962 (1990), p. 508) formulated Bohm-Bawerk's (1889 (1921), p. 118)

discussion of the concept of 'average period of production' in mathematical form as 9 =
(K/1), where K stands for the amount of real capital, I for the rate of investment, and 9
for the average period of production.
7
This can be made clear as follows. Suppose that factor inputs uniformly enter the
production process, and that the absolute (and hence the average) period of production
takes one year which is divided into weeks. The (absolute) waiting time of the first input
employed is 52, that of the second 51, that of the third 50 weeks, and so on. The waiting
time of the last input is only 1 week. This means that the average waiting time of the
production process is 26 weeks, that is, half the average period of production.
8
According to Schumpeter (1954 (1986), p. 847n8), Menger called Bohm-Bawerk's
capital theory "one of the greatest errors ever committed". Presumably, this harsh rejection
to a large extent reflects Menger's disappointment with his student's theory.
9
Cf. also Hayek (1941, pp. 5-6).
16
returned to his more subjectivistic view on capital. 10
If the structure of production is in equilibrium, then it corre-
sponds to the decisions made by the economic agents as regards con-
sumption and saving. That is, the individuals' decisions as regards time-
preferences are expressed in the rate of interest, which governs the profi-
tability of investment projects. In this sense, the rate of interest brings
about intertemporal coordination. However, such intertemporal
coordination does not always need to take place. We shall return to this
issue later. For now it suffices to hint at a difference between pre-
Austrian and Austrian vertical maladjustment theories. In general, vertical
maladjustment theories claim that the phenomena constituting the busi-
ness cycle are caused by a lack of intertemporal coordination. There is
then a lack in correspondence between the structure of production
(investment) and the decisions made by the economic agents as regards
consumption and saving. The pre-Austrian theories did not distinguish
between different stages of production. This means that in their view
either the economic agents save too little, or they save too much. Business
cycles are thus caused by either overinvestment or underconsumption. By
contrast, Menger's less aggregated structure of production also allowed for
investments in 'wrong' (ex-post non-optimal) directions. Such malinvest-
ments constitute vertical maladjustments in the structure of production
because they imply that the industries or stages of production are dispro-
portionately related to each other in vertical order. 11
The capital goods used in a particular stage are assumed to be
highly (though not completely) specific. Capital goods can only be trans-
ferred from one production process to another with great difficulty. The
same applies to the transfer of these goods from one stage of production
10
According to Faber (1986b) and Zuidema (1989), there are two groups of (German)
economists which may also be labelled 'Austrian'. The first group includes economists such
as Von Weizsicker and Orosel. They reformulated the 'period of production' to summar-
ize the structure of production. In this sense they build on the work of Bohm-Bawerk. The
second group includes the followers of Stackelberg and Bernholz, who concentrate on the
relationship between the length of the period of prOduction ('roundaboutness') and the
time structure of different technologies. The present analysis concentrates on the
subjectivistic (Austro-American) followers of both Mises and Hayek. Both groups
mentioned above are not discussed, bec8use their relation to Menger and Bohm-Bawerk
appears too indirect to be meaningful. For an analysis of their respective viewpoints, see
Faber (1986b) and Pellengahr (1986).
11
By contrast, horizontal maladjustment theories claim· that the cycle arises because of
horizontal disproportionalities, that is, disproportions between stages of production of the
same 'rank' (as measured from consumption).
17
to another. Once capital goods have been produced, bought and installed,
the investment is often irreversible. This irreversibility has as a conse-
quence that during a business cycle the adjustment of the capital structure
to its equilibrium position is hampered. It is one of the factors which
make the business cycle a phenomenon with undesired consequences.
Two important criticisms may be advanced against the structure
of production. The first criticism concerns what may be called the problem
of classification. It holds that it is not possible to determine unambiguously
the order of a particular good if it is used for different purposes. This may
be clarified by an example. Consider a particular car which is simulta-
neously used as a producer good (e.g. to deliver consumer goods) and as a
consumer good. The question then arises whether the goods which are
used in the production of the car are goods of the second or the third
order. The second criticism raises a more severe problem and may be
considered a generalisation of the problem of classification. The problem
of circularity holds that the concept of stages of production is meaningless
because production takes place in a circular manner. 12 That is, it claims
that there is no unambiguous 'verticality' which characterizes production
in advanced industrial economic systems. 13 Skousen ( 1990, p. 155) ac-
knowledged that in such systems it is difficult to classify all goods unam-
biguously, but he maintained that this does not mean that the concept of
structure of production becomes meaningless. After all, many capital
goods are highly specific, which means that they cannot be transferred
from one project to another without incurring costs. Furthermore, the use
of the concept makes clear that the heterogeneous capital goods are
mutually dependent, that is, that there are chains of production which can
be lengthened or shortened. The 'price' which is presumed to determine
the lengths of the structure of production is the Wicksellian 'natural rate
of interest'.
12
Economists who rejected the verticality of the structure of production on this ground
are Marshall, Knight, Stigler, Sraffa, and Shackle. For a more detailed analysis of the
problem, see Skousen (1990, pp. 151- 57.
13
As Shackle (1981, p. 239) observed, "[p]roduction in an advanced industrial society
needs for its description a Leontief table of input coefficients, where, in principle ... every
operation or every industry or sector is deemed actually or potentially to contribute means
of some kind to every other, both directly and indirectly. Such sectors as transportation,
the telephone system and the electric power industry plainly have a hand to some degree
in everything that is done by anybody anywhere."
18
2.6. WICKSELL
Knut Wicksell ( 1898) distinguished between two ·different notions of inte-

rest rates, namely the natural and the market (or money) rate of inte-
rest.14 He defined the former as the rate which would arise if real capital
goods would be exchanged directly, without the use of money. This rate
can be seen as the long-run equilibrium rate of interest, which is deter-
mined by the agents' valuation of present to future goods, that is, by their
time-preferences. 15 It expresses the rate at which individuals are pre-
pared to exchange present-period consumption for consumption in future
periods. This means that it brings about equilibrium between the supply
of savings and the demand for investment funds. In such an equilibrium
the natural rate will (roughly) equal the expected yield on newly created
capital. Apart from the natural rate there also exists a rate of interest
which is determined on the loan market. This 'market' (or money) rate of
interest is the rate at which the supply of loanable funds equals the
demand for loans. It is the price to be paid for loans.
The natural rate of interest reflects individuals' time-preferences,
so we may expect that this rate governs the individuals' decisions concern-
ing present and future consumption, and therefore their actions on the
loan market. In general equilibrium the natural rate of interest must
designate the real rate at a level at which the supply of savings and the
demand for investment funds are equilibrated, and the money rate at a
level at which the general price level remains constant. This will only be
the case in a stationary economy. 16 Wicksell thus adopted a stationary
14
As Haberler (1937, p. 36, note 1) observed, "[t]he concept of a 'natural rate' (and
even the term) can be found in earlier English economic writings. "For instance, see Adam
Smith (1776, l.vii,p. 65) and Henry Thornton (1802, pp. 253- 55). Wicksell's student Carl
Uhr (1960, p. 200) opined that Wicksell was not directly exposed to Thornton's ideas, and
that instead the influence was merely indirect, through Ricardo.
15 Wicksell (1898, p. 93, italics in original) stated that "[j]ene Rate des Darlehenszinses,
bei welcher dieser sich gegeniiber den Giiternpreisen durchaus neutral verhilt und sie
weder zu erhohen noch zu erniedrigen die Tendenz hat, kann nun keine andere sein als
eben diejenige, welche durch Angebot und Nachfrage festgestellt werden wiirde, falls man
sich iiberhaupt keiner Geldtransaktionen bediente, sondern die Realkapitalien in natura
dargeliehen wiirden - oder was etwa auf dasselbe hinauskommt, als der jeweilige Stand des
naturlichen Kapitalzinses."
16
Cf. Garrison (1981, p. 62). According to Hayek (1933a (1975), p. 115n), David
Davidson was the first to draw attention to this fact. Cf. also Uhr (1960, p. 279ff.) and
Leijonhufvud (1968, p. 222).
19
state as the benchmark situation of his theoretical framework. 17 This is
not to say that Wicksell (1898) claimed that this benchmark situation
would actually arise. He merely argued that a tendency towards the
equalization of both rates of interest exists (p. 108). 18 Obviously, he had
to account for this tendency, that is, he had to explain the tendency in
terms of equilibrating (hence disequilibrium) processes. For this purpose,
he identified two effects.
The first effect is the real-cash-balance effect, or direct mechanism.
This effect holds that any (exogenous) increase (decrease) in the quantity
of money will lead to a change in the actual distribution of the individu-
als' cash holdings and spendings. The actual distribution then diverges
from the desired distribution, inducing individuals to restore the latter.
Spendings will be increased (decreased). Given the level of output, this
leads to a rise (fall) in prices. Hence, in the long run an increase (de-
crease) in the quantity of money will lead to an increase (decrease) in the
level of prices. In this manner, Wicksell showed that in the long run the
quantity theory of money holds. However, the real-cash-balance effect
does not allow for a link between capital theory and monetary theory.
Garrison (1981, pp. 72 - 73) argued that cash balances are homogeneous,
and can be reallocated rather easily. The transition period during which
such an adjustment takes place is a short, smooth, crisis-free period. By
contrast, capital consists of a set of heterogeneous goods which can often
be used only for rather specific tasks. Once installed, these goods cannot
be transformed or reallocated immediately. Instead, in general disequilib-
rium periods any misallocation in the capital structure will take time to
remove. Wicksell's real-cash-balance effect was thus unable to account for
prolonged transition periods in capital-using economies. The purpose of
linking capital theory with monetary theory was fulfilled by an indirect
effect. This effect does not consider the increase in the supply of money to
be exogenous. Instead, it starts from a divergence between the natural and
the market rate of interest. In Wicksell's view, the natural rate is not
17 It may be argued that Hayek's (1928) notion of intertemporal equilibrium is an

attempt to solve this problem. This notion will be discussed in Chapter 4.
18
Wicksell (1898, p. 108, italics in original): "(m]an kann ... mit Sicherheit erwarten,
dass der Bankzins oder, algemeiner gesprochen, der Geldzins sich schliesslich immer dem
Stande des natiirlichen Kapitalzinses anschliessen wird oder vielmehr - da ja neue
Vetinderungen des natiirlichen Zinsfusses unterdessen eingetreten sein konnen - immer
die Tendenz hat, sich demselben anzuschliessen. Ob aber dies auch mit hinreichender
Schnelligkeit geschieht ... erscheint von vornherein sehr fraglich. "
20
constant, but instead fluctuates. However, banks do not adapt the market
rate of interest continuously. They are presumed to be slow in their
adjustment of this rate to the natural rate. This means that the banks can
hold the former below the latter. In turn, this will induce an expansion of
credit, hence an increase in the supply of money. The fall in the market
rate of interest means that the price of borrowing money has become
lower than the marginal benefits of investment goods. This encourages
investors to increase their investments, using the additional credit. Effec-
tive demand is thus increased, whilst the level of output remains constant.
The result will be a rise in the level of prices. According to Wicksell, this
indirect mechanism links the money market with the capital market (i.e.
market for investable funds). However, this linkage clearly depends on his
capital theory. Although Wicksell adopted the Austrian (Bohm-Bawerki-
an) capital theory, he did not use the notion of structure of production in
his formal analysis of the indirect mechanism. Instead, in his Geldzins und
Guterpreise ( 1898) he simply aggregated this structure by using the notion
'capital' .19 His cumulative process (and business cycle theory) only allows
for overinvestment (because his formal analysis could not distinguish
between types of capital goods). As during this process the market rate of
interest is below (above) the natural rate, the investment in more (less)
capital-intensive methods of production is too cheap. This means that the
capital/labour-ratio will increase (decrease). In other words, too much
(little) capital is used. By contrast, the Austrian notion of 'capital' is a
heterogeneous one, which allows for rna/investments. Therefore, Austrians
argue that the indirect effect should constitute what Garrison (1981, p. 84)
has called a capital-allocation effect. This effect is presumably the hall-
mark of Austrian business cycle theory, and was first discussed by Ludwig
von Mises.
19
According to Garrison (1981, p. 63), Wiclcsell used the symbol 'K' to represent the
sum of the discounted values of the various capital goods. In Wicksell's (1898, p. 120) own
words: "Kennt man ... das in jeder besonderen Industrie, oder vielmehr fiir die Erzeugung
jeder besonderen Art von Genu Jjgiitem verwandte Kapital k, sowie dessen Investierungs-
zeit t, so kennt man auch das ganze industriell beschiftigte Kapital K = rk." He thus used
an aggregated notion of capital K, " ...wobei K = rk natiirlich die Wertsumme der simtli-
chen zur Zeit kapitalistisch investierten Genu Jjgiiter, in irgend einem von ihnen oder in
Geld aufgeschitzt, bezeichnet."
21
3. THE ONSET OF THE AUSTRIAN THEORY
Ludwig von Mises provided three main contributions to Austrian econ-

omics. Firstly, in the 1920s he attacked Barone (1908) views on socialism,
arguing that in such an economic system no rational economic calculation
was feasible. The second contribution consists of his monetary theory, in
which he criticized the quantity theory of money. This criticism particular-
ly concerned the oversimplified representations of this theory, which
concentrated on the general price level. Mises's monetary theory provided
a more 'genetic-causal' view on money. He used this monetary theory to
explain cyclical fluctuations in economic activity. As this explanation
constitutes the first version of Austrian business cycle theory, it forms
Mises's third contribution to Austrian economics. The present study is
concerned with the development in Austrian business cycle theory, so that
it concentrates on the latter two contributions only. This is not to say,
though, that the former is irrelevant for our purposes. As Lavoie (1985)
pointed out, the debate on rational economic calculation under socialism
particularly influenced Mises's views on subjectivism, so that the dis-
cussion of these views in the present chapter contains these influences,
albeit it only implicitly. 1 However, before discussing Mises's business
cycle theory, some methodological issues must be covered in order to
understand the general framework in which it must be interpreted. This
framework particularly concerns Mises's views as regards the nature of
explanation in the social sciences, that is, his 'praxeology'.
3.1. PRAXEOLOGY AND ECONOMICS
Mises (1949 (1966), p. 30) argued that "[t)here are two main branches of
the sciences of human action: praxeology and history." He defined these
two as follows: "History is the collection and systematic arrangement of all
the data of experience concerning human action. It deals with the con-
crete content of human action. It studies all human endeavors in their
infinite multiplicity and variety and all human actions with all their
accidental, special, and particular implications." In contrast, "[p]raxeology
is a theoretical and systematic, not a historical science. Its scope is human
1
For an explicit treatment of these influences, the reader is referred to Lavoie (1985).
23
action as such, irrespective of all environmental, accidental, and individual
circumstances of the concrete acts. ... It aims at knowledge valid for all
instances in which the conditions exactly correspond to those implied in
its assumptions and inferences. Its statements and propositions are not
derived from experience. They are, like those of logic and mathematics, a
priori."2 Praxeology deals with the implications of the action axiom. Mises
(1962 (1978), pp. 5 - 6) regarded this axiom and its implications as incon-
testable and universally valid. This means that "[ t]he theorems attained by
correct praxeological reasoning are not only perfectly certain and incon-
testable, like the correct mathematical theorems. They refer, moreover,
with the full rigidity of their apodictic certainty and incontestability to the
reality of action as it appears in life and history. Praxeology conveys exact
and precise knowledge of real things" (Mises (1949 (1966), p. 39). In his
view, praxeology cannot reach false conclusions, unless some logical error
is made. The only reasonable test of praxeology thus is a test of its logical
consistency. Hence the praxeological conclusions and theorems need not
be tested empirically. 3
According to Mises (1949 (1966), pp. 232 - 34), economics is that
part of the social sciences which is concerned with market phenomena. It
builds on the praxeological theorems. The question then arises whether
the economic propositions are also true and need not be tested. The
answer to this question will depend on the relationship between the
economic theorems and those of praxeology. More precisely, it will
depend on whether the economic theorems can be derived from the
action axiom without the use of auxiliary assumptions.
If the economic theorems could be derived from the praxeological
postulates without using auxiliary assumptions, they would also be incon-
testable. However, man's economic actions take place in a social environ-
ment, the market, which influences his actions. The features of this
environment therefore play a significant role. Praxeologists acknowledge
that their conclusions as regards market phenomena cannot be derived
from the action axiom alone, so that auxiliary assumptions are needed in
order to derive hypotheses concerning market phenomena. Mises (1949
(1966), p. 237) discerned two such auxiliary assumptions, which also
2 For an analysis of the logical foundations of the action axiom, cf. Huussen (1989),
Parsons (1990) and Van Zijp (1991a).
3This is not to say that the theorems always apply to reality. They merely apply under
specific conditions, namely in those circumstances in which the praxeological assumptions
prevail.
24
appear to state the necessary preconditions for the existence of a market.
He argued that this existence implies that" ... there is division of labor and
private ownership (control) of the means of production and that conse-
quently there is market exchange of goods and services." Furthermore, all
coercive and compulsive influences are assumed to be absent (i.e. the
economic system operates under idealizing conditions). For economic
calculation on the market to be rational, he continued, there must also
exist a universally employed medium of exchange, because otherwise "[i]t
would not be possible to reduce all exchange-relationships to a common
denominator" (Mises (1933, p. 102)). Such a reduction is necessary in
order to assess whether some productive activities can be expected to be
profitable (p. 106), i.e. for rational economic calculation to take place.4
Another important assumption, implicitly made by Mises, is that
the market process tends towards a situation of market equilibrium. Mises
(1949 (1966), p. 244) pointed out that "... action ultimately aims at bring-
ing about a state of affairs in which there is no longer action, whether
because all uneasiness has been removed or because any further removal
of felt uneasiness is out of the question. Action thus tends toward a state
of rest, absence of action." Human action is equilibrating, in the sense
that it aims at making itself superfluous. Any successful action (that is, an
action which successfully eliminates profit opportunities) implies an
equilibrating movement, as seen from the viewpoint of the successful
actor. On the level of the individual actor all successful action is equilibra-
ting, because it eliminates profit opportunities. It is in this sense that
praxeologists regard the proposition that action is equilibrating as a priori
true. However, this does not imply that the individual actions are
equilibrating on an economy-wide (or even market) level. In order to
prevent confusion between the individual and the general level, general
equilibrium will be called coordination. 5 The tendency towards
coordination is much disputed among Austrians. Mises (1949 (1966), pp.
244 - 50) assumed that individual equilibrating action would also bring
about coordination. The resulting general equilibrium (which Mises called
the 'evenly rotating economy', ERE) is characterized by the complete
absence of perceived profit opportunities. However, it will never be
reached because it moves each time the data of the economic system
4
Cf. also Mises (1949 (1966) p. 209).
5
The distinction between equilibration and coordination was already used by O'Dris-
coll (1977a) and Selgin (1988), among others.
25
change. In this sense the ERE is a 'moving target'.
The assumption concerning the tendency towards the ERE,
however, does not follow strictly logically from the action axiom (as will
be shown in the next chapter). This means that it need not be a priori
true. Hayek (1937) pointed out that the coordinating tendency depends
very much on the individuals' knowledge and expectations. However,
Mises did not explicitly incorporate expectations into his analysis. Bohm
(1982, p. 52n29) concluded that Mises could only give a vivid picture of
the market process because he left his apriorism behind. Mises's 'pure
logic of choice' framework cannot deal with time and expectations
because it treated preferences as subjective and at the same time as
exogenous (i.e., given in economic analysis). Expectations were only
incorporated implicitly, as were knowledge and knowledge acquisition
(learning) processes. We shall see later that this can be attributed to
Mises's relatively static subjectivism. Replacing this type of subjectivism by
a more dynamic version, as. Lachmann did, makes Mises's conclusion
contingent, to say the least.
Praxeology must thus be supplemented by auxiliary hypotheses in
order to be useful in the explanation of market phenomena. As at least
one of these auxiliary assumptions, the tendency towards coordination, is
empirical as well as contingent, the resulting economic theories are also
contingent. This means that empirical testing should play a role. Although
the praxeological theorems may be true, the economic conclusions need
not be. Hence they should be confronted with empirical reality. Of course,
this should also apply to Mises's business cycle theory, which clearly
belongs to economics, and not to praxeology. Contrary to what is often
asserted as regards his methodological stance, Mises appears to have
recognized this methodological conclusion. In discussing competing
explanations of cyclical phenomena he acknowledged that economic
theories should be tested empirically in order to establish whether they
are not only logically but also empirically valid.6
6 Mises (1928, pp. 40 - 41, my italics) stated that "... die Theorie der allgemeinen
Uberproduktion und die Unterkonsumtionstheorie konnte die theoretische Priifung als
ganz verkehrt abweisen ... Nicht so Ieicht hatte es die Kritik mit einer dritten Gruppe von
Erklarungsversuchen, die die Schwankungen der Konjunktur aus dem periodischen
Wechsel natiirlicher Voraussetzungen des Bodenertrages ableiten wollen. Man kann an
diese Lehren mit den Mitteln der theoretischen Untersuchung allein nicht herankommen.
Es ware denkbar, da {3 solche Einfliisse sich in regelma flger Wiederkehr gel tend machen;
ob es wirklich der Fall ist, kann nur der Versuch lehren, die 1heorie an der Beobachtung zu
verijizieren" Mises thus allowed for empirical testing of economic theorems. Praxeological
theorems need not be tested empirically, because they are supposed to be a priori true.
26
Mises's business cycle theory is a combination of his monetary
theory with the Austrian theory of capital and its notion of the structure
of production. As Wicksell's rates of interest and the structure of produc-
tion were already discussed in the previous chapter, attention is restricted
to Mises's monetary theory.
3.2. MONETARY THEORY
In his earlier works, Mises was mainly concerned with monetary theory.
His Theorie des Geldes und der Umlaufsmittel (1912 (1924)) explains the
existence of money and its functions in the economy. In his opinion the
sole function of money is to be exchanged. As money is scarce, it must be
considered an economic good and like all economic goods it has a price.7
This price will depend on demand for and supply of it, which in turn
depend on the individuals' valuations (pp. 85 - 86). In discussing the
theory of money, Mises concentrated on the demand for money, because
he regarded the money supply as largely determined by the monetary
authorities (i.e., the government and the central bank) and the private
banks.
Money does not satisfy wants directly, it merely facilitates
exchange. If one type of money is more readily accepted by individuals
than another, they will demand more of that type. 8 According to Mises,
the utility individuals derive from using money is equal to the (expected)
utility derived from using the goods bought by it.9 The problem now is
how economic agents determine the amount of money which they are
7
Of course, in order to give this price any meaning, one has to choose a numeraire
other than money itself. Money thus has as many prices as there are other goods.
8 In fact, this desirability of money as a means of exchange is the very cause of its
existence. According to Menger in his Grundsiitze (1871 (1968), pp. 250- 60), money is an
evolved social institution in the same sense as language is. In a barter economy individuals
consider it advantageous to exchange their supply of goods against more easily exchange-
able goods. As an unintended consequence of such human action, money comes into
existence. (In a footnote of almost an entire page, Menger even gave some etymological
views on the origin of the words used in different languages to denote the medium of
exchange). Mises also adhered to this view.
9
Mises (1912 (1924), p. 85): "Nun ist aber ... der subjektive Gebrauchswert des
Geldes, der mit seinem subjektiven Tauschwert zusammenfillt, nichts anderes als der
antizipierte Gebrauchswert der fiir das Geld anzuschaffenden Dinge; seine Gro8e ist zu
bemessen am Grenmutzen der fiir das Geld einzutauschenden Giiter."
27
willing to sacrifice in exchange. In order to solve this problem an agent
needs to know what quantities of other goods one unit of money may buy,
i.e. its purchasing power. This purchasing power forms a bridge between
the utility derived from consumption on the one hand, and the 'inutile'
money on the other. 10 However, this purchasing power will depend on the
demand for money. In other words, if the demand for money depends on
its price, and if the price of money depends on its demand, how can we
determine the latter? Mises recognized this danger of circularity in his
theory. 11 He tackled it by referring to the notion of time. Individuals
determine their demand for money on the basis of its value as determined
in the market yesterday. Obviously this leads to a problem of infinite
regress. The question then remains what determined this 'objective value'
of money at first. Mises solved this problem by indicating that a good can
only become money if it already possesses exchange value based on some
other productive or consumptive function it performs. 12 That is, he
explains the value of money by referring to the non-monetary purposes
which the money good used to have in the past. Money must be or must
have been useful for productive or consumptive purposes other than as
medium of exchange.
The demand for money is just one side of the coin. The other
side is constituted by its supply. Mises regarded this as largely determined
10 Mises (1912 (1924), pp. 85 - 6): "Da dem Geld als solchem jede direkte Beziehung
zu einem menschlichen Bediirfnis fehlt, kann das Individuum sich eine Vorstellung von
seinem Nutzen und mithin von seinem Werte nicht anders bilden, als indem es von einer
bestimmten Kaufkraft ausgeht. "
11
Cf. also Butler (1988, p. 267). However, Yeager (1989) argued that Patinkin (1956)
has shown that this circularity problem is only apparent. According to Moss (1976, p. 27),
Mises confused 'the utility of (holding) money' and 'the utility of services provided by
money'. Patinkin (1956, p. 63) appears to distinguish these different utilities. He incorpor-
ated real balances directly into the individual's utility function by counting real cash balan-
ces as a part of the individual's wealth. Unfortunately, in Patinkin's economy any good can
perform the function of money (by assumption it is the n-th good). According to Hahn
(1965), Patinkin's model does not do justice to the fundamental role played by money in a
money economy, because it neglects money's productive contributions in exchange, and
hence fails to do justice to the difference between a barter and a money economy.
12
Mises (1912 (1924), p. 87) argued that "[a]us der Tatsache, da8 der objektive
Tauschwert des Geldes stets einer Ankniipfung an ein auf dem Markte zwischen dem Geld
und den iibrigen wirtschaftlichen Giitem bereits bestehendes Austauschverhaltnis bedarf,
da das wirtschaftende Individuum anders nicht in der Lage ware, ein Werturteil iiber das
Geld abzugeben, folgt weiter, da8 als Geld nur ein Objekt in Verwendung genommen
werden kann, das in dem Augenblick des Beginnes seiner Tauschmittelfunktion bereits auf
Grund anderweitiger Verwendung objektiven Tauschwert besessen hat. "
28
by the monetary authorities (notably, the government and the central
bank) and the private banking system. However, this does not mean that
these authorities can manipulate the money supply as they like without
causing damage to the economy. This may be explained by Mises's view
that money is not 'neutral', i.e., that monetary changes have effects on
real variables. 13
The nonneutrality of money has led Mises (1912 (1924), 1928 and
1936 (1983)) to criticize the (Wicksellian, or perhaps better Quantity-
theoretical) view that the monetary authorities must stabilize the purchas-
ing power of money (and hence the general price level). One of the basic
tenets of 'mechanical versions' of the quantity theory of money is that,
whenever the supply of money changes, the purchasing power of money
changes inversely, given the demand for money. Money thus only influ-
ences the absolute level of prices but does not change relative prices. The
adjustment process is brought about by the so-called 'real-cash-balance
effect'. Monetary changes are presumed to leave real variables unaffected,
at least in the long-run. In contrast, Mises (1912 (1924), pp. 119 - 20)
claimed that the supply of money cannot be altered simultaneously and
uniformly for all individuals because the additional money cannot be
distributed immediately, simultaneously and proportionately over the
economic agents. As a result, the additional money will not be spent in
the same manner as the money which already existed. The real effects
caused by a change in the nominal supply of money will alter relative
prices and the distribution of income and wealth in the economy. 14 These
changes cannot be studied by using merely a price index, because different
individuals in fact purchase different baskets of goods and so may face
different degrees of change in the purchasing power of their income when
relative prices change (pp. 172 - 77). Obviously, Mises's nonneutrality of
money implies a rejection of the quantity theory of money. As a corollary,
he also rejected Wicksell's (1898) analysis of the 'cumulative process',
because it claimed that the quantity theory of money was valid, albeit only
13
Visser (1971, pp. 409, 429, 432 - 3) argued that pre-war economists defined
neutrality of money in terms of the maintenance of monetary equilibrium. In contrast,
post-war economists study the restoration of general equilibrium, following a monetary
disturbance. For a list of preconditions for neutrality of money in the post-war sense, cf.
Ascbheim and Hsieh (1969, pp. 213 - 15).
14
These real effects have become known as Cantil/on effects, since Cantillon (1755
(1931)) already argued that they may change the velocity of circulation, so that " ... by
doubling the quantity of money in a State the prices of products and merchandise are not
always doubled" (p. 177). See also Schumpeter (1954 (1982), p. 317).
29
in the long run. 15 Mises's analysis implies that the short-run non-neutrality
of money leads to changes which prohibit the return of the economy to its
initial general equilibrium. More specifically, Mises (1912 (1924), pp.
399 - 403) explicitly stressed that the changes in income distribution alters
relative prices and hence leads to changes in the structure of production,
even in the long run. Any process of coordination must therefore incorpor-
ate the process according to which the structure of production will be
adjusted. This short-run adjustment (or disequilibrium) process consists of
the capital-allocation effect. 16 Mises's emphasis on the structure of produc-
tion led him to consider this effect to be more important than the direct,
real-cash-balance effect. We shall analyze it in more detail when discus-
sing Mises's business cycle theory.
Nonneutrality has important implications for monetary policy. As
Butler (1988, p. 283) concluded, Mises claimed that "[m]onetary policy ...
does not affect everyone to the same degree." Therefore, Mises rejected
the stabilization of the price level as a policy aim. In his view the moneta-
ry authorities must refrain from credit expansion in order to make the
equalization of demand for and supply of real capital possible.
3.3. BUSINESS CYCLE THEORY
Mises's contribution to the Austrian theory of the business cycle is that he

linked Austrian capital theory with his theory of money. In his opinion,
the demand for (and supply of) capital makes itself felt in the demand for
(and supply of) moneyY Entrepreneurs must have money in order to buy
the capital goods required for production. They may either use their own
money or they can borrow it on the loan market. In either case the
15 Garrison (1981) has given a detailed comparison between Wicksell's and Mises's
respective views on monetary theory. In his view," ... Wicksell's formulation was a natural
starting point for further developments within the Austrian school" (Abstract). However,
he showed that Wicksell stood very much in the tradition of the quantity theory of money,
which concentrates on the effects of changes in the quantity of money on the general price
level. Austrians, following Mises, reject such an approach because it fails to take the
nonneutrality of money into account.
16 This effect was discussed at the end of the previous chapter.
17 Mises (1912 (1924), p. 348): "Die Kapitalsnachfrage tritt in der Form der Geldnach-
frage auf; der Kapitalbedarf ist scheinbar ein Geldbedarf. Das darf uns iiber das Wesen
der Erscheinung nicht tauschen. Das, was man Geldiiberflu6 und Geldknappheit zu
nennen pflegt, ist in Wahrheit Kapitaliiberflu6 und Kapitalknappheit."
30
market rate of interest determines how much money entrepreneurs will
invest in capital goods, and, therefore, the length of the period of produc-
tion.18 This period need not be the same as the length of time for which
individuals are prepared to postpone their consumption. A business cycle
arises if both periods differ in length.
Mises's analysis starts from his benchmark situation, the ERE. In
this situation individuals are presupposed to act identically as they did in
the previous period. Actions do not change, hence the economy is station-
ary or evenly rotating. Mises (1949 (1966), p. 248) stressed that this a
highly unrealistic condition, and that it "... is merely a tool for our think-
ing. It is not the description of a possible and realizable state of affairs."
After all, in an ERE entrepreneurship does not play any role anymore,
because agents have perfect knowledge, both about the present and the
future. Mises therefore concluded that economists should not study the
ERE, but instead the alleged tendency towards it.
Mises's business cycle theory builds on the monetary explanation
of the cycle as given by the so-called Cu"ency School, which tried to
explain the cycle by reference to the credit expansion resulting from the
issue of bank notes without metallic backing. 19 Mises (1936 (1983), pp. 1 -
2) criticized this school on two accounts. Firstly, he stated that it "... did
not see that bank accounts which could be drawn upon at any time by
means of checks, that is to say, current accounts, play exactly the same
role in the extension of credit as bank notes." He argued that the fact that
Peel's Bank Charter Act was not successful in curbing the credit expan-
sion, was due to the fact that it failed to regulate current accounts. 20
Secondly, the Currency School restricted its analysis to the case where
credit is expanded in only one country. The resulting increases in domestic
prices will hamper exports and encourage imports, which (given initial
external equilibrium) leads to a deficit on the balance of trade. Mises
18
The 'length of the period of production' reflects the 'capital intensity' of the
production processes used. The same applies to the so-called 'roundaboutness' of produc-
tion. In fact, 'roundaboutness', 'length of the period of production', and 'capital intensity'
may be regarded as synonyms. Thus, when the period of production is lengthened, produc-
tion becomes more roundabout, i.e., more capital-intensive.
19
Cf. Mises (1928, p. 39) and (1936 (1983), p. 1).
20
The Peel's Bank Charter Act of 1844 obliged the Bank of England to provide a
complete metallic backing for all bank notes issued in excess of 14 million pounds sterling.
Furthermore, it forbade private banks to increase the amount of bank notes issued.
However, it did not say anything as regards current accounts (Visser (1980), p. 195).
31
(1936, p. 2) concluded that "[u]ltimately the outflow of specie checks the
rise in prices." However, the Currency School did not analyze the case in
which credit expansion takes place in all countries simultaneously.
Mises tried to remedy these shortcomings. In his view, the
demand for and supply of capital manifest themselves in the demand for
and supply of money on the loan market. Mises started his analysis from a
situation in which the money market is in equilibrium. He argued that
credit expansion (if taking place simultaneously in all capitalist countries)
would lead to an increase in the supply of loanable funds. 21 This results
in a decrease in the market rate of interest. In turn, this will trigger of the
capital-allocation effect. Hitherto unprofitable investment projects appear
to become profitable. According to Mises (1912 (1924), p. 361), "... a
reduction of the rate of interest on loans must necessarily lead to a
lengthening of the average period of production." Obviously, this claim
only holds if entrepreneurs expect this reduction to persist long enough to
be able to grasp the profits offered by the reduction?2 Given this expec-
tation, entrepreneurs will try to invest, thereby increasing their demands
for resources and producer goods. The prices of these factors will rise
(including the real wage rate). In turn, prices of consumer goods will also
tend to rise (thereby of course reducing real wages). According to Mises,
these results culminate in a boom. It should be stressed that the boom
need not display overinvestment. As a matter of fact, Mises (1949 (1966),
p. 559) himself observed that "[t]he essence of the credit-expansion boom
is not overinvestment, but investment in wrong lines, i.e., malinvestment."
Mises continued by showing that the upward movement cannot
continue indefinitely. As long as the process of credit expansion is con-
tinued, the crisis can be put off, although this has its limits. Eventually,
the continued process of credit expansion will result in hyperinflation and
a consequent 'flight into real values'. On the other hand, if the credit
21 As Butler (1988, pp. 289- 290) pointed out, "[c]hanges in both the supply of money
and in the demand for it might initiate roughly similar disturbances ... ,but Mises's main
attention goes on changes in supply, which he suggests are more pronounced. Whatever its
origin, the point is that changes in the money relation, that is, the interplay between the
supply of money and the demand for it, bring about changes in prices and wages." The fact
that Mises concentrated on the supply side may be attributed to the monetary situation in
the period 1918- 1924. In this period inflation in Central Europe was extremely high, most
clearly noticeable in Germany. In 1923 the German mark was forced to depreciate in an
accelerating way, due to excessive and accelerating monetary expansion. (I owe this point
to Prof. G. Vandewalle, University of Antwerpen)
22
As will be clear, Mises neglected the crucial role of 'knowledge' and 'expectations'.
32
expansion stops, the market rate of interest will rise, thereby revealing
that the credit expansion has led to unjustified investments. Many firms
will have to curtail their activities (or even close down completely), which
brings about a period of depression. In this depression period it becomes
clear that many investment projects undertaken were in fact unprofitable.
The entrepreneurs under consideration have been misled by the distorted
market rate of interest and have embarked "... upon an expansion of
investment on a scale for which the capital goods available do not suffice"
(Mises (1949 (1966), p. 559). It will take some time before investment
projects prove to be unprofitable. In Mises's (1936 (1983), p. 4) view,
"[t]he capital invested in these [unprofitable] enterprises is lost to the
extent that it is locked in. The economy must adapt itself to these losses
and to the situation that they bring about." If this adjustment process is
completed, the market rate of interest will again equal the natural rate of
interest. 23 The ensuing equilibrium situation (which may differ from the
initial position) will persist unless another process of credit expansion
again distorts the market rate of interest (or unless the data of the system
change). Another business cycle (respectively another adjustment process)
will then set in.
The distortion of the market rate of interest thus plays an import-
ant role. It reflects the Austrian position that prices (and hence the
market rate of interest) contain information as regards consumer prefer-
ences and producer plans. Entrepreneurs act upon the basis of this
information (although their use of non-price information must not be
excluded). If the price system is hampered and entrepreneurs are not
aware of this distortion, prices reflect wrong information. They mislead
entrepreneurs, thereby leading to ex-post non-optimal investment decisi-
ons. These decisions and the resulting ex-post capital misallocation are
brought about by the investors' misperceptions as regards the causes of
changes in the market rate of interest. In this sense, prices play an impor-
tant role in the dissemination of knowledge. However, in the 1920s and
1930s Mises did not study the role of knowledge and expectations. As
Kirzner (1988) stressed, Austrians did not fully appreciate the implications
23
This is not to say that the economy will have returned to its previous equilibrium
situation. As Garrison (1981, p. 130) explicated, "[s]ome of the misallocated capital will
survive this market process. That is, it will not be possible (profitable) for entrepreneurs to
adjust for all investment decisions made during the period that money prices were being
distorted by a monetary disturbance. . .. Capital goods that fit poorly into the overall capital
structure of the new equilibrium will simply loose value; capital goods that do not fit at all
will cease to be capital. "
33
of their arguments in the socialist-calculation debate. As the debate
proceeded, it became gradually clear that the Austrian position centered
on the use of knowledge in society. Mises (1920) already pointed out that
an unhampered price system is indispensable for economic calculation.
However, Kirzner continued, he failed to see (or explicate) its importance
for the use of knowledge in society. It was Mises's successor in the debate,
Friedrich von Hayek, who first stressed the importance of knowledge and
expectations, albeit only in the late 1930s and 1940s?4
Mises (1928, pp. 45 - 46) argued that credit expansion does not
influence the natural rate of interest (as determined by the time-prefer-
ences of individuals) directly, although he acknowledged that there may
be an indirect effect which lowers this rate. Since money is nonneutral,
the credit expansion changes the distribution of income and wealth among
the agents. These agents will have different time-preferences and different
propensities to save. In particular, if commodity prices rise sharper than
wage rates, entrepreneurs will increase their income and wealth whereas
wage earners and salaried people are adversely affected. The latter groups
must then restrict their consumption, whereas the entrepreneurs are able
to expand theirs. Since it is usually presumed that entrepreneurs save
more than wage earners and salaried people, the redistribution of wealth
means that for the economy as a whole the (average) propensity to save
increases, thereby lowering the natural rate of interest. This redistributive
argument is called the doctrine of forced savings. Mises (1928, p. 46 and
1949 (1966), pp. 549 - 50) downplayed the importance of this doctrine
because of its comparative-static nature. He argued instead that a more
dynamic adjustment process should be analyzed.
Mises (1949 (1966), p. 564) claimed that the net result of the
business cycle is impoverishment. In his view some people have increased
their wealth, but"... the immense majority must foot the bill for the malin-
vestments and the overconsumption of the boom episode." It must be
noted that when Mises speaks about impoverishment, he means a relative
and not an absolute impoverishment: the economy could have performed
24 For an account of the economic-calculation debate, see Lavoie (1985). The articles
in which Hayek analyzed the role of knowledge and expectations are 'Economics and
knowledge' (1937), 'The facts of the social sciences' (1943), 'The use of knowledge in
society' (1945), and 'The meaning of competition' (1946).
34
better in the absence of the credit expansion and the consequent boom?5
The question then arises how this relative impoverishment can be pre-
vented. According to Mises (1949 (1966), p. 563), economic policy should
not try to alleviate the depression, because it "... is in fact the process of
readjustment, of putting production activities anew in agreement with the
given state of the market data: the available supply of factors of produc-
tion, the evaluations of the consumers, and particularly also the state of
originary [i.e. natural] interest as manifested in the public's valuations."26
Since the depression is an inevitable consequence of the boom, Mises
proposed to avoid the latter. The appropriate stabilization policy is then
to avoid the expansion of credit. However, in Mises's view the monetary
authorities are bound to increase the money supply because of their
'inflationist ideology'. 27 Therefore, some institutional measures should be
taken. Mises (1949 (1966)) stated, opined that the best institutional
arrangement is the combination of the gold standard with the system of
free banking, in which every bank is allowed to create its own currency
and is free to compete with other banks with regard to attracting lenders
and borrowers, because "... under free banking it would have been imposs-
ible for credit expansion with all its inevitable consequences to have
developed into a regular ... feature of the economic system" (p. 443). He
even claimed that "[o]nly free banking would have rendered the market
25
Unfortunately, the criteria he used for this judgment are rather obscure. Mises
claimed that an economy could have performed better if it did not have experienced a
business cycle. At the same time he admitted that some people are better off because of
the cycle, while the position of others has become worse. Mises's assessment then implies
that in his opinion the loss of utility of the latter is larger than the gain of utility of the
former. However, this presupposes the meaningfulness of interpersonal utility comparison.
It might be better to speak of 'potential impoverishment'.
26
One could call this position therapeutic nihilism. This term was used by William
Johnston (1972, p. 223) to typify the nineteenth-century Austrian intellectual position which
concentrated on diagnosis to the neglect of therapy. However, the term is not entirely
appropriate. Austrians may not want to alleviate the depression, they certainly want to
avoid it by refraining from credit expansion. Furthermore, some Austrians are clearly
willing to prevent the secondary depression, using monetary policy, although they warn that
one should be very careful in applying these measures (cf. Hayek (1978), pp. 210- 11).
27
In Mises's (1949 (1966), p. 578) view, "[p]eople are the more discouraged the greater
their optimism was in the days of the upswing. They have for the moment lost self-
confidence and the spirit to enterprise to such an extent that they even fail to take
advantage of good opportunities. But the worst is that people are incorrigible. After a few
years they embark anew upon credit expansion, and the old story repeats itself." This
explanation of the recurrence of the cycle is rather psychologistic, leaving the changes in
mental attitude unexplained.
35
economy secure against crises and depressions" (p. 443). In his opinion the
government and the central bank, and not private enterprise, cause the
business cycle. As Haberler (1937, p. 65) observed, Mises believed that
without the support of the government and the central bank, the commer-
cial banks can never produce a dangerous credit expansion, because they
would immediately lose cash and become insolvent. 28
Mises's business cycle theory will now be translated in terms of
some methodological characteristics. These can be labelled (1) methodo-
logical individualism, (2) methodological singularism, (3) radical
subjectivism, and (4) methodological dualism.
3.4. MISESIAN METHODOLOGY
3 .4.1. Methodological individualism

Methodological individualism can be defined as the position which holds
that social phenomena must be explained as intentional or unintentional
consequences of the actions of individual human beings, and as natural
givens. Nozick (1977, p. 353) argued that "[t]he methodological individual-
ist claims that all true theories of social science are reducible to theories
of individual human action, plus boundary conditions specifying the condi-
tions under which persons act." Methodological individualism is therefore
a form of reductionism as opposed to holism or, more in particular,
collectivism. Collectivists argue that the notion of an individual is mean-
ingless in the sense that man is a member of a social whole and a product
of social evolution. His actions are constrained by his social environment.
The constraints take the form of norms, values, laws, etc. This culminates
in the view that these constraints and therefore the social wholes must be
logically and temporally prior to the individual and his actions.
Collectivists conclude that economics must study man in terms of col-
lective entities (apart from natural givens).
One of the basic tenets of Mises's business cycle theory is its
methodological individualism. Mises (1949 (1966), p. 42) dismissed the
28Mises (1949 (1966), p. 447) himself also spoke about insolvency. This seems rather
strange, because a bank will only become insolvent if it gives money away, or if it lends
money to debtors who are not creditworthy. This is far from in agreement with general
banking practice. Therefore, the word 'insolvent' might better be replaced by 'illiquid'. It
leads to the conclusion that banks will not go bankrupt. Nevertheless, a check on credit
expansion under free banking will of course remain, because the government will not act
as 'lender of the last resort'.
36
collectivist position as 'vain', because in his view the notions of a whole
and its parts are correlative. Being correlative, the question arises why we
should start from one notion instead of the other? In other words, why
should we employ methodological individualism?
There may be three reasons for adhering to methodological
individualism. Firstly, it may be adopted for ontological reasons. Austrian
economists claim that collective or social wholes are mind-constructs,
which enable us to order the chaotic world around us. These wholes do
not exist in reality, whereas the individual does. Therefore, they do not
possess rankings of preferences: only individuals do. As a corollary, only
individuals can choose and act, not groups.29 This underpinning of me-
thodological individualism may be called ontological individualism: in
reality only the individual exists, not the social whole, and therefore only
the individual acts.
Secondly, if collective wholes exists, this becomes discernible only
in the actions of individuals. Or if we acknowledge that we cannot know
'das Ding an sich', the only thing we can observe is individual actions.
According to Mises (1949 (1966), pp. 42 - 43), we cannot visualize collec-
tive wholes; we can only see a crowd, that is, a multitude of individuals.
Therefore, he concluded, "... the way to a cognition of collective wholes is
through an analysis of the individuals' actions" (p. 42). This reason for
methodological individualism may be called obseTVational individualism. It
is this reason which Mises (as a Kantian) used as a basis for his methodo-
logical individualism.
According to Keizer (1986, p. 24), methodological individualism
may also be defended on normative grounds. That is, by arguing that the
individual should be the measure of everything. This normative individual-
ism argues that one ought to adhere to methodological individualism,
because otherwise the individual will become of minor importance
relative to the social whole. Mises, as a libertarian, also adhered to this
defense, albeit often only implicitly.
Methodological individualism is a form of reductionism, as
opposed to holism. It reduces holistic conceptions, such as the social
whole, to their constituent parts, that is, to individuals. Yet, Mises did not
adhere to other reductionist claims, such as the claim that man's behav-
29
As Keizer (1986, p. 24) argued, "...no collective entity has a real, concrete existence
of its own. They are not living organisms and cannot experience wants. They have no
preferences of their own, apart from those of their individual members. All collective
entities are the sums of their parts, if we include the interrelationships between the parts. "
37
iour must be reduced to chemical processes. This is not to say that he
rejected those claims, as we shall see when discussing his methodological
dualism.
3.4.2. Methodological singularism

The second tenet of Mises's analysis is its methodological singularism. His
praxeology is only interested in the individual because he acts. These
actions are its research object; it tries to unravel the universal in these
concrete actions. One of the universal aspects of acting is that it cannot
be repeated under identical circumstances. In this sense every action
constitutes a singular and unique event. This position is called methodolo-
gical singularism. It regards "[h]uman life ... [as] an unceasing sequence of
single actions" (Mises (1949 (1966), p. 45)). We can distinguish an import-
ant consequence of this methodological singularism.
Methodological singularism implies that the social sciences do not
deal with phenomena which are to a large degree identical and (more
importantly) repeatable. Although Mises would readily agree with the
view that in the abstract all human actions are similar (namely in the
sense that all actors try to attain a future set of circumstances which they
expect to be better than the one that would arise without the action), his
methodological singularism prohibits the treatment of actions as
repeatable (and hence identical). If every action constitutes a singular
event, they cannot be grouped together, as is necessary to use frequency
analysis. Mises thus considered the use of frequency distributions erron-
eous.30 However, since the use of probability calculus merely presup-
poses that the elements of the class under consideration do not differ
30 Mises (1949 (1966), pp. 106- 15) provides a fairly extensive treatment of probability.
He discerned two types of probability, namely class and case probability. The former
means that "... [w]e know or assume to know, with regard to the problem concerned,
everything about the behavior of a whole class of events or phenomena; but about the
actual singular events or phenomena we know nothing but that they are elements of this
class" (p. 107). The calculus of probability provides a presentation in mathematical symbols
for our defective knowledge as regards the behaviour of the members of the class under
consideration. "It neither expands nor deepens nor complements our knowledge. It trans-
lates it into mathematical language" (p. 108). And it does not tell us anything about the
actual singular events, which we did not know yet (p. 108). An example of this kind of
probability is the drawing of balls out of an urn. By contrast, case probability means that
"[w]e know, with regard to a particular event, some of the factors which determine its
outcome; but there are other determining factors about which we know nothing" (p. 110).
It deals " ...with unique events which as such ... are not members of any class .... The case
is characterized by its unique merits, it is a class in itself" (p. 111 ). This means that any
reference to frequency is inappropriate. Mises gives the presidential election of 1944 as an
example of an event to which this type of probability applies.
38
from each other in relevant ways, Mises should have made clear why in his
view the differences are relevant.
3.4.3. Radical subjectivism

From its very beginnings the Austrian School was identified as 'subjecti-
vist'. White (1982, p. 4) even claimed that subjectivism unifies the Aus-
trian school. Subjectivism means that at least some of the data of the
social sciences are constituted by the views of the economic agents. As
Mises (1949 (1966), p. 395) pointed out, "[i]n human action nothing counts
but the various individuals' desires for the attainment of ends. With
regard to the choice of these ends there is no question of truth; all that
matters is value. Value judgments are necessarily always subjective ... "
Praxeology takes human action as its starting point. Action, as
Mises defined it, implies choice. Man chooses between various opportun-
ities and the outcome of this choice will depend on his subjective valu-
ation of the alternatives. These valuations form the facts of the social
sciences. They cannot be 'objectified', because there is no common
denominator. Utility cannot be compared interpersonally or intertempo-
rally.31 It can only be 'measured' on an ordinal scale, not on a cardinal
one. This view is called ordinalism.
The fact that the data of the social sciences can be called subjec-
tive has another corollary, namely the dispersion of knowledge. However,
Mises did not consider the individual's knowledge (and expectations). It
was Hayek who emphasized the important role of knowledge and expecta-
tions. Post-Misesian Austrians recognized that Mises's notion of
subjectivism was unable to grasp the importance of expectations. Lach-
mann (1982a, p. 37) argued that Mises failed to grasp the opportunity of
incorporating expectations into his analysis because subjectivism" ... meant
to him no more than that different men pursue different ends .... The ends
themselves ... we have to regard as 'given'." Mises's analysis can be
interpreted as a pure logic of choice. It did not permit him to consider the
incorporation of expectations into his analysis an important step forward
(cf. Mises (1949 (1966), p. 21)). According to Bohm (1982, p. 46), this
result follows from the fact that "[t ]he framework of the logic of choice
does not allow for considerations of time" (p. 47, italics in original),
although it is perhaps more correct to state that it does not allow for the
31
Of course, this latter feature does not mean that an individual cannot compare
utility of present and future goals at a given point in time.
39
analysis of change. In this respect the Hayekian, Kirznerian and Lachman-
nian theories concerning the acquisition and dissemination of knowledge
can be interpreted as attempts to elaborate on Mises and to fill the gap in
his analysis.
O'Driscoll and Rizzo (1985, p. 22) emphasize the distinction
between what they call static and dynamic subjectivism. In their view,
static subjectivism considers the mind" ... a passive filter through which the
data of decision-making are perceived. To the extent that this filter can be
known, the whole process of decision-making is perfectly determinate." In
other words, the pure logic of choice is situationally deterministic. The
Robbinsian economizer is the prototype in economics of the static-
subjectivistic decision-maker. Dynamic subjectivism, on the other hand, "...
views the mind as an active, creative entity in which decision-making
bears no determinate relationship to what went before" (italics in original).
This form of subjectivism states that the actor sees his own decision-
making as indeterminate and, therefore, cannot be in a position to predict
his actions. The reason for this is twofold. Firstly, the individual cannot
know his future knowledge, and therefore his future actions must be
unknown (hence indeterminate from the individual's point of view).
Secondly, the ability to foresee one's own decision at a certain point in
the future logically precludes the ability to decide at that point in time
because this would mean that the decision can be known and therefore
already must have been made (p. 25). In the dynamic-subjectivistic view,
decisions cannot be treated as flowing inescapably out of the objective
circumstances. It emphasizes, as Shackle (1972) argued, the creativity and
indeterminacy of human decisions. The static-subjectivistic position, on the
other hand, hardly implies any choice at all. Whereas in a static-subjecti-
vistic framework individuals are faced with a given means-ends framework
without any learning process, in a dynamic-subjectivistic environment they
continuously adapt their knowledge and expectations to changing circum-
stances. According to O'Driscoll and Rizzo (1985, pp. 9 - 10), "Austrians
have stressed the view of the market process as an engine of discovery ...
Learning, for Austrians, is not merely plugging in new values of variables
in an otherwise unchanging learning function. It involves a structural shift
in knowledge, i.e., a change in the learning functions themselves. The
market process is both the source and the manifestation of these struc-
tural changes ... " Given this distinction between static and dynamic
subjectivism, the respective analyses of Lachmann (1982a) and Bohm
(1982) show that Mises adhered to static subjectivism. We shall see later
40
that the Austrian revival in the 1970s can be regarded as incorporating
dynamic subjectivism.
3. 4. 4. Methodological dualism
A fourth major tenet of Misesian theory is its methodological dualism,
which is the position that the social sciences should not adopt the research
methods of the natural sciences. According to Mises (1949 (1966)), this
methodological dualism follows from the fact that the data of the social
sciences cannot (yet) be explained by the natural sciences. He argued that
"[c]oncrete value judgments and definite human actions are not open to
further analysis", even though we may "... assume or believe that they are
absolutely dependent upon and conditioned by their causes. But as long as
we do not know how external facts - physical and physiological - produce
in human minds definite thoughts and volitions resulting in concrete acts,
we have to face an insurmountable methodological dualism" (p. 18, italics
in original). This dualism thus centers on the impossibility to reduce the
social sciences to the natural sciences.
3.5. CONCLUSIONS
Mises's business cycle theory is a monetary theory, in the sense that the
cycle is caused by monetary factors. An expansion of credit, in the form of
either the issue of bank notes or the unlimited use of current accounts,
reduces the market rate of interest below the natural rate of interest.
Entrepreneurs are then inclined to expand their investments, thereby
lengthening the 'period of production', that is, investing in more capital-
intensive equipment. As soon as the credit expansion comes to an end,
these investments will prove to be unprofitable. The economy then has to
adapt itself again to the new (and correct) market rate of interest. The
resulting equilibrium will only be disturbed if credit is expanded again.
This means that Mises's business cycle theory regards the cause of the
cycle to be an exogenous one. The economy is considered to be stable in
the sense that eventually it is supposed to return to equilibrium, that is,
there is a tendency towards coordination (although Mises does not
explicate why this tendency should exist). Only exogenous disturbances
can hamper this tendency, albeit only temporarily.
The cyclical process is irreversible in the sense that it takes time
to discover that the new investments are not profitable. And as time is
41
irreversible (cf. Mises's methodological singularism), so are the boom and
the resulting crisis and depression. This led him to conclude that the
depression constitutes a necessary adjustment process, in which wrong
(unprofitable) investment projects are abandoned. There is only one way
of preventing the adverse effects of the cycle, namely by preventing the
credit expansion altogether.
It should be noted that Mises's analysis does not aggregate the
entrepreneurs' behaviour. He allowed for a variety of economic agents,
such as entrepreneurs, wage earners and salaried people. Producers are
also distinguished by the type of goods they produce (first-order, second-
order, etc.). This clearly can be explained by his methodological individua-
lism. However, combined with his subjectivism, this leads to some vague-
ness as regards the manner in which investment projects are undertaken
or abandoned, and as regards the speed of the adjustment process. Mises
did not consider such vagueness a serious objection, arguing that the
unknowability of other people's valuations implies that social scientists
cannot use the precise, mathematical tools of the natural sciences. This
point was to be elaborated by Hayek.
42
4. HAYEK'S YEARS OF HIGH THEORY
4.1. INTRODUCTION
Hayek's business cycle theory draws heavily on that of Mises. The concept
of the 'structure of production' provided Hayek with a clear picture of the
way in which market economies are organized. Furthermore, Mises's
monetary theory rejected the quantity theory of money because money
should be considered nonneutral (Hayek {1929 (1966), p. 16); cf. Chapter
3). 1 Although Hayek also contributed to monetary theory, his research on
business cycle theory is more important for our purposes. His ensuing
concentration on the role of knowledge in economics may even be
regarded his major contribution to Austrian economics. His emphasis on
knowledge has two corollaries. Firstly, it provided Austrian business cycle
theory with a dynamic benchmark, substituting intertemporal equilibrium
for Mises's evenly rotating economy. Secondly, eventually this facilitated
the discussion of the coordination problem.
This chapter aims to provide a reconstruction of Hayek's work in
the period 1928 - 1946. This work starts with the replacement of Mises's
equilibrium construct by an intertemporal general equilibrium concept.
This equilibrium concept was used in Hayek's business cycle theory.
However, it did not distinguish between individual and general equilib-
rium. His 1937 article 'Economics and knowledge' provided this dis-
tinction and enabled Hayek to consider the so-called coordination prob-
lem. In the 1940s this problem became the focus of Hayek's research
efforts. In particular, he posed the question why there would be a ten-
dency towards the coordination of individual plans. Hayek tried to solve
this problem, which will be called the Hayek Problem, by following a
1
Hayek severely criticized Wicksell for concentrating on the general price level. In
Hayek's (1933a (1975), pp. 114- 15) view, • ...Wicksell does not recognize here a monetary
tendeing, independently of changes in the price-level, to break down the equilibrium
system of barter economics: so long as the stability of the price level is undisturbed,
everything appears to him to be in order. Obsessed by the notion that the only aim of
monetary policy is to explain those phenomena which cause the value of money to alter, he
thinks himself justified in neglecting all deviations of the process of money-economy from
those of barter-economy, so long as they throw no direct light on the determination of the
value of money: and thus he shuts the door on the possibility of a general theory covering
all the consequences of the phenomena which he indicates. • He considered Wicksell's
(1898, p. 97) justification 'incomprehensible' (p. 114n).
43
research agenda (the Hayek programme) which consists of the Hayek
Problem and some methodological constraints. This programme will also
be reconstructed in this chapter.
4.2. INTERTEMPORAL EQUILIBRIUM
Hayek (1933a (1966), p. 42n) stated that his monetary business cycle
theory must be placed in a general equilibrium framework, because such a
framework incorporates general interdependencies. The use of such an
equilibrium construct implies that the theory must use the logic of com-
parative-static equilibrium theories, which means that the business cycle
theory can only demonstrate that disturbances must be exogenous: the
system will always react by creating a new equilibrium (pp. 42 - 43). The
exogenous disturbance(s) must return time after time in order to explain
the recurrence of the cyclical movement. Contrary to Mises, Hayek
regarded this a highly unsatisfactory feature of equilibrium business cycle
theories. In his view, business cycles must be explained endogenously? He
tried to accomplish this by introducing time, which forced him to expand
his framework.
As was shown in Chapter 2, Wicksell's (1898) theoretical frame-
work was applicable only in the context of a stationary economy. More-
over, Mises (1912 (1924)) had also adopted a stationary benchmark. As
early as 1928, Hayek rejected both static equilibrium concepts. This is not
to say that he rejected the equilibrium concept per se. On the contrary, he
claimed that "... the concept of equilibrium is just as indispensable a tool
for the analysis of temporal differences in prices as it is for any other
investigation in economic theory" (1928 (1984), p. 75). The question
merely concerns how to incorporate the time element in the equilibrium
concept without having to assume that all economic data remain the
same. Mises's static equilibrium was an evenly rotating economy, in which
there were no changes in prices and production. Hayek criticized this
notion by observing that "... even a self-replacing economy cannot present
the same picture at every moment in time. On the contrary, the same
2 I shall use the term 'endogenous' in the Schumpeterian sense. Schumpeter (1954
(1982), p. 745) considers 'endogenous' business cycle theories to imply that " ... each phase
of the cyclical process is induced by the conditions prevailing in the preceding one."
Conversely, 'exogenous' then means that each 'cycle' is caused by a disturbance from
outside the system.
44
processes can be repeated within it only periodically.... It ... follows that,
in such an economy, the factors which are operating upon the prices of
the same sort of goods at different points in time [within the same time
period] - and hence also these prices themselves - will be different" (p.
73). Hence, Mises's 'evenly rotating economy' appears inconsistent on its
own account. Hayek argued that it is better to allow for prices to change
over time, even in a benchmark situation. He replaced it by an inter-
temporal equilibrium concept in which "... [a]ll that needs to be assumed
... is that the wants and the means of production existing at every point in
time are known to the individual economic subjects at the time at which
they frame their economic plan for the period as a whole" (p. 76). Knowl-
edge thus plays a crucial role in this equilibrium concept. Hayek argued
that a large number of the changes in the economic data are known
beforehand. Rational economic agents will take these changes into
account, and - given their knowledge - they will choose the best course
action open to them. As long as the agents know all the relevant changes
in the data, they can maintain equilibrium. As early as 1928 Hayek consi-
dered perfect knowledge and perfect foresight the defining characteristics
of an intertemporal equilibrium notion. However, he did not distinguish
between individual and general equilibrium, which implies that he was not
yet able to appreciate the importance of the coordination problem.
4.3. HAYEK'S EARLY BUSINESS CYCLE THEORY
Hayek's early work on business cycle theory took place in the period 1929
- 1933. It encompasses his Geldtheorie und Konjunkturtheorie (1929,
published in English in 1933 as Monetary Theory and the Trade Cycle
(1933a)) and the lectures he gave in 1930-31 at the London School of
Economics (published as Prices and Production in 1931}. Additionally, he
also wrote some articles on the subject. 3 In this work he treated the
business cycle as a disequilibrium phenomenon. That is, agents are suppo-
sed to have imperfect knowledge and imperfect foresight. This means that
they will make expectational errors and decide to undertake unprofitable
3
These articles are 'A note on the development of the doctrine of "forced savings"'
(1932), 'The present state and immediate prospects of the study of industrial fluctuations'
(1933d), and, more importantly, his Copenhagen-lecture 'Price expectations, monetary
disturbances and malinvestments' (1933c). Page numbers refer to their respective reprints
in Hayek (1939).
45
investments. According to Hayek (1933c, p. 141), the imperfection of
knowledge is in fact a prerequisite for the business cycle. The question
then is why agents make such expectational errors and, additionally, why
different agents make similar mistakes. In Hayek's (1933c, p. 141) view,
there are two reasons why agents should make similar expectational
errors. The first is an exogenous one, and refers to some psychological
state of mind (e.g., Pareto's 'waves of optimism or pessimism", Keynes's
'animal spirits'). Hayek did not dismiss this reason, but he did not con-
sider it a crucial one. Instead, he argued, it is more likely that the agents
are misled by following guides and symptoms which as a rule prove
reliable. The price system provides such a rule, in the sense that economic
agents base their actions on the prices which they observe. As Hayek
(1933c, p. 141) recognized, "... it may be that the prices existing when they
[i.e., the agents] made their decisions and on which they had to base their
views about the future have created expectations which must necessarily
be disappointed." The prices on which agents base their actions may be
distorted. The most important price in this regard is the market rate of
interest. It may differ from the natural rate and if this difference is
unknown and unexpected, a business cycle will arise. The question then is
why do the two rates differ?
Hayek (1933a, pp. 95 - 98) discerned two reasons for such a
difference. Firstly, the natural rate may change, while the market rate
remains unchanged. He argued that the processes resulting from such a
change in the natural rate of interest can be better interpreted as a mere
adjustment process to a new equilibrium situation, instead of a business
cycle, although they do not differ from the latter in relevatn aspects.
Secondly, the market rate may change, given the natural rate. Then,
monetary factors cause the difference between the two rates. In particular,
the difference will be caused by a credit expansion. Hayek claimed that
the resulting processes constitute a business cycle. In other words, in
Hayek's terminology a business cycle is - by definition - caused by monet-
ary factors. But how can these factors change the market rate of interest?
Hayek (1933a, pp. 148 - 49) discerned three factors which may cause such
a change while leaving the natural rate unchanged: ( 1) changes in the
volume of cash, caused by the in- and outflow of gold, (2) changes in the
volume of money, as regulated by the central banks, and (3) the creation
of deposits by private banks. He considered the first empirically less
46
important and the second influence a rather special case.4 In his view,
there is no need for presupposing a deliberate lowering of the market rate
of interest. Therefore, he concentrated on the third influence. The
creation of credit by private banks is seen as the main reason why the two
rates of interest differ.
The question then is why private banks should want to expand
credit. Hayek (1933a, pp. 144- 45) rejected the idea (as implicitly adopted
by Mises) that the credit expansion is an exogenous event. In his view, it
is an inevitable consequence of the existing credit organization (pp. 146 -
48), resulting from the fact that it is in the interest of a private bank to
lend as much as possible (pp. 151 - 52). The volume of its loans will
depend on the volume of its deposits. An increase in the latter will raise
the reserve ratio, thus enabling the bank to increase its loans. However,
such a simple relationship between loans and deposits does not apply to
the banking system as a whole. The credit supplied by one bank can be
used to buy resources and intermediary products. The seller of these
resources and products will deposit the received amount in his own bank,
whose lending capacity then increases. If the second bank also adheres to
a fractional reserve ratio, it will expand credit. This process can go on
indefinitely, as long as the credit is not used in a way which leads quickly
to the market for consumer goods (as in the latter case the credit is not
deposited with another bank) (pp. 158 - 59). Hayek concluded that this
process of credit expansion enables the banking system as a whole to "...
grant credit to an amount several times greater than the sum originally
deposited" (p. 160).
We can now turn to the business cycle itself. Suppose that the
money supply increases according to the process just described. The
additional money then enters the economy on the credit market. As a
result, the market rate of interest falls below the natural rate, and hence
becomes distorted. Agents will subsequently raise their demands for loans,
presumably for consumption as well as investment purposes, although
Hayek's business cycle theory concentrates on the latter. Hayek (1933c, p.
4 Hayek (1933a, pp. 150) acknowledged that ".. .it is possible to assume, with Professor
Mises, that the Central Banks, under the pressure of an inflationist ideology, are always
trying to expand credit and thus provide the impetus for a new upward swing of the Trade
Cycle; and this assumption may be correct in many cases .... But before deciding in favour
of this special assumption ... we have to ask whether, in some other part of our credit
system, such extensions may not take place automatically under certain conditions -
without the necessity for any special assumption of the inadequate functioning of any part
of the system. "
47
142) presupposed that investors expect that "... the supply of capital [and
thus the market rate of interest] will for some time continue at the
present level." This expectation leads them to increase their investment.
As Hayek noted, "[i]t is only some such assumption that will justify the
employment of additional capital to start new roundabout [i.e. capital-
intensive] methods of production which, if they are to be completed, will
require continued investment over a further period of time" (p. 142). That
is, investors must not know that the market rate of interest is 'distorted'.5
They must have imperfect knowledge as regards this distortion, otherwise
there will be no business cycle. They must be unable to solve the 'signal
extraction problem', which means that they must not be able to extract
the natural rate of interest from the market rate.
If the market rate of interest falls below the natural rate, then
planned savings are not large enough to cover planned investments. This
difference between planned savings and planned investments will persist
as long as the credit expansion continues. The gap is filled by what Hayek
(1931, p. 18), like Mises, called forced savings. 6 During the expansion
process agents compete for the factors of production. This will tend to
raise their prices and those of producer goods. Those agents who see their
incomes rise as a consequence of the increases in factor prices (e.g.
increases in their wage rates), are able to exert a higher demand for
consumer goods. But the supply of these goods has declined, because
production factors have been transferred from the consumer to the pro-
ducer goods industries. Then the prices of the consumer goods will rise,
forcing agents who have not experienced a rise in income to curb their
consumption, or equivalently to raise their (ex post) savings. The increases
in prices of consumer goods, combined with spreading optimism, will
induce agents to continue to invest. In order to make these increasing
investments possible, however, more credit is necessary, because of the
rises in product and factor prices. If the credit expansion ceases, the
market rate of interest will rise. This will make it clear that the invest-
ment projects which were undertaken with the additional credit, are
unprofitable. Investors have invested in too capital-intensive methods of
5 This was also the case in Wicksell's explanation of business cycles. However, Wicksell
adopted Spiethoffs view that these cycles were caused by real phenomena. That is,
business cycles are caused by the fact that the market rate of interest does not immediately
adjust to changes in the natural rate of interest. Cf. Chapter 2.
6
For a discussion of different views on the 'doctrine of forced savings', see Hayek
(1931 (1935), pp. 17- 19) and (1932 (1939), pp. 183- 97).
48
production. The projects under consideration must be abandoned. As in
Mises's theory, the elimination of these projects takes place during the de-
pression. Credit flows back to the banks, which means that their degree of
liquidity increases. This lays the foundation for a new boom.7
It would appear that Hayek considered the business cycle an
inevitable process. One should not try and alleviate its effects once it is on
its way. However, in Profits, Interest and Investment (1939) he made an
important distinction between an ordinary mild depression and a 'second-
ary depression' (pp. 176 - 77). Whereas in the 1930s his views on the
policy measures to be taken against a secondary depression are not easy
to disentangle, in 1978 he stated that "[s]uch a 'secondary depression'
caused by an induced deflation should of course be prevented by approp-
riate monetary counter-measures" (p. 210)). 8 These measures must be
directed against a further shortening of the period of production. This led
him to suggest that the new credit be made available to the producers,
because they will invest, stop the shortening of the period of production
and therefore end the secondary depression. As knowledge is dispersed
throughout the economy, however, two problems arise: (1) it is difficult to
ascertain whether the market rate of interest in fact is equal to the natu-
ral rate; (2) it is almost impossible to know the extent to which credit
must be expanded.9
Like Mises, Hayek advocated the 'free banking' system as the best
way to avoid the adverse effects of the cycle. Already in 1937 he played
with this idea, although his major work on the subject was published not
until the 1970s. 10 In his 1976 pamphlet 'Denationalization of Money' (p.
99) he concluded that "[t]he abolition of the government monopoly of
money was conceived to prevent the bouts of acute inflation and deflation
which have plagued the world for the past 60 years. It proves on examin-
ation to be also the much-needed cure for a more deep-seated disease:
the recurrent waves of depression and unemployment that have been
represented as an inherent and deadly defect of capitalism." 'Currency
1
The fact that during the depression fixed capital (e.g. machinery) is transformed into
free capital ('loanable funds') appears to be a characteristic of all vertical maladjustment
theories. However, the reverse need not be the case, that is, not all theories exhibiting this
characteristic are vertical maladjustment theories. For instance, cf. Hawtrey (1913 (1970)).
8
Cf. Barry (1979, p. 165).
9
Cf. Machlup (1977, pp. 23- 4).
10
Hayek (1937b (1971), p. 77).
49
competition' is seen as the best way to prevent credit expansion and
therefore business cycles. 11
4.4. THE COORDINATION PROBLEM
As Foss (1989, p. 12} argued, Hayek's business cycle theory of the early
1930s is an application of his (1928) notion of perfect foresight intertem-
poral equilibrium to "... the phenomena connected with 'investment'" (as
Hayek {1937, p. 42n) himself called it). It explained cyclical fluctuations as
disequilibrium phenomena, resulting from actions based on incomplete
information about the nature of changes in the market rate of interest.
However, Hayek recognized that his 1928-definition of intertemporal
equilibrium was too stringent, in the sense that plans and actions may be
mutually consistent (coordinated} even if agents do not have complete
information and perfect foresight. This raised the question as to what
knowledge is necessary and sufficient for coordination to arise?
This question was addressed in Hayek's (1933c) Copenhagen
lecture and, more specifically, in his 1936 Presidential Address to the
London Economic Club and published in 1937 as 'Economics and know-
ledge' .12 In the former he already pointed out that the main difficulty
with the traditional approach to economics was its abstraction from time.
However, he had not yet redefined general equilibrium in terms of coordi-
nation. Hence he could not yet appreciate the problems posed by explain-
ing disequilibrium adjustment processes in terms of knowledge acquisition
processes. Although he presupposed that there would be a tendency
towards coordination, he acknowledged that this presupposition formed
the fundamental problem economics must solve. In his 1937 article he
sketched a possible way in which a solution to this problem could be
found.
Hayek (1937) started by describing the framework in which the
solution to the coordination problem should be found. This framework
consists of two equilibrium concepts, each applying to a different level:
11
For a critical analysis of Hayek's 'currency competition', see Visser (1991).
12 In the 1930s and early 1940s Hayek also wrote on capital theory. However, these
writings do not seem to have had much impact on the Hayek programme. Moreover, he
never incorporated his new insights on this issue into his business cycle theory. Therefore,
the writings will not be discussed.
50
that of the individual and that of society as a whole. Agents are said to be
in equilibrium when they cannot improve their actions, given their knowl-
edge. In individual equilibrium an agent's actions are optimal (with regard
to the plan upon which they are based). On the other hand, Hayek
observed, one cannot distinguish a plan for society as a whole. In his view,
only individuals can make plans. Nevertheless, their actions can only be
said to be in equilibrium if they can be understood as part of one plan (p.
36}. Therefore, general equilibrium must refer to all individual plans, or
rather, to the multitude of individual equilibria. 13 Optimality for the
system as a whole is characterized by optimality for all agents. Such a
general equilibrium presupposes that no individual plan is frustrated, i.e.,
that they are all mutually compatible. If this is the case, then there is no
endogenous reason why individual plans will be frustrated. This means
that there is no coordination problem. The question then becomes which
preconditions must be fulfilled for the general equilibrium to exist.
At first sight the situation of general equilibrium appears to
require perfect knowledge and perfect foresight on the part of the econ-
omic agents. In 1928 Hayek had already defined general equilibrium in
these terms. But in 1937 he changed his general equilibrium concept by
suggesting that agents do not need to know everything. 14 Instead, they
must possess merely all relevant knowledge (p. 42}, and their plans must
be based on identical (though potentially false) expectations of the same
set of external events (p. 38). Coordination is then defined as the situation
in which there are no endogenous reasons why individual plans will be
frustrated. It will arise if all agents have all relevant knowledge. The
question then arises, what is relevant knowledge? And how can economic
agents obtain this sort of knowledge?
The question concerning the tendency towards general equilib-
rium centers on the agents' knowledge acquiring (learning) and expecta-
tions formation processes. Hayek (1937) criticized general equilibrium
analysis for not explaining how such a general equilibrium could be
attained (pp. 45 - 46). In his view, this type of analysis in fact assumes
away the very problem. This is not to say, though, that he proposed to ban
the notion of general equilibrium from economics altogether. Instead, he
13
Cf. Butos (1986, p. 334).
14
Caldwell (1988) explains this change by referring to the capitalism-socialism debate
in which Hayek was engaged during the 1930s. This debate made clear that the economic
problem facing society is not a technical optimization problem, but instead the problem of
how to make of as much as dispersed knowledge as possible.
51
opined that it is possible to render general equilibrium analysis empi-
rically meaningful by means of "... definite statements about how knowl-
edge is acquired and communicated" (p. 33). Or, to put it differently, the
problem is not whether general equilibrium exists, but rather whether a
tendency towards it exists. This means that Hayek sought to explain why
in discoordination situations a dynamic process would take place, which
would lead the economy towards a general equilibrium. In his own words,
"... the assertion that a tendency toward equilibrium exists ... can hardly
mean anything but that, under certain conditions, the knowledge and
intentions of the different members of society are supposed to come more
and more into agreement [i.e. become more and more coordinated] or, to
put the same thing in less general and less exact but more concrete terms,
that the expectations of the people and particularly of the entrepreneurs
will become more and more correct" (p. 45)). This raises the question
under what circumstances such a tendency towards coordination will exist?
Hayek could not give an answer to this. He readily admitted that "... I am
now getting to a stage where it becomes exceedingly difficult to say what
exactly are the assumptions on the basis of which we assert that there will
be a tendency toward [general] equilibrium and to claim that our analysis
has an application to the real world. I cannot pretend that I have as yet
got much further on this point" (p. 48). Nevertheless, he did go a little
further by stating that "... the relevant knowledge which he [the economic
agent] must possess in order that equilibrium may prevail is the knowl-
edge which he is bound to acquire in view of the position in which he
originally is, and the plans he then makes" (p. 53). The agents must base
their plans on knowledge which they would acquire during and because of
their actions. Hayek thus believed that economic agents learn their way
into coordination. However, he did not indicate the way in which agents
could learn this knowledge. This is not to say that he did not implicitly
ascribe some form of learning to the economic subjects. As Boland (1982,
p. 69) indicated, Hayek assumed that these subjects learn inductively.
Since no reliable inductive logic exists, such learning need not lead to
correct knowledge.
Although Hayek could not provide all the answers, he had stated
the issues which needed further investigation in order to explain the
dynamic adjustment process between two successive equilibria. As he
clarified later, the main conclusion of his 1937 article was that "... the task
of economic theory was to explain how an overall order of economic
activity was achieved which utilizes a large amount of knowledge which
52
was not concentrated in any one mind but existed only as the separate
knowledge of thousands or millions of different individuals" (Hayek (1964,
pp. 91 - 92) ). This reflects that adopted the view that knowledge is
dispersed. This dispersion of knowledge means ( 1) that agents have incom-
plete knowledge, and (2) that they do not posses identical knowledge, i.e.,
that information across agents and markets is heterogeneous. 15
Butos (1986, pp. 341 - 42) observed that "Hayek, in a sense, starts
out by constructing an equilibrium model of the business cycle but ends
up in 1937 with a different set of issues, concerns, and a new research
agenda." This agenda constitutes the Hayek programme. 16 It tries to
answer the question why there would be a tendency towards coordination.
The first attempt to provide an answer to this question is Hayek's (1939)
analysis of the Ricardo effect, with which he intended to explain why a
boom would turn into a depression, that is, why a disequilibrating move-
ment would be transformed into an equilibrating (coordinating) one.
However, the ensuing debate showed that his contemporaries did not
appreciate the meaning of the effect. This led Hayek to elaborate the
underlying research agenda more explicitly. But before analyzing the
debate on the Ricardo effect we must make some remarks on Hayekian
capital theory. More in particular, we must first consider the incorporation
of capital in the Hayek programme.
4.5. CAPITAL, INVESTMENT, AND DISCOORDINATION

DYNAMICS
In the 1930s Hayek had not only been concerned with business cycle
theory. He had also devoted his research efforts to capital theory. In
15
Hayek (1942b, p. 280) argued that "... the term 'subjective' stresses [an] important
fact ... : ... the knowledge and beliefs of different people, while possessing that common
structure which makes communication possible, will yet be different and often conflicting
in many respects. . .. It only exists in the dispersed, incomplete, and inconsistent form in
which it appears in many individual minds, and this dispersion and imperfection of all
knowledge is one of the basic facts from which the social sciences have to start."
16
The term 'Hayek programme' is related to Boland's (1986) Popper-Hayek pro-
gramme as well as Caldwell's (1988) Hayek Problem. The former is a research pro-
gramme, not only aiming to solve the Hayek Problem (that is, to explain discoordination
dynamics) but also to do so in a way which is consistent with Popper's criticial-rationalistic
methodology. The Hayek programme as defined below does not adopt this latter property.
It is a reconstruction of Hayek's attempts to explain discoordination dynamics and of his
prescriptions for the best ways to accomplish such an explanation.
53
particular, he had tried to defend Austrian (i.e. Bohm-Bawerkian) capital
theory against Frank Knight's criticisms. As his defence did not have
much success, he tried to provide Bohm-Bawerk's theory with new and
sound foundations, which culminated in his The Pure Theory of Capital
(1941}Y In this book Hayek argued that capital theory in general proved
unsatisfactory because it had not been studied in the context of industrial
fluctuations (pp. v and 4 - 6). He tried to remedy this shortcoming by
considering its 'coordination foundations'. This means that he incor-
porated "... the interrelations between the different parts of the material
structure of the process of production, and the way in which it [i.e. the
structure of production] will adapt itself to changing conditions" (p. 3}.
This is not to say that Hayek aimed to provide a discoordination analysis
of changes in that structure. On the contrary, he tried to devise an
analytical framework which could be adopted in such an analysis. As was
already shown in Chapter 2, Austrian capital theory interprets 'capital' as
a set of heterogeneous goods, which are distinguished by the particular
use which is made of them. The capital goods are assumed to be highly
(though not completely) specific, in the sense that its possibilities to be
shifted from one production process to another are limited. Moreover,
capital goods are distinguished as to their distance from final consump-
tion. The same holds for the stages of production which compose the
structure of production. This structure is interpreted as a reflection of past
production plans. In turn, investment decisions are the reflection of
changes in these plans. According to Hayek (1941, p. 22), any analysis of
changes in the structure of production must focus on the problem whether
the investment decisions are mutually consistent. After all, he concluded,
"[ t]he essential problem remains that of whether the plans of different
agents will tally and will accordingly all stand a chance of being successful,
or whether the present situation carries the seed of inevitable disappoint-
ment to some, which will make it necessary for them to change their
plans." Thus, any analysis of changes in the structure of production must
focus on the coordination problem as regards the investors' decisions.
Hayek thereby recognized that the question whether these investors will
learn their mistakes is an empirical one, which cannot be answered on a
17 Hayek's capital theory will not be discussed in detail because this leads too far
astray from our main objective, the reconstruction of the Hayek programme.
54
priori grounds. 18 He nevertheless firmly believed that agents will be
successful in their learning endeavours, and hence that they will bring
about a tendency towards coordination.
In the years following the first publication of his Pure Theory,
Hayek was engaged in the debate on the Ricardo effect. This debate con-
cerned the cyclical fluctuations in the capital structure. It indicated that
Hayek's contemporaries did not appreciate his framework, in the sense
that they adhered to comparative-static analysis. The debate led Hayek to
shift his research activities towards economic methodology (and later
social and political philosophy). In turn, this prohibited him from" ... going
on with a further elaboration of the explanation of industrial fluctuations
... " (1941 (1950), p. v). The intended elaborations of his analytical frame-
work were never to be written.
4.6. THE DEBATE ON THE RICARDO EFFECT
In 1939 Hayek published his Profits, Interest and Investment, a collection of

essays written between 1929 and 1939. The first of these essays, which
carried the same title as the collection, discussed the so-called Ricardo
effect. This effect evoked a fierce debate between Hayek (1939 and 1942a)
and Wilson (1940) and Kaldor (1942). The differences which existed
during the debate were not resolved by it. The debate is a prime example
of miscommunication between economists, as has been argued by O'Dris-
coll (1977a) and, more in particular, Moss and Vaughn (1986). This
section will give a brief account of the Ricardo effect and the ensuing
debate. But the main purpose will be to interpret it as the next step in the
execution of the Hayek programme, that is, as an attempt to provide (1)
an explanation of the dynamic process towards spontaneous coordination,
and (2) a more explicit account of the turning points of business cycles.
18
In an important footnote Hayek (1941, p. 23nl) added that the claim of the
existence of such a tendency would be "... strictly true only if we are thinking of a single
deviation of a particular element in a situation which is otherwise in equilibrium, that is on
the assumption that all other expectations are confirmed. If more than one element turns
out to be different from what was expected, the relation is no longer so simple." This
statement leaves room for the acknowledgement that if more than one plan is frustrated,
there may not be a tendency towards coordination.
55
4.6.1. The Ricardo effect
The Ricardo effect derives its name from Ricardo's (1817, Ch. I, Section
V) statement that "... with every rise in the price of labour, new tempta-
tions are offered to the use of machinery." Hayek (1939 (1950)) claimed
to adopt this argument, and argued that it could explain the upper and
lower turning points of business cycles. His analysis started by assuming a
credit expansion which leads to malinvestments and hence to an undue
lengthening of the structure of production. Eventually, in the later stages
of the boom, there will not be enough resources to maintain the longer
(more 'roundabout') structure of production. Agents must then decide
whether to produce for the near or the more distant future. Stated in a
somewhat simplified manner, they must choose whether to produce
capital or consumption goods. However, Hayek continued, at these later
stages "... the prices of consumers' goods do as a rule rise and real wages
fall" (p. 11). 19 In turn, this increases the profit margins in the consumer
goods industries. The use of labour-intensive methods of production will
become more profitable, resulting in capital being replaced by labour. It
may be objected that capital goods themselves are the product of labour,
which means that the decrease in wages must lead to a proportional
decrease in the price of capital goods. Of course, this objection only
applies in general equilibrium. As we shall see later, this is not the situ-
ation Hayek was interested in. He maintained that "... a much higher rate
of profit will now be obtainable on money spent on labour than on money
invested in machinery" (p. 14)?0 This higher rate of profit will have two
consequences, both arising because of the Ricardo effect. Firstly, the
effect will cause a tendency to use more labour with the existing machin-
ery (by working overtime, double shifts, etc.). Secondly, new machinery
will be of a less expensive, less labour-saving or less durable type. The
Ricardo effect thus ensures that labour-intensive methods of production
become more profitable and will be used. This, in turn, reduces the
demand for investment goods. The second objection against the Ricardo
effect concerns the agents' expectations as regards the price increases.
The working of the effect implies that agents expect the relatively higher
19
It should be noted that Hayek (1939 (1950), p. 8) defined the real wage rate as the
ratio of the money wage rate and the prices of the consumer goods produced by the
labour under consideration.
20
This emphasis on the rate of profit may be invoked by Hansen's (1932, p. 335)
remark that this rate is the relevant variable.
56
product price and the relatively lower real wages to persist for a period of
time that is long enough to make it worthwhile to change their methods
of production. Hayek (1939 (1950), pp. 16 - 18) acknowledged this. More-
over, he also adressed the case in which the agents expect that the prices
of consumer goods continue to rise. This will merely reinforce the effect,
because the profits to be gained in the short term will then continue to
rise (due to the continuing fall in real wages).
The next step in Hayek's argument is to show that the process
described above will turn the boom into a recession. For this to be the
case, the decline in demand for investment goods (caused by the falling
profitability of employing capital) must more than offset its increase due
to the rise in demand for consumers' goods and the resulting demand for
greater production capacity. Whether this will be the case depends on the
reactions of agents to the continuation of the rise in product prices.
According to Hayek, the rising rate of profit in the consumer goods indus-
tries will eventually reach a level at which most agents will try to expand
their consumer output as quickly as possible. In Hayek's views, this means
that they will invest in more labour, while keeping the capital endowment
of the firm constant. The methods of production will become less and less
capital-intensive ('shorter'), eventually causing a fall in total demand for
investment goods. As a result, the investments goods industry will fall into
a recession. 21
Wilson (1940, p. 172) questioned the validity of Hayek's argu-
ment: "Would it not be possible for the system to remain in equilibrium at
a high level of activity?" In his view, the initial credit expansion leads to a
larger (more capital-intensive) capital structure. This means that the
economy is richer with a larger per capita capital stock that could produce
enough income to generate the savings to maintain itself. Then there is no
reason why a depression would emerge (Moss and Vaughn (1986, p. 556)).
However, Hayek (1933 (1939), p. 180) had already addressed this issue by
21
Hayek's (1939 (1950)) analysis assumed that the market rate of interest will not
change. It is obvious that as soon as we depart from this assumption the market rate of
interest will increase because of the rise in profits (which causes an increase in the demand
for credit), or because the expansion of credit is ended. This increase in the market rate
will curtail production in those industries that cannot earn profits at this higher rate. In
tum, "... incomes and the demand for consumers' goods and profits in the consumers'
goods industries would cease to rise" (p. 32). This process ultimately causes the boom to
tum into a recession. Hayek tried to show that that the boom will come to an end even if
the market rate of interest were to remain constant. In his view, he had already shown in
his Prices and Production (1931) that the boom would cease in a situation in which the
market rate of interest was perfectly flexible.
57
stating that this was merely a theoretical and not a practical possibility.
He regarded such an event highly unlikely because its preconditions were
too stringent. More in particular, money must be neutral, in the sense that
changes in its supply must leave the structure of relative prices unaffected.
4.6.2. The debate on the Ricardo effect

Wilson's 1940 review article aimed his criticisms of the Ricardo effect
against its empirical relevance and its logical consistency. He built on the
work of Kaldor (1939) for this latter issue.
Hayek (1939) had used a numerical example in which he tried to
show the result of a fall in real wages on the rate of profit earned on
various kinds of labour. An increase in the rate of profit in the consumer
goods industries would lead to a considerably higher increase of the rate
of profit on labour employed for shorter periods. Wilson (1940, p. 170)
thought this example to be "clearly unrealistic and misleading by implica-
tion", because it referred to investment periods of shorter than one year
while in the 'real world' only investment for longer periods is relevant.
Using Hayek's example, he showed that "... the consequent changes in
relative profitability [on investments for longer periods] will be too small to
be taken into account at all." Furthermore, Wilson stated that it is highly
unlikely that producers would change their methods of production because
of short-run fluctuations in real wages and in the marginal cost of borro-
wing. He quoted Hicks's (1939, p. 226) remark that "... interest is too
weak for it to have much influence on the near future; risk is too strong
to enable interest to have much influence on the far future" (p. 178).22
Hayek (1942a, pp. 131 - 32) addressed Wilson's criticism by intro-
ducing two concepts which measure the proportions in which capital and
labour are combined. It must be stated in advance that he interpreted the
firm as managing several investment projects. This means that it is faced
with different rates of return on different investments. As O'Driscoll
(1977a, p. 99) stated, it will try to equalize them at the margin (excluding
risk differences). For a firm to be in equilibrium, these internal rates of
return must be equal at the margin. However, in a discoordination
22
Obviously, this point of criticism concerns the agents' knowledge and expectations. If
agents think that a given change in the market rate of interest reflects a change in the
natural rate, they will react. Thus, Hicks and Wilson presume that agents know that the
market rate of interest does not equal the natural rate. However, this knowledge may not
be available in situations of discoordination. Hicks and Wilson assumed away the very
problem in which Hayek was interested.
58
situation, there may be differences between firms as regards their mar-
ginal rates of return. Thus, the model of the single firm does not apply to
the economy as a whole. The former may be in equilibrium, but the latter
is not coordinated. Hayek (1942a, p. 141) stressed the point that the
situation under consideration is not an equilibrium situation, but rather
one "... in which the causes of continuous and cumulative change are
inherent." Therefore, prices need not always be equal to marginal costs. In
this disequilibrium setting Hayek discerned two forces which tend to fix
the same price at different levels. The first or real force is constituted by
a given and only slowly changeable output of consumer goods, on the one
hand, and on the other, a relatively fixed propensity to consume on the
part of the consumers. Together both factors determine a ratio between
the prices of consumer goods and the prices of the factors of production.
The second or monetary force consists of an elastic supply of money
which (in general disequilibrium) tends to determine this ratio at a
different level. The extent to which both levels differ (that is, the extent to
which prices are distorted by monetary factors) depends on how fast diffe-
rences in prices are transmitted to the rest of the economy (via an
increase in incomes and the resulting rise in consumers' demand): "... the
speed at which an increase of incomes leads to an increase in the demand
for consumer goods limits the extent to which by spending more money
on the factors of production we can raise their prices relatively to those of
the products" (p. 143). Mter all, if there are lags (frictions) in an econ-
omy, money may not be neutral.
For another criticism directed against Hayek's business cycle
theory Wilson used Kaldor's (1939) conclusion that "... a fall in real wages
will increase the scale of investment (until marginal real wages in the new
equilibrium are the same as before), but it will leave the method of pro-
duction adopted unchanged" (p. 49). Wilson (1940, pp. 173 - 74) opined
that this diametrically opposed conclusion results from the fact that both
economists presuppose a different production function. In his view, Hayek
(implicitly) assumed either a nonhomogeneous production function in
which the marginal rate of substitution changes in favour of direct labour
as the scale of output increases (cf. Figure 1), or else a rising supply
schedule of credit ('loanable funds') to the firm, whereas Kaldor (explicit-
ly) assumed the production function to be linear and homogeneous, which
means that the marginal rate of substitution between labour and capital is
59
Indirect labour independent of the scale of
output (cf. Figure 2). 23 Accor-
ding to Wilson, one might
argue that relaxation of the
assumption of a constant mar-
ket rate of interest will validate
Hayek's argument. Then, the
marginal cost of borrowing will
increase. However, this means
0 N' that the supply curve of credit
drr~ct labour
(S) must be rising. Further-
Figure 1
more, the demand curve for
credit must shift to the right
(cf. Figure 3). This is possible
ind lrect labour
only if total investment increa-
~·
ses (from C to E). As capital
intensity can only be reduced
after the marginal cost of bor-
rowing has increased, it must
be accompanied by an increase
in total investment. The sub-
stitution effect can never offset
the scale effect. Or as Wilson
0 H' direct labour
(1940, p. 176, italics in original)
Figure 2
himself put it: "It is logically im-
possible for a reduction in depth
interest rat~
to offset an increase in width,
and given the supply schedule of
loanable funds, an increase in
consumers' outlay can never,
under these circumstances, lead
to a fall in the demand for
A
capital goods." Wilson thus
dismissed Hayek's business
cycle theory and, more in par-
0 investments/ savings ticular, his Ricardo effect as
Figure 3 incorporating a logical incon-
23
The analysis in terms of the figures has been derived from Blaug (1962 (1990)).
60
sistency. Kaldor (1942, p. 377) repeated this criticism and concluded that
"[t]his is the fundamental point, which knocks the bottom out of Professor
Hayek's new theory of the trade cycle, quite apart from any arbitrariness
or unreality of the assumptions on which it is based."24 However, as was
said above, Wilson and Kaldor analyzed a situation of general equilibrium
(coordination). In contrast, Hayek rejected such a comparative-static and
general-equilibrium approach. In turn, he criticized Kaldor (1939) and
Wilson (1940) for not facing the effects of the limitation in real resources
at the end of the boom. 25 Agents must then choose whether to provide
for the near or for the more distant future. This problem centers on the
question whether the future profits can offset the profits obtainable in the
short run, the latter constituting a component of the costs to be made
when pursuing long-term profits. Hayek (1942a, pp. 147 - 48) argued that
the item of the forgone short-term profits "... represents the costs of extra
waiting which the more capitalistic methods involve and which nowhere
enters the calculations of Mr. Kaldor and Mr. Wilson." He thus used the
notion of opportunity costs and (implicitly) claimed that costs are subjec-
tive.26 In his view, agents will choose the short-term, more labour-inten-
sive method of production because "... if the profits which might be made
in the near future are not obtained, they (and perhaps a certain amount
of permanent business) will be lost for good to a competitor" (p. 149).
Furthermore, the distant future will be more uncertain, which constitutes
another incentive for agents to seize short-term profit opportunities.
The discussion ended with the opponents sticking to their guns,
whereby Wilson and Kaldor were generally considered to have been victo-
rious. In 1967, however, the debate would be re-opened.
4.6.3. The re-opening of the debate: Hicks's 'The Hayek story'

J.R. Hicks's 'The Hayek Story' (1967) adressed the question who was right
in the 1930s, Hayek or Keynes. He described the great impact Hayek
made with his first English book, Prices and Production (1931). This book,
24
Kaldor preferred to use the term 'Concertina effect', because he argued that it could
not be found in Ricardo and because the effect merely implies that the capital-intensity of
production decreases in the upswing of the cycle and increases in the downswing.
25
In 1969 he explained that this limitation, at least partly, arises from the fact that
firms will find their ability to raise more capital limited by a rising supply curve of loanable
funds.
26
This subjectivistic notion of costs is in agreement with Hayek's later work on the
subjective nature of the economic data (including the agents' knowledge).
61
based on the 1930-31 Tooke Lectures at the London School of Econ-
omics, was an enormous success, even though several referees severely
criticized it. 27 As Hicks (1967, pp. 204 - 5) indicated, British economists
found it hard to understand it: "... what emerged, when we tried to put the
Hayek theory into our own words, was not Hayek. There was some inner
mystery to which we failed to penetrate. ... It is not so much that it was
rejected; it slipped through our fingers." Hayek's attempt to show what the
'mystery' was obviously did not succeed. And his later efforts hardly did
have an audience (Hicks (1967, p. 205)).
But what was Hayek's 'mystery', in Hicks's view? What was it that
made Hayek's theory so unintelligible to British economists? Hicks argued
that the essence of any dynamic theory is its lag. In his view, the lag in
Hayek's theory was a consumption-lag. In the Hayek model credit expan-
sion will lead to a fall in the market relative to the natural rate of inter-
est. This induces investments to rise, which implies - given Hayek's
assumption of full employment of production factors - a rise in prices of
investment goods, including labour. Therefore, nominal wages will rise.
Normally, according to Hicks, this must lead to higher demand for
consumer goods. However, not so in the Hayek model. "In spite of the
rise in wages the demand for consumption goods does not rise; so the
prices of consumption goods do not, at this stage, rise. This is how he is
able to maintain that there is a rise in the prices of producers' goods,
relatively to the prices of consumers' goods, ... This is the lag ... which
gives the Hayek model its peculiar slant" (Hicks (1967, p. 208)). And in
Hicks's view, this lag is clearly inadmissible. He argued that the Hayek
theory could only be saved by transforming it into a growth theory.
Hayek (1969b, pp. 279 - 80) dealt with Hicks's criticism by distin-
guishing between a single addition to the money supply and a continuing
inflow of money. He agreed that in the case of a single addition the
implied lag and the consequent price distortion are indeed temporary
phenomena. A continuing increase (at a constant percentage) in the
supply of money, however, will change relative prices for longer periods,
due to the nonneutrality of money. This change in relative prices will
depend on the place where the money enters the economy and on the way
it spreads through the economy. "The prices affected later never will catch
up with those affected first" (p. 280). Distorted relative prices will distort
investments. When the inflow of money stops, prices of investment goods
27 For instance, see Sraffa (1932) and Hawtrey (1932).
62
at this stage will fall, while prices of consumer goods continue to rise for
some time. According to Hayek, some of the investment will prove less
profitable than before, and therefore ex-post non-optimal. It is here that
the Ricardo effect is seen to become operative.
To summarize, Hayek's business cycle theory is a theory of the effects of

distortions of relative prices. As O'Driscoll (1977a, p. 153 - 4) concluded,
"[h]is theory is in the Cantillon tradition, which, broadly speaking, empha-
sizes distribution effects. Hayek's hypothesis concerns where and how
injections of money and credit enter the economy. He looked to private
investment as the key variable." This analysis of the effects of increasing
money and credit must hypothesize on the place where the additional
money and credit enters the economy (namely on the producer side), and
on the way in which it spreads through-out the economy. Furthermore, it
must also take the agents' knowledge and expectations into account. The
resulting analysis aims to be dynamic, capable of explaining discoordina-
tion dynamics. However, Hayek's business cycle theory was too incompre-
hensible for his British contemporaries to be appreciated. His Ricardo
effect (in itself controversial) operates only in discoordination situations.
It was widely misunderstood because Hayek's opponents interpreted it in
a comparative-static framework, which means that capital intensity can
only fall if total investment increases. With the benefit of hindsight, Moss
and Vaughn (1986, p. 551) concluded that Hayek aimed at more than just
comparing the 'Pure Logic of Choice' under varying sets of relative prices.
He wanted to explain the business cycle, or rather, "... why the end of the
boom phase of the trade cycle nearly always consists of a depression."
Why would rational agents, pursuing their own interest, make decisions
which, when reconciled with those of other agents, would cause the end of
the boom and the subsequent crisis? "... [W]hat are the microeconomic
foundations for the 'spontaneous disorder' that seems to characterize the
business cycle?" (p. 553)). In their view, "... Hayek's point is that precisely
because the entrepreneurs are mistaken about the true real savings of the
consumers, a time must come when they have evidence of their mistakes
and take steps to correct them" (p. 563). In this sense, the business cycle
is a phenomenon which is the unintended consequence of individual
rational action, due to imperfect knowledge and imperfect foresight.
Moss and Vaughn thus do not agree with Blaug's (1962 (1990), p.
545) conclusion that the Ricardo effect is "... only another instance of the
vice of neoclassical economics: the hasty application of static theorems to
63
the real world." By contrast, they hold that "... this is the last accusation
one could logically hurl at Hayek's analysis. The very reason why Hayek
encountered so much difficulty in communicating his message was precise-
ly that he was not presenting an exercise in comparative statics, but was
rather hypothesizing a particular adjustment process where the final
equilibrium state depended upon the particular path of adjustment
followed in the economy" (p. 546) ). They go further by arguing that "[i]t
was by responding to Kaldor's formidable criticisms of that mechanism
[i.e., the Ricardo effect] that Hayek finally realized why the comparative-
static approach based on perfect information was totally out of step with
the type of phenomena Hayek was trying to model" (p. 548)). His aim to
explain the dynamic causal process in time is clearly impossible in a
comparative-static equilibrium analysis. Moreover, in Hayek's (1942c}
view economists should not interpret economic relationships as being of a
mechanical nature. Both positions led him to write a series of articles on
economic methodology in which he attacked scientism, defined as the
position which leads to the "slavish imitation of the method and language
of [natural] Science" (p. 269). These articles form the first step in what
Caldwell (1988) has called Hayek's transformation, i.e., his turning away
from technical economics to other fields of research. 28
4.7 THE HAYEK PROGRAMME
In the early 1940s, Hayek tried to convince his fellow-economists of the

inappropriateness of the methods of the natural sciences when applied to
28
Caldwell (1988, pp. 532- 33) suggested that Hayek's (1941) views differ significantly
from those he explicated in 1937. He argued that "[t]hough he was increasingly cognizant
of the limitations of [general] equilibrium analysis as the 1930s progressed, Hayek never
abandoned the belief ... that economic analysis must make use of some concept of
equilibrium. Now [i.e. in 1941, p. 17] we find, to the contrary, that equilibrium analysis is
only 'prepatory' to the 'ultimate goal' of causal-genetic explanations of economic processes
as they take place in real time" (p. 532). However, Caldwell's implication of a shift in
Hayek's views on the role of equilibrium analysis is somewhat misleading. After all, Hayek
(1937, p. 33) bad already argued that "... my main contention will be that the tautologies,
of which formal equilibrium analysis in economics essentially exists, can be turned into
propositions which tell us anything about causation in the real world only in so far we are
able to fill those formal propositions with definite statements about how knowledge is
acquired and communicated." In other words, (general) equilibrium analysis is useful only
in so far it can be made empirically meaningful by adding statements as regards the
agents' knowledge acquiring processes, that is, by explaining ('genetic-causally') the
tendency towards coordination. In this sense, formal equilibrium analysis can be only
'prepatory' to the 'ultimate goal' of telling us anything about causation in the real world.
64
the social sciences. He wrote a series of articles, first published in Econo-
mica between 1942 and 1944 (and republished in The Counter-Revolution
of Science (1955 (1964), Part One)), in which he outlined his views on the
methods to be used by social scientists. Furthermore, these views were
supplemented in two papers on the complexity of the social sciences and
the ensuing restricted degree of explanation, both republished in Hayek's
(1967a) Studies in Philosophy, Politics and Economics. It should be noted,
however, that in the meantime he had changed his views about the
methods in the natural sciences, due to Popper's criticism on Hayek's
notion of 'scientism'. This means that he adopted Popper's methodological
monism, although he continued to argue that natural and social phenom-
ena differ in degree.
The research programme which emerged from these works will be
called the Hayek programme on explaining dynamics. It consists of two
elements: ( 1) the problem it tries to solve, that is, the Hayek Problem;
and (2) its methodological constraints. We may observe three such
constraints, namely methodological individualism, subjectivism, and the
compositive method.
4.7.1. The Hayek Problem

As early as 1935, in his 'The nature and the history of the problem' (as
regards economic calculation), Hayek distinguished between two types of
problems, namely economic and technological problems (pp. 3 - 8). The
latter problems are so-called 'engineering problems'. They presuppose an
unambiguous ranking of preferences, which enables the formation of a
consistent plan. Moreover, they can be solved by mere optimization under
constraints techniques (Pure Logic of Choice). But Hayek argued that the
economy as a whole does not have such an unambiguous ranking of prefe-
rences, unless all individual plans are compatible (i.e. unless general
equilibrium prevails). Therefore, transposed to the economy as a whole,
the engineering or technological problem only emerges in general equilib-
rium. However, this was not the problem in which Hayek was interested.
Instead, he studied situations of general disequilibrium (discoordination),
which are characterized by inconsistent individual plans. In such a situ-
ation, he argued, it is impossible for any one mind to decide which resour-
ces must be used for which purposes. Therefore, the economic problem
cannot be solved "... by first communicating all this knowledge to a central
board which, after integrating all knowledge, issues its orders" (Hayek
(1945, p. 84)). This means that he rejected the meaningfulness of the
65
metaphor of the auctioneer. Instead, he continued, "[w]e must solve it by
some form of decentralization.... We need decentralization because only
thus can we insure that the knowledge of the particular circumstances of
time and place will be promptly used." In this view, the economic problem
in society then becomes" ... how to secure the best use of resources known
to any of the members of society, for ends whose relative importance only
these individuals know. Or, to put it briefly, it is a problem of the utiliz-
ation of knowledge which is not given to anyone in its totality" (p. 78)?9
The problem he tried to solve was how this decentralized system could
bring about a state of affairs, in which all plans are coordinated. That is,
the Hayek Problem is: Why should the knowledge of agents eventually
become correct? Or stated in a more general manner: why should there
be a tendency towards the coordination between all individual agents'
plans? This question centers on the ways in which they acquire and
communicate knowledge, and on how they form expectations.
Hayek had concluded from the economic calculation debate of
the 1930s that economists should try to solve the Hayek Problem by
analyzing the capitalistic (decentralized) market system. The debate itself
had not resulted in an answer to the question how knowledge is acquired
and communicated. Therefore, it could not explain the tendency towards
coordination. In turn, this means that it could not explain the business
cycle in terms of the acquisition and communication of knowledge by
economic agents. Hayek attempted such an explanation in 1939. He was
able to make clear the importance of expectations. However, he did not
explicate how agents learn, nor did he give an insight into how they
formed their expectations. He merely presupposed that economic agents
eventually would learn the 'correct' state of the system. Obviously, this
presupposition enabled him to evade the questions to be answered. Thus,
he could not give a full explanation of discoordination dynamics.
Hayek did not immediately address the problems yet to solve, but
first outlined the methodological constraints of the Hayek programme.
These constraints are methodological individualism, subjectivism (or the
dispersion of (types of) knowledge), and methodological dualism (or the
compositive method).
29
Obviously, this is merely a new formulation of an old problem. As Hayek (1945
(1949), p. 78) acknowledged, it is the problem of a rational (efficient) economic organiz-
ation which was somewhat obscured by the mathematical refinements of economic theory,
in particular by the use of mathematics. Hayek's formulation should be considered new in
the sense that it emphasizes the crucial role of knowledge.
66
4. 7.2. Methodological individualism
Hayek's analysis is often interpreted as methodological-individualistic.
This interpretation seems to be supported by a sequence of articles
published under the title 'Scientism and the study of society'. In these
articles Hayek rejected the idea that 'social wholes' (such as society,
capitalism, government) have an existence of their own. In his view, these
wholes "... are no more than constructions, and ... can have no properties
except those which follow from the way in which we have constructed
them from the elements ... " (1943a, p. 45). This rejection of 'holism' or
'collectivism' thus appears to be based on ontological (or, at least, obser-
vational) individualism. It leads Hayek to conclude that explanations of
discoordination dynamics must run in terms of changes in individual plans
and actions.
Economics presupposes that individual agents act in order to
attain an end, that is, that they act rationally (purposively). But this does
not mean that social phenomena should be explained as the intended
consequences of purposive behaviour. On the contrary, Hayek (1942c, p.
288) claimed that the explanation of conscious, rational and purposive
behaviour is the task of psychology. By contrast, the social sciences should
concentrate on the explanation of social phenomena as the unintended
consequences of these rational actions. The phenomena are unintended or
undesigned because no individual aims at establishing or creating them. 30
In this sense Hayek's explanations are so-called spontaneous-order expla-
nations which explain social phenomena as spontaneously and uninten-
tionally brought about by individual rational action. In this sense these
explanations obey the principle of methodological individualism. However,
this principle is not unambiguous.
Lukes (1973, p. 124) argued that there are several forms of
methodological individualism. Their central tenet is the assertion that "...
all attempts to explain social and individual phenomena are to be rejected
... unless they refer exclusively to facts about individuals." Nozick (1977, p.
353, italics in original) specified this assertion by stating that "[t]he
methodological individualist claims that all true theories of social science
are reducible to theories of individual human actions, plus boundary
conditions specifying the conditions under which people act." Hayek (1946,
p. 6) himself argued that the basic contention of his methodological posi-
3
° Ferguson (1767 (1978) p. 122) already described such social phenomena as • ... the
result of human action, but not the execution of any human design. •
67
tion (which he called 'true individualism') is "... that there is no other way
toward an understanding of social phenomena but through our under-
standing of individual actions directed towards other people and guided
by their expected behavior." In other words, individual economic agents
are not isolated, because they must decide in a so-called decision situ-
ation.31 Nevertheless, it is clear that in Hayek's business cycle theory only
individuals can decide. Hence, methodological individualism is a methodo-
logical constraint in the Hayek Programme. However, the question can be
raised why individuals make expectational errors? That is, does some
'objective reality' exist against which their actions can be assessed?
Economic agents are presumed to act rationally. This means that
they will do the best they can, given their knowledge. In this sense they
act according to the logic of the decision situation. Given their knowledge,
the agents' decisions are ex-ante optimal. However, as Hayek's business
cycle theory shows, this does not mean that they cannot make any errors.
They may invest in too capital-intensive methods of production, so that
their decisions are ex-post non-optimal. Hayek explained the differences
between the ex-ante intentions and the ex-post outcomes by distinguishing
between the observer's (objective) decision situation and the agents'
(subjective) decision situation. That is, he adopted subjectivism as a
methodological constraint.
4.7.3. Subjectivism
Before addressing the issue of the types of knowledge, we must first make
it clear what we mean by saying that Hayek's analysis is subjectivistic.
Mises had already pointed out that preferences are subjective, in the
sense that they depend on the economic subject holding them. Given the
framework of static and dynamic subjectivism, as discussed in the previous
chapter, we can discern an important difference between Mises's and
Hayek's subjectivism. As Lachmann (1982, p. 37) observed, the former did
not consider the incorporation of expectations (and knowledge acquisi-
31 Therefore, it is perhaps better to speak of methodological situationalism. This

methodological principle holds that illtetactions instead of actions form the crucial
elements in society. This means that the individuals' decision situations are very important.
As Knorr-Cetina (1981, p. 15) stated, "... methodological situationalism ... challenges
methodological individualism for the simplifying assumption that the locus of social action
is the individual human being ... "In her view, "[m]ethodological situationalism has replaced
the model of the individual actor as the ultimate unit of social conduct by a conception
which incorporates the reciprocity and the situated character of social action. • This
conception is the decision situation.
68
tion) into his analysis a step forward because subjectivism "... meant to
him no more than that different men pursue different ends. ... The ends
themselves ... we have to regard as 'given'." Mises's analysis of the agents'
decision-making process is a form of pure logic of choice or optimization
under constraints, in which the ends and means were subjectively deter-
mined and assumed beyond the analytic touch of the scientist (and hence
'given'). This may explain why Mises did not incorporate processes of
knowledge acquisition and expectations formation into his analysis. In
turn, this led him to neglect the coordination problem which is inherent in
the step from individual to general equilibrium. By contrast, Hayek's
(1942c) analysis is more (though not completely) dynamic. He explicated
the importance of the incorporation of knowledge and expectations when
stating that "[s]o far as human actions are concerned the things are what
the people acting think they are" (pp. 277 - 78, italics in original). In this
sense, the facts of the social sciences are subjective. But given this subjec-
tivism, how can outside observers assess whether the outcome of some
action is non-optimal?
In the previous section it was asserted that Hayek explained the
ex-post non-optimality of the agents' decisions by distinguishing between
two types of knowledge. These types are ( 1) the knowledge of the econ-
omic agent and (2) that of the outside observer. The latter type of
knowledge entails scientific knowledge concerning the general structure of
the economy. The former consists of entrepreneurial insights and the
knowledge of the particular circumstances of time and place. Each agent
is presumed to have superior knowledge of his own situation, including
the market in which he operates. The distinction between these types of
knowledge is crucial in Hayek's analysis, because it allows for a distinction
between the truth status of the agents' knowledge and the role played by
this knowledge in the decision-making process (Boland (1978, p. 251)).
The role of the agent's knowledge in the decision-making process is a
purely logical one. It provides the economic agent with a sufficient and
logically consistent explanation of the world he faces (Boland (1978, p.
251)). This means that the agent's problems can be solved using tech-
niques of optimization-under-constraints, that is, according to the pure
logic of choice. In this sense the decision-maker decides in a single-exit
decision situation.
The explanation of social phenomena as unintended consequences
of human action appears to be inconsistent with Hayek's assumptions as
regards the dispersion of knowledge and the subjectivity of the economic
69
agents' knowledge. After all, such an explanation presupposes that the
social scientist must also know the intentions (or preferences) of the
individual economic agents. The question thus arises how the social
scientist can obtain this knowledge if it is dispersed and if it cannot be
held by any single mind. Hayek (1942c, p. 277) solved this problem by the
method of Verstehen (understanding). He argued that "... in his conscious
decisions man classifies external stimuli in a way which we can know
solely from our own subjective experience of this kind of classification.
We take it for granted that other men treat various things as alike or
unlike just as we do, although no objective test, no knowledge of the rela-
tions of these things to other parts of the external world justifies this. Our
procedure is based on the experience that other people as a rule (though
not always - ... ) classify their sense impressions as we do. ... But we not
only know this. It would be impossible to explain or understand human
action without making use of this knowledge." In this view, the social
scientist can understand the actions of the agents in terms of underlying
plans, otherwise a teleological explanation would be impossible. The
method of understanding, or Verstehen, is built on the presumption that
human beings classify their sense impressions more or less similarly. This
provides the social scientist with a 'key' which enables him to translate
observed behaviour into statements about the plans on which such beha-
viour is based.32
4.7.4. The compositive method

Hayek's 'Scientism and the study of society' (1942c, 1943a and 1944b) is a
protest against what he called the scientistic prejudice, which led social
scientists to adhere to the methods of the natural sciences. The rejection
of scientism can be interpreted as a corollary of three issues, namely
Hayek's subjectivism, his use of the method of Verstehen, and his view on
the method of discovery in the natural sciences. Since the former two
have already been discussed above, this section concentrates on the
method of discovery.
In Hayek's (1943a, p. 42) opinion, "... natural scientists ... are used
to seek first for empirical regularities in the relatively complex phenom-
ena that are immediately given to observation, and only after they have
found such regularities to try and explain them as the product of a
32It appears that Lachmann (1970, p. 30) also adopted this view. However, Popper
(1957, p. 138) had argued that the method of Verstehen can be used to derive hypotheses,
but not as a method of justification, because these hypotheses may be false.
70
combination of other, often purely hypothetical, elements (constructs)
which are assumed to behave according to simpler and more general
rules." The laws and specific circumstances concerning the system as a
whole are obtained inductively, that is, by generalizing observations. The
behaviour of the elements is derived from the behaviour of the system as
a whole, because the latter can be observed while the former cannot. The
main task of the natural sciences thus is to recognize the particular as an
instance of an inductively derived general rule. Hayek (1942c, p. 287)
called this method the analytic method. Transposed to the social sciences,
it holds that the agents' behaviour is explained in terms of inductively
obtained developmental laws for society as a whole. 33
According to Hayek (1964 (1967a)), the social sciences cannot
discover 'laws', because "... the conception of law in the usual sense has
little application to the theory of complex phenomena, and ... therefore
also the description of scientific theories as 'nomologic' or 'nomothetic' ...
is appropriate only to those two-variable or perhaps three-variable
problems to which the theory of simple phenomena can be reduced but
not to the theory of phenomena which appear only above a certain level
of complexity". He thus attributed the absence of genuine (universally
valid) laws in the social sciences to the greater complexity of the data of
the social sciences. This greater complexity can be expressed in terms of
the "... number of elements of which an instance of the pattern must
consist in order to exhibit all the characteristic attributes of the class of
patterns in question" (1964 (1967a), p. 25), whereby a pattern reflects
some order or regularity. This led Hayek to reject the analytic method
and to propose the synthetic or compositive method instead. 34
33
Popper (1968 (1972), p. 185, italics in original) criticized Hayek for accepting "...that
positivism or scientism is the only philosophy appropriate to the natural sciences." He poin-
ted out that in fact the natural sciences do not use the analytic method, and redefined
scientism as "... the imitation of what certain people mistake for the method and language
of science" (1957 (1976), p. 105n, italics in original). In Popper's view, scientism thus
incorporates a misconception of the methods of the natural sciences. In the Preface of his
Studies in Philosophy, Politics, and Economics (1967a) Hayek agreed with this redefinition,
and swapped his methodological dualism for Popper's methodological monism. However,
he maintained that the natural and social sciences differ, because in his view the latter deal
with more complex phenomena than the former. The sciences thus differ in degree, not in
kind.
34
Hayek (1952b (1979) p. 67n, italics in original) explains that he "... borrowed the
term compositive from a manuscript note of Carl Menger, who, in his personal annotated
copy of Schmoller's review of his Methoden der Sozialwissenschaften (Jahrbuch jar
Gesetzgebung, etc., n.f. 7 (1883), p. 42), wrote it above the word deductive used by Schmol-
ler." The term indicates that there is a close resemblance between Menger's (and hence
71
The compositive method starts from the rationality postulate.
Without delving deep into epistemological matters, it can be noted that
Hayek considered it to be a classification scheme, with which perceptions
are interpreted in teleological terms. 35 That is, the postulate reflects that
choices are made in decision situations, and that they can be 'understood'
as part of a plan of some agent. As discussed above, the method of Verste-
hen yields such understanding. 36 According to Hayek (1942c, p. 277), it
leads to infallible results, because as a rule human beings classify their
sense impressions similarly. The method of Verstehen may therefore serve
as a method of justification.37
The second phase in Hayek's compositive method is a deductive
and explanatory (in the sense discussed above) one, in which hypothetical
models are constructed in order to reproduce patterns of social interac-
tion. A class of patterns can thereby be defined as (sets of) algebraic
equations and hence as mathematical models. 38 They reveal states of the
macro-system, deduced from the introspectively derived knowledge about
the agents' decision situations (i.e. from micro-knowledge). These states
are the unintended consequences of the agents' purposive actions. They
are feasible because the micro-elements (agents) of the macro-systems
interact and are structurally connected. The deductions should therefore
include the specification of these structural connections and the relevant
Hayek's) and Mill's respective views on explanation in the social sciences, since the latter
had opined that "[i]n social phenomena the Composition of Causes is the universal law"
(1843, Book VI, Chapter vii, Section 1).
35 Huussen (1990) argued that Hayek's epistemological views are based on Mach's
(1905) 'empirio-criticism' ('neutral monism'). Mach (1905 (1920), p. 6) advanced that all
perceptions may be ordered in two ways, namely in a mental (psychological, teleological)
or physical order (context). These orders (contexts) can be regarded as means of classifica-
tion. Hayek (1952, p. 3) criticized Mach's implicit assumption that there is a one-to-one
correspondence between these orders, thus rejecting Mach's 'physical reductionism'.
36 'Understanding' aims to obtain knowledge about other minds, whereas 'explaining'
concerns social interaction.

37 It should be noted, however, that other Austrians, such as Mises and Rothbard, do
not accept Hayek's epistemological foundation of the rationality postulate. As Huussen
(1989) showed, the former considered the rationality postulate to be a Kantian synthetic a
priori proposition. Rothbard (1976, p. 24), on the other hand, took an 'Aristotelian and
neo-Thomist' stance, claiming that the rationality postulate is a 'self-evident', Aristotelian-
essentialistic law of reality (cf. Van Zijp (199la)).
38Hayek (1864 (1967a), p. 24) argued that "[e]very algebraic equation or set of such
equations defines ... a class of patterns, with the individual manifestation of this kind of
pattern being particularized as we substitute definite values for the variables."
72
interaction mechanism. In Hayekian business cycle theory, the vertical
structure of production specifies such structural connections, whereas the
process of competition is the relevant interaction mechanism. The expla-
nation of social phenomena can focus on two aspects, namely their
formation (origin, genesis) and their functioning (operation). Hayek
argued that these aspects should be taken into account, because in the
social sciences they are interrelated: the origin of social phenomena
concurs with the manner of their functioning. 39 The explanation of such
phenomena thus involves a genetic aspect.40 As Hayek (1942b, p. 289nl)
pointed out, the length of the evolutionary process is of no importance in
this regard, nor does it matter whether it is often repeated.
Hayek (1943b, p. 68) argued that the deductive stage yields
macro-systems (patterns, models). These are to be regarded as 'aprioristic'
classification schemes, with the introspectively obtained micro-knowledge
as their elements. The schemes are 'superimposed' on the perceptions, in
the sense that they are used to bring order in these perceptions. This
means that they are empirically meaningless (i.e. do not have empirical
content) though cognitively significant. The (empirically) empty 'boxes'
(elements) of these classification systems must be filled with empirical
data in order to establish its empirical validity, whereby the data consist
of initial conditions (including the agents' decision situations) and some
ceteris-paribus clause. This procedure of filling the 'boxes' was to be
undertaken in the third step of the compositive method. Given the classi-
ficatory nature of economic models, however, discorroborative test results
should not lead to their rejection. Hayek ( 1943b, p. 73) opined that they
can merely show whether the model under consideration is relevant. That
is, the model (classification scheme) may be 'discorroborated' by the facts
for two reasons. Firstly, its initial conditions do not occur in reality. In this
case, Hayek argued, the theory is considered to be empirically irrelevant
for the phenomena to be explained, as it was not designed to explain that
situation. Secondly, the theory may not apply to the particular case under
39
According to Hayek (1967b, p. 101), "... the institutions did develop in a particular
way because the co-ordination of the actions of the parts which they secured proved more
effective than the alternative institutions with which they had competed and which they had
displaced." Cf. also Menger (1883 (1969) p. 88).
40
Hempel (1965 (1970) p. 447) defined a genetic explanation as an explanation which
"... presents the phenomenon under study as the final stage of a developmental sequence,
and accordingly accounts for the phenomenon by describing the successive stages of that
sequence."
73
study because it does not take account of a sufficient number of condi-
tions. In other words, some omitted variables may have exerted an
influence on the phenomenon to be explained. In this second case the
ceteris paribus clause has been violated. The theory is then merely incom-
plete. Hayek concluded that a theory should only be rejected if it is
internally inconsistent. The role of empirical testing in economics is
therefore a very limited one.
According to Hayek (1942c, p. 289nl), the complexity of social
phenomena only allows social scientists to draw up generic schemes or
'explanation sketches'. According to Hempel (1965 (1970), p. 238), an
explanation sketch "... consists of a more or less vague indication of the
laws and the initial conditions considered as relevant, and it needs 'filing
out' in order to turn into a fully-fledged explanation." Since they describe
social phenomena in broad and somewhat vague terms, the predictions of
such sketches are difficult (impossible?) to discorroborate. In Hempel's
view, the sketch should therefore contain suggestions for further research
which must transform the sketch into a complete deductive-nomological
(D-N) explanation.41 Hayek, on the other hand, opined that social phe-
nomena cannot be explained in full detail because of the dispersion of
knowledge. That is, he maintained that social phenomena depend on too
many elements (individual agents) and are therefore too complex to be
explained in full.
In the 1960s, Hayek (1964 (1967a) and 1967b) elaborated his
views on the generic nature of social theories, and particularly on the
feasibility of predicting social phenomena. He thereby adopted Popper's
description of science as a hypothetical-deductive (H-D) system. Hempel
(1965 (1970), p. 367) pointed out that the H-D model of explanation, and
particularly its D-N version, implies the so-called symmetry thesis, which
holds that explanation and prediction have the same logical structure.42
Given this symmetry thesis and given Hayek's adoption of the H-D model,
41 Such an explanation subsumes an explanandum (phenomenon to be explained)
under general laws (covering laws). Hempel (1965 (1970), pp. 336 - 37) pointed out that
the deductive-nomological (D-N) model of explanation explains an explanandum E in
terms of an explanans S, which consists of particular facts C 1, C 2, ... Ck and general laws
Ll,~····Lr.
42 It should be noted that Hayek did not consider Hempel's (1965 (1970)) D-N model
of explanation relevant for the social sciences, because in his view social phenomena
cannot be subsumed under general laws. Furthermore, the D-N model aims to specify the
explanans fully. Hayek's adherence to generic schemes indicates that he did not regard
such full explanation of social phenomena feasible.
74
the limited feasibility of explaining social phenomena indicates that
prediction is also subject to the limitations imposed by the complexity of
the social sciences. Analogous to his views on the feasibility of explana-
tion, though, Hayek (1967b, p. 28) maintained that (social-)scientific
predictions are possible only in generic terms, yielding pattern predic-
tions.43 He advanced that "... the prediction that a pattern of a certain
kind will appear in defined circumstances is a falsifiable (and therefore
empirical) statement. ... The circumstances or conditions in which the
pattern described by the theory will appear are defined by the range of
values which may be inserted for the variables of the formula. All we
need to know in order to make such a theory applicable to a situation is,
therefore, that the data possess certain general properties (or belong to
the class defined by the scope of the variables)" (p. 28). Hayek mentioned
Walrasian general equilibrium theory as an example of a pattern. He
claimed that "(s]ince the theory tells us under which conditions a pattern
of this sort will form itself, it will enable us to create such conditions and
to observe whether a pattern of the kind predicted will appear" (p. 36).
He concluded that the resulting pattern prediction is testable, although its
empirical content will be small. Hence, social scientists should be modest
with respect to their predictive (and explanatory) claims.
According to Paque (1990, p. 291), Hayek's views on pattern pre-
diction reflect a belief that general-equilibrium theory (describing a class
of patterns) could, at least in principle, be discorroborated in empirical
tests. He rejected this belief, on the grounds that such a refutation is
meaningful only if some exogenous restrictions are placed on the theoreti-
cal framework. For instance, without any (exogenous) knowledge about
the agents' decision situations it is impossible to determine whether their
plans are coordinated and whether the economy has reached a Pareto
outcome, because it is impossible to know whether perceived prices are
equilibrium or disequilibrium prices. Paque (1990, pp. 292 - 93) subse-
quently reinterpreted Hayek views, and concluded that ( 1) economic
model building is much more complicated than in the physical sciences
because of the complexity of the phenomena involved; (2) econometric
testing is subject to more constraints and cannot achieve as much as
43
Patterns can be described by (sets of) algebraic equations, in the sense that any
equation gives a class of patterns. Individual manifestations (instances) of such patterns
(models) are obtained if the (exogenous and endogenous) variables in the equation(s) are
given definite values. A pattern prediction then is not a prediction in the usual sense of the
word; it predicts that a certain model will be valid in particular circumstances.
15
testing in the physical sciences (mainly because the test environment
cannot be controlled); and (3) economists should not expect economic
theories to deliver precise forecasts which could serve as a solid basis for
guiding policy decisions. This view on testing will be considered part of
the Hayek programme.
4.7.5. Summary
By the mid 1940s Hayek's analysis had come a long way. In 1928 he had
replaced Mises's static concept of the evenly rotating economy by his own
perfect knowledge and perfect foresight equilibrium. In 1937 he distin-
guished between individual and general equilibrium. This enabled him to
recognize and tackle the coordination problem. In order to solve it he
outlined a research programme, which we have called the Hayek pro-
gramme. It aims at explaining discoordination dynamics. Rational individ-
uals are presumed to make decisions which are ex-ante optimal. This
means that these decisions are governed by the logic of the situation, as
perceived by the agents concerned. However, the knowledge used in the
decision process need not be correct. This leads to unexpected and often
undesired outcomes of their actions. Agents are frustrated in their at-
tempts to fulfil their plans. According to Hayek, a learning process will
then start. They are supposed to identify their past mistakes, and to learn
from them. This means that their knowledge and their (perception of the)
decision situations have changed. The logic of these situations thus leads
to revisions of the agents' actions.
This process of revision can go on as long as coordination is not
attained. It led Hayek to adhere to situational dynamics (Boland ( 1978, p.
251)). This means that the agents' behavioural changes result from
changes in their knowledge. Situational dynamics thus interprets changes
in their actions as the result of their learning process. Agents are only
supposed to learn from the undesired outcomes of their actions. 44 The
Hayek programme thus tries to solve the Hayek Problem by explaining
44 As Boland (1978, p. 251) puts it: "... it is by means of ... unintended consequences
that an actor may learn that his knowledge is false." It should be noted, however, that
Boland uses the term 'unintended consequences' in the same sense as we have used
'undesired results'. We have chosen the latter because it should not be confused with
'unintended consequences' in the meaning of 'social phenomena'. Agents will only change
their actions if they discover more efficient means (given their preferences}, and not
because some social phenomena (such as language, market, or money) have arisen. That
is, the rationality postulate implies that undesired results are always unintended, but
unintended consequences need not be undesired.
76
discoordination dynamics in terms of the rational responses by agents to
changes in their knowledge. This means that the Hayek programme
should stress the importance of learning. Given the dispersion of knowl-
edge, this learning should be ensured by a decentralized process or
system. Hayek suggested that the market system and the competitive
process are most suitable for this task. They constitute the next direction
in which he directed his research efforts.
4.8. COMPETITION, LEARNING AND THE HAYEK

PROBLEM
Interpreting social phenomena as unintended consequences of individual

action does not solve the Hayek Problem. It does not say anything as to
why a tendency towards general equilibrium should occur. Such a ten-
dency means that there is a tendency for the decisions of economic agents
to become 'correct' in the objective sense (ex-post optimal). In other
words, it will depend on their learning behaviour. As was already stated,
Hayek sought the solution to the economic problem facing society in
complete decentralization of decision-making. In his view, economic
agents would learn fast enough as to bring about a tendency towards
coordination. Such learning behaviour presupposes, though, that agents
are capable of obtaining knowledge about the correct relative profitability
of the alternative actions which they can undertake. This knowledge must
be communicated at such speed that agents can make objectively correct
(i.e. ex-post optimal) decisions. The question then is, which form of
decentralization will ensure such rapid communication of the relevant
knowledge? According to Hayek (1945 (1949), p. 85), "... in a system in
which the knowledge of the relevant facts is dispersed among many
people, prices can act to co-ordinate the separate actions of different
people in the same way as subjective values help the individual to co-
ordinate the parts of his plan." The decentralized market and the con-
comitant price system form this decentralized system which should ensure
the communication of all relevant knowledge. The process of competition
(as opposed to the situation of perfect competition) is presumed to ensure
the implied learning of the 'correct data'.
In 1946 Hayek delivered the Stafford Little Lecture at Princeton
University, in which he pointed out that the term 'competition' had
changed in meaning. He argued that 'perfect competition' had nothing to
77
do with 'competition' as used in ordinary language. It merely denotes a
situation, a specific market structure. In contrast, competition in the
ordinary sense of the word implies change. In this sense, one might even
argue that competition in the ordinary meaning is completely absent
under conditions of 'perfect competition'. Hayek (1946 (1949), p. 93)
denied that this is merely a semantic issue. In his view, it blurs the
fundamental task of economics, namely the analysis of the use of knowl-
edge in society. If the analysis of perfect competition and the concomitant
role of the Walrasian auctioneer is abandoned and replaced by the
analysis of the process of competition, then "[ t ]he problem becomes one
of how the 'data' of the different individuals on which they base their
plans are adjusted to the objective facts of their environment (which
includes the actions of the other people) .... we have now to deal not only
with several separate sets of data of different persons but also - and this is
even more important - with a process which necessarily involves continu-
ous changes in the data for the different individuals" (pp. 93 - 94 ). The
process involving continuous changes in the data is either a learning
process or a process of changes in the interrelationships between econ-
omic agents. These changes cannot be analyzed in the framework of
'competitive equilibrium', because this framework is merely a "tautological
method". It can only derive conclusions which "... are implicit in its
assumptions: the desires and the knowledge of the facts, which are
assumed to be simultaneously present to a single mind [such as the
auctioneer], determine a unique solution" (p. 93). The state of perfect
competition is incapable of dealing with situations in which knowledge is
dispersed. In contrast, the process of competition is not. Hayek therefore
concluded that we should concentrate on the latter.
Hayek's conclusion raises the question as regards the nature and
significance of competition as a process. He pointed out that the analysis
of this process need not assume the production of given commodities at
given marginal costs. Instead, its outcome (and hence the solution of the
economic problem in society) is unknown, because the process involves
the agents' continuous attempts to do better than their competitors.
Competition between entrepreneurs then means that they will try to
discover new ways of doing (new) things better (p. 101). This means that
Hayek abandoned the assumption that ends and means are given. Instead,
the ends-means framework is also allowed to change. This reflects a more
dynamic subjectivism than that previously used by Mises and Hayek
78
himself.45 He argued that if (at least some of) the attempts to do (new)
things better are successful, then the competitors will adopt the new
practices, copying the successful methods. This adoption means that they
must have learned the new methods themselves. In Hayek's view the price
system plays a crucial role in this respect. Unless distorted by external
influences, it allows the diffusion of information because it indicates the
most efficient use of resources known to any agent. After all, if some
producer were to invent a new way of producing his products, he could
underbid his competitors. The plans of the latter would then be frustrated,
because the data (i.e. the methods of production) have changed. This
unforeseen change requires the competitors to adapt their methods of
production as well, or else they will be forced out of the market.
Hayek {1946 (1949)) criticized his contemporaries for neglecting
the process of competition and for merely analyzing states of (general)
equilibrium. As we have seen in the debate on the Ricardo effect, he
stressed the need to analyze the process of discoordination dynamics,
because it determines the new equilibrium situation which (given the
absence of other unforeseen changes in the data) will eventually be
attained. The process of competition is considered to be most important
during the phase of (general) disequilibrium, because in this phase the
competitors must adapt their actions in order to stay in business. Hayek
concluded that competition as a process ensures that "... it is most unlikely
that, without artificial obstacles which government activity either creates
or can remove, any commodity or service will for any length of time be
available only at a price at which outsiders could expect a more than
normal profit if they entered the field" (p. 105). In other words, he main-
tained that there will be a tendency towards coordination. However, as in
1937, he was still unable to answer the question how such a tendency
comes about. After all, discoordination prices do not convey complete and
correct information. This means that there is no reason why agents should
learn the 'correct' data of the economic system. They need not learn from
their mistakes, because under changing conditions adjustment does not
imply improvement.
Hayek himself would not solve the Hayek Problem. By the end of
45
According to Lachmann (1982) and Bohm (1982), the Austrian revival adopted
some sort of dynamic subjectivism, in which agent use their active minds to order and
formulate ends, allocate means to them, make and revise plans, determine whether the
plans were successfully carried out, an subsequently adapting the plans. It can be argued
that this type of subjectivism builds on the subjectivism implicit in the Hayek programme.
79
the 1940s he became increasingly interested in political philosophy. As the
above quotation from his 1946 article already indicates, he acknowledged
that governmental policies can create or remove the conditions under
which competition as a process will flourish. In the late 1940s and early
1950s he envisaged the danger of such removal happening. This led him
to warn that "... we should worry much less about whether competition in
a given case is perfect and worry much more whether there is competition
at all" (p. 105). He redirected his research activities into the field of
political philosophy, thereby elaborating his popular book The Road to
Serfdom (1944). In his 1947 article '"Free" enterprise and competitive
order', based on a discussion held at the first conference of the Mont-
Pelerin Society, he made the final step away from technical economics
and economic methodology towards epistemology and political philosophy.
Moreover, he undertook a final attempt to solve the Hayek Problem by
elaborating a neuro-psychological theory of learning. In the late 1960s he
would return to technical economics by writing 'Three elucidations of the
Ricardo effect' (1969). However, this defense was a mere riposte in the
re-opened debate on the effect.
4.9. CONCLUSIONS
Hayek elaborated Mises's business cycle theory by replacing its bench-

mark (twice) and endogenizing the disturbances causing the cycle. The
former enabled him to presuppose that all agents have imperfect knowl-
edge. The dispersion of knowledge forces them to form expectations on
the basis of this incomplete knowledge. Given the distortion of the market
rate of interest, investment decisions will thus be based on false infor-
mation. These decisions will then be ex-post non-optimal, leading agents
to invest in too capital-intensive methods of production. As soon as they
discover their mistakes they will try to correct them. However, it takes
time to dismantle the unprofitable investments, and so will the depression
(i.e. the adjustment process in which unprofitable investment projects are
abandoned).
Hayek's business cycle theory could not clearly explain why there
is a tendency towards coordination. This led Hayek to concentrate on the
explanation of this tendency. The ensuing research agenda has been called
the Hayek programme. It contains three methodological constraints,
namely (1) methodological individualism, (2) subjectivism (or the disper-
80
sion of knowledge), and (3) methodological dualism (or the acceptance of
the compositive method). Its central problem is the Hayek Problem, which
tries to answer the question how to explain discoordination dynamics in
terms of rational individual action. The process of competition in a
decentralized market is interpreted as the means of communicating all
relevant knowledge, thus ensuring the tendency towards coordination.
However, Hayek did not elaborate this idea further, and hence did not
solve the Hayek Problem. Instead, he became increasingly interested in
the relationship between governmental policy and the process of competi-
tion. Therefore, he concentrated on political philosophy. It robbed
Austrian economics of its most prominent proponent. However, Hayek's
influence on economics had already diminished by 1946. The 1950s and
1960s were to be dominated by the Hicks/Hansen version of Keynesian
macroeconomics. As Caldwell (1988, pp. 537 - 38) noted, the followers of
one of the founders of neoclassical theory, Carl Menger, were soon
regarded as "... just another fringe group that was criticial of mainstream
practices." Only a few economists remained loyal to the ideas of Mises
and Hayek. Their contributions to the execution of the Hayek Programme
will be discussed in the next two chapters.
81
5. THE YEARS IN THE WILDERNESS
5.1. INTRODUCTION
The formulation of the methodological constraints of the Hayek pro-

gramme, as given in the previous chapter, took place when Austrian
thought did not occupy the central place in economics it had taken in the
early 1930s. The success of Keynes's General Theory of Employment, Inte-
rest, and Money proved detrimental to the Austrian tradition. As early as
1940 Ludwig Lachmann, a student of Hayek's at the London School of
Economics in the 1930s, acknowledged that the eclipse of Austrian
business cycle theory could be regarded as a consequence of the "...
ascendancy of the doctrines of Mr. Keynes and his followers ... " (1940, p.
179). He added that "[b ]y 1940, its most faithful adherents have to admit
to themselves that few of the high hopes it held out in the halcyon days of
the early 1930s have been fulfilled" (p. 179). However, it seems too far-
fetched to contribute all of the eclipse of the 'Austrian School' to Keynes.
As was mentioned in the previous chapter, the Austrian capital and
business cycle theory was attacked in 1935 by Knight. Hayek's defence
could not convince his opponent, which led Hayek to state his case more
fiercefully in The Pure Theory of Capital (1941 (1950)). Again this work
failed to convince the profession. As a matter of fact, the Keynes-Hayek
debate had already been lost by the latter. The Austrian School had alrea-
dy become a fringe group in neoclassical economics. Additionally, the
debate on the Ricardo effect made clear that the Hayek Problem was not
widely appreciated as an important problem. Hayek's reaction to the
(from his point of view unfavourable) outcome of the debates consisted in
turning towards economic methodology. He wrote 'Scientism and the
study of society' (1942 - 44), which was discussed in the previous chapter.
Some other Austrians also wrote articles on economic methodology, in
which they repeated and elaborated the arguments of Mises and Hayek.
As Dolan (1976, p. viii) however remarked, by that time the Austrian re-
search programme could only be pursued in relative obscurity. This
became even more so when Hayek's research efforts moved towards non-
economic fields of research. In Lachmann's (1982, p. 630) words, "... when
in 1950 Hayek left the London School of Economics to join the distin-
guished Committee on Social Thought in Chicago and devote himself
entirely to political philosophy, Austrian economists lost their most
83
inspiring leader. For them, the next 25 years or so were years in the
wilderness."
This chapter is intended to describe the work of Ludwig Lach-
mann, Israel Kirzner, and Murray Rothbard from the 1940s to the 1960s.
It covers the period between Hayek's 'Economics and knowledge' (1937)
and the Austrian revival starting in the 1970s. 1 During this period the
Austrians were mainly concerned with either clarifying their point of view
(thus trying to set their contemporaries straight), or addressing key issues
concerning the Hayek Problem.
5.2. INNOVATION AND THE AUSTRIAN BUSINESS

CYCLE THEORY
Lachmann's early work appears to be dominated by Keynesian economics

and its adverse effects on Austrian business cycle theory? Unlike many of
his contemporaries at the London School of Economics, Ludwig Lach-
mann did not adopt Keynes's theory. 3 On the contrary, he even downplay-
ed it by ascribing its success to the fact that "... it is a most vivid descripti-
on of a peculiar historical situation ... " (1940, p. 179). In 1939 he argued
that the Austrian theory intended to explain a primary depression, in
which there is no lack of effective demand. The Keynesian alternative did
not explain this situation, but instead concentrated on the secondary
depression (in which there is such a lack}. Although Keynesian and
Austrian methodologies differed substantially, Lachmann {1939, p. 68)
concluded that both theories should be regarded as complementary, in
that they explain different situations. He therefore regarded the Key-
1
Hayek's work on competition and the use of knowledge in society has already been
discussed in the previous chapter. Therefore, some overlap in time exists between the two
chapters.
2
Lachmann wrote several articles to reverse the tide. For instance, Lachmann (1937)
addressed the issue of liquidity-preference and its relationship with fundamental uncertain-
ty; and Lachmann and Snapper (1938) tried to oppose the Keynesian view on the cyclical
fluctuations in commodity stocks by showing that empirical analyses of these fluctuations
could not distinguish between the Keynesian and the Austrian view.
3Abba Lerner, Nicholas Kaldor and perhaps John R. Hicks all adopted Keynes's
views. This means that three of the 'big four' turned Keynesian, Hayek being the only
exception. For an account of this eclipse of Austrian thought by one of the economists who
experienced it, see Lachmann (1982).
84
nesian policy conclusions as valid in the case of a secondary depression. 4
Lachmann did not blame Keynes for the entire eclipse of the
Austrian theory. He argued that the intellectual roots of the earlier
Austrian theory had also contributed to it. The earlier versions of the
theory had started from the assumption of full employment which was
very much criticized in the 1930s. Hayek (1939) had removed it by starting
his analysis in a situation of unemployment and overcapacity. Lachmann
(1940) pursued a more or less similar goal. He argued that Keynes's
theory gave the impression (but nothing more than that) of being more
dynamic than that of the Austrians. He claimed that the Austrian theory
of the business cycle merely appeared to be too static to the non-Austrian
contemporaries. 5 Like Hayek (1939) he therefore tried to show that" ... the
Austrian theory is essentially dynamic, and ... that any appearance to the
contrary in its first presentation was really due to the upbringing of its
protagonists to whom Walrasian equilibrium conditions appeared the
natural jumping-off ground for all excursions into the real world."6 Lach-
mann not only stressed Hayek's (1939) rejection of the assumption of full
employment, but he also tried to restate the theory, trying to make it
more dynamic by eclectically incorporating Schumpeter's (1939) analysis
of technological change ('innovation') into Hayek's business cycle theory.7
Hayek (1933a) had already allowed for real causes of the Austrian busi-
ness cycle, but he had preferred to call the resultant processes mere 'ad-
4 Seventeen years later, Lachmann (1956, pp. 125- 26) would still hold this view.
5 Mises's 'evenly rotating economy' was a static equilibrium concept. However, Hayek's
1928 'perfect knowledge and perfect foresight intertemporal equilibrium concept' incor-
porates time. Nevertheless, it appears plausible that Hayek (1933a) confused his readers by
starting his analysis in general equilibrium, just like comparative-static analysis.
6 Lachmann (1940, p. 269) clarified the dynamics of Austrian analysis by referring to

its presumption that the 'investment operation' is non-reversible This means that "[o]nce
'free capital' has been converted into buildings and machinery, any failure of events to
conform to expectations will upset everything."
7
In Schumpeter's (1939, p. 88) view, 'innovation' means combining factors of produc-
tion in new ways which were not known before, which means that new knowledge has been
'created'. An innovation leads to increased consumer and producer spending. Spreading
through the economy from the point where they originated, these increases create the
boom. However, this process will not go on indefinitely, because when completely dis-
seminated, the effects of the innovation will cease. Lachmann's fellow-Austrians did not
consider technological change in this respect, although Hayek (1936b) did address the
question whether a central planning authority could more adequately use the opportunities
for increasing 'general wealth' made available by technical development than individual
entrepreneurs (which, in Hayek's view, it did not).
85
justments to a new general equilibrium position', instead of 'business cy-
cles'. Lachmann (1940) did not make such a distinction. 8 In his view, the
monetary system is not necessarily the main causal factor of the business
cycle. It may nevertheless play an important role during the boom, becau-
se it· can finance the innovation-induced investment projects by providing
firms with credit. Moreover, entrepreneurs may undertake innovative
investments because they are misled by distortions in the price system.
Lachmann (1940, p. 192) pointed out that a 'cheap money policy' may
cause such distortions. Entrepreneurs then mistakenly innovate, leading to
a rapid intercyclical increase in labour productivity, which may result in
increased unemployment. In Lachmann's view, this 'mistaken innovations
theory' can explain the industrial crises in the nineteenth and early
twentieth century. However, he admitted that "[i]t is ... not easy to account
for the crisis of 1929 by the help of the Austrian theory" (p. 195).
Lachmann's 1940 article failed to solve the problems implicit in
Hayek (1939), because it did not explicate the agents' learning behaviour
and their expectations formation processes in the course of the business
cycle. In this sense the Hayek Problem had yet to be solved. In addition,
Lachmann also failed to revive Austrian business cycle theory. Other
economists ignored it altogether, and even Lachmann himself never
returned to the 'mistaken innovations model'. Hence Hicks's (1965, p.
185) poetic conclusion that the Austrians' voices were "... drowned in the
fanfare of the Keynesian orchestra" clearly applies to Lachmann (1940).
5.3. KNOWLEDGE AND EXPECTATIONS
In 1943 Lachmann returned to the analysis of expectations in 'The role of

expectations in economics as a social science', which contains the results
of research he had already undertaken during 1939 - 40.9 This article had
two aims. Firstly, it made a plea for limiting the goal of economics as a
social science. This limitation is related to Lachmann's subjectivistic
interpretation of knowledge and expectations. He argued that expectations
depend on the agents' past experience, thereby stressing that this experi-
8Perhaps this is an early indication of his later rejection of (the tendency towards)
general equilibrium.
9 In 1941 Lacbmann published an article in which be discussed the difficulties in
measuring 'capital'. We shall neglect it because it is not very important for our purposes.
86
ence consists of the agents' interpretation of past observable events (p.
18). These interpretations are subjective, in the sense that they depend on
the subject holding them. They cannot be regarded as outcomes of any
determinate process. The same applies to the process according to which
the individuals transform their experience into expectations (p. 14). In
Lachmann's view, this means that the individuals' knowledge and expecta-
tions can neither be explained nor predicted (pp. 15 - 16). This conclusion
follows from his view that explanations in the social sciences should not
be probabilistic. 10 This indeterminateness also led him to reject general
equilibrium theory. 11
Lachmann also discarded prediction as a goal of economics. He
argued instead that the goal of economics should be more limited. Econo-
mics should merely make economic activity intelligible in terms of purpo-
sive human actions (and the plans, knowledge and expectations on which
they are based). But the subjectivity and indeterminateness of the econo-
mic agents' interpretations and expectations appear to hamper this goal.
After all, how can we explain economic activity in terms of human action
and expectations if the latter are subjective and cannot be regarded as
outcomes of determinate processes? Lachmann (1943, p. 18, italics in
original) implicitly referred to the method of Verstehen when claiming that
"[w]e now realise that ultimately it is the subjective nature of these beliefs
which imparts determinateness to expectations as it is their mental nature
which renders them capable of explanation." The (social) scientist can
make human action intelligible because he can interpret such action as
purposive and goal-directed. In this view, economics merely provides an
interpretation scheme, that is, a way of interpreting social reality. 12
The second aim of Lachmann (1943) concerns the role of expec-
10
Lachmann (1950) stated in his inaugural lecture at the University of Witwatersrand
that "[a]ll attempts to smuggle in predictability by the large back-door labelled 'the Law of
Large Numbers' are bound to fail since human events lack the quality of 'randomness'
essential for this purpose" (p. 170). For a discussion of the introduction of randomness in
economics, see Morgan (1990). For a discussion of competing 'approaches' in
econometrics, see Darnell and Evans (1990).
11 According to Lachmann (1943, pp. 15 - 16), general equilibrium analysis is a 'pure

logic of choice'. It interprets human action as a "...quasi-automatic response to an external
stimulus. " It thus treats experience interpretation and expectation formation processes as
determinate.
12
This view was also endorsed by Hayek (1942 - 44). In his The Sensory Order (1952)
he would provide some epistemological (or perhaps neuro-psychological) foundations of
this position.
87
tations in the Austrian explanation of cyclical fluctuations. 13 This article
appears to have been much influenced by the debate on the Ricardo
effect. In the course of this debate Wilson (1940) and Kaldor (1942) had
criticized Hayek's business cycle theory by arguing that investment does
not depend on fluctuations in the market rate of interest. Lachmann
(1940) also argued that investment depends on longer term considerati-
ons, and in his 1943 article he adhered to this view. Phrased in terms of
the elasticity of expectations, this means that he considered interest expec-
tations fairly inelastic. 14 The capital market then does not react upon
every short-run fluctuation in the market rate of interest, because it is
characterized by agents who do not believe that these fluctuations reflect
fluctuations in the natural rate of interest. They do not believe that the
marginal rates of profit to be earned on the already undertaken invest-
ment projects have changed. Lachmann (1943, p. 23) observed that "[i]f
inelastic expectations are really as frequent and important as some writers
would have us believe, an interesting problem arises with regard to the
interpretation of Wicksellian theory, more particularly in its Austrian
version. ... Without fairly elastic expectations there can ... be no crisis of
the Austro-Wicksellian type." 15 He even suggested that this theory might
only be applicable to economies which are in the early stages of
industrialisation or which undergo rapid technical progress because under
these circumstances "... long-run forces have not yet had time to take
shape." However, this scepticism as regards Austrian business cycle theory
does not mean that Lachmann dismissed it. He merely concluded that
13 In his 1978 interview with the Austrian Economics Newsletter Lachmann explained
that "[i]twas Rosenstein-Rodan who in discussing Austrian trade cycle theory with me said
'Ah, yes, but whatever happens in the business cycle is in the first place determined by
expectations.'" Rosenstein-Rodan had been an assistant to Hans Mayer, who held Menger's
chair in the University of Vienna.
14 Hicks (1939, p. 205) defined "... the elasticity of a particular person's expectations of
the price of commodity X as the ratio of the proportional rise in expected future prices of
X to the proportional rise in its current price." Thus, it is the sensitivity of expectations as
regards (proportional) future price changes for (proportional) changes in current prices.
15 Strangely enough, this argument did not seem to take Hayek's (1939) analysis into
account, which asserted that the cycle could also take place if the market rate of interest
was to remain constant. His analysis assumed money wages to be given. An increase in the
demand for consumer goods (caused by an increase in consumer credit) then raises their
prices and lowers the real wage rate. Profit margins increase and investment will rise. Of
course, as Hayek (1939, pp. 16 - 18) recognized, entrepreneurs must expect that the rela-
tively higher product price and the relatively lower real wages persist for a period of time
that is long enough to make it worthwhile to change their method of production.
88
(Austrian) economists should make it clear why elastic expectations would
prevail, that is, why entrepreneurs are inclined to mistakenly interpret
changes in the market rate of interest to reflect changes in the natural
rate of interest. This conclusion provoked a reaction by Mises.
Mises (1943) intended "... to recall that the 'Austrians' are not
guilty of having neglected an essential assumption of their theory" (p.
252). He was clearly not prepared to dismiss the Austrian theory because
of Lachmann's criticism. Instead, he defended the assumption of elastic
interest-expectations by referring to the entrepreneurs' inability to distin-
guish between long-term and short-term changes in the market rate of
interest, that is, by referring to their inability to solve the 'signal extraction
problem' correctly. He explained this inability by arguing that the absolute
level of the market rate need not be low. The rate is only relatively low,
that is, relative to the level which it would have reached in a period of
progressive inflation. This level equals the natural rate of interest plus a
compensation for the rate of inflation. However, since only the market
rate of interest is observable, entrepreneurs must use this rate as a proxy
for the natural rate. If the proxy is distorted, investment decisions will be
ex-post non-optimal, and there will be malinvestments! 6
Lachmann did not write a rejoinder, which may indicate that he
acknowledged the validity of Mises's argument. However, he did not
appear to be satisfied with the contemporary Austrian analysis. More in
particular, he concentrated on the factors inducing price expectations to
change during the economic process. His 'Note on the elasticity of expec-
tations' (1945) tried to address the relationship between actual prices and
price expectations. It formed the first Austrian attempt to specify the
circumstances in which expectations were likely to change (excluding
Hayek's (1939) remarks on expectations).
16
O'Driscoll (1977a, p. 106) formulated Mises's remarks as follows: Assume that nt is
the natural rate of interest at time t, it is the nominal or market rate at that time, and 'i is
the real rate. Assume also that APt represents the actual rate and APt e the expected rate of
inflation (formed at t-1). By definition, it = 'i + APt. In coordination the equality nt = it
( = 'i + APt) holds. If applied to the future this means that the equality it + APt+ 1 = nt +
APt~ 1 is the relevant equilibrium condition. Now suppose that a credit expansion reduces
i below n. In addition, suppose that individuals correctly anticipate future changes in the
price level (APt~ 1 = tJPt+ 1). According to O'Driscoll, this means that it + APt~ 1 < nt +
APt~ 1 • As long as the expansion of credit continues, this inequality will apply because i
then remains below n. This means that capital-intensive methods of production appear
more profitable than they in fact are. As O'Driscoll concluded, "[t]he fact that market
participants may succeed in protecting themselves against the effects of a generally
depreciating currency provides no basis for concluding either that inflation will be neutral
in its effects on the allocation of resources or that capital malinvestment can be avoided."
89
Lachmann wrote his 1945 article to criticize Lange's (1944) at-
tempt to try and measure this elasticity. Lange had assumed that "... each
person [acts as if he] forms some idea about the most probable value and
the 'practical range' of the expected price" (p. 30). He thus assigned
probability distributions to expected prices. By computing a certainty-
equivalent, Lange tried to obtain an 'objective' measure of the degree of
uncertainty of price expectations. Lachmann's (1945, p. 253) reaction was
to reject this mathematical and mechanical interpretation of the elasticity
of expectations, as it would imply that each and every price change leads
to a proportionate change in price expectations, given a certain degree of
elasticity. 17 He pointed out that the elasticity of prices must not be treated
as given. That is, it must not be interpreted mechanistically, because it
depends on the (free) choices of individuals. Instead, it may well depend
on the relationship between the initial price-expectations and the actual
outcome. If prices fall outside the range, expectations are disappointed.
Agents will then react by adjusting the range. Conversely, if prices fall
within the range, individuals consider their expectations to have come
true. In that case, they will not alter the range of expected prices. More-
over, expectations may even have a stabilizing effect. If prices move
towards the fringe, their rise (fall) will be hampered by inelastic expectati-
ons. Closer to the fringe people will be more inclined to sell (buy),
because they do not expect the rise (fall) in prices to continue.
Lachmann's views on the ways in which the expected-price range
will be adjusted implies that he assumed that individuals form expecta-
tions on the basis of past (price) experience. An individual's expected
price range is changed if past prices fall outside this range. This obviously
means that individuals learn inductively. As it did in Hayek's case, this
poses a problem, namely the Problem of Induction (cf. Boland (1982)),
which holds that there is no reliable inductive logic. As was already shown
in the previous chapter, there is no reason why individuals should form ex-
post optimal expectations on the basis of past experience. However, it is
not clear whether Lachmann (1945) already recognized this. Nevertheless,
17 In the 1940s and 1950s he would continue to stress this indeterminateness. In 1947
he explicitly rejected the ways in which his contemporaries incorporated expectations into
their analyses. He argued that Hicks's elasticity of expectations, Shackle's potential surprise
function, and Lange's practical range of expectations were meaningless, because "[n]one of
these theories came to grapple with the central fact of a dynamic world: the human acts of
interpretation by which men try to keep abreast of the changes in need and resources. All
these authors disregard the fact that man casts the material of his knowledge in the mould
of expectations" (1947, p. 142). As this knowledge is indeterminate, so are the expectations.
90
he made at least a step towards transforming the Hayek Problem. The
transformation itself would only be made in 1947.
5.4. PROCESS ANALYSIS
One of the major issues in Lachmann's work in the 1940s on capital

theory is the analysis of the differences between Austrian and neoclassical
capital theory. Lachmann argued that the neoclassical view on capital
restricts its analysis in important ways. In particular, the neoclassical
assumption of the homogeneity of capital prohibits the study of the
problem of malinvestments. In order to overcome this restriction Mises
and Hayek had claimed that 'capital' consists of heterogeneous goods.
Lachmann (1947) also adhered to this view. He argued that "(e]very capi-
tal instrument is designed for a purpose" (p. 203). This purpose is the
fulfilment of specific tasks in a firm's production process. However, it can
only perform them if it is combined with other capital goods. The capital
goods used must be considered complementary, because they allow the
execution of the firm's plan. "Factor complementarity presupposes a plan
within the framework of which each factor has a function" (p. 200). In
contrast, factor substitution refers to change, that is, to the adjustment of
a plan. "Substitutability indicates the ease with which a factor can be
turned into an element of an existing plan" (p. 200), thereby changing the
manner in which the plan is carried out (or the plan itself). Lachmann
(1947, p. 204) concluded, that "(t]he theory of capital has therefore every
reason to occupy itself with the network of production plans." The incom-
patibility of plans will induce entrepreneurs to change them. This will also
change their investment decisions. These decisions will prove to be com-
plementary when the different stages in the structure of production are
complementary. A discoordination situation will arise if these decisions
are not coordinated. This led Lachmann (1947, p. 210) to the methodo-
logical conclusion that the complementarity of investment decisions should
be studied. In his view this should be done by "... a new type of sequence
analysis which enables us to follow up, sector by sector, the chain of
changes set in motion by the impact of the original change." This type of
analysis, called process analysis, analyzes sequences of discoordination
situations in terms of (changes in) the individuals' knowledge and their
91
decision situations! 8 In Lachmann's view the problem thus became how to
explain sequences of discoordination situations. In this sense he rejected
the Hayek Problem, which sought to establish the validity of the claim
concerning the existence of a tendency towards coordination. Process
analysis, or situational dynamics, was considered to be the appropriate
method of analysis.
Questions of methodology were also addressed in Lachmann's
inaugural lecture (1950) and his review of Mises's Human Action (1951).
In the former he again rejected the notion of 'probabilistic explanation' by
claiming that the 'Law of Large Numbers' does not apply in the social
sciences, thus proposing to eliminate probability theory from economics.
Secondly, he stressed the specific Austrian view that capital goods are "...
future consumption goods in statu nascendi, and [that] their valuation re-
flects the pattern of time preference between the various combinations of
consumption goods of different degrees of futurity" (p. 104, italics in origi-
nal). And thirdly, he criticized Mises (and Hayek) for not explicating their
views on the processes according to which knowledge is acquired and
expectations are formed. Nevertheless, he acknowledged that the views
which were implied by Mises's analysis of the market process are ade-
quate. That is, he also adhered to the view that the market process
stimulates "... the spreading of knowledge through the promotion of those
capable of interpreting market data and of thus transforming them into
market knowledge, and the elimination of those who cannot read the signs
of the market" (p. 103). Lachmann (1954) addressed this issue of the
spreading of information, but his analysis did not add much to what was
already contained in Hayek (1945) and (1946).
5.5. CAPITAL, EXPECTATIONS, AND PROCESS ANA-

LYSIS
Austrians have a long tradition in capital theory. Menger, Bohm-Bawerk,

Mises and Hayek all wrote on the subject. Their theories can be divided
18 In adopting the term 'process analysis' Lachmann presumably followed Hayek's

(1941, p. 18n1) suggestion to eliminate the terms 'statics' and 'dynamics'. Hayek had
argued that "...the only relevant distinction is between two methods, that of logical analysis
of the different plans existing at one moment ('equilibrium analysis') and that of causal
analysis of a process in time. For this distinction the terms statics and dynamics seem
altogether inappropriate, and it would probably be better if they were to disappear entirely
from economics. "
92
into two broad categories. 19 Firstly, Menger and Mises interpreted the rate
of interest as determined purely by considerations of time-preference. In
contrast, Bohm-Bawerk also allowed for productivity as a reason for the
existence of interest. Hayek (1941, p. 420) more or less took this latter
standpoint. In contrast, Lachmann's Capital and Its Structure (1956)
returned to the former approach and later, Kirzner (1966, pp. 100 - 1)
would also endorse this purely-subjectivistic approach.2°
In his 1956 book Lachmann elucidated 'process analysis' by ex-
plaining that it is "... a causal-genetic method of studying economic chan-
ge, tracing the effects of decisions made independently of each other by a
number of individuals through time, and showing how the incompatibility
of these decisions after a time necessitates their revision" (p. 39).21 Not
surprisingly, Lachmann (1956, pp. xv - xvi) intended "... to emphasize the
transmission of knowledge, the interaction of minds, as the ultimate agent
of all economic processes". The main transmission mechanism in decen-
tralized market economies is the price system. In discoordination prices
do not reflect all information, that is, they are distorted. Entrepreneurs
are then faced with a 'signal extraction problem', which will lead them to
make expectational errors. They must then interpret prices (i.e. they must
answer the question 'what information do prices reflect?'). Consequently,
they must form their expectations on the basis of incomplete knowledge.
Such expectations are not likely to be ex-post optimal. This non-optimality
does not only arise because of the incompleteness of information. It also
exists because" ... today's knowledge may be out of date tomorrow, hence
no longer a safe guide to action" (p. 22) ). The economic agents are then
likely to make expectational errors. This means that (at least some of) the
individuals' plans are frustrated. They will learn from their past mistakes
and try to avoid them in the future. However, there is no reason to
suppose that their learning will lead to ex-post optimal expectations,
19
For a discussion of Menger's versus .BOhm-Bawerk's capital theory, cf. Endres
(1987), Zuidema (1989), and Garrison (1990).
20
The Austrians' contributions to capital theory are only considered if they directly
bear on the Hayek Problem (either as formulated by Hayek or as its transformed version).
21
Process analysis thus does not allow for strategic, game-theoretic interactions,
because decisions must be made independently, that is, without knowledge about the other
players' actions. According to Lachmann, Hicks (1939), Lindahl (1939) and Lundberg
(1937) had already expounded this method.
93
because we cannot derive from the past what will happen in the future?2
Lachmann considered the future indeterminate and concluded that there
is no reason to presume a tendency towards coordination. 23 Instead, he
considered the future 'kaleidic', unknowable (though not unimaginable),
and too volatile to know. The market process need not lead to coordinati-
on, because prices in discoordination situations do not reflect correct
information. As had already become apparent in 1947, Lachmann repla-
ced the Hayek Problem (i.e., the problem of explaining the tendency
towards coordination) by the problem how to explain discoordination
situations in terms of previous situations of discoordination. It should be
noted that this attempt to transform the Hayek programme was not suc-
cessful. Austrian economists continued to presuppose the existence of a
tendency towards coordination. In this respect Lachmann's view thus
deviates from that of other Austrians. 24
22
Lachmann (1956, pp. 67- 68) recognized that prices not only convey past events but
also expectations held in the present or past: "It is precisely the economic function of
forward markets to spread knowledge not about what is or has been, but about what
people think will be. In this way, while the future will always remain uncertain, it is poss-
ible for the individual to acquire knowledge about other people's expectations and to
adjust his own accordingly, expressing his own views about future prices by buying or
selling forward, thus adding his own mite to the formation of market opinion as expressed
in forward prices. In other words, forward markets tend to bring expectations into
consistency with each other. They are on the side of the stabilizers." Obviously, the prices
on forward markets can only convey information about what people in the present think
what will happen in the future.
23
Egger (1986, p. 62) even argued that Lachmann (1956) rejected the equilibrium
approach of Hayek's Prices and Production. However, Lachmann in fact also implicitly
recognized a general equilibrium, because he referred to the interpersonal inconsistency of
plans (p. 40). Such a reference implies the notion of interpersonal consistency, i.e. general
equilibrium. Mises and Hayek used general equilibrium as a benchmark and as the starting
point of their analysis. They merely presupposed a tendency towards such an equilibrium,
although they rejected the position that it could ever be attained. Lachmann (1956, p. 40)
went even further and claimed that ".. .in process analysis ... we need no such assumption."
Thus, he differed from Mises and Hayek in the sense that he rejected their presupposition
of a tendency towards general equilibrium, although he still used such an equilibrium
position as some sort of benchmark.
24 It is sometimes argued that the rejection of the existence of a tendency towards
coordination can be attributed to Shackle. Boland (1978) rationally reconstructed Hayek's

work by identifying what he called the 'Hayek Problem'. Boland's definition of this
problem incorporates the notion of situational dynamics which holds that behavioural
changes result from changes in the individuals' knowledge and/or from intended or
unintended changes in their situation. Thus, it differs from our definition by eliminating
Hayek's presumption of a tendency towards coordination. This difference can be explained
by Boland's attempt to replace Hayek's inductive epistemology by Popper's concept of
objective knowledge. After all, inductive learning implies that there is no reason to
presume that the individuals' expectations will become correct. In this regard Boland
94
According to Lachmann (1956), the task of business cycle theory
is to list all possible causes and therefore his approach may be characteri-
zed as 'eclecticist'. He believed that it is quite possible that business cycles
are caused by other factors than credit expansion. He regarded malinvest-
ment and underconsumption theories not as mutually exclusive, but rather
as complementary. Using Hicks's distinction between weak and strong
booms25 , he concluded that "... we do not maintain that the Austrian
theory could explain every and any industrial fluctuation that has ever
occurred. Such a claim would of course be incompatible with our plea for
eclecticism. The Austrian theory is a theory of the strong boom, it deals
with its causes and consequences. Undoubtedly, weak booms which ended
when consumption failed to keep in step with production have occurred in
history; America from 1929 to 1932 seems a prominent example. To
account for them a different kind of model is required. All we contend
here is that an underconsumption theory, which might account for the end
of a weak boom, is not exactly a suitable instrument for analyzing strong
booms. And there is now good historical evidence to show that strong
booms were a more or less regular feature of the expanding world
economy in its 'normal' conditions from 1870 to 1914" (p. 113).26 Thus, the
Austrian theory is intended to explain strong booms and the undercon-
sumption theory explains the recessions following weak booms. Lachmann
thus argued that the Austrian business cycle theory need not always be
applicable to reality. 27
(1978, p. 251) refers to Shackle's rejection of any such presumption. However, as was
shown, Lachmann (1947) had already rejected Hayek's presumption, whereas at that time
Shackle formulated his mechanistic 'potential surprise function'. Therefore, it is perhaps
more correct to ascribe the transformation of what we have called the Hayek Problem to
Lachmann. This transformation thus constituted a movement from our to Boland's 'Hayek
Problem'.
25
Hicks (1950, p. 107) distinguished between weak booms "which die of their own
accord" and strong booms "which are killed by hitting the ceiling". Or, "[t]he main reason
for the collapse of a weak boom is the insufficiency of its accelerator; the main reason for
the collapse of a strong boom is the insufficiency of real resources to sustain it" (p. 133).
26
According to Egger (1986, p. 58), Lachmann is unique among Austrians (differing
sharply with Rothbard, Mises, and Hayek) in identifying 'America from 1929 to 1932' as a
weak boom.
27
As was shown in Chapter 3, Mises (1928, pp. 40 - 41) also argued that economic
theories need not be universally applicable. Whether a theory applies in a specific situation
must be made clear by confronting the theory with empirical evidence. However, Mises
(1928) had rejected the underconsumption theory on theoretical grounds. He even main-
tained this position after the Great Depression. Cf. his (1949 (1966), pp. 301 - 3). By
95
To summarize, Lachmann started from Hayek's 1937 article 'Economics
and knowledge' by studying the role of expectations in economics. How-
ever, he considered knowledge and expectations to be subjective, in the
sense that they could not be explained by any determinate process. He
concluded that economists should not adopt mechanistic rules to describe
the individuals' processes of expectations formation and knowledge
acquisition. Instead, he continued, they should merely try to make social
phenomena intelligible in terms of purposive human action. He therefore
limited the goal of economics. Morever, he also rejected Hayek's pre-
supposition of the existence of a tendency towards coordination. Accord-
ing to Lachmann, there is no reason to presuppose such a tendency. In his
view, economics should adopt process analysis (or situational dynamics).
That is, it should adopt a causal-genetic method of explaining sequences
of discoordination situations.
In the 1960s two students of Mises took the (by then rather small)
center stage of Austrian economics, namely Murray Rothbard and Israel
Kirzner. The next section discusses the former's attempt to apply the
Austrian business cycle theory to the Great Depression, and section 5.7
expounds Kirzner's contribution to the analysis of market processes.
5.6. ROTHBARD'S HISTORICAL STUDY
In the early 1960s Rothbard published two books, namely Man, Economy,
and State (1962) and America's Great Depression (1963)?8 In the former he
undertook an attempt to provide an "... old-fashioned treatise on econ-
omic 'principles'" (p. vii), in which he explicitly adopted Mises's Human
Action (1949) as his starting point (p. xi). This means that Rothbard took
Mises's 'evenly rotating economy' as a benchmark, returning to a static
equilibrium concept. Consequently, he could not incorporate processes of
knowledge acquisition and expectations formation into his discussion of
economic principles, and hence could not address the Hayek Problem.
contrast, Lachmann (1951, p. 423) stated that "[u]nderconsumption crises are not imposs-
ible, but they are unlikely to be frequent."
28This latter work explained the Great Depression of the 1930s as a consequence of
the government's expansion of the supply of money. Thus, Rothbard extended the
applicability of the Austrian business cycle theory, compared with Lachmann's (1940)
argument that the theory was intended only to explain late nineteenth-century and early
twentieth-century business cycles. Hayek (1969b) would adopt the latter position.
96
Rothbard's (1963) position closely resembled that of Mises (1949).
The latter had argued that business cycles are caused by exogenous
disturbances. As was shown in Chapter 4, Mises was criticized by Hayek
for not having developed an endogenous theory of the business cycle. This
criticism was rejected by Rothbard (1963, p. 36, italics in original). He
argued that "[p]rocesses are either analyzed correctly or incorrectly; the
only test of any analysis is its truth, not whether it is exogenous or endog-
enous. If the process [of credit expansion] is really exogenous, then the
analysis should reveal this fact; the same holds true for endogenous
processes. No particular virtue attaches to a theory because it is one or
the other. "29 He continued by arguing that the "[r ]ecurrence [of the boom]
stems from the fact that banks will always try to inflate credit if they can,
and government will almost always back them up and spur them on" (p.
37}. Monetary expansion is thus the cause of the business cycle.
Like other Austrians, Rothbard argued that the detrimental
effects of the business cycle should be avoided, either by enacting a law
against credit expansion (i.e. against fractional-reserve banking), or by
establishing a system of free banking, combined with the return to a com-
modity standard. 30 Additionally, he argued that the best depression policy
was to adopt a policy of laissez faire, because only such a policy would
facilitate the market's recovery process (pp. 25 - 29). Of course, this
presupposes that agents are able to learn the 'objectively correct' state of
affairs (cf. Chapter 4). However, Rothbard could not provide this policy
conclusion with sound foundations, because he could not make clear how
such learning would take place and why the newly acquired knowledge
would also hold for the future. Like Mises, his adoption of the static ERE
framework rendered it impossible to address the Hayek Problem.
29
The question whether a process is endogenous or exogenous depends, of course, on
whether the economic theory under consideration explains this process, or whether it
'merely' asserts its existence. Rothbard's statement implies that he does not envisage a role
for business cycle theory to explain the process of credit expansion as the result of
individual optimizing (rational) actions.
30
If fractional-reserve banking is abolished, private banks are not able to increase
their lending potential by lending to each other. See also Friedman (1948, p. 247).
97
5.7. COORDINATION AND THE ROLE OF ENTREPRE-
NEURSHIP
Israel Kirzner, another student of Mises at New York University, pub-

lished a textbook, Market Theory and the Price System (1963), which
emphasized the process of competition (instead of the state of perfect
competition). 31 In this work he argued that in a market economy the
problem of coordination would find its solution in the market process. 32
"The key role is played by market prices" (p. 38, italics in original). More
in particular, "[m]arket prices guide individual decision makers toward
decisions that tend to consider implicitly all the relevant conditions
prevailing in the market." This means that Kirzner considered market
prices to be sufficiently good guides to bring about coordination. Of
course, he acknowledged that "[ t ]he absence of perfect knowledge implies
that some (probably most) of the resultant bids and offers, on a given
trading day, will be made in error" (p. 113). More precisely, "... the mista-
kes that can be made are of two possible kinds. First, bids and offers may
be mistaken because they unwittingly pass up superior opportunities (the
particular market participants are ignorant of) in favor of the inferior oppor-
tunities ... Second, bids and offers may be mistaken because they deli-
berately pass up desirable opportunities in the e"oneous belief that still more
attractive opportunities can be secured ... " (p. 114, italics in original). These
errors will lead to learning processes. Individuals will revise their plans for
future trading, using the information gained from their market experi-
ences (p. 123). However, Kirzner is in line with the Austrians' views on
the indeterminateness of these learning processes by not specifying them.
He merely claimed that the errors are corrected on the basis of past
experience, thus presuming that individuals learn inductively. The question
then is why the new courses of action should be ex-post optimal. Accor-
ding to Kirzner (1963, p. 244), "[t]he direction of adjustment ... will always
be toward the elimination of those disappointments generated at the prior
set of prices. Market agitation will proceed in this way initiating changes
31 By 1963 Kirmer had already published his The Economic Point of View (1960). This
book is best interpreted as a textbook on economic methodology. It does not aim to solve
(part of) the Hayek Problem (neither in its initial nor in its revised version.
32 Kirmer (1963, pp. 36 - 38) divided the coordination problem into three sub-
problems. These concern (1) the goods and services to be produced, (2) the combination
of resources to be used, and (3) the reward to each participant in the economic process.
98
in consumption and production in a continual tendency away from existing
inconsistencies among decisions." In other words, Kirzner argued that
there would be a tendency towards coordination, because individuals
would try to eliminate all known profit opportunities. In his view, entre-
preneurship plays an essential role in bringing about this tendency. "It is
the ceaseless search by entrepreneurs for such profit opportunities that
prevents the continuation of existing market activities - in other words, it
is the search for profits that renders such a market state one of disequili-
brium [or rather discoordination]." This may suggest that Kirzner (1963)
did not take into account that the changes in decisions may themselves
disturb the social environment in which other individuals decide. How-
ever, the reverse is true. He argued that "... adjustments in one area will
impinge on other areas and will eventually be reflected in the adjustments
subsequently made. These subsequent adjustments may of course affect, in
turn, still other areas as well as the area where the very first adjustment
was itself made." This means that a disturbance may create profit oppor-
tunities which did not exist before, so that it is in fact discoordinating.
Nevertheless, Kirzner maintained that there will be a tendency towards
coordination, neglecting the discoordinating effects of actions. As Hayek
(1937} already had indicated, the question whether an economy tends
towards coordination can only be answered empirically. Despite this
recognition, many Austrians (Mises, Hayek himself, Rothbard, Kirzner)
adhere to the view that such a tendency exists, even though they cannot
establish it on logical, and have not done so on empirical grounds.
5.8. SUMMARY AND CONCLUSIONS
In the late 1930s and 1940s Austrian thought lost considerable ground.
The rise of Keynesianism was one of the major causes of its eclipse.
Additionally, the debates on capital theory and on the Ricardo effect were
to have desastrous consequences for Hayekian business cycle theory.
Nevertheless, a few economists tried to revive this theory. In particular,
Lachmann emphasized the Hayek programme by pointing out the central
place knowledge and expectations should have in economics. Hayek had
left the problem of how to explain the tendency towards coordination
unresolved. He had admitted that he could not explain why an inductive
learning process should lead to correct knowledge and ex-post optimal
expectations. Obviously, there is no need why inductive learning should
99
lead to correct knowledge, because there is no reliable inductive logic.
The Hayek Problem can then only be resolved by substituting inductive
learning by a non-inductive method of learning. Lachmann differed from
Hayek in the sense that he rejected the presumption that a tendency
towards coordination exists, thus transforming the Hayek Problem into the
problem of situational dynamics. Obviously, this transformation allowed
him to presume that individuals learn inductively. After all, if there is no
reason why a tendency towards coordination should exist, then there is no
reason why the entrepreneurs' knowledge will become increasingly
correct. Hence there is no reason why an inductive learning procedure
should be rejected. However, this is possible only because the Hayek
Problem was transformed into the problem of explaining situational
dynamics. Lachmann proposed process analysis as the correct method for
explaining discoordination situations in terms of previous discoordination
situations. As a corollary, he argued that business cycle theory should also
adopt this method. Thereby he considered the Austrian business cycle
theory ill-suited to account for secondary depressions, such as that of the
late 1920s and early 1930s. In contrast, Rothbard (1963) tried to explain
the crisis of 1929 and the ensuing depression in terms of this theory. He
adopted the Misesian version with its static benchmark and the exogenous
disturbances. As was already shown in Chapter 4, this theoretical frame-
work cannot be used to solve the Hayek Problem, because it deals with
knowledge and expectations only implicitly.
Kirzner addressed the problem why a tendency towards
coordination should exist, following a different tack. Mises had remarked
that the role of the entrepreneur in the market process is crucial. This led
Kirzner to focus on this role. He pointed out that the discovery and
elimination of profit opportunities would lead to a tendency towards
coordination. However, he neglected the discoordinating features of the
entrepreneurial actions, which consist in changing the decision situations
of other individuals. Nevertheless, his views of the market process as a
discovery process would become very important in the revival of Austrian
economics.
100
6. THE AUSTRIAN REVIVAL
6.1. INTRODUCTION
In the 1970s the Austrian School of Economics experienced a revival. As

was shown in the previous chapter, its ideas had almost been forgotten for
approximately three decades. This means that they had to be rediscove-
red. This rediscovery of Austrian economics was set in motion by two
publications by John R. Hicks. 1 In 1967 he raised the question as to who
had been right, Keynes or Hayek. In this way he undoubtedly directed
attention towards the work of latter. Additionally he discussed Hayek's
business cycle theory, concluding that it could only be saved by transfor-
ming it into a theory of growth. In Chapter 4 it was shown that Hayek had
denied this conclusion, and that he subsequently gave three elucidations
of the Ricardo effect. This is another way in which Hayek's theoretical
views may have reached other economists. Despite Hayek's remarks,
though, Hicks persisted in his conclusion and carried out the transfor-
mation himself. In 1970 he published 'A Neo-Austrian growth theory',
which contained (as he called it) "a first sketch" of the theory as presented
in Capital and Time (1973 (1987), p. 6n). 2 This latter work carried the
sub-title 'A Neo-Austrian Theory', thus drawing attention to Hayekian
ideas. Hicks's interpretation of Austrian economics was not generally
appreciated. Not only Hayek rejected Hicks's attempts to revise Austrian
economics, but Israel Kirzner's Competition and Entrepreneurship (1973)
also dismissed his 'Neo-Austrianism'. Kirzner severely criticized all
analyses which are restricted to general (or even partial) equilibrium for
disregarding the market process and the role of the entrepreneur. This
emphasis on the role of the entrepreneur was particularly important in the
attempt to increase the plausibility of the assumption of a tendency to-
wards coordination. However, Austrian economists would never prove the
existence of such a tendency.
Austrian economic thought had been discarded for almost three
1
O'Driscoll (1977a, pp. 1 - 6) already pointed out Hicks's role in the revival of Aus-
trian economics.
2
Note that I will use the tenn 'Austrian' to denote the views of contemporary
followers of Hayek. This use was suggested by Lachmann (1977, p. 28) in order to
distinguish 'real Austrian thought' from Hicks's 1973 analysis.
101
decades, and many of its central ideas had fallen into oblivion. Its revival,
as Caldwell (1982, p. 117) noted, has therefore been largely" ... a doctrinal
exegesis of the masters" in order to restore their ideas. This is not to say
that the revival merely consists in repeating the ideas of Mises, Hayek,
and Lachmann. Israel Kirzner in particular has elaborated the Austrian
notion of competition as a process of discovery by entrepreneurs, as
opposed to the neoclassical situation of perfect competition. Additionally,
Lachmann's work on subjectivism has been important in stressing the role
and subjectivity of knowledge and expectations, and in pointing out the
flaws of 'the mechanical analogy'. O'Driscoll and Rizzo (1985) continued
his work by distinguishing between static and dynamic subjectivism. This
distinction is useful in interpreting human decision-making. In this chapter
it will be argued that the Austrian revival centers on these two aspects: it
is a combination of 'competition as a process' and 'dynamic subjectivism'.
In this sense the revival consists most of all of attempts to make plausible
why a tendency towards coordination exists.
This chapter is organized as follows. Section 6.2 discusses Kirz-
ner's work on competition as a process and the role of the entrepreneur
in this process. In Kirzner's view, the process of competition is a learning
process. In particular, two types of learning processes can be distinguish-
ed. These will be considered in section 6.3. This is followed by a dis-
cussion of the Kirznerian view on the entrepreneurial discovery process
(section 6.4}. This view has been criticized on several accounts. The most
important criticism argued that Kirzner disregarded discoordinating
tendencies. In reply Kirzner redefined the concepts 'entrepreneurship' and
'alertness'. These redefinitions will be discussed in section 6.5. It led to
the view that entrepreneurs are creative, in the sense that they try to
bring about correspondence between their image of the future and the
realized future. This view is based on the assumption that individuals have
different abilities, which is discussed in section 6.6. The 'uniqueness of the
individual' explains why individuals differ in their abilities to grasp profit
opportunities. The Austrian market process can only operate successfully
if it allows for such differences. In section 6. 7 we turn from micro- to
macroeconomics by considering the revival of Hayekian busines cycle
theory. This mainly consists of attempts to reformulate Hayek's theory in
modern terminology. Such reformulations are crucial if Austrian thought
is to become an alternative for Keynesian, Monetarist, or New Classical
theories. However, the reformulations have been stated in verbal and
diagrammatical 'language'. Given the contemporary emphasis on math-
102
ematical model-building, it is likely that Austrian economics will only be
considered an alternative to contemporary doctrines, if it can be math-
ematically formalized. Such mathematical formalization thus appears to
be a necessary precondition for the success of its revival. On the other
hand, the use of mathematics in the social sciences is strongly disputed
among Austrians. This issue will be addressed in section 6.8. Even though
Austrian economics has not been formulated in the form of a mathemat-
ical model, there has been an attempt to test it empirically. It was tried to
establish whether the Hayekian business cycle theory is consistent with the
empirical facts for the period 1959 - 1981 (section 6.9). Section 6.10 con-
tains some concluding remarks.
6.2. THE SYNTHESIS OF COMPETITION AND

ENTREPRENEURSHIP
In the late 1940s Hayek had emphasized the role of the price system as a
means to transmit information. Coordination prices transmit all relevant
information, whereas discoordination prices reflect incomplete informa-
tion. The question then rises why economic agents will learn their way
into coordination if they base their actions on unreliable signals, such as
discoordination prices. Hayek (1937) had already acknowledged that there
are no a priori (i.e. economic-theoretical) reasons why this should be so.
The Austrian revival, however, would try to solve Hayek's unresolved
issue in an a priori manner. The first attempt was undertaken by Kirzner
(1973), who combined the Hayekian construct of (a tendency towards)
intertemporal coordination with the Misesian notion of the entrepreneur. 3
According to Kirzner (1973), profit opportunities reflect the fact
that 'something' is being sold at different prices in two markets. They
point out that there is room for mutually beneficial trade, which is
foregone because of a lack of knowledge on the part of economic agents.
That is, profit opportunities reflect coordination failures. 4 In turn, this
means that every opportunity grasped removes some coordination failure
3
Mises (1949 (1966), pp. 252- 53) interpreted entrepreneurship as a function which" ...
is not the particular feature of a special group or class of men; it is inherent in every
action and burdens every actor."
4
As K.irmer (1973, p. 222, italics in original) stated, " ... entrepreneurial profit oppor-
tunities exist wherever there is scope for more complete coordination of individual plans."
103
(ceteris paribus, of course). In this sense entrepreneurial activities are
coordinating. The question whether a tendency towards coordination
exists thus amounts to the question why entrepreneurs will be successful
in discovering and grasping profit opportunities. Kirzner thereby pointed
out that dis'coordination prices (that is, price discrepancies) stimulate
entrepreneurial alertness. He argued that entrepreneurs search for new
knowledge (profit opportunities), motivated by the desire for profit. Their
alertness leads them to discover profit opportunities, which they are
presumed to discover by 'peering through the fog'. Kirzner (1973, p. 74)
thus suggested that entrepreneurs merely respond to already existing
opportunities, that is, they do not create them. In this sense, Kirznerian
entrepreneurship is alert, active, but non-creative. 5 The already existing
opportunities can be regarded as arbitrage opportunities, whereby
arbitrage also includes productive activities (p. 85).6
The fact that entrepreneurs discover profit opportunities does not
mean that they do not make mistakes. The entrepreneurial perception of
these opportunities may not be correct in the sense that the actions may
not be ex-post optimal. The market then penalizes such mistaken entre-
preneurial activity. Entrepreneurs who are penalized suffer losses. The
competitive (or rivalrous) process is to ensure the elimination of mistaken
entrepreneurial activity. Only successful entrepreneurs will stay in busi-
ness, since they are rewarded for their correct perceptions by making a
profit. Kirzner (1973) concluded that this process of elimination elimina-
tes 'bad' (mistaken) entrepreneurial activities and hence reduces the price
differentials between markets, setting in motion a movement towards the
coordination of plans. He claimed that if we "... broaden our theoretical
vision of the individual decision-maker from a 'mechanical' Robbinsian
economizer to Mises's homo agens, with the universally human entrepre-
neurial elements of alertness in his makeup, we can cope with the task of
explaining the changes which market forces systematically generate" (p.
72, italics in original).
5 Loasby (1982, p. 119) criticized Kirmer's view, because the latter did not allow for
any role of the imagination of the entrepreneur in creati11g the future, and hence profit
opportunities. Shackle (1973) did allow for such a role, leading Loasby to conclude that
"Kirmer's entrepreneurs are alert, Shackle's are creative."
6 Cf. also Mises (1949 (1966), p. 329), who already argued that entrepreneurs "...
discover discrepancies between the height of the prices of the complementary factors of
production and the anticipated future prices of the products, and they are intent upon
taking advantage of such discrepancies."
104
The competitive process thus brings about individual learning
processes. Since learning involves changes in the respective ends-means
frameworks of individual decision-makers, the concept of the 'Robbinsian
maximizer' is inappropriate for the analysis of the process of competition,
because it assumes that the ends-means frameworks are given.7 Any analy-
sis which is limited to Robbinsian maximizing decision-makers fails to
grasp the meaning of the competitive process. Additionally, such decision-
makers are also assumed to be price-takers, which means that individually
they cannot cause changes in prices. 8 In Kirzner's view, this leaves unex-
plained how the market process proceeds, i.e., how prices change. Neo-
classical analysis circumvents this problem by adopting the metaphor of
the Walrasian auctioneer. Kirzner rejected this metaphor, and substituted
the price-taking Robbinsian maximizer by the price-setting entrepreneur
(p. 116). This raises the question as to how entrepreneurs decide to
change their prices.
As was shown in Chapter 3, O'Driscoll and Rizzo (1985) distin-
guished between static and dynamic subjectivism. Static subjectivism holds
that the process of decision-making is a determinate one, in the sense that
choice can be causally explained. In economics the Robbinsian maximizer
is the prototype of the static-subjectivistic decision-maker. Its choice con-
sists of four factors, namely "... ( 1) the ordinal ranking of goals or wants,
(2) knowledge of the relationship between courses of action (or commod-
7
Kirmer (1973, p. 35) criticized the passivity of the Robbinsian economizer, which
neglects the entrepreneurial element in human action. Instead, he argued, economic
subjects do more than passively select the best alternative in a given ends-means frame-
work. They are active, in the sense that they learn from their past actions and change their
ends-means framework accordingly. Kirmer therefore described entrepreneurship as active,
and human, whilst Robbinsian maximizing behaviour is passive, and mechanical. Cf. also
Shackle (1966, p. 130). Kirmer was not the only economist who discussed the role of
entrepreneurship in the market process. Mises (1949 (1966)) and Schumpeter (1934) had
also analyzed it. Kirmerian entrepreneurship is similar to that of Mises, albeit that Mises
could not depict the entrepreneurial actions as constituting a learning process. Kirmerian
entrepreneurship differs from that of Schumpeter, because both economists adopted
different equilibrium constructs. Schumpeter adhered to a stationary construct, whilst
Kirmer adopted the Hayekian intertemporal general equilibrium construct. As a result,
Schumpeter's entrepreneurship is innovating, in the sense that it disturbs the stationary
equilibrium. From an intertemporal point of view, however, this innovation can be
interpreted as the discovery of a profit opportunity between on the one hand present
markets of factors of production, and on the other hand future markets for produced
goods. For a discussion of the differences, see Kir1.ner (1971 (1979)) and (1973).
8
Kirmer stressed this property of the Robbinsian economizer in 'Equilibrium versus
market process' (1976a, p. 119), and 'Hayek, knowledge, and market process' (1975 (1979),
p. 19).
lOS
ities) and want satisfaction, (3) knowledge of prices, and (4) knowledge of
the income constraint" (O'Driscoll and Rizzo (1985, p. 28)). These factors
are the constituent parts of choice, in terms of which choice can be
decomposed. Such decompositions do not constitute causal explanations
because causes must be temporarily prior to their effects, whereas consti-
tuent parts are not prior to the whole. This means that static subjectivism
does not fulfill its task of explaining choice, because it does not specify its
causal determinants. As a consequence, the Robbinsian maximizer will
not do for the explanation of entrepreneurial choice. Instead, O'Driscoll
and Rizzo (1985, p. 28) argued that "[t]he point at which the ranking and
perceptions are finalized is, or constitutes, the point of decision." That is,
the decision of an individual decision-maker consists of the choice of (and
not within) the relevant ends-means framework. Static subjectivism consi-
ders this framework as given. Hence it cannot explain its choice. By
contrast, dynamic subjectivism "... views the mind as an active, creative
entity in which decision-making bears no detenninate relationship to what
went before" (p. 22), so that choice (i.e. the ranking of ends) cannot be
explained deterministically. 9 It is interpreted as a result of man's 'free
will'. In this manner dynamic subjectivism solves (or rather evades) the
problems posed by static subjectivism. As Kirzner (1975 (1979)) recog-
nized, its introduction shows that the tendency towards coordination is
indeterminate, since it depends on (indeterminate) entrepreneurial
choice. 10 The tendency therefore cannot be established on a priori
grounds. At most, it can be made plausible. Kirzner would undertake such
an attempt to increase this plausibility in his 'Knowing about knowledge: a
subjectivist view of the role of information' (1976b (1979)) and 'Alertness,
luck, and entrepreneurial profit' (1978b (1979)). Both papers address the
problem as regards entrepreneurial knowledge acquisition processes.
9 Latsis (1976) distinguished between single-exit and multiple-exit decision situations.

In the former the outcome of the decision-making process is completely determinate. In
the latter type of decision situation it is indeterminate. Static subjectivism thus considers
any decision situation to be a single-exit situation, whilst dynamic subjectivism holds that
choice takes place in multiple-exit situations.
10 In his paper 'Hayek, knowledge, and market processes' he argued that " ... insight
into the entrepreneurial element in human action does not by itself assure us that people
necessarily learn the correct facts of their situations from their market experiences. While
the recognition of universal human alertness provides grounds for presuming learning, it
does not, it may seem, guarantee discovery of the truth" (p. 29). Nevertheless, he opined
that " ...people possess a propensity to discover what is useful to them" (p. 29).
106
6.3. TWO TYPES OF KNOWLEDGE ACQUISITION PRO-
CESS
Kirzner {1973) argued that entrepreneurs search for profit opportunities,

which seemed to suggest that the entrepreneurial knowledge acquisition
process is a planned activity. But he in fact distinguished between two
types of knowledge acquisition processes. The first consists of the delibe-
rate and rational search for knowledge. It considers the acquisition of
knowledge as the result of a deliberately planned search activity, which is
conducted in precisely the same way in which all other economic activities
are carried out, namely by equating marginal costs and benefits.
Consequently, 'ignorance' is merely a situation of optimal imperfect know-
ledge which arises from non-zero costs of information acquisition. A
change in this optimal state will only take place if there has been a
change in the marginal benefits or costs of searching knowledge. How-
ever, as Shackle (1955, pp. 17 - 18) and Boulding (1968, p. 146) already
pointed out, the deliberate search for knowledge is caught in a paradox.
Knowledge will only be bought if it is not already possessed, that is, if its
content is not known. The buying of knowledge thus means the buying of
an unknown good. This raises the problem of how to determine the
marginal costs and benefits of such unknown knowledge. In other words, a
rational and deliberate search activity for specific knowledge can only
take place if at least some information about the marginal costs and
benefits of that knowledge is already present. Obviously, this information
cannot have been gathered by a rational and deliberate search process,
because this would again presuppose some information about marginal
costs and benefits. Hence, planned search activity presupposes knowledge
which is not acquired by such an activity. This Shackle-Boulding paradox,
as Kirzner {1976a, p. 142) called it, holds that any planned learning
process presupposes undeliberately acquired knowledge. Any planned
learning process presupposes a spontaneous learning process. In Kirzner's
view, "... a very great volume of one's awareness of one's environment,
and of one's expectations concerning the future, is the result of learning
experiences that occu"ed entirely without having been planned" (p. 142,
italics in original). Such spontaneous earlier learning underlies any delibe-
rate search activity, because the latter cannot take place without the
former. This makes clear that 'mainstream economics' fails to appreciate
the role of entrepreneurship, because it can only analyze optimizing
behaviour and hence deliberate learning, thus ignoring spontaneous
107
learning (which may, of course, be followed by a deliberate-learning pro-
cess). In what follows we shall concentrate on the latter.
According to Kirzner, the entrepreneurial discovery process may
be subdivided into two distinct events, namely confrontation and percep-
tion. Entrepreneurs must first be confronted with already existing profit
opportunities, which they may subsequently perceive. The question
concerning the tendency towards coordination may now be rephrased as
'Why should either confrontation or perception take place?' In this regard
two possibilities emerge. Firstly, both confrontation and perception may
be a matter of sheer luck. However, this would render the notion of
'entrepreneurship' and 'alertness' meaningless. Secondly, they may both be
a matter of entrepreneurial ability. In this case one might argue that this
ability is a factor of production, so that profit can be regarded as a reward
for this ability. The usual marginality calculus then applies, so that the
economy is already in general equilibrium. The rejection of the latter then
appears to be mistaken. Kirzner addressed this problem in his 'Alertness,
luck, and entrepreneurial profit' (1978b {1979)).
Kirzner (1978b, p. 158) recognized that neither sheer luck nor
entrepreneurial ability alone suffice to explain the spontaneous discovery
of profit opportunities, although they both play a role. An entrepreneur
must be lucky to be confronted with new profit opportunities, but such a
confrontation does not mean that he will perceive them. In Kirzner's view,
perception requires alertness. Entrepreneurs must be alert in order to
perceive a specific discoordination of plans. Both luck and alertness are
necessary preconditions for discovery, and hence for the tendency towards
coordination. Furthermore, he maintained that entrepreneurial alertness
should not be regarded as merely another factor of production, because
"... the distinctive aspect of entrepreneurial activity [is] its inability to be
compressed within the [general] equilibrium conception of the market" (p.
155). The reason for this inability is the fact that "[m]arket entrepre-
neurship reveals to the market what the market did not realize was
available or, indeed, needed at all" (p. 181). Entrepreneurship discovers
opportunities which were not known and, more importantly, about which
nothing was known. This latter property of entrepreneurial discoveries
means that they cannot be interpreted as the outcome of any optimization
process, because their marginal costs and benefits are unknown. Economic
subjects cannot determine the marginal productivity of 'entrepreneurship'
and hence cannot equate its marginal benefits with its marginal costs.
To summarize, both luck and alertness play a crucial role in the
108
discovery of profit opportunities. Because one cannot assess the marginal
costs and benefits of alertness, it cannot be regarded as a factor of
production. Hence Kirzner argued that the general equilibrium framework
cannot capture the entrepreneurial element in human action. Moreover,
the notion of the Robbinsian maximizer does not allow for the incorpora-
tion of 'learning', because it presumes that the ends-means framework is
given. Therefore, Kirzner (1978b (1979), p. 158, italics in original) conclu-
ded that "[t]o see the individual as entrepreneur one must ... see the
decision as encompassing also the very identification of the ends-means
configuration itself, within which action is being conducted."
6.4. CREATIVE ENTREPRENEURSHIP
Kirzner's view on entrepreneurship as alertness has been criticized by

several economists, including some fellow-Austrians. In particular, Murray
Rothbard criticized Kirzner for disregarding the crucial importance of
fundamental (i.e. Knightean) uncertainty. 11 He argued that Kirzner's
entrepreneurship was reduced to mere alertness, whereas Mises's
entrepreneurship was defined in terms of fundamental uncertainty. "Mises
conceives of the entrepreneur as the ucertainty bearer.... To Kirzner, on
the other hand, entrepreneurship becomes reduced to the quality of alert-
ness; and uncertainty seems to have little to do with the matter." 12 White
(1976) also criticized Kirzner for not recognizing the importance of
uncertainty and the creative part played by the entrepreneur. In an
attempt to address these criticisms, Kirzner ( 1982) altered his description
of entrepreneurship and alertness. He allowed the entrepreneur to be
11
Knight (1921 (1948)) distinguished between risk and uncertainty. He argued that
"[t]he practical difference between the two categories, risk and uncertainty, is that in the
former the distribution of the outcomes in a group of instances is known (either through
calculation a priori or from statistics of past experience), while in the case of uncertainty
this is not true, the reason being in general that it is impossible to form a group of
instances, because the situation dealt with is in a high degree unique" (p. 233, italics in
original). This means that an uncertain future is unknowable and unpredictable, although
not unimaginable. By contrast, a risky future situation might be predicted. It can therefore
be known to some degree. (Note, as Knight himself already did, that it is possible for
uncertainty to become risk. This will be the case when over time sufficient instances of the
phenomenon under consideration have occurred to enable an individual to draw up a
probability distribution.) Cf. Mises's methodological singularism (Chapter 3).
12
Rothbard in his unpublished paper 'Professor Hebert on entrepreneurship', cited by
Kirmer (1982, pp. 141 - 42).
109
creative in the sense that he envisaged an important role for the entrepre-
neur's formulation of an image of the future. Fundamental uncertainty
forces the entrepreneur to formulate this image, on the basis of which he
then acts. His actions will be more successful if the image of the future is
more accurate. This motivates him to bring about a correspondence
between his image and the future (1982, p. 149). Alertness then was
redefined as "... this motivated propensity to formulate an image of the
future man's alertness" (p. 149, italics in original). This redefinition allows
entrepreneurship to play a creative role. However, there is a problem in
this regard. High (1982, p. 164) pointed out that the redefinition of
alertness leaves unexplained why entrepreneurial losses exist. He argued
that Kirzner failed to appreciate the importance of 'judgment', which he
described as "... the mental process of assigning relevance to those things
we already know. In an uncertain world, judgment influences the images
we form of the world, and it chooses among courses of action whose out-
comes are not known" (p. 167). This leaves us the following picture.
Alertness is needed in order to perceive possibilities for arbitrage (in a
broad sense, including trade and production). But not all of these oppor-
tunities are profitable. Hence entrepreneurship needs judgment in order
to assess whether a given arbitrage opportunity is also a profit opportun-
ity. Profits are the reward for good judgment, whilst losses are the punish-
ment for bad judgment. This means that the tendency towards coordina-
tion is a tendency for entrepreneurial judgments to become increasingly
correct. This view is opposed by Lachmann and his followers, who reject it
as misplaced determinism. Instead, they describe the market as an evol-
utionary process which consists of both coordinating and discoordinating
tendencies.
6.5 THE MARKET AS AN EVOLUTIONARY PROCESS
In their explanations of discoordinating forces Austrians usually refer to

changes in the data. The situation of coordination is considered to be a
moving target because the data (values, technology, and resources)
continue to change. 13 But there may also be endogenous reasons why a
situation of coordination will never be reached. These have been stressed
13 Cf. Hayek (1937, p. 45), Mises (1949 (1966), p. 250), Rothbard (1962 (1970), p. 262)
and Kirmer (1973, p. 81).
110
by the so-called 'evolutionists'. They hold that entrepreneurs learn new
ends and new means to achieve these ends. Such learning thus implies
that the ends-means frameworks of the entrepreneurs change. In tum, this
means that the relevant general equilibrium situation changes: 'the target
moves' for endogenous reasons. Moreover, as Loasby (1982, p. 116) obser-
ved, "[t]he market process redistributes income, and a redistribution of
income entails a redefinition of equilibrium." This provides another
endogenous reason why the target moves. Hence, entrepreneurship not
only has coordinating but also discoordinating effects. High (1986, p. 118)
concluded that "... characterizing the market as a strictly equilibrating [i.e.
coordinating] process ... leaves unexplained those changes in tastes,
technology, and resources that upset previously compatible plans. These
changes are the result of entrepreneurship and are as endogenous to the
market as any other entrepreneurial change." Entrepreneurship turns out
to be discoordinating as well as coordinating. This does not mean that the
market process consists of mere chaos, about which we cannot say any-
thing. High (1986, p. 116) interpreted the process as an evolutionary pro-
cess. Like all other evolutionary processes, it is open-ended, in the sense
that its outcome cannot be determined beforehand and that it cannot be
explained deterministically. In this view the process of competition will
induce economic subjects to try and improve their circumstances, so that
it is an evolutionary order which stimulates progress. This order is an
unintended consequence of purposive action of individual economic
agents. It can only thrive because of the diversity of wants, knowledge,
and expectations. 14 The resulting diversity in experiments means that error
is an inherent and unavoidable feature of the market process. As Fehl
(1986, p. 83) observed, "... it is just the very diversity of individual imagin-
ations which - together with general rules in the Hayekian sense and other
institutional arrangements- constitutes economic order."
The interpretation of the market as an evolutionary process
undermines its usual welfare-theoretic justification as (increasingly)
bringing about coordination and efficiency. However, Loasby (1982, p.
129) had already provided a different justification by arguing that "[t]hose
who believe in the market system because of its superior ability to gener-
ate and use knowledge cannot legitimately assume the perfection of
knowledge (implying some variety of economic determinism), which
14
According to Loasby (1982, p. 120), "[c]ompetition is important, not only in provi-
ding a stimulus to seek such opportunities but also in encouraging creative ideas and the
diversity of experimentation, which is necessary to progress. •
111
advocates of other systems use to bolster their case. An advocate of
human freedom - which is necessary a freedom to choose wrongly, even
foolishly - can never be certain that he is right on the grounds of effi-
ciency, which provide the conventional terms of debate. It is one of the
strengths of his position that he can freely admit that he may be wrong."
In this way the freedom to choose (wrongly), not its welfare-theoretically
optimal outcome, becomes the justification of the market process.
6.6. THE DIFFERENT ABILITIES OF INDIVIDUALS
It is, of course, merely a matter of definition (classification) whether the

evolutionary view is considered to belong to Austrian economics. Accord-
ing to Garrison (1985, p. 313n10 and 1987, p. 340n12), Austrians reject
this view, and maintain that a tendency towards coordination exists. They
concede that it is impossible to prove that the 'invisible hand' is at work
(in its coordinating role), so that the presupposition of the existence of
the tendency rests on 'belief. 15 In his view, Kirzner (1979) had provided
the justification of the Austrian belief by showing that its validity "...
depends critically on the nature of the institutional arrangements. So long
as the arrangements are such that expectations consistent with underlying
economic realities are rewarded and expectations inconsistent with those
realities are penalized, the tendency can be expected to prevail" (1986a, p.
97). He thereby claimed that a tendency towards coordination will exist if
'incapable' investors will make losses, and hence will be forced out of the
market because of a lack of investable funds. However, this conclusion
only holds either if investors differ in their ability to invest successfully, or
if past success guarantees future success. This latter reason clearly would
imply an inductive fallacy. The fact that the past investment decisions of a
particular economic agent have been ex-post optimal does not mean that
this will also be the case in the future. Past success does not guarantee
future success. This means that investment decisions need not become
increasingly ex-post optimal. In a comment on Loasby (1982), Garrison
(1982) therefore defended the tendency towards coordination by arguing
that individuals differ in their abilities to invest profitably. He stated that
"... the idea that different people know different things is inherent in the
15 Garrison (1985, p. 313nl0) argued that both Austrians and Keynesians base their
theoretical frameworks on beliefs; the former believe that the 'invisible hand' is at work,
whereas the latter believe that it is not.
112
very concept of an individual. Differential knowledge remains a puzzle
only to those economists whose understanding has been numbed by the
continual contrary-to-fact assumptions of identical individuals and ident-
ical firms as represented by the corresponding homothetic indifference
curves and production functions" (p. 137). He concluded that individuals
differ from each other as regards their knowledge, and hence they differ
in their abilities to grasp profit opportunities and to invest successfully.
The fact that individuals differ in their knowledge and abilities is implied
by the concept of the 'individual'. In other words, Austrians reject the so-
called Principle of Indifference, which applies when there is no reason to
presume that the elements of the class under consideration differ from
each other in any relevant aspect. Stated differently, the homogeneity
postulate does not apply to economic agents.
The rejection of the Principle of Indifference allows some entre-
preneurs to be more successful than others. It thus plays an important
role in Austrian microeconomics. However, it also appears to have an
important consequence for the Austrian macroeconomic method of expla-
nation. Mter all, if economic agents cannot be regarded as members of
the same homogeneous class, then it is not allowed to describe their
behaviour in terms of frequency distributions, because such distributions
presuppose that the elements of the population under consideration do
not differ from each other in any relevant aspect. Hence, Austrians
appear to have concluded, probabilistic explanations should not be used in
economics (and other social sciences). This may explain why their theories
are always presented verbally (sometimes supplemented by graphs). Math-
ematics and probabilistic explanation are considered to be of limited
value (to say the least). However, rejection of 'frequentialist probability
calculus' does not mean that one should adopt only deterministic (in the
sense of non-probabilistic) explanations. As Hacking (1989, pp. 164 - 65)
observed, probability need not be interpreted in a frequentialist (or
'objective') manner. Instead, it can 'merely' reflect 'degrees of belief.
Such a subjective interpretation means that "... a statement of probability
is the speaker's own assessment of the extent to which he or she is confi-
dent of a proposition" (p. 165). Such an interpretation may very well be
consistent with Austrian subjectivism. Moreover, Hacking (1989, p. 165)
observed that the resultant 'subjectivist probability calculus' is constrained
"... by exactly the same mathematical rules as govern the frequency
conception of probability." It may therefore be possible for Austrians to
use probability calculus.
113
6.7. THE REVIVAL OF HAYEKIAN BUSINESS CYCLE
THEORY
Lachmann (1956 (1978), p. 113) and Hayek (1969b, p. 174) argued that
Austrian business cycle theory aimed to explain the industrial fluctuations
between 1870 and 1914. Latter-day Austrian economists maintained that
the theory can also explain the industrial fluctuations as they occurred
during the interbellum and after World War II. In their view Hayekian
business cycle theory provides a better insight into monetary dynamics
than its Keynesian, Monetarist, or New Classical counterparts. The main
reason given for this is the fact that it builds on Austrian capital theory.
Its structure of production enabled Hayek to interpret industrial fluctu-
ations in terms of rna/investments. This interpretation was not generally
appreciated. It appears that non-Austrians either rejected or did not
understand Austrian capital theory. On the other hand, Austrians failed to
explain their views. This failure was in part due to the fact that at least
some assumptions were not explicated; the theory was 'too vague'. During
the 'years in the wilderness' (and already in the late 1930s and 1940s)
there was a communication problem between Austrians and non-Aus-
trians. This problem has played a major role in the Austrian revival,
because it evoked many attempts to solve it. This is not surprising, given
the fact that the profession had almost forgotten the ideas of many Aus-
trian economists. Furthermore, many Austrian ideas had been incorpora-
ted in other research traditions, and in particular in neoclassical 'main-
stream' economics. 16 As a result, these ideas were interpreted in the
latter's comparative-static framework, thus neglecting the coordination
problem between stages of production altogether. 17 In their attempts to
revive their approach Austrian economists tried to solve the communica-
tion problem. This has not only led to an extensive exegesis of the works
of the Austrian masters, but also to different attempts to start a discourse
16 As Leijonhufvud (1981, p. 133) observed, many strands of economics share identical

roots. He showed that the Wicksellian distinction between the natural and the market rate
of interest was not only adopted and elaborated by the Austrians but also by the early
Keynes and by his Cambridge and neoclassical followers. In tum, Wicksell adopted Bohm-
Bawerk's capital theory, albeit in a modified (more aggregated) form. Additionally,
Menger is usually credited (jointly with Jevons and Walras) for starting the 'marginalist
revolution'.
17 See the debate on the Ricardo effect (Chapter 4) and Hicks's 'Neo-Austrian'
reinterpretation of Austrian economics (this chapter).
114
between Austrians and non-Austrians in order to identify and eliminate
misunderstandings. These attempts were undertaken by comparing Aus-
trian economics with other research traditions. Such comparisons are
possible only if the rival traditions are 'translated' into a common langua-
ge. As Garrison (1982, p. 137) indicated, the Austrian revival not only
attempts to refine and elaborate the views of the masters, it also tries to
'translate' its views into modern and more appropriate terms in order to
facilitate comparison with other traditions. In turn, the comparisons also
intended to show the merits of Austrian economics. It was often argued
that this tradition was able to explain phenomena which could not be
analyzed by its rivals. In this sense the revival comprises attempts "on the
strategic front." 18
O'Driscoll is presumably the first to have contributed to the
revival of the Hayekian business cycle theory. His Economics as a
Coordination Problem (1977a) restated the Hayekian business cycle theory
as a discoordination phenomenon. Hayek's work is considered to provide
"... a basis for a radical alternative to the 'neoclassical' paradigm of
efficient allocation with timeless production, perfect anticipations, costless
exchanges, (almost) instantaneous attainment of equilibrium, and a world
of no institutions" (p. xx). But before this alternative could be developed
any further, Hayek's theory had to be restated into modern terms.
Garrison has been concerned mainly with this 'translation' of
Misesian and Hayekian macroeconomics} 9 In his 1978 exposition of
'Austrian macroeconomics', he used the concept aggregate production time
(APT). This concept has a value and a time dimension. That is, it consists
of the multiplication of value (i.e. capital invested) and time (i.e. the
length of time which the particular investment goods are invested)?0
18
Garrison's (1982) title 'Austrian economics as the middle ground' can be regarded
as an attempt to rid 'Austrianism' of its image as 'radical economics' (which carries the
connotation 'contrary to common sense').
19
In private correspondence Garrison (letter of August 30, 1991) explained that in
general he tried to revive and develop capital-based macroeconomics, thereby propounding
the view that Austrian macroeconomics is more reasonable than its Keynesian, Monetarist,
or New Classical counterparts.
20
Garrison (1978, p. 6) recognized that his concept is merely an indication of the
extent to which capital is 'tied-up' in the production process. It cannot be measured di-
rectly. It should also be noted that the APT differs from Bohm-Bawerk's 'averageproduc-
tion time' or 'average period of production'. In fact, Garrison's (1978, pp. 6- 7) opined that
"[m]any of the problems of Bohm-Bawerk's capital theory had their roots in his use of the
average period of production: Because the denominator of his average was the value
115
Using the APT, Garrison subsequently restated the Austrian views on
capital and intertemporal coordination along IS/LM lines. 21 Additionally,
he provided a comparison between the Austrian and the Keynesian view.
An important difference between these views concerned the role of the
rate of interest. Following Wicksell, Austrian economists emphasized that
the interest mechanism serves to coordinate (ex-ante) saving and (ex-ante)
investment decisions. In t~e framework of the structure of production, this
means that any coordination failures result in rna/investments. Industrial
fluctuations are interpreted as resulting from such failures. By contrast,
the Keynesian 'liquidity-preference' view holds that the role of the rate of
interest is to equate money demand and money supply. The saving-invest-
ment equality is considered to be an (ex-post) identity, hence it is impossi-
ble (nor necessary) for the interest rate to coordinate saving and invest-
ment decisions. 22 The Keynesian analysis thus fails to capture the prob-
lems on the capital market which arise if the market rate of interest is
distorted (i.e. differs from the natural rate).
Garrison {1981) compared the respective monetary dynamics of
Mises ('capital-allocation effect') and Patinkin ('real-cash balances effect'),
starting from Wicksell. This comparison tried to establish in what ways
Austrian economics differed from its neoclassical counterpart. Garrison
argued that one of the main differences between Mises and Patinkin
concerns the level of aggregation. Mises and the other Austrians adopted
Wicksell's analysis in a rather disaggregated framework. This is clear from
their interpretation of 'capital' as a set of heterogeneous goods, and of the
'structure of production' as a complex and multidimensional set of inter-
relationships between producer goods. By contrast, Patinkin's neoclassical
approach incorporated the concept of 'homogeneous capital'. This means
that problems of malinvestment cannot be addressed. A presumably
related issue concerns the respective informational assumptions made by
dimension of the structure of production (value reckoned in labor units), and because
changes in the numerator of his average are typically accompanied by changes in the
denominator in the same direction, the direction of change in the average period of
production is generally ambiguous." By considering the aggregateproduction time, no such
problem of ambiguity is implied.
21
Caldwell (1982, p. 136n44) remarked that "[i]n a conversation, Garrison noted that
his attempt to translate Mises's [and Hayek's] business cycle theory into the IS-LM
framework was not universally appreciated. Fellow Austrians were displeased that he was
working with aggregates; and his professors at the University of Virginia disparaged his
model as untestable."
22
Cf. Leijonhufvud (1981, p. 135).
116
Mises's followers and the neoclassical adherents to Patinkin's views. In
Garrison's view, the latter have often assumed that individuals have
perfect knowledge and perfect foresight. Conversely, Austrians stress the
scarcity of information which is inherent in complex and decentralized
system such as in market economies. Economic agents obtain (at least
part of) their information from the price system. They are bound to be
less than perfectly informed, because discoordination prices do not reflect
all available information. In this sense the agents are only as informed as
the price system will allow them to be. According to Garrison, the differ-
ences in aggregation level and in informational assumptions are responsi-
ble for the differences between Austrians and neoclassicals as regards
'forced savings', 'distribution effects', and 'neutrality of money'.
A third attempt to illuminate Austrian macroeconomics and to
translate it into more modern terms was undertaken by Bellante and
Garrison (1988). They chose to reformulate the Austrian views in terms of
the 'Natural-Rate' (and Phillips-curve) approach of Monetarism. More in
particular, they focused their attention on Hayek's and Friedman's
respective views on monetary dynamics. According to Bellante and
Garrison (1988, p. 232), "... the two stories are themselves not substitutes,
but complements." Mter all, Hayek (and the other Austrians) analyzed
such dynamics in the face of the intertemporal coordination of decisions
concerning the structure of production (and hence capital goods), whereas
Friedman (and the Monetarists) addressed intertemporal substitution on
the labour market, presuming that (for some reason) the capital goods
sector is not influenced by monetary phenomena (via the market rate of
interest). A second difference between these two 'stories' concerns the
neutrality of money. Monetarists assume away the income-distributional
effects of a monetary expansion, whereas Austrians instead emphasize the
importance of these effects. Stated differently, Monetarists assume that (at
least in the long run) money is neutral: 'helicopter money' is distributed
immediately and proportionately among the economic agents. Money is a
'tight joint', because in the long run it is presumed to lead to changes in
monetary phenomena only. Conversely, Austrians stress the importance of
the way in which money enters into and spreads through the economy.
Such 'Cantillon-effects' disturb relative prices and hence create distortions
in the structure of production. Money may (or rather, will) have real
effects, hence it is a 'loose joint'.
The attempts to translate Austrian economics have been accom-
panied by attempts to devise an alternative, dynamic-subjectivistic
117
research programme. Rizzo (1982) and Langlois (1982) formulated
Austrian economics in terms of the Lakatosian research programme.
O'Driscoll and Rizzo (1985) undertook a more elaborate attempt. Its
important elements are the creativity in human decision-making and the
indeterminateness of the future. Furthermore, they adopted Hayek's
compositive method which interprets social phenomena as unintended
outcomes of individual rational action (pp. 19 - 20). They identify three
effects which constitute the Hayekian boom (pp. 205 - 6). 23
The above suggests that the translation of the Austrian theory has
been completed, at least in verbal and diagrammatical terms. It thus
appears that nothing stands in the way of a correct understanding of this
theory. However, given the contemporary emphasis on model building, an
important translation problem remains. It concerns the formalization of
the Austrian business cycle theory into mathematical 'language', which
many Austrians do not regard as useful.
6.8. AUSTRIAN ECONOMICS AND MATHEMATICAL

FORMALIZATION
The question whether economics should use mathematics appears to be

rather archaic. The main economic research traditions all adopt the
heuristic prescription of mathematical formalization, which holds that they
23
These effects are a discount effect, a cost effect, and a derived-demand effect.
Consider the following standard present-value formula:
PV = S 1 I ( 1 + r ) + S 2 I ( 1 + r )2 + ... + S n I ( 1 + r )n,
where S i is the expected stream of quasi-rents in period i, and r is the market rate of inte-
rest. The discount effect arises because entrepreneurs use r as a discount rate in the
computation of PV. The cost effect arises from the fact that r is also the price to be paid
for loans. In both cases, a fall in r increases the profitability of investment projects. The
derived-demand effect, by contrast, makes investment in some capital goods more profita-
ble, while at the same time reducing the profitability of investment in other capital goods.
If entrepreneurs interpret a fall in r as reflecting an increase in the propensity to save (i.e.
a fall in the natural rate of interest), then they will increase the investments in those capi-
tal goods which yield consumption output in the more distant future (goods of type 1).
Given full employment at the end of the boom, this means that they bid for those factors
which are currently employed in the production of consumption goods and capital goods
which yield consumption goods in the present (goods of type 2). However, if the fall in r is
caused by an expansion of credit, then the structure of production is distorted. Hence the
derived-demand effect may cause rna/investments, given expectational errors.
118
should formulate their views in the form of mathematical models.24 Most
Austrian economists are unwilling to follow this practice. Their position is
arguably best described by Mises (1949 (1966)) and (1953 (1977)). He
stated that the future is fundamentally uncertain and cannot be known. In
particular, the choices of entrepreneurs are confronted with fundamental
(Knightean) uncertainty, i.e., a lack of knowledge about the future. Mises
(1953 (1977), p. 98) argued that this means that there is no reason to
presume that in the future individuals will act like they have done in the
past. In other words, there are no 'behavioural constants'. In Mises's view,
mathematical economists hypothesize that such constants exist. But
"[w]hether this hypothesis is true ... can only be ascertained afterwards, i.e.
by historical experience" (p. 98). Sharply distinguishing between pure
economic theory and economic history, Mises objected that in this sense
the 'mathematical method' "... proves to be a method dealing with the
data of economic history" (p. 99). However, this critique appears to be too
harsh, because mathematical equations may merely reflect what has
already been stated in verbal language, albeit in a more formal and more
precise manner?5 Mises acknowledged this, but he opined that "... these
equations differ entirely in their practical applicability as well as in their
cognitive reference from the equations of mechanics" (p. 99). It thus turns
out that he did not reject the use of mathematics per se, but rather the
mechanistic interpretation of mathematical equations. In his view, such an
interpretation implies that the actions of individuals are considered to be
determinate. This conflicted with his libertarian position which holds that
24
Following Katmer (1991, p. 18), 'formalization' is defined as " ... the development
and analysis of relations among variables that constitute part (or all) of an economic
model." As he noted, "[i]t is not necessary that these relations be expressed in mathema-
tical form." In tum, 'mathematization' can be defined as formalization in mathematical
terms. We shall restrict our analysis to the latter type of formalization.
25
It is not clear whether Mises saw any merit in such a translation from verbal to
mathematical 'language'. Suppes (1968, pp. 653 - 56), however, distinguished seven
advantages of (mathematical) formalization. Firstly, formalization is helpful in "[t]he
clarification of conceptual problems or the building of an explicit logical foundation."
Secondly, formalization may bring out explicit meanings of concepts. Thirdly, it may
standardize terminology and techniques. Fourthly, it permits the development of a more
general theory, without the inclusion of inessential details. Fifthly, by explicating the essen-
tial assumptions formalized theories may attain a greater degree of 'scientific objectivity'.
Sixthly, formalization sets out the precise conditions which are required for the analysis
under consideration. In this manner it may prohibit the ad hoc addition of new assump-
tions. And seventhly, it " ... makes possible an objective analysis of what are the minimal
assumptions necessary for statement of the theory." Debreu (1984, p. 275) added linguistic
convenience and the ability to obtain deeper understanding and analytical extensions that
might not otherwise be possible.
119
man's free will causes his choices to be indeterminate.26
The issue of the interpretation of mathematical equations can be
generalized. According to Garrison (1991, p. 57), mathematical economics
"... is inherently silent on the issue of cause and effect." This means that
"[s]ystems of equations can be suitably employed to describe the conse-
quences of human action, but such mathematical descriptions are inhe-
rently blind to notions of intentionality and causality" (p. 56). They can
describe the mutual determinacy of variables, which results from human
action, but they cannot incorporate the underlying element of human
volition. Garrison thus downplayed the potential of mathematical econ-
omics. He emphasized the fact that the mathematical equations must
always be interpreted (either implicitly or explicitly) in verbal language in
order to establish the direction of causality. This will not be denied by
many economists. However, as Weintraub (1985, p. 146) pointed out, this
process of interpretation may raise some difficulties, in the sense that the
availability of mathematical techniques may affect the questions which are
addressed. By concentrating on entrepreneurial decision-making, the
coordination problem and the structure of production, Austrians appa-
rently do not restrict their questions in such a way. 27 Their insistence on
the dispersion of knowledge and the (dynamic) subjectivity of the econ-
omic data leads them to emphasize the limits of the usefulness of math-
ematical analysis. They warn against any overzealous use of the math-
ematical techniques available. This does not mean that they reject mathe-
matics and abstractness per se. Instead, they maintain that there are many
problems in economics which cannot be addressed in mathematical
language, mainly because this would involve an oversimplification of these
26
However, if the use of mathematics is combined with probability calculus, it allows
for the indeterminateness of the outcome of individual events (choices) whereas the overall
outcome of these events exhibit a pattern in the form of a probability distribution. Some
Austrians (e.g. Hoppe and Herbener) reject this approach, on the grounds that the
distribution of the error term cannot be known. Other Austrians (e.g. Garrison) are more
pragmatic and do not want to eliminate all mathematical economics and econometrics.
Garrison (classroom conversation, July 10, 1990) is even willing to 'grant' the normal dis-
tribution of the error term to 'the econometricians'. But he warns against an interpretation
of statistical tests which does not leave room for type-! and type-11 errors. This means that
he merely warns against 'bad' econometrics. For related criticisms, see also Leamer (1983).
27
As mathematical economics can only deal with (Knightean) risk, and not with
fundamental (Knightean) uncertainty, this means that it is not suited to capture the
entrepreneurial element in market processes.
120
problems. 28 Hence, Garrison (1991, p. 60) opined that "[t]he appropriate
imperative is much milder in both substance and tone: Do not allow the
applicability of mathematical and statistical methods to define the scope
of economics."29 Thus, mathematical analysis need not be rejected from
an Austrian point of view, although its scope remains limited.
Thalenhorst and Wenig (1984) attempted to formalize Hayek's
(1931} theoretical framework, in particular his notion of intertemporal
equilibrium. They restricted their attention to a stationary intertemporal
equilibrium, in which no net investment takes place, and in which rates of
profit do not differ between stages of production. The rate of profit is the
relevant decision variable in the entrepreneurial decision about the
duration 0 of a production process. Thalenhorst and Wenig (1984, p. 216)
assumed that a rise in this duration brings about an increase in the level
of final output. They measure output in units of some standard basket of
consumer goods, which may be interpreted as the consumption of some
representative agent, thus solving the problem of aggregation over com-
modities (p. 216}. Another simplification concerns the number and length
of the stages of production. There are nO stages, which each comprise a
subperiod of duration 1/n (p. 216). In order to switch from discrete to
continuous time analysis, Thalenhorst and Wenig (1984, p. 222) assumed
that n approaches to infinity. Production in each stage takes place by
many firms, which each operate in that particular stage only. These
assumptions suffice to generate an unambiguous vertical price structure,
28
In his The Pure Theory of Capital (1941 (1950), pp. viii - ix) Hayek adopted this
position when stating that "]t]he mathematical form of expression is of assistance where it
helps us to deal with a greater number of variables than can conveniently be dealt with in
ordinary language. But the power of mathematical tools ... also has its limits. And the
problems with which we have to deal here are so complex that I soon found that, in order
to make them amenable to exact mathematical treatment on a plane where I could even
attempt it, I had to introduce much more drastic simplifications than seemed compatible
with the object." In other words, he did not mathematically formalize the coordination
problem because of the inherent oversimplification. Of course, non-Austrian economists
also warn against such oversimplification. Cf. Hutchison ( 1977, p. 88).
29
He added that "[m]ost economists if confronted explicitly with this recommendation
would, I suspect, accept it, many believing that it simply goes without saying. Implicitly,
however, the recommendation is systematically rejected - as judged by the extent to which
the applicability of these methods have in fact been allowed to dictate subject matter" (p.
60). Mises (1949 (1966), p. 355) also complained that the formulation of general-equilib-
rium states had unduly dominated the attention of economists. See also McCloskey's (1991,
p. 12) attack on the practice of economists, who (like mathematicians) conduct ".. .a search
through the hyperspace of conceivable assumptions" (p. 11).
121
which Hayek called the 'price fan'. 30 Prices and wages in this fan are
assumed to be perfectly flexible, which ensures immediate coordination.
This reflects that Thalenhorst and Wenig's (1984) model can serve as a
benchmark model only. It should be used as a starting point, from which
to depart in discoordination dynamics.
In the benchmark situation of perfect coordination production
takes place in the form of what Hayek (1931a, p. 40) called 'synchronized
processes of production'. This means that in each time period new
processes of production are started at a constant rate (intensity) x. Given
the equal time length of all stages of production, this means that in each
period x processes of production change from one stage to the next, and
hence that x processes become completed (p. 218). Under these condi-
tions, the number of processes under way is xn.
The duration n reflects an entrepreneurial decision, which
depends on the market rate of interest, among others. Thalenhorst and
Wenig (1984, p. 223) assumed that there is such a unique rater, which is
determined at an economy-wide credit market. This rate represents the
opportunity costs of one unit of money invested in the process of produc-
tion. Given the assumptions about the market structure in each stage of
production, firms will take the prices of their output, p, and input, w
(wages), as given by the market. That is, they maximize their profits per
production process (II), given p and w. By definition, maximum profit is
obtained if a lengthening of the duration of a production process does not
increase profits, that is, if (dii/dO) = 0. Thalenhorst and Wenig showed
that the optimal duration n· is a decreasing function of both the market
rate of interest r and the real wage q (p. 224 ). They also analyzed the
determinants of the rate of profit p. Given the optimal duration l'l* and
hence the optimal returns of output pb(Q*), this rate is determined by the
market rate of interest and the real wage rate (p. 225). 31
The model of Thalenhorst and Wenig (1984) contains four endo-
30 Hayek (1931a, p. 78) stated that production in its successive stages may be thought
of " ...as a fan, the sticks of which correspond to the prices of the different stages. If more
demand is concentrated to the one extreme - consumers' goods - the fan opens, the
differences between the stages become larger, and goods gravitate towards the stages
where higher prices are obtained, that is, towards the stages nearer consumption." Hayek
(1931, p. 79n1) recognized that at this point the simile of the fan may become misleading,
because "[t]he opening of the fan is ... accompanied by a reduction of the number of stages
of production, i.e.,of the number of sticks" (p. 79).
31
For a mathematical derivation and more elaborate analysis of this result, see
Thalenhorst and Wenig (1984, pp. 224- 25).
122
genous variables, namely the price level p, the nominal wage rate w, the
market rate of interest r, and the duration 0. Consumption demand,
money stock, and velocity of money are exogenous variables. This is also
the case with the intensity x, although this variable was endogenized in the
final part of Thalenhorst and Wenig's paper. The resulting model showed
that in equilibrium real variables are independent from monetary vari-
ables. This appears to be in contradiction with Hayek's business cycle
theory, which holds that monetary injections generate malinvestments.
Thalenhorst and Wenig (1984, pp. 233 - 34) explained that this result
arises because of the fact that their model only allows comparative-static
analyses, whereas Hayek's statement "... is concerned with disequilibrium
situations in which - according to him - changes in the quantity of money
disturb rather than support the process of adjustment to equilibrium."
That is, Thalenhorst and Wenig's model only includes comparative-static
monetary neutrality, whereas Hayek was concerned with monetary neu-
trality in the sense of monetary equilibrium.
Thalenhorst and Wenig showed that Hayek's equilibrium con-
struct can be mathematically formalized in a consistent (logically valid)
manner, but they acknowledged that this formalization assumes away
profit opportunities which can be gained from reallocating capital. It thus
remains silent on the Hayek Problem and hence on the dynamics of dis-
equilibrium situations (p. 234).32 Nevertheless, there are no reasons why
it may not serve as a basis for further research in this regard. Such
research should then concentrate on the dynamic (disequilibrium)
response on a monetary injection in the credit market.
32
Thalenhorst and Wenig (1984, pp. 227 - 28) suggested an adjustment process
through the real wage rate. They argued that adjustments in this rate let the rate of profit
p approach the market rate of interest r. That is, if the profit rate is larger (smaller) than
the market rate of interest, a unit of money used in production has a higher (lower) yield
than a unit of money lent on the credit market. Disregarding differences in uncertainty and
risk, it is then profitable for entrepreneurs to start more (less) production processes in
each period, that is, to increase (decrease) the intensity x. Under conditions of full
employment, this presupposes a rise (fall) in the real wage rate, which continues until the
gap between p and r is closed. This adjustment process not only guarantees that the real
wage rate is at its equilibrium position, but also that the intensity of the production process
is at its full-employment level. If the latter situation does not hold, so that the intensity is
below its full-employment level, then the real wage rate will be bid down. The costs of
production then fall, increasing the profit rate above the market interest rate. In tum, this
induces entrepreneurs to invest more in production and hence to increase the intensity. It
should be noted, however, that these adjustment processes are only partial, in the sense
that they assume the market rate of interest and the price level as given. The authors
recognized that the analysis of disequilibrium dynamics (and hence of the Hayek Problem)
must consider these variables as endogenous.
123
To summarize, an important aspect of the Austrian revival is its concern
with the role of mathematics in economics. Although mathematization
and abstractness are not rejected per se, they are approached with con-
siderable caution. In particular, Austrians reject the alleged practice of
(mathematical) economists to restrict their attention to problems which
can be formulated mathematically. However, it may well be that the
Austrian research tradition can benefit from some formalization. It should
at least make clear whether the Hayekian business cycle theory is logically
valid. This appears possible only if a complete lag structure and an
expectations formation process can be specified. 33 Thalenhorst and
Wenig's (1984) model may be useful in this regard. However, Austrians
have been very sceptical towards empirical testing. They argued that any
discorroboration of a logically valid theory need not result in its refuta-
tion, because it may result from a misspecification of the initial condi-
tions, or from the fact that the ceteris-paribus clause did not apply. Discor-
roboration therefore cannot constitute a criterion of refutation. In turn,
empirical testing can only establish whether the theory under consider-
ation has been relevant for the test period. Such a test has been carried
out by Wainhouse (1982) and (1984).
6.9. TESTING HAYEKIAN BUSINESS CYCLE THEORY
The empirical validity of Hayekian business cycle theory is confronted

with two problems. Firstly, as Haberler (1937 (1946), p. 43, italics in
original) observed, a necessary condition for Hayek's theory to hold is that
"... the demand for consumer goods does not rise pari passu with the
creation of credit and the rise in demand for capital goods." Hicks (1967,
p. 208) opined that this condition reflected a lag in consumer expendi-
tures. He argued that such a consumption lag is theoretically unsatisfac-
tory because there is no reason to presume why a rise in income would
not immediately lead to a rise in consumption. Hence he concluded that
this feature of the Hayekian business cycle theory made it very implaus-
33 As was shown in the previous chapter, Lachmann (1943) had already criticized the
Austrian theory of the business cycle for neglecting the role of expectations. Neither Mises
nor Hayek ever specified the process according to which expectations are formed, although
they implicitly recognized its relevance. Egger (1986, p. 65) nevertheless concluded that
"[t]here is some doubt that the Austrian cycle theory takes the problem of the formation of
expectations sufficiently serious." This neglect may result from the Austrians' radical
subjectivism and their adherence to descriptive realism.
124
ible. As was shown in Chapter 4, Hayek (1969b) rejected Hicks's interpre-
tation. He did not consider the conception of a 'lag' as useful. Instead, he
pointed out that the additional money flows into the economy in the form
of an expansion of credit, and he presumed that this additional credit
would be used for investment purposes. It will not immediately spread
throughout the economy. 34 On the other hand, Foss (1990, p. 6) main-
tained that the conception of a lag can be useful. Referring to Hayek
(1932), he argued that "Hayek's argument is ... dependent ... on the time
path of real factor incomes relative to the linear representation of produc-
tive activities [i.e. the vertical structure of production] and the lags this
implies" (italics in original).35 The structure of production can only be
changed gradually, in the sense that its change takes time. Foss's interpre-
tation is substantiated by Hayek's (1931a (1935), p. 88) claim that "... for
some time, consumption may even go on at an unchanged rate after the
more roundabout processes have actually started, because the goods which
have already advanced to the lower stages of production, being of a highly
specific character, will continue to come forward for some little time."
Money will not spread immediately throughout the economy, and there-
fore it takes time to disturb the structure of production. Analogously, it
also takes time for entrepreneurs to learn their mistakes and to adjust
their actions.
The second problem as regards the empirical validity of Austrian
business cycle theory concerns the role of empirical testing. Austrian do
not generally appreciate its importance, but instead maintain that falsifica-
tion cannot be a criterion for refutation, because it can never be ascertai-
ned whether the discorroboration concerns the main hypothesis or one of
the auxiliary assumptions. They conclude that econometric testing can
34
Hayek (1969b, p. 173) compared a situation of continuous credit creation to a
situation in which a viscous liquid, such as honey (read: money), is poured into a vessel
(the economy): "... if the stream hits the surface at one point, a little mound will form
there from which the additional matter will slowly spread outward." That is, there is no
such thing as 'helicopter money'. The question about the length of the production lags is
identical to the question why the liquid under consideration only spreads slowly, that is,
why it is such a viscous one.
35
Hayek (1932, p. 242) stated that "... even the money used for the purchase of new
capital goods must ultimately be paid out to the factors which make these new capital
goods. But they will rise to the full extent only when all the new money has passed
backwards through the successive stages of production until it is finally paid out to the
factors. There will, therefore, always be a considerable lag between the increase in the
money used for productive purposes and the corresponding increase in the incomes of the
factors -and consequent increase in the demand for consumers' goods."
125
neither verify nor falsify economic theory. However, even if it is conceded
that the Hayekian business cycle theory is logically valid (which is debata-
ble in itself), then it is not clear whether it is also empirically valid for all
time and places. As the theory was originally intended to explain the 'typi-
cal' nineteenth and early-twentieth century business cycles, the question
arises why it should also be empirically valid for present-day industrial
fluctuations. Econometric testing may at least indicate whether this claim
is supported by empirical 'evidence' and hence is justified. Such testing
necessitates a full specification of (1) the relevant elasticities, (2) lag-
structures, and (3) an expectations formation process.
Wainhouse (1982) and (1984) attempted to test Hayek's business cycle

theory empirically. 36 He thereby concentrated on Hayek (1933a) and
(1931a). As correlation does not imply causation, Wainhouse had to make
clear how the former could be translated into the latter. Following
Granger (1969) and (1980), he used the notion of Granger-'causality'. 37
This 'causality' can be described as follows: if the addition of a variable Y
to the information set helps to improve the forecasts of variable X (that
is, reduces the variance of X given 10 ) , then that variable Y Granger-
'causes' X. Wainhouse added that "[t]he choice of the information set(s)
and time series considered as the conditional environment within which
an empirical test of causality is conducted is based on economic theory"
(p. 69). In this manner the problem of 'measurement without theory' can
be avoided. The Hayekian theory can now dictate what propositions (and
hence what directions of causality) must be tested. Wainhouse (1984, pp.
47- 53) formulated the following nine propositions: 38
(1) Changes in the supply of savings are independent of changes in

the supply of bank credit.
36
This section will draw only on his 1984 article, which consists of a summary of his
1982 (unpublished) thesis. Unfortunately, I have not been able to analyze the latter.
37 Granger-'causality' is not based on some philosophical definition of cause and effect.
Instead, it merely reflects correlation. As Sims (1972, p. 543) observed, it is a sophisticated

version of post hoc ergo propter hoc, which means that it is a sophisticated version of a
logical fallacy (which of course remains a logical fallacy). The fact that the term 'causality'
is a misnomer was already acknowledged by Granger and Newbold (1977, p. 225), who
admitted that 'temporally related' would be more appropriate. Therefore, the term will be
put between quotation marks.
38
In the remainder of this section only the test results are discussed. For a discussion
of the tests themselves, see Wainhouse (1984).
126
(2) Changes in the supply of credit lead changes in the rate of inter-
est. Furthermore, changes in credit and interest rates are inversely
related.
(3) Changes in the rate of change of credit lead changes in the output
of producer goods.
(4) The ratio of producer goods prices to consumer goods prices
tends to rise after the initiation of a credit expansion.
(5) The prices of producer goods closest to final consumption tend to
decline relative to the prices of producer goods further away from
the consumer good in the production scheme.
(6) The prices of consumer goods rise relative to the prices of pro-
ducer goods, reversing the initial shift in relative prices.
(7) Toward the end of a Hayekian trade cycle, unemployment should
increase first in producer goods industries and then, with some
lag, in consumer goods industries.
(8) Employment will expand in consumer goods industries as relative-
ly more labour resources are applied both in response to the fall
in real wages and in an effort to satisfy consumer demand.
(9) Around the cycle peak, inflation in raw materials prices will
exceed that in consumer goods prices.
Proposition 1 reflects Hayek's view that the cycle is credit-induced, and

that money (and hence credit) is non-neutral. The rationale behind it is
that business cycles result from the fact that the market rate of interest
lies below the natural rate. Hayek had claimed that changes in the supply
of credit are independent from real economic influences, as reflected in
the level of savings. Wainhouse (1984, p. 57) observed that Hayek's theory
refers to the desired (and not the actual) supply of savings. Since this is an
unobservable variable, Wainhouse substituted it by several alternative
proxies. He concluded that "[t]he calculated F-values and results of the F-
tests for these [null] hypotheses uniformly support the notion that changes
in savings do not G·-cause changes in credit" (p. 58). However, the
proposition was rejected for the personal savings and deposits series. This
"... seems to be associated only with the addition of personal consumption
expenditures (PCE) on durable goods to the savings series" (p. 58, italics
in original). Wainhouse explained this association as a result from the fact
that these types of goods are often purchased with borrowed money. He
concluded that the tests corroborate proposition 1.
The rationale for the first part of the second proposition is that
127
the increase in the supply of credit precedes the fall in the market rate of
interest. Wainhouse's (1984, p. 59) tests indicated that this proposition
should be accepted at both the 99% and 95% confidence levels for fifty-
five of the fifty-seven cases considered (three measures of credit, nineteen
interest rates). An additional test confirmed that the relationship between
changes in the supply of credit and changes in the rate of interest was
'unidirectional'. Furthermore, the second part of proposition 2 stated that
the changes in the variables were inversely related. It turned out that
"[t]he majority of the signs (thirty-eight of fifty-seven) are positive ...
The[se] mixed results ... suggest that the credit series used in the analysis
represent realizations of the interaction of supply and demand. There
seems to be some evidence that, with respect to the nature of the credit
variables, we are dealing with a classical identification problem" (p. 60,
italics in original). Nevertheless, Wainhouse (1984, p. 60) concluded that
"[b]oth propositions 1 and 2 receive significant support from the data ... "
However, given the evidence with respect to the identification problem,
this conclusion may not be warranted as far as proposition 2 is concerned.
Disregarding this possibility, the conclusion amounts to the observation
that the causes of a credit-induced cycle as identified by Hayek were
present in the period under consideration. The question then is whether
these causes have really induced a Hayekian business cycle, that is,
whether there is a relationship between credit and output. This question is
addressed by proposition 3.
Proposition 3 concerns the non-neutrality of credit. According to
Wainhouse (1984, p. 60 - 61), "... we expect to observe, among the forty
measures of the output of producer goods examined, an association
between the credit sensitivity of an industry and its causal relationship to
accelerations in credit. The expected relationship between accelerations in
credit and changes in the output of producers' goods obtains in 102 of the
120 cases examined (85 percent)." Proposition 3 and hence the non-neu-
trality of credit need not be rejected, whereby he claimed that "... unidi-
rectional causality runs from credit to output" (p. 61).
The tests of the first three propositions suggest that the causes of
the Hayekian business cycle may well have been present in post-war
United States. The question whether the dynamics of this type of cycle
can characterize the industrial fluctuations in the U.S. for the period
under consideration was addressed by testing the propositions 4 - 6.
After having identified periods of credit expansion Wainhouse
tested whether such periods were followed by changes in relative prices.
128
The first test concerned the changes in relative prices of producer versus
consumer goods. Hayek had assumed that in the beginning of the boom
the former typically rise faster than the latter. Wainhouse examined 162
cases (6 credit expansion episodes and 27 prices). He concluded that in
110 instances Hayek's conjecture was corroborated (p. 63). The ratio
which was mentioned in proposition 4 indeed tended to rise after the initi-
ation of a credit expansion. But if this tendency existed, then Hayek's
theory implied that proposition 5 should also be valid. Wainhouse's ( 1984,
pp. 63 - 64) test examined 300 cases, of which "... 213 conform to the
behavior predicted by proposition 5, either coinciding with (145) or
lagging behind (68) the onset of the credit expansion." In the 87 remaining
cases the proposition was not corroborated.
As was shown in Chapter 4, Hayek had argued that the malinvest-
ment, the ex-post non-optimal lengthening of the structure of production,
and the concomitant process of forced savings will eventually lead to a
shortage of consumer goods. Proposition 6 accounts for this feature of the
latter stages of the boom in Hayekian business cycle. According to
Wainhouse (1984, p. 64), this proposition is also supported by the data,
especially those associated for the U.S. credit expansion periods of
October 1964 to 1967 and July 1977 to 1980.
Wainhouse (1984) only presented and discussed the test results of
propositions 1 - 6. He did not (directly) test propositions 7 - 9. Instead, he
argued that "... to the extent that movements in real resources are con-
comitant with changes in relative prices, the results of relative price
movements - propositions 4 - 6 - do suggest support for propositions 7 - 9"
(pp. 65 - 66). However, Wainhouse's tests contain two omissions. Firstly,
as he himself already recognized, his analysis does not incorporate an
expectations formation process (p. 66). And secondly, the analysis at
industry-level neglects the problem of circularity. As was shown in Chap-
ter 2, this problem refers to the fact that production need not be organ-
ized vertically, but instead may be 'circular' in the sense that in principle
every industry can produce goods which serve as an input for all other
industries. That is, there is a problem of determining which industries
produce lower-stage and which produce higher-stage goods. Unless this
problem is solved, it appears impossible to put Hayek's theory to a severe
test. Unfortunately, it is not clear from Wainhouse (1984) whether and
how he addressed this problem. Nevertheless and despite these remaining
problems, Wainhouse's tests are important in the Austrian revival because
they suggest that the Hayekian business cycle theory may be empirically
129
valid over the period after World War II. Unfortunately, Wainhouse's
(1984) description of his test results is rather incomplete, and does not
contain the mathematical model on which the tests were based.
6.10. CONCLUSIONS
The revival of Austrian economics is a revival of dynamic subjectivism.

Austrian economists have increasingly stressed the creative activity of the
human mind in ranking preferences, acquiring knowledge, forming expec-
tations, and shaping the future. In particular, Lachmann, O'Driscoll and
Rizzo have played an important part in this revival. They have thus
changed a situational constraint of the Hayek programme. Creativity also
forms an essential feature of entrepreneurship, although Kirzner's earlier
analyses somewhat obscured the issue. Whereas Mises had pointed out
that 'fundamental (or Knightean) uncertainty' is the main characteristic of
entrepreneurial action, Kirzner instead concentrated on 'alertness'. In his
later work he acknowledged that alertness is necessary only if individuals
are confronted with Knightean uncertainty. In his view, as in that of many
other Austrians, competition between individual economic agents brings
about a tendency towards coordination. In this sense entrepreneurship is
presumed to be coordinating. This presumption cannot be maintained on
a priori (economic-theoretical) grounds. In fact, it rests on a 'belief that
individuals eventually will learn what is useful to them. Differences
between individuals in this regard are explained by referring to differences
in the individuals' abilities (including knowledge) to seize profit opportu-
nities. Lachmann and his dynamic-subjectivist followers (High, Loasby)
reject this belief and argue that entrepreneurial action is discoordinating
as well as coordinating. They interpret the process of competition as an
evolutionary process.
As was shown in Chapter 4, the Austrian (or Hayekian) business
cycle theory asserts that the boom is set in motion by a difference
between the market and the natural rate of interest, due to credit expan-
sion. The resulting malinvestments persist for some time because of the
time-consuming nature of production. During the depression, they are
eliminated, so that Austrians consider this phase of the business cycle as
an inevitable consequence of the boom. This theory may be confronted
with two problems. The first problem concerns the underlying Austrian
capital theory, and particularly concerns the question whether production
130
is a predominantly vertical or circular process (see Chapter 2). The
second problem concerns the logical validity of Austrian business cycle
theory. Non-Austrian economists question this validity, presumably
because the theory has never been formalized mathematically. However,
as was shown above, Austrian economists are sceptical with respect to the
usefulness of mathematical economics, although this does not mean that
they do not reject it per se. In their opinion, formalization may easily lead
to overaggregation and hence oversimplification, blurring the complexity
of the social world. Furthermore, it may also increase the danger of inter-
preting economic relationships mechanistically. Such an interpretation is
inconsistent with the Austrian subjectivism. In particular, mathematical
formalization tends to eliminate the role of entrepreneurship in the
market process. Mter all, one of the characterizing features of
entrepreneurship is the fact that it faces Knightean uncertainty. Such
uncertainty seriously hampers any mathematical formalization, because it
reflects a complete absence of knowledge of events which may happen in
the future. The underlying dynamic subjectivism thus appears to prohibit
any mathematical formalization, even as a macroeconomic approximation.
The absence of a fully specified mathematical formalization of
Austrian business cycle theory means that the theory is difficult to test
empirically. Wainhouse (1984, p. 65) therefore undertook only a" ... broad-
brush empirical examination of Hayek's theory of the trade cycle ... ",
which indicated that the theory may be empirically valid for the period
considered. However, many Austrians argue that such empirical testing
cannot establish whether a theory is empirically valid. They claim that
there are no crucial tests, because a single hypothesis can only be tested
in conjunction with a large number of other hypotheses. Any discorrobor-
ation may then be attributed to one of these other hypotheses, including
the ceteris-paribus clause. In the Austrian view, the complexity of the
social world and the ensuing dispersion of knowledge prohibit a complete
specification of this clause. In turn, this means that empirical discorrobor-
ations need not be interpreted as falsifications, so that falsification cannot
be a practical criterion for refutation. Instead, Austrians argue that a
theory must only be refuted if it is logically invalid. The Austrian criterion
of refutation is thus constituted by logical inconsistency.
131
PART II. NEW CLASSICAL BUSINESS CYCLE THEORY
7. THE ROOTS OF NEW CLASSICISM
7.1. INTRODUCTION
In the last two decades business cycle theory has been experiencing a
remarkable survival. Whereas it seemed as if it had disappeared from the
research agenda in the 1960s/ it reemerged when New Classical Econo-
mics came up with a new explanation of industrial fluctuations. Although
its views on monetary policy evoked much criticism, in the second half of
the 1970s New Classicism has nevertheless assumed a prominent place in
the profession.
The purpose of this chapter is to outline the roots of the New
Classical research programme. These roots are very diverse in nature. For
instance, in their business cycle theories New Classicals distinguish
between shocks and transmission (or propagation) mechanisms in a way
which bears close resemblance to the approach of Frisch (1933) and
Slutzky (1937). Their emphasis on the role of knowledge and expectations
in economics appears to build on the work of Friedrich von Hayek. New
Classicals also adopt general equilibrium analysis, which means that their
work can be regarded as .standing in the Arrow-Debreu tradition. Despite
these influences it will be argued in this chapter that New Classicism finds
its roots in Monetarism. 2 The Monetarist connection is (almost) obvious,
given the early New Classical attempts to formalize and test Friedman's
Natural Rate Hypothesis.
This chapter will describe the connection between Monetarism
and New Classical Economics, starting in section 7.2 with a brief dis-
cussion of the most important formulations of the quantity theory of
money. It describes Fisher's transactions approach and Pigou's and
Marshall's (Cambridge) cash-balances approach and their respective
money demand functions. Keynes's liquidity-preference approach will also
be addressed. Section 7.3 analyzes Milton Friedman's (1956) restatement
1
Cf. Bronfenbrenner (ed.) (1969).
2
Cagan (1989, p. 195) described Monetarism as the research tradition which is based
on " ...the view that the quantity of money has a major influence on economic activity and
the price level and that the objectives of monetary policy are best achieved by targeting the
rate of growth of the money supply. • The term was first used by Brunner (1968). For a
short history of Monetarism, see Cagan (1989).
135
of the money demand function, including the Permanent Income Hypoth-
esis. This hypothesis holds that the agents' present demand for money
does not depend on their present income, but instead on their permanent
(i.e. expected life-time) income. This criticism had some important impli-
cations for the variability of the velocity of money (V) and, in turn, for the
effectiveness of monetary policy. Friedman's restatement amounts to the
proposition that individuals do not suffer from money illusion, so that only
real variables determine the demand for real balances. In 1958 Phillips
discovered an empirical relationship between a nominal and a real
variable, the so-called Phillips Curve. This relationship and its respective
interpretations by Phillips himself, Lipsey (1960) and Samuelson and
Solow (1960) are discussed in section 7.4. Section 7.5 addresses Fried-
man's Natural Rate Hypothesis, which holds that the long-term Phillips
Curve is vertical. In contrast, the short-term curve may be negatively
sloped because the economic agents may base their actions on incomplete
information. Such incomplete information causes expectational errors.
Phelps's (1967) incorporation of these expectations led to the so-called
expectations-augmented Phillips Curve, which is discussed in section 7.6.
Section 7. 7 contains some conclusions.
7.2. THE QUANTITY THEORY OF MONEY AND ITS

CRITICS
Presumably the best known and most influential version of the quantity
theory of money is Irving Fisher's (1911) equation of exchange, formulated
mathematically as MV = PT. Fisher presumed that causality runs from
money (M) to the other variables. In fact, he argued that the ultimate
('normal', long-run) effect of a change in M would be a proportionate
change in the general price level (P), although he allowed for short-term
changes in velocity (V) and 'trade' (1) to cause the business cycle (pp. 73
and 159)).3 The proportional long-term relationship between M and P
reflects the so-called (neo- )classical dichotomy, according to which money
is presumed not to influence real variables.
Fisher's formulation of the quantity theory of exchange adopts the
transactions approach. It emphasized the role of money in circulation (and
hence in transactions), and thus stressed its medium-of-exchange aspect.
3 According to Fisher such short-term transition periods may take as long as ten years.
136
In contrast, the Cambridge cash-balances approach by Pigou (1917) and
Marshall (1923) concentrated on the role of money as a store of value.4 It
can be formulated as M = kPy, where y is national income in constant
prices, and k reflects the time duration of the flows of final goods and
services which money could purchase. 5 The 'Cambridge-k' thus represents
the fraction of income which individuals actually hold or desire to hold in
the form of money.6 This means that Pigou (1917) and Marshall (1923)
transformed the quantity theory from a theory about prices into a theory
about the demand for money. As Friedman (1971, p. 10) observed, this
suggests an analysis of monetary phenomena in terms of supply and
demand conditions. The money supply crucially depends on the monetary
authorities, the private banks (under the restrictions imposed by the
monetary authorities) and the public.7 The demand for real money
(M/P), on the other hand, is determined by the public only. In what
follows we concentrate on these demand conditions.
In Fisher's view, the demand for money (Md) depends on its
purchasing power and hence on the actual price level (P). The volume of
transactions (1) also exerts an influence on Md, as do the expected rate of
inflation (pe = (~/dt)(1/P)) and some institutional factors (u) which
determine V (hence k). Fisher's view can then be formulated as Md = Md
(P, y, p e, u), where the real national income y is a measure of T. 8 In
contrast, the Cambridge-approach of Pigou and Marshall stressed money
as a store of value, that is, as an asset. Its formulation of the money
4
Pigou (1917, p. 54) claimed that his formula "... is a somewhat more effective engine
of analysis. It focusses attention on the proportion of their resources that people choose to
keep in the form of titles to legal tender instead of focussing it on •velocity of circulation'.
This fact gives, as I think, a real advantage, because it brings us at once into relation with
volition - an ultimate cause of demand - instead of with something that seems at first sight
accidental and arbitrary."
5 Friedman (1971, p. 7) noted that Fisher's V differs from Pigou's. The former applies
to all transactions whereas the latter merely concerns the payments for final goods and
services. Cf. also Bordo (1989, p. 152).
6
As Friedman (1971, p. 9) indicated, k may be defined as either the desired or the
actual ratio of money to income. If the former definition is used, then M must be defined
as the desired quantity of money. On the other hand, if k is defined as the actual ratio,
then M gives the actual quantity of money. In either case, k = 1/ V.
7
Cf. Friedman and Schwartz (1963b, pp. 776 - 98), Cagan (1965}, Friedman (1971, p.
11) and (1989, pp. 9 - 11).
8
The mathematical formulations of the respective demand functions for money have
been derived from Visser (1980, p. 121).
137
demand function therefore includes the opportunity costs of holding
money, that is, the rates of interest of other assets. It can be formulated
as Md = Md (P, y, r, pe, u), where r gives these rates of interest.
Keynes's (1923) work on monetary theory can be interpreted as
an elaboration of the Cambridge cash-balances approach. He downplayed
the relevance of its long-term implications, and argued instead that in the
short run a change in M will also affect k and y. 9 His analysis emphasized
those factors which could disturb the proportionality of the relationship
between M and P, thereby attributing great significance to expectations
about interest rates. His demand function for money can be formulated as
Md = Md (P, y, r- r, r
Y:), where r- denotes the difference between the
interest rate and some 'normal' rate (speculation motive), and Y: is the
expected nominal income (transactions- or perhaps better finance motive).
One of the most important elements of his analysis is his assumption that
the economy can be in a stable equilibrium in which its resources (includ-
ing labour) are underemployed. In such conditions of underemployment
the rate of interest is considered to be low. Since Keynes (1936 (1983), p.
203) opined that the rate of interest is a psychological ('conventional')
phenomenon, which depends on "... the prevailing view as to what its
value is expected to be", monetary policy will be ineffective as a means to
increase effective demand in situations of underemployment. After all, if
the interest rate is low, individuals will then believe that r < They will r.
hoard their cash balances, reducing V (increasing k). This situation is
Keynes's famous liquidity trap. It implies that k is highly volatile. In the
1950s this view on k's volatility was criticized by Milton Friedman, who
presupposed that the demand function for money does not comprise
present income but instead permanent income (or wealth).
7.3. FRIEDMAN'S RESTATEMENT OF THE QUANTITY

THEORY
Friedman (1956) adopted the cash-balances version, implying that the

quantity theory is a theory about the demand for money. His money
demand function starts from the premise that money is an asset. Its
9
It is in this regard that he stated that the ".. .long run is a misleading guide to current
affairs. In the long run we are all dead. Economists set themselves too easy, too useless a
task if in tempestuous seasons they can only tell us that when the storm is long past the
ocean is flat again" (p. 80, italics in original).
138
demand therefore depends on the various rates of interests which can be
earned on other assets, whereby the term assets not refers to financial
assets but also to all other goods. Keynes had already allowed the rate(s)
of interest to play a role in the demand for money, but he had neglected
the importance of wealth for this demand. 10 Friedman (1956, p. 4)
opined that money is one of the forms in which economic agents can
decide to hold their wealth. In his view, the appropriate measure of
wealth is permanent income, measured as a discounted, present-value
stream of payments which are derived from an existing stock of wealth
(including human wealth, such as the income-earning ability of the
individuals under consideration), so that it "... includes all sources of
'income' or consumable services" (p. 4). According to Friedman (1957, pp.
25 - 26), the Permanent Income Hypothesis (PIH) holds that present
income (Y) and present consumption (C) both consist of a permanent (YP,
CP) and a transitory (Yt, Ct) component. The respective permanent
components are functionally related, whereby the ratio of permanent
consumption to permanent income, c, depends on the rate of interest (or
sets of rates of interests) (r), the ratio of nonhuman wealth to income
(W), and the portmanteau variable u. 11 The demand function for real
balances can then be formulated as: MJP = Md/P (YP, W, pe, r, u).
Friedman (1956) subsequently addressed three issues (pp. 15 -
17). Firstly, he argued that the demand for money is highly stable (and
more stable than the Keynesian consumption function). More importantly,
he presumed that it plays a very important role in determining crucial
variables such as the level of money income or of prices. Secondly, he
claimed that the supply of money is affected by at least some factors
which do not determine the demand for money. The stability of the
demand function for money then allows economists to solve identification
problems, in the sense that they can trace out of the effects of these
changes in money supply. The third issue concerns the form of the
demand function for money. According to Friedman, Keynesians base
their underemployment analysis on the assertion that the demand for
money is infinitely elastic at a 'small' positive interest rate, leading to the
liquidity trap. Friedman denied this infinite interest elasticity of money
°
1
Keynes (1923, p. 45) recognized that wealth may play a role in the demand function
for money, but his factual analysis does not take wealth into account.
11
The variable u includes factors which determine the consumer's tastes and prefer-
ences, the technological conditions of production, etc.
139
demand and hence the liquidity trap. In his view, monetary policy can
influence real variables, albeit only in the short run. In the longer run,
though, prices are flexible, so that changes in M will merely lead to
changes in P. Consequently, long-term monetary policy can only affect the
general price level by causing more (or less) inflation.
Friedman's views imply that there can be no long-term relation-
ship between monetary and real variables. Nevertheless, in 1958 A.W.
Phillips found such a relationship, namely between on the one hand the
rate of change in the nominal wage rate (w) and on the other the unem-
ployment rate (U) and the rate of change of this rate (u). This relation-
ship was called the Phillips Curve. As Solow (1978, p. 147) recognized, the
curve is not formally linked to Keynesian analysis, but it was welcomed as
filling a gap in the Keynesian system, in the sense that it was interpreted
as reflecting the rigidity of (real) wages. 12 The curve came to be
regarded as an exploitable trade-off which could be used in economic
policy. Before addressing its explanation by Monetarists, we must first
describe this empirical relationship and some of its other interpretations.
7.4. INTERPRETATIONS OF THE PHILLIPS CURVE
In 1958 A.W. Phillips conjectured that the tighter the labour market, the
more rapidly employers had to increase nominal wages in order to attract
new employees and retain the ones they already employed. He therefore
hypothesized "... that the rate of change of money wage rates in the
United Kingdom can be explained by the level of unemployment and the
rate of change of unemployment ... ", thus presupposing that causality runs
from real to nominal variables (p. 284). Assuming downward wage sticki-
ness, Phillips (1958, p. 283) discovered that his empirical data supported
the hypothesis that the (hyperbolical) relationship between the rate of
unemployment (U) and the rate of change of money wage rates (w) was
remarkably stable over the period analyzed. This relationship was called
the Phillips Curve. Moreover, Phillips also found that the latter variable
was (negatively) related with the rate of change in the unemployment rate
(u), in the sense that if the unemployment rate increases (decreases),
money wages increase less (more) for every rate of unemployment (p.
12 For a more elaborate and critical analysis of (the reasons of the acceptance of)
Phillips's fmdings, see Wulwick (1987, particularly Part II, pp. 841 - 43).
140
290). This means that there are anti-clockwise cyclical movements around
the Phillips Curve. According to Hoover (1988, p. 23), this curve is
therefore best interpreted as reflecting stable (constant) combinations of
U and w, given that u is zero. As it concerns the economy as a whole, the
movements around the curve can be called 'macro-loops'. Phillips (1958)
explained these loops by conjecturing that employers will expect the
demand for their product to rise if unemployment decreases. They will
anticipate this rise by hiring more labour, which in turn is supposed to
increase the nominal wage rate. In other words, Phillips (1958) explained
the macro-loops in terms of micro-loops.
Lipsey (1960) objected that if the demand-expectations of the
employers are correct, the economy will simply attain a situation in which
unemployment will be lower than it was in the initial situation. 13 This
means that there will be no cyclical movements on single markets, and
hence no micro-loops (p. 21), so that the relationship between the rate of
unemployment (U) and the rate of change in the nominal wage rate (w)
on a single market must be stable and unambiguous. However, this is not
to say that there cannot be macro-loops. Lipsey (1960, pp. 22 - 23) based
his alternative explanation of the Phillips Curve and the macro-loops "...
on the hypothesis that the recovery affects different markets at different
times while the fall in effective demand is, at least during the early stages
of the recession, most evenly distributed." Furthermore, he assumed that
wages rise faster in markets with excess labour demand than they fall in
markets with excess labour supply (p. 18). As a result, the rate of change
of the nominal wage rate for the economy as a whole may rise, even if the
overall rate of unemployment remains unchanged (p. 18). 14
Lipsey's analysis reflects an awareness of the empirical and
variable nature of the Phillips Curve. He recognized that it is a relation-
ship between economy-wide averages, and that it cannot be regarded as
stable. In fact, he opined that the analysis of the curve "... points a general
warning against the procedure of accepting statistically fitted relations
without relating them to models of market behaviour" (p. 23, italics in
original). He concluded that "[t]he macro-curve will thus be useful for
prediction providing that the same sort of inter-market inequalities
continue to occur. Great care must be taken in using the curve to predict
13
Other important critics were Hines (1971) and Mackay and Hart (1974).
14
If the former markets are denoted by the subscript A and the latter by B, then the
above implies that w = (wA + w8 )/2 > 0 even if U remains unchanged.
141
what would happen if the level of U were held constant for some time for,
if this were done, the degree of inter-market inequality in excess demand
would be expected to change considerably." This point was neglected in
the later literature on the relationship between U and w. It appears that
Samuelson and Solow (1960} played an important role in this regard.
They reformulated the Phillips Curve by substituting the rate of change in
prices (p) for the rate of change in nominal wages (w), thus implying that
prices and nominal wages are very closely connected. As the reduction of
both the rate of inflation and the rate of unemployment are usually con-
sidered to be objectives of economic policy, this reformulation facilitated
the reinterpretation of the curve as an exploitable policy trade-off. 15
Policymakers are considered to be able to choose a combination of infla-
tion and unemployment which they regard as desirable. It was thought
that this could be done without any repercussions in the longer run. This
led to the incorporation of the Phillips Curve into macroeconometric
models during the 1960s and 1970s. As Hoover (1988, p. 24) noted, the
curve was widely used in the analysis of government macroeconomic
policy. But when it could not accurately chart the relationship between the
rate of inflation and the rate of unemployment in the period of stagflation
of the early 1970s, Friedman's criticisms of the curve were fairly rapidly
accepted.
The oversimplified interpretation of the Phillips Curve as an
exploitable trade-off thus considers it to be a short-run but nevertheless
stable relationship. It implies that individuals suffer from money illusion,
because real variables (unemployment) are related to monetary variables
(inflation). This posed a problem for Friedman whose reformulation of
the quantity theory implied that money is neutral, at least in the long run.
The standard interpretation of the Phillips Curve thus challenged his
quantity-theoretical position, and induced him to undertake an alternative
explanation of the relationship as found by Phillips. Before discussing this
alternative, we must first return to Friedman's monetary studies. 16
15 It should be mentioned that Samuelson and Solow (1960, p. 193) themselves

rejected such an interpretation. They opined that "[i]t would be wrong, though, to think
that our Figure 2 menu [cf. their p. 192] that relates obtainable price and unemployment
behavior will maintain its shape in the longer run. What we do in a policy way during the
next few years might cause it to shift in a definite way. "They mentioned the possibility of
moving the Phillips Curve to the left by institutional reforms.
16 For a more elaborate analysis of these studies, see Hirsch and De Marchi (1990,
Chapter 10).
142
7.5. FRIEDMAN AND THE PHILLIPS CURVE
Friedman (1958 (1969)) tried to establish whether changes in the supply

of money are correlated with changes either in the general price level or
in real output. Over longer periods (back to the early 1880s) he found a
one-to-one relation with the former. However, this relationship was not
mechanically rigid. Friedman (1958, p. 174) opined that two 'disturbances'
prevent such rigidity: (1) changes in real output, and (2) changes in the
desired real cash balances as a fraction of the public's income. His
characterization of changes in output as 'disturbances' implies that he
already presumed that under 'normal' (long-term) conditions changes in
M do not affect real output (p. 184). There is then no relationship
between the rate of unemployment ( U) and the change in the general
price level (p ), which means that the long-term Phillips Curve must be
vertical. The rate of unemployment is then on its long-term equilibrium
level. Friedman (1968, p. 8) defined this natural rate of unemployment
(NRU) as" ... the level that would be ground out by the Walrasian system
of general equilibrium equations, provided there is embedded in them the
actual structural characteristics of the labor and commodity markets,
including market imperfections, stochastic variability in demands and
supplies, the cost of gathering information about job vacancies and labor
availabilities, the cost of mobility, and so on." 17 This rate can best be
interpreted as a long-term, perfect knowledge and perfect foresight equili-
brium.18 As such, it is a benchmark against which the actual performance
of an economy can be assessed. 19
The NRU leaves the relationship as found by Phillips unex-
17
It should be noted that Friedman's analogy with Wicksell's notion of a 'natural rate
of interest' is inadequate. Whereas the NRU may be seen as the long-term general equili-
brium rate of unemployment, Wicksell's natural rate of interest merely refers to monetary
equilibrium. The two types of equilibrium need not to coincide. (I owe this point to Mr.
Botha from the Centre of Economic Analysis in Pretoria, South Africa)
18
However, the NRU need not reflect a unique general equilibrium nor does it have
to be 'path-independent'. This path-dependence means that the equilibrium value of a
variable depends on its (disequilibrium) values in the past. This phenomenon is called hys-
teresis. For a detailed treatment of hysteresis, see Cross (ed.) (1988).
19
Phelps (1968 (1970), p. 124 - 25) argued that " ... any actual economy is almost
continuously out of equilibrium, so we need also to study wage and price dynamics under
arbitrary conditions." He continued by claiming that "[t]he Phillips curve studies of the past
decade have done this ... "Thus, the Phillips curve is seen as a disequilibrium phenomenon,
while Phillips himself presumably would have argued that it represents long-run stable
combinations of the nominal wage rate change and the unemployment rate.
143
plained. It merely suggests that given the assumption about the long-term
neutrality of money the Phillips Curve must be interpreted as a short-term
disequilibrium relationship. It does not specify the equilibrating mechan-
ism which is supposed to operate in such a disequilibrium situation. In
their analysis of the cyclical (hence short-term) influence of money,
Friedman and Schwartz (1963 (1969), p. 222) were aware of the need to
specify such a transition mechanism.20 Their 'tentative sketch' of this
mechanism considered "... the concept of cyclical fluctuations as the
outcome of balance sheet adjustment, as the effects on flows of adjust-
ments between desired and actual stocks" (p. 234 ). The cyclical move-
ments in the economy arise from the fact that "[ t ]he stocks serve as
buffers or shock absorbers of initial changes in rates of flow, by expanding
or contracting from their 'normal' (i.e. 'natural' or 'desired') state, and
then slowly alter other flows as holders try to regain that state" (p. 234 ).
The mechanism thus presupposes that the increase in the supply of money
disturbs the cash balances of individuals. This means that the increase
must be unexpected, otherwise rational individuals would already have
taken it into account. Hence expectations should play a crucial role in the
explanation of short-term dynamics. However, as Hirsch and De Marchi
( 1990, p. 225) argued, "[e ]xpectations enter into the original formulation
of the money demand relation, but their centrality is appreciated only as
Friedman, in the early 1960s, began to explore a transmission (or adjust-
ment) mechanism." The problem then is how to formalize an expectations
formation mechanism. Phelps (1967) advanced a formalization of such a
mechanism. 21
20
They argued that "[a] fully satisfactory explanation of the minor movements would
require an explicit and rigorously stated theory, which could take the form of a series of
simultaneous differential equations describing the reaction mechanism of the economy,
together with a specification of the joint distribution function of the random disturbances
impinging on it, and a specification of the systematic disturbances that could be introduced
into it. Our belief that money plays an important role in minor movements is equivalent to
asserting that some of these differential equations would contain the stock of money as a
variable; that disturbances in the stock of money are among the random or systematic dis-
turbances impinging on the system; and that these disturbances alone would be capable of
generating a path for such major economic variables as money income, prices, output, and
the like, comparable to the path they actually follow during mild depression cycles" (p.
223).
21
It is interesting to note that Phelps (1968, p. 682) observed that "[c]ontinental
economists like von Mises (1953, pp. 418- 20) always emphasized the role of expectations
in the inflationary process", thereby referring to the 1953 English translation of Ludwig von
Mises's Die 1heorie des Geldes und der Umlaufsmittel (1912 (1924)). As was shown above,
though, Mises's static subjectivism prevented him from elaborating the role of expectations.
144
7.6. THE INCORPORATION OF EXPECTATIONS
Phelps's (1967) analysis used as a benchmark the natural rate of unem-

ployment, which he defined as the rate at which the actual rate of inflati-
on equals the expected rate of inflation (p. 255). It implies that expecta-
tions are correct, and hence presupposes that individuals must have
perfect knowledge and perfect foresight. They have then no reason to
adjust their expectations about inflation. Moreover, there are no endoge-
nous reason for the rate of inflation to change. Phelps's version of the
NRU is therefore more correctly called the non-accelerating inflation rate
of unemployment (NAIRU). Phelps presupposed that this NAIRU holds
only in the long run. His short-term analysis reveals a quite different
picture because individuals will then base their expectations and decisions
on incomplete information. The analytical problem was how to model
their expectations formation process and the incomplete information used
in it. Phelps (1967, pp. 262 - 63) solved the first part of this problem by
adopting the Adaptive Expectations Hypothesis (AEH), which had already
been used by Koyck (1954) and Cagan (1956). This hypothesis holds that
expectations are weighted averages of the past values of the variable in
question. They are revised by a fraction (a) of the forecast error. This
revision takes place only gradually, hence 0 < a < 1. The hypothesis can
be formalized as follows: 22
(7.1) E(xt) - E(xt_ 1) = a [xt_ 1 - E(xt_ 1)]
where xt is the value of variable x at time t, and E(xt) is the expectation of

variable xt formed at time t-1, with 0 < a < 1. This equation reflects that
economic agents revise their expectation of x at time t by some fraction of
the forecast error of their expectation of x at time t-1. The same expres-
sion must hold during the previous periods. By recursively substituting the
(unobservable) left hand side by the (observable) right hand side of the
equation, one obtains:
(7.2) E(x,) = ax,_1 + a(l - a) x,_2 + a(l - a) 2 xt_ 3 + ... +

+ a(l - a)" xt-n-1 + (1 - a)"+1 E(xt-n-1)
As time elapses, the latter term approaches to zero because 0 < a < 1.
22 Cf. Begg (1982, p. 23).
145
In the long run E(xt) is constituted only by observable terms and is
therefore itself observable. The fact that 0 < a < 1 indicates that expec-
tational errors will exert their influence for some time, although in the
long run expectations will have adjusted completely and hence will be
correct. This means that the AEH could be used in an explanation of the
negative slope of the Phillips Curve for longer periods of time. However,
although the problem of specifying an expectations formation mechanism
was solved, expectational errors presuppose that agents have incomplete
knowledge and/ or foresight. Therefore, Phelps's specification was still
incomplete, because it did not contain the agents' information sets.
Phelps's introduction to the so-called 'Phelps volume' (1970)
suggested a way in which the information sets could be specified. In 1968
he had already hinted at the possibility of spatial mismatching between
jobs and people, due to information costs. In his 1970 introduction he
drew a picture of an economy which consists of several labour markets, or
'islands'. These 'islands' are imperfectly linked both physically and infor-
mationally. Information about the other islands thus travels slowly. As a
result, economic agents are presumed to have all current information
about the island on which they live (and offer their labour services), but
incomplete information about the economy as a whole. These assumptions
amount to an information set of the agents in which all current local
information is included but which contains only lagged (with one period)
and hence incomplete global information. Stated differently, information
is heterogeneous across markets and homogeneous across agents on the
same market. In this 'island parable' agents must form expectations on the
basis of incomplete short-term global information. In the case of a change
in their local nominal wages they are faced with an interpretation (or
signal extraction) problem, namely whether this change should be attri-
buted to a change in their real wage rate or to a change in the general
price level. The incomplete information on which their expectations are
based, will lead (at least) some agents to solve this signal extraction
problem incorrectly, in the sense that they form incorrect estimates of the
rate of inflation. This means that they will respond to nominal changes as
if they were real in nature. It thus appears that they suffer from money
illusion, whereas in fact they merely interpret the change in nominal wage
wrongly because of insufficient information. This insufficiency then
explains the relationship between the level of unemployment and changes
in the nominal wage rate, that is, the negatively sloped Phillips Curve. It
indicates that the agents' actions are ex-post non-optimal, and will be revi-
146
sed when the presently lacking global information becomes available.
Phelps's analysis, or rather his use of the AEH, is somewhat
problematical on two accounts. Firstly, the hypothesis does not explain the
magnitude of the adjustment parameter (a) by an economic theory.23 It
simply postulates that new information is only gradually fed into the
expectations. It therefore allows for systematic expectational errors, which
may be easily corrected, so that it may not be consistent with the rational-
ity postulate. 24 Moreover, Phelps's version of the AEH is entirely back-
ward-looking, and it presupposes that expectations are formed by using
only the past values of the variable under consideration. Economic agents
thus neglect currently available information about other variables. Again,
this may be inconsistent with the rationality postulate. 25 Secondly, Phelps
(1970) suggested that general equilibrium analysis is the appropriate
method of analysis, whereby he argued that the NRU is path-depend-
ent.26 New Classicals addressed both problems, but they did not adopt
Phelps's latter position, as will be shown in the next chapters.
7.7. SUMMARY AND CONCLUSIONS
Friedman's work on monetary theory stands in the tradition of the

quantity theory of money. But his restatement of Fisher's formulation took
the form (not the content) of the Cambridge equation. It contains two
basic premises. Firstly, he subscribed to Fisher's view as regards the
primacy of money, and secondly, he interpreted money as an asset. This
latter point enabled him to substitute Keynes's formulation of the money
demand function by his own, in which wealth or permanent income plays
23
For a more detailed criticism of the AEH, cf. Begg (1982, pp. 25 - 26) and Pesaran
(1989, pp. 17- 19).
24
Hahn (1986, p. 281) remarked, though, that the fact that "...an agent will not persist
in expectations which are systematically disappointed ... "does not necessarily imply that " ...
agents have expectations which are not systematically disappointed ... "
25
Pesaran (1989, Ch. 9) provided an extension of the AEH, which takes account of
information on other variables than the one to be predicted.
26
He argued that " ... the transition from one equilibrium to the other tends to have
long-lingering effects on the labour force, and these effects may be discernible in the
equilibrium rate of unemployment for a long time. . .. [T]he natural rate of unemployment
at any future date will depend upon the course of history in the interim. . .. [S]uch a
property is sometimes called hysteresis" (1972, p. xxiii).
147
an important part. As permanent income fluctuates less than current
income, the demand for money will be more stable than the Keynesians
had argued. More importantly, Friedman argued that the interest elasticity
of the demand for money was not infinite. In his view, monetary policy
may therefore be effective, at least in the short run.
The discovery of the Phillips Curve and its subsequent interpreta-
tion as an exploitable trade-off have been severely criticized by Friedman.
He opined that changes in the supply of money must find their way in
changes either in the general price level or in real output, given the
relatively constant velocity of money. The Phillips Curve appeared to
suggest the latter effect, but Friedman rejected this relationship. Instead,
he propounded the Natural Rate Hypothesis, which in very general terms
holds that real variables are influenced by real variables only. However,
the NRH left the Phillips Curve unexplained. Friedman and Phelps both
gave such an explanation in terms of expectational errors. In doing so,
Phelps formulated an analytic, non-stochastic, partial-equilibrium model in
which economic agents form their expectations adaptively. It resulted in a
negatively-sloped short-run Phillips Curve which in the long run becomes
vertical at the 'natural rate- of unemployment' (NRU, or better NAIRU).
He suggested, however, that his analysis was unsatisfactory, because it did
not adopt a general-equilibrium framework.
Phelps's analysis forms the background against which New Classi-
cal general-equilibrium analysis must be interpreted. He asked whether
informational imperfections could lead an economic system which oper-
ates in a mutually consistent way into large-scale employment fluctuations.
This question redirected Lucas's research efforts towards general equilib-
rium analysis. Additionally, Lucas would address two possible drawbacks
of the analysis of Phelps. As was shown, the AEH may be inconsistent
with the rationality postulate. In 1961 John Muth had already used a
different expectations formation hypothesis .. His Rational Expectations
Hypothesis (REH) became one of the hallmarks of New Classicism. Fur-
thermore, as a first approximation Phelps (1968) had used a non-stochas-
tic model. Instead, New Classicals would change this property by adopting
Frisch's (1933) and Slutzky's (1937) 'stochasticism'. As Lucas (1981, p. 7)
stated, "... the needed ingredients for a general-equilibrium formulation
seemed to be readily at hand." They 'merely' had to be combined.
148
8. THE RISE OF NEW CLASSICAL ECONOMICS
8.1. INTRODUCTION
From the 1970s onwards New Classical Economics (NCE) attracted much
attention among economists. Its assumptions and theoretical conclusions
evoked much controversy, which resulted in an extensive literature on the
methodology of the NCE. One of the earliest analysis was that of Mad-
dock (1979), summarized in his (1984), who depicted the development of
the NCE and provided a reconstruction along Lakatosian lines. He argued
that the initial goal of the 'Rational-Expectations theorists' was to test the
Natural Rate Hypothesis. However, in the mid 1970s the research pro-
gramme experienced a problem shift towards the theory of economic
policy, and more in particular towards the neutrality proposition, which
holds that anticipated changes in the money supply will not affect real
output and unemployment. A second approach in the discussion of New
Classical methodology is that of Klamer ( 1984), who concentrated on the
style of argument (rhetorics) of the NCE! He also suggested that the
central claim of this tradition is the neutrality proposition, thus in this
regard concurring with Maddock. Boland's (1986) critical analysis of the
so-called disequilibrium foundation of equilibrium economics is the third
discussion of the NCE, although its object is not restricted to this research
programme. However, his criticisms of what he called Macroeconomics,
Rational Expectations, Stochasticism, and Instrumentalism directly apply
to the NCE, as do his criticisms of the assumption of continuously clear-
ing markets. A fourth analysis of the NCE was provided by Hoover
(1988), who made a comparison between Friedman's and New Classical
methodology. He concentrated on the former's partial and the latter's
general equilibrium analysis, and concluded that Friedman's economics is
Marshallian in nature, whereas the NCE is a Walrasian type of monetar-
ism. Hoover's notion of 'Walrasian analysis' was criticized by Hirsch and
De Marchi (1990), who considered it too restrictive and instead proposed
the distinction between 'Deweyan' and 'Millian' analysis.
Given the propensity of economists and economic methodologists
to differ, one might expect that the various studies of the emergence and
1
Klamer's unpublished dissertation provided the basis for his (1984). It was written
more or less simultaneously with Maddock (1979).
149
/
/
development of New Classicism would inevitably lead to differences
concerning the central claims of this research programme. However, there
has been a remarkable agreement on these claims, although emphasis
differs. There is a consensus as regards the definition (or classification) of
theNCE, which can be characterized by the following assumptions:
(1) the assumption of continuous market clearing;

(2) the Lucas supply function (including some version of the NRH);
(3) some version of the Rational Expectations Hypothesis; and
(4) some assumption about the information set of the individuals.2
This chapter will analyze what versions of these assumptions were used by
the early contributors to the programme. The analysis will be restricted to
the work of Robert Lucas, Leonard Rapping, and Thomas Sargent.
The chapter aims to describe the rise of the NCE, without yet
providing a methodological assessment. 3 It is organized as follows. Section
8.2 discusses the formalization of the Natural Rate Hypothesis (NRH) by
Lucas and Rapping (1969a and 1969b). They adopted a partial-equilib-
rium model in which the concept of the 'representative individual' was
incorporated. Such a framework does not allow for the analysis of econ-
omy-wide interdependencies. Phelps (1970) noted that partial-equilibrium
analysis could therefore not be used to account for the negative slope of
the Phillips Curve. He introduced instead the so-called 'islands parable',
discussed in the previous chapter. Moreover, Lucas and Rapping used the
AEH, which is in danger of being inconsistent with the rationality postu-
late. This problem was solved by substituting the Rational Expectation
Hypothesis (REH) for the AEH. The former will be discussed in section
8.3. The change from partial-equilibrium to general-equilibrium frame-
work is addressed in section 8.4. Section 8.5 subsequently discusses an
econometric issue which arose because of the substitution of the REH for
the AEH. The REH implied that economic agents would include econ-
omic policy in their expectations. An anticipated change in policy then
results in a change in expectations, and therefore in a change in behav-
iour. Lucas criticized 'orthodox' methods of policy evaluation because they
neglected these behavioural changes. The exposition of the so-called
'Lucas critique' is followed in section 8.6 by a discussion of Lucas's at-
2 Cf. also Barro (1981b, p. 41).
3 Such an assessment will be given at the end of Chapter 9.
150
tempt to test the NRH empirically. This attempt evaded the issues con-
cerning econometric policy evaluation which Lucas himself had raised. In
contrast, Sargent's empirical tests, which were performed at about the
same time as that of Lucas, did take Lucas's critique into account. Before
discussing these tests, a short detour will be made by outlining their
'prehistory'. This discussion of Sargent's early work will make clear that
Friedman's NRH amounts to the same thing as the long-term framework
of Fisher's solution to the Gibson paradox. Section 8.8 contains some con-
clusions.
8.2. THE FORMALIZATION OF THE NRH
The first step towards New Classical Macroeconomics was made by Lucas
and Rapping (1969a).4 They attempted to provide "... the rationalization
in supply-and-demand terms of the observed correlation between unem-
ployment rates and rates of inflation, or Phillips curve" (p. 20). 5 Friedman
and Schwartz's (1963b (1969)) 'tentative sketch' had already offered a
rationalization in verbal terms. Moreover, Phelps (1967) had shown that
the Adaptive Expectations Hypothesis could be used as the transition
mechanism from the short-term to the long-term Phillips Curve. However,
he had not been able to include both types of curves in his mathematical
model. Lucas and Rapping (1969a) tried to solve this problem by formal-
izing the NRU in such a way as to allow for short-term deviations from
this rate. In their view, "... an adequate model must contain both a short
run and a long run. There are, then, three features which we feel a model
of the labor market (or, more broadly, the production-employment sector)
should possess. First, it should incorporate the neoclassical feature that
for a fixed capital stock the aggregate supply schedule (relating the price
of goods to real output) will become perfectly inelastic over a long period
of stable aggregate demand. Second, the model should imply an elastic
short-run aggregate supply function consistent with the observed fluctu-
ations in real output and employment in the face of shifting aggregate
4
Page numbers refer to the 1981 reprint of Lucas and Rapping (1969a).
5 They expressed their dissatisfaction with "[r]ecent attempts to give a theoretical base
to the Phillips curve [which] have been based largely on a view of the labor market as
dominated by collective bargaining, where bargaining outcomes bear no explicit relation to
supply-and-demand forces" (p. 20).
151
demand. Finally, the transition from short-run to long-run market equilib-
rium should be described in full" (pp. 20- 21). In other words, the vertical
long-term Phillips Curve (i.e. the NRH) was accepted as the relevant
framework in which to model the short-term Phillips Curve. The model
which they used was a partial-equilibrium, perfect-competition model of
the labour market. Its most important aspect is the aggregate supply
function, which has also become known as the Lucas supply junction. 6
This function describes long-term as well as short-term output relations.
As Lucas and Rapping's labour demand function is a fairly straightforward
application of the marginal productivity theory, the discussion here will
concentrate on their supply function. 7
Lucas and Rapping (1969a, p. 24) formulated their theory of the
supply of labour in terms of the choice between goods and leisure, facing
a single household (or individual) in a competitive market. The house-
hold's supply of labour was assumed to depend on current nominal wages,
current prices, the present value of future real wages, the present value of
future prices, the real rate of interest, and initial asset holdings (non-
human wealth). By assuming homogeneity of degree zero in all arguments
Lucas and Rapping subsequently deflated the nominal variables by
current prices. The formulation of the household's supply function of
labour reflects that it must not only choose between current goods and
current leisure, but also between current labour supply and future labour
supply. That is, it must intertemporally allocate its labour supply.
An important step in Lucas and Rapping's (1969a) analysis was to
interpret the household as a representative household. In this manner the
household's supply of labour could be transformed into an aggregate
supply of labour. But this procedure also entails an important narrowing
of the problems which can be analyzed. By using the concept of the
representative household Lucas and Rapping implicitly assumed that all
individuals belong to the same 'class' and face similar optimization
problems (which do not differ from each other in relevant aspects). The
6 Lucas (1981a, p. 5) himself gave credit to Rapping by stating that its proper name
should have been the Lucas-Rapping supply function. In the literature on the NCB,
however, it is called the Lucas supply curve. For several versions of this curve, see Minford
and Peel (1981).
7
Lucas and Rapping's (1969a) labour demand function is based on an aggegate
production function with constant elasticity of substitution, with constant returns to scale,
with labour-augmenting technological change, and with a homogeneous real capital stock.
Additionally, it is assumed that labour is a freely variable input, i.e., there are no adjust-
ment costs in varying labour input.
152
concept is a hypostatisation in the sense that it treats aggregates and index
numbers as if they obey the principles of microeconomics. It is implicitly
defined as the mathematical mean of the group (or subsystem) as a
whole. In turn, this means that it does not allow for distributional effects.
This emphasis on aggregates disregards their composition and its changes.
The effects of redistributions between economic agents then fall outside
the scope of analysis. Additionally, Lucas and Rapping (1969a, p. 21)
assumed competitive labour markets in which a 'Walrasian auctioneer'
ensures that (partial) equilibrium will prevail. This eliminates the
coordination problem, as discussed in Chapter 4, from the domain of
economics.
The supply function which was used by Lucas and Rapping
(1969a) assumed that the aggregate labour supply depends on the current
real wage rate (w.), the anticipated real wage rate based on information
available at time t (w.), the nominal wage rate (r.), and the deflated
market value of initial non-human wealth (a.). In contrast to Friedman
(1957), Lucas and Rapping (1969a) eliminated the wealth effect from this
labour supply function, because they considered it to be empirically
negligible. 8 Fluctuations in unemployment are then caused only by
changes in prices and wages.
Following Friedman's (1957) Permanent Income Hypothesis
(PIH), Lucas and Rapping (1969a, p. 26) interpreted the (current and
expected) real wage rate as consisting of two components, namely a
permanent and a transitory one. The representative household bases its
labour-supply decision on some notion of its 'normal' real wage rate. This
rate is 'measured' by the wage-rate expectation of the representative
household, w.·. Given the fact that this household is defined as the
mathematical mean of the system as a whole, this means that the econo-
my-wide permanent or 'normal' real wage rate is identical to the expected
real wage rate. The transitory component indicates the degree in which
the current real wage rate differs from the permanent rate (w.- w.). If the
former is higher than the latter, the supply of labour will be increased,
because the households will intertemporally substitute their labour activi-
8
Lucas and Rapping (1969a, p. 50nl2) admitted that the available evidence is not
unanimous in this regard. When reflecting on his 1969a paper with Rapping, Lucas (198la,
p. 3) also gave a theoretical reason for excluding the asset effect. He argued that reduced
employment caused by wealth increases is perceived as a positive phenomenon, whereas
reduced employment in depressions are seen as negative phenomena. This means that a
fall in the supply of labour which is induced by a rise in wealth cannot constitute cyclical
unemployment.
153
ties. Of course, the reverse also applies.
The transitory component in the real wage rate and in the general
price level, and hence the cyclical changes in the supply of labour, can
only be determined if the permanent component is known. The latter is
reflected in the household's expectation, and hence an assumption about
the household's expectations formation mechanism is required. Lucas and
Rapping (1969a, pp. 27- 28) adopted the Adaptive Expectations Hypoth-
esis (AEH), which was already discussed in the previous chapter. It
enabled them to reformulate their labour supply function into an unem-
ployment-rate function, which expresses deviations from the permanent
(or 'normal') rate of unemployment in terms of deviations from the
permanent real wage rate and permanent price level. This implies some
sort of money illusion on the part of the economic agents. Mter all, a
change in the general price level would lead to a change the unemploy-
ment rate. This property follows from the assumption that the agents form
their expectations adaptively. Given the AEH, the NRH will only hold in
the long run.
The analysis of Lucas and Rapping (1969a), and more in parti-
cular their version of the AEH and the consequent money illusion, is in
danger of being inconsistent with the rationality postulate. In a second
article, which was published in 1969, they acknowledged that their version
of the AEH was based on" ... an unreasonable stubbornness on the part of
the households: if a sustained inflationary policy is pursued by the govern-
ments [i.e. if the rate of inflation continues to increase], households
following the adaptive rule will continue forever to underpredict future
prices" (p. 344, italics in original). They changed it by choosing a lag
structure which fitted their data best. As Maddock (1979, p. 198) noted,
the lag structure was thus inductively derived, and not imposed on econ-
omic-theoretical grounds.
Lucas and Rapping (1969a and 1969b) had thus formalized the
idea of Friedman (1968) and Phelps {1967) that the negatively sloped
Phillips Curve reflects short-term expectational errors. In the long run
expectations will be correct, thus making the long-term Phillips Curve
vertical. Furthermore, their empirical tests had indicated that "... statistical
Phillips curves are highly unstable over time, and this instability is far too
serious to be dismissed by a vacuous reference to structural change ... "
(1969b, p. 349). They concluded that the curves "... are a weak foundation
on which to base policy decisions" (p. 349), thus dismissing the Samuelson-
Solow suggestion that the curves reflect exploitable trade-offs. This
154
constituted a formidable attack on 'orthodox' stabilization policy. How-
ever, their results were not generally accepted.
Lucas and Rapping's analysis was criticized on several accounts.
Firstly, Rees (1970) criticized their assumption that labour markets
continuously clear. He argued that economic agents will only withdraw
their labour supply if they expect the current real wage rate to be tempor-
arily below its 'normal' level. However, he did not ascribe such views to
the unemployed during a prolonged depression, such as the Great Depres-
sion in the 1930s. He rejected the model's implication that "... unem-
ployment arises from the recalcitrance of suppliers and not from defi-
ciencies in aggregate demand" (p. 309).9 In a reply Lucas and Rapping
(1972, p. 60) argued that in their (1969b) model real GNP (and hence
aggregate demand) does influence unemployment. They acknowledged
that it does not enter in the unemployment rate function, but as this func-
tion is one of three structural equations in a three-equation system, it is
also necessary to take the other equations into account. Since unemploy-
ment is a function of real wages, and real wages are a function of real
GNP per household, unemployment is determined by real GNP. 10 How-
ever, when they concentrated on the effect of lagged unemployment on
actual unemployment, it turned out that for the period 1930 - 1945 the
persistence in unemployment model could not be fully explained by price
expectations behaviour (p. 62). Thus Lucas and Rapping's (1972) model
was confronted with an anomaly. In his 1973 test Lucas would circumvent
this anomaly by incorporating lagged income as an explanatory variable in
the aggregate supply curve. 11
The second problem which arose from Lucas and Rapping's
(1969b) paper concerns the framework in which they had conducted their
analyses. As they themselves already recognized, their partial-equilibrium
models disregarded general interdependencies in the economy. In his later
work Lucas would rationalize this neglect by adopting Phelps's (1970)
9
Rees (1970, p. 309) also pointed out that Lucas and Rapping's use of the term
'Phillips Curve' is misleading, because Phillips had assumed that the direction of causation
runs from unemployment to changes in nominal wages. Lucas and Rapping followed Fried-
man's reinterpretation and reformulation by considering the relationship between unem-
ployment and real wages, with causality running from the latter to the former.
10
See Lucas and Rapping (1969b (1981)), equations (27) and (17) respectively.
11
The test itself will be discussed in more detail in Section 8.6.
155
'islands parable'. 12
Thirdly, Lucas and Rapping had used the AEH, which (even in its
inductively improved version) remained unsatisfactory from a theoretical
point of view because it is in danger of being inconsistent with the
rationality postulate. Therefore, Lucas would substitute it by Muth's
(1961) Rational Expectations Hypothesis (REH). However, the adoption
of Phelps's parable would change the content of the REH. In time it
would lead to Lucas's critique of standard methods of econometric policy
evaluation.
8.3. THE RATIONAL EXPECTATIONS HYPOTHESIS
Presumably the best-known characteristic of New Classical analysis is the

Rational Expectations Hypothesis (REH). Muth (1961, p. 315) had advan-
ced the hypothesis that expectations are essentially the same as the
predictions of the relevant economic theory. Or, as he formulated more
exactly, "... expectations of firms (or, more generally, the subjective
probability distributions of outcomes) tend to be distributed, for the same
information set, about the prediction of the theory (or the 'objective'
probability distributions of outcomes)" (p. 316). He argued that if econ-
omic theorists could predict better than the economic agents, they must
have superior foreknowledge of the predicted outcome. But if they have,
then the question arises why some entrepreneur does not study economics
in order to sell his better predictions? After all, it would be rational for
optimizing agents to use economic theory if this would lead to (higher)
profits. By linking economic theory and economic practice, Muth took a
modest stand for economists with regard to the predictive powers of their
theories. However, if a theorist wants to obtain definite outcomes of the
assumed expectations formation mechanism (as Muth did), he must
specify a 'correct' model of the economy. Furthermore, he must make
assumptions with regard to" the content of the agents' information sets.
12 Lucas (1981a, p. 6) opined that "[t]he best thing that happened to Rapping's and my
[1969a] paper was that Edmund Phelps came across it and a number of related papers by
others at a time when he himself was working on similar problems. . .. Rapping and I had
been thinking in sectoral terms typical of at least the more econometrically oriented
macroeconomic tradition. We viewed ourselves as constructing a model of the 'wage-price
sector', potentially suitable for combining with other models of other 'sectors' to provide a
model of the entire economy. . .. Phelps, as is evident from his introductory essay to the
volume, was thinking in general-equilibrium terms ... "
156
Therefore, Muth {1961, p. 317) used ("[f]or purposes of analysis ... ") a
'specific form' of the REH in a partial-equilibrium analysis, in which be
assumed that the random disturbances are normally distributed. Individu-
als were also assumed to know ·these distributions, which means that on
average their expectations are correct. That is, the expectations of the
~
'representative individual' are correct, ~nd thus must be based on all

relevant information. Lucas (1972a) transposed the REH to a general-
equilibrium context, which lias an important implication with regard to
the individuals' information sets. This implication will be discussed in the
next section, but first we must distinguish between types of REH.
The general formulation of the REH leaves room for several
interpretations. A taxonomy may be formulated which discerns a strong
and a weak form of the REH. 13 Both forms assume different information
sets. The strong form holds that the representative agent knows at least:
(i) the 'true' structure of the model economy,

(ii) the 'true' values of the parameters in that economy, and
(iii) the past values of the relevant variables. 14
Stated differently, the strong form holds that expectations (taken separ-
ately) may be incorrect, but that they are correct in the aggregate. The
expectational errors are supposed to cancel out, presumably due to the
'Law of Large Numbers'. 15 As Muth (1961, p. 316 - 17) assumed that in
the aggregate agents do not make larger expectational errors than econ-
omic theory, his version of the REH may best be represented as a strong
form. The weak form of the REH is merely a restatement of the rational-
13 For an analysis of a more elaborated taxonomy, see Snippe (1986- 87).
14 Cf. Grossman (1980, p. 10). Runde and Torr (1985, p. 220) argued that the use of
the concept of the 'representative agent' means that possible differences between agents
are disregarded. These differences may concern (1) their abilities to gather and process
information, and/or (2) their forecast functions. New Classicals presume that agents
belong to the same class, in the sense that they do not differ from each other in any rel-
evant aspect. McCallum (1979, p. 287n4) has defended the New Classical use of the REH
by noting that "[a]rguments based on expectational differences across individual consumers
or firms amount to objections to macroeconomics, not rational expectations. "
15
Haltiwanger and ·Waldman (1989, p. 620) have shown that a situation in which
expectations are rational in the aggregate (because deviations from the average expectation
cancel out) may differ from the situation in which all agents hold rational expectations (i.e.
in which there are no deviations from the mathematical mean), depending on the nature of
interaction between economic agents. The NCE disregards such interactional effects by
restricting its analyses to the representative agent.
157
ity postulate. Agents are assumed to optimize the information on which
they base their decisions. Obviously, if information is not costless, the
optimal information need not be sufficient to lead to Muth-rational
('correct') expectations. Moreover, the weak form may suffer from an
indeterminacy, as it will be impossible to determine the optimal 'amount'
of information. Information optimization presupposes expectations about
its marginal returns and costs. However, expectations (being informed
predictions) already presuppose information. In other words, expectations
are needed in order to optimize information, while information is needed
to form expectations. This may appear to be a problem of circularity, but
it is not. It is a problem of infinite regress, because the information
needed for expectations formation is of a different kind than the informa-
tion for which the expectations are needed. Therefore optimizing informa-
tion involves infinite regress, leaving the optimality of the information
gathered by the individuals unexplained and hence undetermined. This
can also be phrased in terms of the Shackle-Boulding paradox (see
Chapter 6), which holds that any planned (rational) knowledge acquisition
process presupposes undeliberately acquired knowledge. Hence the
economics of information, as originated by Stigler (1961), fails to explain
how the marginal benefits of information can be equated with the mar-
ginal costs. In this sense the REH presupposes the existence of non-
deliberate search procedures which it cannot explain. At the same time,
and despite this drawback, it also appears to be the most sophisticated
and least ad-hoc formalization of an expectations formation mechanism.
The content of rational expectations crucially depends on the
model in which the REH is incorporated. The hypothesis cannot stand on
its own, because otherwise it cannot be asserted which expectations are
rational. It must therefore be analyzed in a model. 16 Muth's (1961) model
had been a partial-equilibrium model, and so were those of Lucas and
Rapping (1969a and 1969b). In two subsequent papers Lucas adopted a
general-equilibrium framework, and more in particular, Phelps's 'islands
16 This led Buiter (1980, p. 38) to conclude that "[t]he hypothesis appears to be in
danger of being consistent with any conceivable body of empirical evidence, because the
assumption of optimal use of the available information cannot be tested independently of
an assumption about the available information set." Buiter's irrefutability critique (as
Kamath (1989, p. 222) labeled it) amounts to the view that the REH is irrefutable, and
hence non-scientific in a Popperian sense. It presumes, however, that a single hypothesis
can be tested. This runs counter to the Duhem-Quine thesis, which holds that only
combinations of hypotheses can be tested, and that therefore crucial tests of isolated
hypotheses cannot exist. This thesis seriously weakens Buiter's critique.
158
parable', instead. 17
8.4. FROM PARTIAL TO GENERAL EQUILffiRIUM
In his 'Expectations and the neutrality of money' Lucas tried to give "... a
simple example of an economy in which equilibrium prices and quantities
exhibit what may be the central feature of the modern business cycle: a
systematic relation between the rate of change in nominal prices and the
level of real output" (1972a, p. 66)} 8 In order to discuss this Phillips-
Curve relationship he formulated a mathematical model of an economy in
which individuals do not have 'money illusion'. Moreover, prices are
assumed to be perfectly flexible and market-clearing, and individuals form
their expectations according to the REH. In this setting the fluctuations in
real output (and employment) must be attributed to the fact that individ-
uals have incomplete information. Furthermore, Lucas (1972a) assumed
that individuals are price-taking agents, which means that a single agent
cannot influence (or rather, has an infinitely small effect on) the price on
his market. This assumption rules out any price-setting behaviour, which
means that Lucas's model needed to specify a mechanism by means of
which disequilibrium prices are changed into equilibrium (market-clear-
ing) prices. He 'solved' this coordination problem by assuming that
"[w]ithin each market, trading by auction occurs, with all trades transacted
at a single, market clearing price" (p. 68). The 'Walrasian auctioneer' thus
ensured that all prices are completely flexible, and that all markets clear
continuously. The resulting general equilibrium does not imply that the
actions are ex-post optimal in some absolute sense. That is, it does not
need to be a Natural Rate Equilibrium (NRE) which is characterized by
17
It should be noted (as De Marchi (1990) did) that the articles referred to were
already written in 1970, although they have been published as late as 1972. Presumably the
first, entitled 'Expectations and the neutrality of money', was received by the Journal of
Economic Theory on September 4, 1970, whereas the second, 'Econometric testing of the
Natural Rate Hypothesis', was presented at a conference on Octo~r 30-31, 1970. Given
the years of publication, we shall refer to them as Lucas (1972a) and (1972b) respectively.
Although there is hardly any indication (except the dates mentioned) as to which paper
was written first, the order given above will be maintained. The reason for this is a logical
one. Lucas (1972a) tried to formalize Phelps's parable, using the REH, whereas in his
(1972b) he addressed some (econometric) difficulties in testing models which include this
hypothesis. As model specification preceeds empirical testing, Lucas presumably formu-
lated the model before directing his attention towards the problems inherent in testing it.
18
Page numbers refer to the 1981 reprint of Lucas (1972a).
159
perfect knowledge and perfect foresight. The actions may well prove to be
ex-post non-optimal, because in the process of acting individuals have
obtained new (better) knowledge. Such a general equilibrium concept is
called a Rational Expectations Equilibrium (REE). 19 It merely holds that,
given their information sets, individuals form (ex-ante) optimal (and
rational) expectations.
Given the REH, the content of the expectations is determined by
the information which is contained in the individuals' information sets.
Lucas (1972a, p. 68) postulated a two-period overlapping-generations
model in which the individuals belonging to each new generation are sto-
chastically distributed over two markets, with fraction 9/2 going to one
and 1-(9/2) going to the other market. The allocation variable 9 is
unknown, except indirectly via prices. After the distribution of agents over
both markets, no communication is possible between the markets. There
are three goods: labour, output, and money. Output is only produced by
the young, and it cannot be stored. Neither output nor money can be
inherited. In such a framework, exchange is only possible between the
young and the old. The former sell their output against the money which
the old have held over from the preceding period (p. 68). As in Phelps's
parable, the agents must act on the basis of incomplete global and com-
plete local inforJDation, so that information is heterogeneous across mar-
kets. As the agents are presumed to be identical (p. 68), information is
homogeneous across agents on the same market. Money (m) is assumed
to be the exogenous, primary variable. Its supply (m 1 = ·m~, where m 1
applies to period 1 and m 0 to period 0) changes at random. Like the
changes in 9, the actual changes x are also indirectly known through
changes in the general price level, although again this information will
only become available in the next period. Both 9 and x are serially inde-
pendent, which means that past deviations from their mean do not reveal
any information as regards current and future deviations. Lucas (1972a, p.
73) assumed that the agents know the 'true' probability distributions of
changes in the money supply, and of changes in the allocation of individ-
uals over both markets. The agents thus know the 'true' distributions of
monetary as well as real disturbances.2° Moreover, as the global informati-
19
According to Radner (1989, p. 317), the REE is characterized by the fact that
individals learn from price changes. That is, in the REE individual decision-makers take
account of 'the potential informational feedback of prices'.
20 This is merely another way of stating that expectations are formed rationally in the
sense of Muth (1961), as the latter's definition (which was given in section 8.3) reveals.
160
on of the previous period is known, agents also know mo- In conclusion,
"... the state of the economy in any period is entirely described by three
variables m, x, and 9" (p. 73). That is, the price p can be expressed as a
function of (m, x, 9). The dynamic behaviour of the economy as a whole
can then be obtained by describing the successive constellations of these
three 'state variables'. The economic agents form their expectations on
the known distributions of these variables. The state variables fluctuate at
random, randomly disturbing p. The disturbances may have either a
monetary or a real nature. The agents thus face a signal extraction prob-
lem: they m'!st determine how much of a change in their local price must
be attributed to real and how much to monetary disturbances. As global
information is lagged one period, the agents are bound to make expec-
tational errors. They will mistake monetarily-induced price changes for
changes which result from real disturbances. As Lucas (1972a, p. 78)
stated, "... monetary changes have real consequences only because people
cannot discriminate perfectly between real and monetary demand shifts."
Economic agents then merely behave as if they suffer from money illusi-
on. However, Lucas (1972a, p. 79) continued by arguing that such expecta-
tional errors cannot be systematic, because rational individuals will form
their expectations according to the REH. This means that the 'classical'
neutrality proposition can be maintained in a manner which explicitly takes
account of expectations.2 1 Anticipated changes in the money supply will not
induce changes in real variables, whereas unanticipated monetary changes
will merely have transitory effects on these variables. This has important
consequences for the effectiveness of monetary policy. As Lucas (1972a,
pp. 78 - 79, italics in original) stated, "[s]ince [the agents'] ability to discri-
minate [between nominal and real disturbances] should not be altered by
a proportional change in the scale of monetary policy, intuition suggests
that such scale changes could have no real consequences." Hence, system-
atic monetary policy does not influence real economic activity. Monetary
disturbances are merely capable of influencing real variables because the
economic agents do not have sufficient information to solve the signal
extraction problem adequately. In turn, this means that monetary policy
can only be successful by 'fooling' people.
By combining Friedman's NRH, Phelps's 'island parable', and
Muth's REH, Lucas (1972a) provided an example of an economy in which
21 This proposition holds that monetary disturbances merely lead to changes in

nominal (and not in real) variables.
161
the existence of a Phillips Curve is consistent with the absence of money
illusion. His model exhibited the four characteristics mentioned in the
Introduction, which became integral parts of the modelling standard of
New Classical Economics. As De Marchi (1990, p. 33) concluded, Lucas's
(1972a) paper set the modelling style. However, the content of the REH
changed in the process of combining these propositions. Muth (1961) had
used the hypothesis in a partial-equilibrium framework, which implies that
in order to form (on average) optimal expectations, (the aggregate of)
agents must know the structure of their local market. This means that they
must know how the other market participants will act. In a partial-equi-
librium context, this implicit assumption may be a plausible one: 'hog
farmers are likely to know the hog market' (and if they do not, they will
go bankrupt). In Lucas's (1972a) general equilibrium framework an analo-
gous assumption is presupposed. As was shown in section 8.3, he argued
that the average of expectations will only be incorrect if in the aggregate
agents do not anticipate a change in the supply of money. Ex-post
optimality is caused only by the incompleteness of information about the
source of disturbance. 22 Again, just as in Muth's analysis, this presupposes
that the agents know (in a probabilistic sense) what the state of the
system will be. In Lucas's (1972a) general-equilibrium analysis this system
is the economy as a whole. He thus implicitly assumed that on average
the agents know the structure of the economic system as a whole. It turns
out that Lucas's transposition of the REH from a partial-equilibrium to a
general-equilibrium context has an important implication for the assumpti-
on about the agents' information sets.
The substitution of the REH for the AEH not only changed the
content of the agents' information sets, it also invalidated a common
22
According to Boland (1986, pp. 120- 21), rational-expectations theorists assume that
all individuals use the same infallible method of inductive learning. He opined that
expectational errors may not only be attributed to inadequacies of the information sets but
also to inadequate methods of learning. However, one may defend the REH by arguing
that both sources of expectational errors are 'observationally equivalent'. That is, mere
observation cannot distinguish between expectational errors caused by inadequate methods
of learning, and errors due to inadequate information sets. The point here is that any
mathematical formalization, including that of expectational errors, presupposes abstraction
and simplification. The RE-theorists do not claim to explain why economic agents make
expectational errors (which can be said to belong to the realm of psychology), but instead
try to formalize such errors, irrespective of their source.
162
econometric technique of evaluating economic policy. 23 This invalidation
has been called the 'Lucas critique'.
8.5. THE LUCAS CRITIQUE
In 1970 Lucas presented a conference paper which was published in 1972

as 'Econometric testing of the Natural Rate Hypothesis' (1972b}. In this
paper Lucas observed that the NRH is challenged by observations that a
trade-off between changes in the inflation rate and changes in the level of
real output exists. The question is whether the NRH can be formulated in
a testable form, and if it can how should the test be conducted?
Lucas (1972b) claimed that the answer to the first part of this
question is affirmative. After showing the possible inconsistency of the
AEH with the rationality postulate (see Chapter 7), he combined the
NRH with the REH, which led to the joint NR/RE hypothesis (p. 93}. 24
Lucas subsequently added an aggregate demand schedule, which he inter-
preted as a policy rule, that is, as "... a (possibly randomized} rule giving
the current value of xt as a function of the state of the system" (p. 96,
italics in original)}. The solution of this model expressed real GNP (yt) in
terms of present and lagged nominal GNP (xt and xt_ 1 respectively) and a
random error term (17t). Lucas showed that the coefficient in this equation
depends on the parameters in the policy rule. This raises a problem as
regards the appropriate way of testing this type of models. Lucas (1972b,
p. 99) opined that the common methods of testing would take the sum of
the coefficients of nominal GNP as a measure of the long-run effect of a
once-and-for-all (permanent) demand shift. Suppose that this shift is
caused by nominal factors, such as a change in the rate of inflation. The
joint NR/RE hypothesis then predicts that real GNP will not change,
because economic agents do not suffer from money illusion. This means
that the sum of the coefficients of xt and xt_ 1 must be zero. According to
Lucas, this test has been the 'standard' test of the NRH. He argued that
the REH renders this test inappropriate, because the coefficients of xt and
23
Lucas (in Klamer (1985, p. 38)) remarked that in the 1960s he already knew the
REH. However, he "... dido 't understand then how fundamental a difference it made
econometrically. I didn't realize that if you took it seriously you had to rethink the whole
question of testing and estimation. I guess no one else did either, except for Muth."
Sargent (in Klamer (1985, p. 61)) concurred with this view.
24
Page numbers refer to the 1981 reprint of Lucas (1972b).
163
xt_ 1 contain the parameters of the policy rule. This means that a change in
monetary policy will alter these coefficients, presumably changing their
sum. As a result, the new policy "... cannot be evaluated by simply sum-
ming parameters implied by some previous, now irrelevant policy" (p. 99).
That is, in an economic system in which economic agents form their
expectations according to the REH, an anticipated change in monetary
policy will immediately be incorporated in the (rational) expectations.
This means that the agents will adapt their actions accordingly, thus
changing the parameters of the behaviourial equations (and hence the
structure of the economy). Any method of testing the joint NR/RE
hypothesis which takes the parameters in the model as invariable is
invalid because of the so-called 'cross-equational restrictions'. Lucas
(1981a, p. 9) concluded that "[i]f the theory of 'Expectations and the Neu-
trality of Money' was the correct way to formulate the Friedman-Phelps
natural-rate hypothesis, then it was evident that the econometric methods
then being applied to test this hypothesis were entirely missing the point."
This criticism of the econometric method under consideration is called the
Lucas critique.25 Lucas (1972b, pp. 99 - 100) subsequently outlined an
alternative test procedure.
Sargent (1971) also invalidated a common method of testing the
NRH (or as he called it the 'accelerationist' thesis of Phelps and
Friedman). Empirical tests of the NRH had assumed that economic
agents form their price expectations adaptively. The price expectation
then is a weighed average of past relative changes in the general price
level. Substitution of this price-expectation function into a Phillips-Curve
relationship yields an equation which expresses the present relative
change in nominal wage rate (awtfwt_ 1) in terms of past inflation rates
(M't_JPt-i- 1), and some other variables (p. 33).26 The NRH is supposed to
25 The Lucas critique can be interpreted as an instance of Goodhart's Law, which in its
broadest sense holds that 'social laws' are subject to change if one attempts to exploit
them. Lucas (1981a, p. 10) observed that Sargent (1971) had already achieved the objective
of making clear that 'orthodox' distributed-lag tests could not be used to test the NRH
(the so-called 'Lucas critique'). He added that he himself " ...did not know this at the time
.... "In private correspondence, Sargent recalled that his (1971) paper was written without
any knowledge about Lucas's paper. Lucas and Sargent thus discovered the 'Lucas critique'
simultaneously, independently, and along different lines.
26
Sargent formulated this function as follows:
164
be corroborated if the coefficient of the past inflation rates, a, and hence
the public's inflation expectation, is close to unity, whereas it is considered
to be discorroborated if it is closer to zero. Estimation of the parameters
is possible only if some restriction is added. According to Sargent, "[a]l-
most always, the constraint that has been imposed is that the distributed
lags in [the AEH] sum to unity" (p. 34). This constraint is justified by the
argument that eventually economic agents will fully incorporate an unex-
pected change in the rate of inflation into their price-expectations. Under
the AEH this will only hold if the new inflation rate remains unaltered for
quite some time. Sargent claimed that in reality this is never the case.
Instead, it is "... most appropriate to ask what sort of expectations-genera-
ting scheme would be reasonable in the light of the actual behavior of the
rate of inflation during the period being studied" (pp. 34 - 35). Rational
economic agents will adopt an expectation formation mechanism which
minimizes their expectational errors. The most reasonable restriction
which can be imposed on the sum of the distributed lags is then the one
which is compatible with the observed behaviour of the rate of inflation.
Sargent (1971, pp. 35 - 36) subsequently showed that the appropriate
restriction is that this sum is less than unity. 27 In turn, this means that the
unity-restriction leads to underestimation of a, so that "... those estimates
tell us virtually nothing about the validity of the accelerationist thesis" (p.
37).
To summarize, Lucas ( 1972b) tried to answer two questions. The first

question asked whether the combination of the NRH and REH (the
NR/RE hypothesis) could be formulated in a testable form. The answer
proved to be affirmative. The testable form turned out to pose a problem
for orthodox policy-evaluation techniques. The Lucas critique showed that
these techniques could not be used for rational-expectations models.
Sargent (1971) also reached this conclusion, albeit along somewhat
different lines. Both argued that changes in policy parameters influence
behavioural parameters. Economic agents will take changes in the policy
where w,
is the nominal wage rate in period t, P, is the general price level in t, u, is the
unemployment rate in that period, f(U,, ... )is the short-run Phillips curve with Of/ OU < 0
and with the sequence of dots representing a list of other variables, and l1 is an unobser-
vable random variable.
27
He argued that if the unity-restriction would be appropriate, then the actual rate of
inflation "... would display extremely strong serial correlation or 'drift'" (p. 36). As this is
not the case, he concluded that the distributed lags sum to less than unity.
165
rule into account by changing their actions. Lucas (1972b, p. 98)
subsequently asked "[h]ow (if at all) can models of this [NRH/REH] class
be tested?" He suggested a test procedure which would take the relation-
ship between policy parameters and behaviourial parameters into account,
but he did not carry out such a test. Instead, his 1973 attempt to test the
NR/RE hypothesis evaded the problems implied by policy changes.
8.6. LUCAS'S PARTIAL TEST OF THE NRH
Lucas (1973) undertook an attempt to test his version of the NR/RE

hypothesis. He divided aggregate supply into a normal (secular) and a
cyclical component, Ynt and Yet respectively. The former follows a trend
which is identical for all markets z. The cyclical component, which consists
of deviations from this trend, differs between markets. Contrary to Lucas's
previous models, the equation describing this component contained devia-
tions in lagged income as an explanatory variable. This inclusion was a
reply to Rees's (1970) criticism that the model of Lucas and Rapping
(1969b) could not account for persistence. Lucas (1973) thus assumed that
expectational errors have a drawn-out effect on aggregate output, which
accounts for persistence in the economic time series under consideration.
As he had done in his 1972-models, Lucas (1973) explained the
Phillips Curves in terms of expectational errors as regards the nature of
changes in local prices. He assumed again that economic agents know all
past deviations of aggregate supply from its trend value, and all past
demand shifts (p. 134).28 They can use this information to construct a
'prior' distribution of the general, economy-wide price level Pu which was
assumed to be normal, with mean~ and variance az.
The deviations of
the actual local price Pt(z) from Pt are also normally distributed with zero
mean and variance -??9 Rational economic agents will use all available
information in their expectations formation process, hence price-expecta-
tions will depend on both Pt(z) and Pt. Lucas added a function in which
changes in aggregate demand (xt) were due to either real (yt) or nominal
(Pt) factors. The process which describes the changes in xt consists of two
components, namely its mean ~ and a random disturbance variable ut. The
28
Page numbers refer to the 1981 reprint of Lucas (1973).
29
The distributions J and .fl are assumed to be independent from each other.
166
test equation which Lucas (1973, p. 136) subsequently derived is:
(8.1) Ya = -n6 + n~x, + A.yc,t-l
where Tis a measure of the response of real output to changes in aggre-

gate demand. The first term of the right-hand side of this equation
reflects the influence of the average change in x, on deviations of aggre-
gate supply from its trend. As the variable u, is a random variable, this
average change is also the anticipated change in aggregate demand. The
second term on the right-hand side of (8.1) gives the actual change in
aggregate demand, with the same parameter but with opposite sign. The
equation thus implies that anticipated changes in x, do not have any effect
on the deviation of real output from its trend, whereas unanticipated
changes in aggregate demand will affect the latter with magnitude T.
Lucas (1973, p. 137) conjectured that economic agents will be more easily
surprised if changes in the nominal component of x, are less frequent.
Economic policy can then more easily 'fool' people. This means that the
'tradeoff between inflation and output is more favourable, and that the
Phillips Curve is flatter. He thus argued that the variance of x, was
inversely related to the slope parameter 1r.
Lucas's test over the period 1952 - 1967 concentrated on eighteen
countries. Two types of nominal income behaviour could be distinguished:
Argentina and Paraguay had experienced highly volatile and expansive
policies, whereas the policies of the remaining sixteen countries had been
relatively smooth and moderately expansive. It turned out that the vari-
ances of nominal GNP of the former were at least ten times those of the
other countries, whereas their 1r's were smaller (by a factor of ten) than
those of the latter. His conjecture thus was corroborated, although this
result depended heavily on the two extreme Latin-American cases.
Lucas's (1973) test can be criticized on two accounts. Firstly, as
De Marchi (1990) pointed out, the test did not contain all the knowledge
derived in Lucas's own process of discovery. In particular, Lucas did not
consider his own critique on conventional econometric testing, although in
his 1972b-article he had already proposed a solution to the ensuing prob-
lem of testing the NRH. This means that the test was only a 'partial' test
of the knowledge which was available at the time. However, this is not to
say that the Lucas's critique did not play any role in Lucas's (1973) test.
In fact, the awareness of it made clear that the problems which were
implicit in testing the joint NR/RE hypothesis had to be avoided. Lucas
accomplished this by using international data while at the same time
167
assuming changes in the policy rules to be absent. This enabled him to
study the effect of different policy regimes without having to address the
Lucas critique.
A second criticism which may be levied against the test concerns
the fact that Lucas included past deviations from the natural rate (of
output) as an explanatory variable for the persistence in the temporary
(i.e. cyclical) component of aggregate supply. Such an inclusion was not
founded on economic theory.
To conclude, Lucas's test had evaded his own critique on econometric

policy evaluation by concentrating on inter-country differences in the
slope of their Phillips Curves and in the variability of nominal GNP.
Furthermore, he used a model in which persistence was only explained by
including lagged deviations of output from its trend. In contrast, Sargent's
(1973a) and (1973b) tests took (Sargent's own version of) the Lucas cri-
tique into account. He also showed that the rate of interest could be
incorporated in the New Classical framework. Before discussing his tests,
a short detour will be made in order to outline their 'prehistory'.
8.7. SARGENT'S TESTS OF THE NRH
8.7.1. Fisher's solution to the Gibson paradox

Sargent's work in the late 1960s and early 1970s was concerned with the
relationship between the general price level and the (nominal and real)
rate of interest. In his 'Commodity price expectations and the interest
rate' (1969) he addressed what Keynes (1930 (1965), pp. 198 - 208) had
called the 'Gibson paradox'. 30 'Classical' theory implied that a rise (fall) in
the interest rate would produce deflationary (inflationary) pressures
because of the gap between desired savings and investment, and the
ensuing fall (rise) in effective demand. A.H. Gibson found that the data
did not correspond to this pattern. 31 He showed that in reality the interest
rate and the general price level tend to move together. Classical theory
was therefore confronted with an anomaly. Presumably the best known
°
3 For a list of important articles and books on this paradox, see Sargent (1972, p.
212n1).
31 In particular, see Gibson (1923).
168
solutions are those of Keynes (1930 (1965)) and Fisher (1930 (1961). 32
Sargent (1969) concentrated on the latter's suggestion to distinguish
between the nominal and the real interest rate. In particular, Fisher had
stated that in equilibrium the former equals the latter plus the rate of
inflation (p. 43). 33 In his view, causality ran from the rate of inflation to
the rates of interest (pp. 36 - 37).34 In a situation of perfect foresight, a
change in the rate of inflation will be anticipated. In response, rational
economic agents will alter their actions in such a way as to leave the real
rate of interest unaffected. However, Fisher observed that economic
agents do not have perfect foresight. The ensuing unanticipated change in
the rate of inflation will not be translated immediately into a change in
the nominal interest rate. There will be a transition period during which
the real rate of interest will be 'distorted'. Fisher explained this incom-
plete adjustment by arguing that "... between price changes and interest
rates a third factor intervenes. This is business, as exemplified or
measured by the volume of trade. It is influenced by price change and
influences in turn the rate of interest" (p. 429). Rising prices and lagging
remunerations for the factors of production cause producers to earn larger
profits. This will induce them to invest more, thus increasing the demand
for credit. In turn, this leads to a rise in the nominal rate of interest
(1925, p. 180). For reasons of convenience, this effect will be called
32 Keynes (1930 (1965)) adopted the Wicksellian distinction between the 'natural' (r )
0
and the market rate of interest (rm)· His explanation can be stated in six propositions: (1)
rm is very sticky in comparison to r0 • This means that it cannot maintain equilibrium
between saving (S) and investment (I) (p. 203). (2) r0 exhibits long-term movements
(extending over decades) because of the fact that the annual increase in the capital stock is
relatively small with respect to this stock (p. 204). (3) If these movements are upward, then
rm < r0 for quite some time. In this case I > S. If r0 is falling, then I < S (p. 204). (4) If I
> S, then the price level will fall, and vice versa (see Keynes's 'Fundamental Equations'
(viii.) and (x.) (p. 138). This phenomenon ".. .is a slight, long-continued drag in a particular
direction," and does not cause a credit cycle (p. 204). A fall (rise) in r0 thus has two
effects. Firstly, it eventually induces a fall (rise) in rm. Secondly, and more importantly, it
will ensure that I<(>) S. In tum, this induces a fall (rise) in the general price level. The
interest rates and this price level thus move together. Propositions (1) and (2) amount to
what may be called proposition (2a), which holds that the market rate lags behind the
natural rate. This view was already held by Wicksell, who considered it to be the cause of
the business cycle (see Chapter 2). For a more detailed discussion of Keynes's and other
solutions of the paradox, see Visser (1980, pp. 151 -54).
33
Page numbers refer to the 1961 reprint of Fisher (1930).
34
As will be shown later, Fisher also allowed for the reverse causal relationship.
169
Fisher's indirect effect. 35
Fisher (1925, p. 184) and (1930, p. 419) calculated the length of
the transition period, describing the adjustment process as a distributed
lag function. He found that the lag weights of the distributed lag function
declined slowly, and that the transition period was therefore rather long
(1930, p. 427). His estimations suggested that it could take ten to thirty
years before the effects were fully incorporated in the nominal interest
rate (p. 438).
8. 7 .2. The implausible length of the transition period

Sargent (1969) interpreted Fisher's distributed lag function as an expecta-
tions formation mechanism. Given the calculated length of the transition
period, this interpretation implies that it would take ten to thirty years
before economic agents correctly anticipate the new rate of inflation and
act accordingly. Sargent concurred with Cagan (1965, p. 257) that this is
very implausible. He recalculated the length of the transition period, using
an equation in which the nominal rate of interest rt is explained in terms
of (changes in) aggregate output xu relative changes in the real money
supply mt·, the rate of inflation Pu and a random disturbance term ut. This
recalculation yielded similar lengths of lags as found by Fisher. Sargent
explained these lengths in terms of what he called the extrapolative effect,
which holds that an increase (decrease) in the general price level induces
economic agents to expect prices to rise (fall) still further. This hampers
the adjustment process, thus lengthening the transition period. Additional-
ly he derived another result from his estimates. The parameters of the
equation which was tested indicated that there was also a shorter-term
regressive component in the process of expectations formation. This
means that economic agents expected a rise (fall) in the rate of inflation
35 This effect should not be confused with Wicksell's indirect mechanism, which holds
that an expansion of the money supply will lead to an increase in the general price level,
via the (market or nominal) rate of interest. This means that causality runs from the
interest rate to the price level. Fisher's indirect effect relates the rate of inflation (and
hence the change in the price level) to the nominal rate of interest via effective demand.
Causality thus runs from (changes in) the price level to the interest rate. Fisher (1930
(1961), p. 443n21) recognized that he was not the first to discover this indirect effect. He
stated that "Prof. Knut Wicksell was one of the first to recognize the influence of interest
rates upon prices. . .. Prof. Alfred Marshall, Prof. Gustav Cassel, Rt. Hon. Reginald
McKenna, Chairman of the Midland Bank of London, Mr. R.G. Hawtrey, of the Treasury
of Great Britain, and many other well known economists, bankers, and business men have
emphasized that business activity is influenced and may be largely controlled by manipula-
tion of the discount rate" (p. 443n21).
170
to be followed by a fall (rise) in this rate. Sargent (1969, p. 138) argued
that such expectations were rational because of the cyclical properties
II •••
of price movements over the period under consideration."

It appears that Sargent was not satisfied with the results of his
1969-explanation of the length of the lags in Fisher's expectations forma-
tion mechanism. In his 'Anticipated inflation and the nominal rate of
interest' (1972) he criticized the mathematical model which Fisher had
used.36 He claimed that in his empirical work Fisher had implicitly
assumed that the anticipated rate of inflation does not affect the real rate
of interest, not even in the transition period (pp. 212 - 13). This assump-
tion can be formalized as:
(8.2) r, "" p, + 1t 1
(8.3) p, "" a + e:,
where rt is the nominal rate of interest, Pt is the real rate of interest, '~"t is
the anticipated rate of inflation, a is a constant, and Et is a stochastic term
which is uncorrelated with the nominal rate of interest. The second
equation implies that a change in the anticipated rate of inflation will
affect only the nominal rate of interest, leaving the real rate unchan-
ged.37 Sargent doubted the plausibility of this relationship and tried to
establish whether the equations given above "... can in general be taken to
characterize correctly the relationship between anticipated inflation and
the nominal rate of interest" (p. 213). In disequilibrium (or transition)
periods the actual rate of inflation is not fully reflected in the nominal
rate of interest. This implies that the real rate of interest must also have
changed (systematically), and hence that it cannot be described by equa-
tion (8.3). Sargent concluded that the above equations should be replaced
by one or more other equations in order to avoid the implausible explana-
tion of the extremely long lags in terms of the process of expectations
36
It should be noted that Sargent presumably wrote his 1972-article in 1970 as an
attempt "... to formalize for students the relationships among the various hypotheses
advanced in Milton Friedman's AEA presidential address (1968)" (1987, p. 117n1).
37
It should be stressed that at least in his theoretical work Fisher already pointed out
that the second equation does not hold in the short run, i.e., during the transition period
(which can be very long, as was already shown). Hence it is misleading to label this
equation "Fisher's formula" because this suggests that Fisher not only considered it to be
valid in equilibrium, but also during the transition periods.
171
formation. The question then arises as to the correct way of modelling the
relationship between the nominal rate of interest and the anticipated rate
of inflation. This question was addressed in Sargent's 'Interest rates and
prices in the long run' (1973a).
8.7.3. Muth-rationality as a criterion for plausibility

Sargent (1973a) started his analysis by performing a test of the length of
the transition period, based on the equations (8.2) and (8.3). He com-
bined them with the AEH, which yielded equation (8.4):
~ .llP,.
(8.4) r, = ex + y L..., ,_, ___, + e,
i=O pt-i-1
where '• is the nominal rate of interest, a is a constant, the sum is the
adaptively formed expected rate of inflation, and e. is a stochastic term
which is uncorrelated with the nominal rate of interest. Sargent's esti-
mated this equation for the U.S. in the period 1870-1940.38 The resulting
estimates of the 'decay parameter' >.. were close to unity. He concluded
that "[t]hese estimates corroborate the main outlines of Fisher's findings"
(p. 392). But these findings did not convince him that the lags were really
very long. He maintained that "[w]hile Fisher's explanation of that [i.e. the
Gibson] paradox formally 'works', the implied lags in forming expectations
do seem extraordinarily long." This implies that plausibility enters as a cri-
terion for the acceptance (or rejection) of empirical results and the
underlying propositions. It is interesting to note that in his previous
articles Sargent had not given any criterion for plausibility. This suggests
that his rejection of Fisher's empirical estimates rested on a priori beliefs.
However, in 1973 he acknowledged that "[t]o say that the estimates of the
weights obtained by that [i.e. Fisher's adaptive-expectations formation]
procedure are implausible apparently means that they do not resemble
the weights that really characterize the process by which people seem to
form expectations about future rates of inflation" (p. 392). Fisher's
estimates can only be considered implausible if superior (extraneous)
38 The estimates are derived by a search procedure. As Sargent (1973a, p. 391n5)

explained, "[o]ur procedure here was first to search over A.'s ranging from .1 to .9 at steps
of .1. Having found the value of X, say ~. that, among these nine values of X, delivered the
smallest residual variance, we then searched again over [~ - .09, ~ + .09] at steps of .01
for the ~ associated with the minimum residual variance. This value was taken as our
estimate of X."
172
knowledge about the 'true' expectations formation mechanism is available.
The problem is how to obtain such knowledge. According to Sargent
(1973a), Muth (1961) had provided a source of information about the
length of the lags by arguing that the forecasts of the economic agents will
be identical (in a probabilistic sense) to those of statistical and economic
theory. Sargent proposed to use these rational expectations as a "...
yardstick against which we will judge the 'plausibility' of the expectations
implied by Fisher" (p. 393). This means that he implicitly accepted the
proposition that the REH adequately describes "... the process by which
people seem to form expectations about future rates of inflation." His
subsequent tests yielded some indications that Fisher's expectations
formation mechanism was not in accordance with the REH, and hence
that his explanation was not 'plausible' (in the sense of Muth-rational. 39
Sargent (1973a, p. 402) concluded that the ability of Fisher's equation
(8.3) to explain the 'Gibson paradox' was seriously weakened, because it is
difficult both to accept this explanation and to maintain that the extraordi-
nary long lags in expectations estimated by Fisher are (Muth-)rational.
This conclusion raised the problem of providing a Muth-rational explana-
tion of the paradox.
8. 7 .4. The inclusion of omitted variables

Sargent (1973a) suggested the following solution to the problem he
himself had posed. He argued that in his empirical work Fisher had only
taken into account a one-way causal relationship between the rate of
inflation and the nominal rate of interest, with causality running from the
former to the latter. In this sense the inflation rate appears to be
exogenous. Sargent argued that this is unduly restrictive, and that it would
be more appropriate to transform the interest rate into an endogenous
variable. He suggested to test Fisher's model for feedback from the
nominal rate of interest to the rate of inflation. This model can be repre-
sented as a special case of (8.5):
1-a(L) -b(L) I r, e,
(8.5)
1 -c(L) 1-d(L) P, u,
where a(L ), b(L ), c(L ), and d(L) are one-sided polynomials in the lag
operator L, and where ut and et are mutually independent white noises (p.
39
For a description of these tests, the reader is referred to Sargent (1973a).
173
405). If there is no feedback from rt to p.., then c(L) must be zero. Refer-
ring to an early version of Sims (1972), Sargent argued that this model
could be tested by analyzing equation (8.6):
'"2
(8.6) rt = 'E
j=-ml
hi Pt-i + vt
where m 1 and m 2 are positive parameters, the h/s are the estimated
distributed lag parameters, and vt is a statistical residual. Adopting the
Granger-Sims notion of causality, the existence of a feedback from the
nominal interest rate to the rate of inflation means that future values of
the latter are correlated with current values of the former. 40 Hence Sar-
gent tested the hypothesis that hj = 0 for all j < 0. The test results
suggested that "... an explanation of the interest-inflation relationship that
does not permit feedback from interest to inflation is probably unduly
restrictive" (p. 422).41 It thus appeared that feedback should be allowed
for. However, if in equation (8.5) c(L) does not equal zero, some
interpretational problems emerge, because "... it will no longer be 'ratio-
nal' to form expectations of inflation by looking at current and lagged
rates of inflation alone, since current and past rates of interest are of
some help in predicting subsequent rates of inflation" (p. 427). Therefore,
Sargent proposed another interpretation of his test results. This interpre-
tation holds that what appears to be feedback from interest to inflation is
caused by some omitted variables which influence both r and p (p. 427).
Sargent then built a model in which the variable 'aggregate demand'
(represented by 'changes in the money supply') was included, and in which
both the interest rate and the rate of inflation were endogenous.
Sargent's (1973a) model described a closed economy with one
good which was produced according to a linear-homogeneous production
function in both labour and capital. The nominal rate of interest r and the
rate of inflation p are mutually determined. Sargent generated artificial
(annual) data which indicated that the nominal rate of interest r moved in
the same direction as the rate of inflation p. That is, the generated data
reflected the Gibson paradox. The values of the regression coefficients of
40 For a brief discussion of Granger-Sims 'causality' ,see Chapter 6.
41
Sargent (1973a, pp. 425- 26) recognized that Fisher had argued that causality need
not always run from the rate of inflation to the interest rate, but that the reverse influence
may also occur.
174
r on future and past values of p were similar to the estimates which
Sargent had derived from historical data.42 However, the anticipated rate
of inflation ?r was assumed to be constant (pp. 438 - 39), which reflects
the fact that Sargent was more interested in the 'long-term' relationship
between the nominal rate of interest and the rate of inflation than in the
'short-term' adjustment process.43 The assumption implied that the long
mean lags between the nominal rate of interest and the rate of inflation
cannot be explained in terms of long lags in the process of expectations
adjustment. Sargent conjectured instead that "[ t)he key reason that the
Gibson paradox may infest the data generated by the model is the failure
of wages and prices to adjust sufficiently quickly to keep output always at
its full-employment level" (p. 442). In his view, "... it does not seem
necessary to stress differences between nominal and real rates of return in
order to explain the Gibson paradox" (p. 442). This conclusion implies
that real rates of return (or interest) would behave similar to nominal
rates. To determine whether this implication was empirically valid,
Sargent regressed two measures of the real return on equities on whole-
sale commodity price inflation for the period 1871 - 1929 (pp. 444 - 45).
His results corroborated his prediction that movements in the anticipated
rate of inflation cannot explain the relationship between the nominal rate
of interest and the actual rate of inflation (that is, the Gibson paradox).
He concluded that the explanation of this paradox should focus on the
relationship between movements in real rates of return and the price
level. However, this relationship is somewhat problematical, because it
relates a real variable with a nominal variable. This implies that rational
economic agents suffer from money illusion. This is a similar problem to
the one Lucas had faced concerning the Phillips Curve. Lucas's inter-
pretation of this relationship as a consequence of incomplete knowledge
and incomplete foresight also proved useful for Sargent in his attempt to
explain the Gibson paradox. This paradox was also interpreted as a short-
run phenomenon which arises because of the fact that economic agents
42
There was only one exception, namely the coefficient on the current value of p,
which in the simulations proved to have become larger.
43
When showing the way in which his model works, Sargent ( 1973a, p. 435) had noted
that •[a]ssuming that the system is dynamically stable, the final resting place for all
variables will be the same as if 1r had remained at its steady-state value throughout the
adjustment process; but the path to steady-state equilibrium may be much different. • The
assumption that 1r remains unchanged thus means that the path to the final resting place is
not considered to be relevant. In tum, this implies that Sargent was only interested in the
'final resting place' of his model.
175
make expectational errors.44 In the long run, Sargent (1987, p. 117n1)
argued, Friedman's NRH and Fisher's two-equations model are two sides
of the same coin.
8. 7.5. Sargent's tests

It is interesting to note that Sargent (1969), (1972) and (1973a) did not
refer to Lucas's work. This indicates that the latter's formalization of the
NR/RE hypothesis did not exert any influence on Sargent's pre-1973
work.45 Sargent's 'Rational expectations, the real rate of interest, and the
natural rate of unemployment' (1973b) connected his work to that of
Lucas, as "[t]he argument in this paper is heavily dependent on the analy-
sis of the natural rate hypothesis carried out by Lucas in a series of
papers.'146 This suggests that as late as 1973 Lucas's work had a sudden
and substantial impact on Sargent, presumably because the latter's work
had reached a stage in which Lucas's contributions proved useful.
Sargent (1973b) started from his 1973a-conclusion that Fisher's
'long-run' framework and Friedman's NRH amounted to the same thing.
However, the former's tests had been inappropriate because they implied
that the lengths of the lags in the process of expectations adjustment were
implausibly long. This raised the question of how the NRH could be
appropriately tested. Sargent (1973b) attempted to outline the proper
ways in which such a test could be executed. Additionally, he performed
two such tests. As the explanation of the inappropriateness of Fisher's
tests was already discussed in Sargent (1973a), we shall concentrate on the
44 The substitution of the AEH by the REH played an important role in this regard.
As Sargent (1987, p. 117n1) observed, "[w]hileworking with adaptive expectations ... one
obtains only weak or 'long-run' versions of Friedman's hypotheses. Switching to rational
expectations (or 'perfect foresight') leads to much more precise and more immediate
versions of Friedman's hypotheses. This led Neil Wallace and me to produce Sargent and
Wallace (1975) and Sargent (1973b)."
45 This also becomes clear from Klamer's (1985) respective discussions with Lucas,
Rapping and Sargent. Lucas argued that Sargent and he "... didn't talk very much during
the two years he [Sargent] was there" (p. 33). Rapping confirmed that Sargent "... did not
pay much attention to what Bob [Lucas] and I were doing. He did not talk with Lucas
much" (p. 225). Sargent himself stated that when he met Lucas at Carnegie-Mellon, he
didn't know what Lucas was up to, and he didn't completely understand his work. In fact,
he ".. .learned from Lucas mostly by reading his stuff" (p. 60).
46 Sargent (1973b (1981), p. 162) thereby referred to Lucas and Rapping (1969a), and
Lucas (1972a), (1972b), (1973), and (1976). The page numbers of Sargent (1973b) refer to
the 1981 reprint.
176
ways in which he proposed to test the NRH.
Sargent (1973b) assumed the following equation:
q
(8.7) Un, = p(p, - Ep,l6,_ 1) + L l;Unt-i + u,, p< 0
i=l
where Unt is the unemployment rate (which serves as a reverse index for
the natural logarithm of real output Yt minus a constant k), Pt is the
natural logarithm of the price level, Ept is the expected value of Pt formed
at time t-1, (Jt is the information set at time t, A is a parameter, q is the
period of relevant lagged variables, and ut is a normally distributed
random disturbance term. The forecast of this term cannot be improved
by including components of the information set which is available at the
time. This means that ut obeys E(utlfJt-h ut-h ut-2, ... ) = E(utlut-h ut-2, ... ).
Sargent (1973b, pp. 175 - 76) assumed that ut can be described by the
following process:
(8.8) IYI < 1
where ~ut is a normally distributed and serially uncorrelated random

variable. Using (8.7) and (8.8) Sargent derived equation (8.9):
q
(8.9) Un, = (J.. 1 + y)Un,_ 1 + L (J..; - y l;_ 1)Unt-i - y l 9 Un,_9 _1 +
i=2
+ p(p, - Ep,j6,_t) - py(p,_t - Ep,-116,_2) + ~~~~
The expectation of this variable conditional on all past rates of unem-

ployment and all other information available at (Jt_2 gave the equation
from which Sargent derived his tests (p. 176):
q
(8.10) E(Un,IUnt-1' ... , Unt-q-1' elt-2 = 0·; + yli-1)Unt-i + L (l; - y.
i=2
- y lqUnt-q- 1 - y pE[(p,_ 1 - Ep,
where (Jh is a subset of the information set (Jt· Based on this equation,
Sargent proposed two tests. The first proposal concerned the case in
which ut is serially correlated, and follows the following nth order process:
177
It
(8.11) ut = LY!lt-i + ~~~~

i=l
where ~ has the same properties as in equation (8.8). If the expectation of

(8.10) is taken conditional on past values of the rate of unemployment
and the subset of information 8h-h then equation (8.11) indicates that the
lagged error in the price expectations remains. Given the fact that ut is
lagged n periods, this means that the conditional expectation of Unt can
be written as:
(8.12) E(Untl Unt-l' Unt_ 2 , ••• , Unt-n-q' 61t-~t-l)
= E(UntiUnt-1' Unt_ 2, .•• , Unt-n-q)
This equation implies that the forecast of Unt based upon past observa-
tions of this unemployment rate cannot be improved by including compo-
nents of the information subset 8h-I· These components will then have
coefficients which do not differ significantly from zero when they are
added to a regression of Unt upon lagged values of itself. This proposition
can be tested. Sargent (1973b, p. 176) added that "[t]he higher the order
of serial correlation in the u's, the more periods components of 8t must be
lagged to warrant the implication that their coefficients are zero."
The second test proposal of Sargent (1973b) concerned the case
in which ut is not serially correlated, and hence in which the A.'s are zero.
This means that the economic agents cannot improve their forecasts by
including more information. That is, the expected value of Unt formed on
any subset 8h-I of the full information set 8t-I equals zero. Mathematically,
this can be formulated as E(Unt j8h_1) = f3(pt - Ept IBtl- I) + ut = 0. This
equation can then be tested empirically by regressing the rate of unem-
ployment upon components of the information subset. Sargent in fact
performed this test twice. He first regressed the rate of unemployment
against its own lagged values and a subset 81t_ 1 which consisted of the
lagged price level {pt-h Pt-2, Pt-3 and Pt4 ) and the lagged nominal wage rate
(wt-h wt_2, wt_3 and wt4 ). Testing the null hypothesis that the coefficients on
these lagged variables do not differ significantly from zero, he found that
the regression corroborated this hypothesis at the 95% confidence level.
However, his second regression which included a larger subset of 8t led to
178
a quite different conclusion.47 Sargent again tested the null hypothesis
that the coefficients of the components of the enlarged subset under
consideration have zero coefficients, but this time he found that the null
hypothesis must be rejected at the 99% confidence level. The Natural
Rate Hypothesis was thus discorroborated. However, Sargent cautioned
not to reject the NRH too hastily.
Sargent (1973b) mentioned four circumstances in which a rejec-
tion of the NRH would be premature. Firstly, his test had assumed that
there is no serial correlation in the u's. If in fact such serial correlation
exists, then the test is biased towards rejection of the null hypothesis.
Secondly, he argued that individuals may rationally form expectations on
the basis of a smaller subset than the one used in the second test. He
stated that in that case "... the essence of the natural rate hypothesis could
stand unrefuted even though tests using large subsets 8tt_1 find systematic
effects of 8tt_1 on Un.." He thereby referred to the first test, which had
considered a smaller subset and which did not allow for the rejection of
the null hypothesis. However, this defense is confronted with the problem
that it does not explain why rational economic agents would not use infor-
mation which can improve their expectational performance. The third
defense of the NRH against the failure of the test to corroborate this
hypothesis held that the u's may have been correlated with components of
8tt_ 1• Such a correlation means that the coefficients of the components will
be biased upwards, and hence that the regression will be biased towards
rejection of the null hypothesis. Finally, Sargent took a position which
resembles 'sophisticated falsificationism'. He argued that "... it has not
been shown that an autoregression for unemployment yields ex ante
predictions of unemployment inferior to those of a particular structural
macroeconometric model that embodies a particular aggregate supply
theory other than the natural rate hypothesis" (pp. 177 - 78). He con-
cluded that there is no way of knowing whether such a better alternative
theory exists until a so-called 'horse race' is held. The second test which
Sargent (1973b) performed consisted of such a horse race.
47
This subset included "... values of the logarithm of the money supply (currency plus
demand deposits), seasonally adjusted (m ), the federal and state and local government
deficit on the national income accounts basis (Dej); and the logs of the GNP deflator,
seasonally adjusted (p), of the implicit deflator for personal consumption expenditures (pc),
of the average hourly wage rate in manufacturing, seasonally adjusted (wry, of government
purchases of goods and services (g), of total federal and state and local government
employment, seasonally adjusted (ng), and of GNP {y). Each of these arguments is
included lagged one, two, and three periods" (p. 178).
179
Sargent's second test involved a comparison of the predictions of
equation (8.7) with those of the following equation:
q
(8.13) Un, = L A. 1Un,_; + p(p, - Ep,l6,_ 1) +
i=l
+ P(l - u)(Ep,l6,_ 1 - p,_ 1) + u,
This equation differs from (8.7) as it contains the third term on its right-
hand side. This term represents the influence of the difference between
the price forecast Ept based on information available at time t-1, minus
the price at t-1. Equation (8.7) and hence the NRH implies that a = 1. In
contrast, equation (8.13) shows that" ... if a < 1 (a > 1), then increases in
the systematic part of the rate of inflation decrease (increase) the unem-
ployment rate, contrary to the natural rate hypothesis" (p. 180). Sargent
tested both equations for the United States over the period 1952:1 -
1970:4, using quarterly data. The 'horse race' was thus held between the
joint NR/RE hypothesis and the hypothesis that the systematic part of the
rate of inflation affects the rate of unemployment. The test itself pointed
in the direction of rejection of the former. However, Sargent added that
such a rejection would not be based on an 'unusually' high confidence
level (p. 186).48 He stated that the evidence "... would not be sufficiently
compelling to persuade someone to abandon a strongly held belief in the
natural rate hypothesis" (p. 187). Prior beliefs thus play a crucial role in
the assessment whether the discorroboration of the null hypothesis should
lead to its rejection.
8.8. CONCLUSIONS
New Classical Economics arose in the late 1960s and early 1970s from the
attempt of economists such as Lucas, Rapping and Sargent to formalize
Friedman's (1968) contention that the long-run Phillips Curve would be
vertical. By 1973 its main proponents were Robert E. Lucas, Jr., and
Thomas J. Sargent, who had succeeded in the formalization. Lucas
combined Friedman's NRH with Phelps's islands parable, thereby elimin-
48 This raises the question why an unusual high level is required, and why a usually
high level does not suffice? The answer to this question presumably is that the aim of New
Classical model-building is to build 'satisfactory' models which meet several other (non-
empirical) criteria. These will be discussed in Chapter 10.
180
ating distributional effects from his analysis. Furthermore, he assumed
that e~ectational errors may only be due to inadequate information sets.
That is, he assumed that economic agents know the correct structure of
the economy. Sargent had concentrated on what Keynes had called the
Gibson paradox. He rejected Fisher's analysis because of its implication as
regards the lengths of the lags in the process of expectations adjustment,
and he used Muth's REH as a criterion for plausibility. Like Lucas's
explanation of the Phillips Curve, Sargent's explanation of the Gibson
paradox ran in terms of insufficiently quick price and wage adjustment,
which were also due to incomplete information sets and imperfect fore-
sight. His analysis implicitly adopted the NRH as its framework.
The introduction of the REH led Lucas and Sargent to criticize
standard econometric methods of policy evaluation. This 'Lucas critique'
implied that past tests of the combination of the NRH and the REH were
invalid. As a corollary, testing of the joint NR/RE hypothesis was yet to
be undertaken. Lucas's 1973 test used international data in such a way as
to evade the recommendations based on the critique. In the same year
Sargent developed another way of testing the NR/RE hypothesis and
carried out such a test. It indicated that forecasts of the rate of unemploy-
ment could be improved by taking a number of other variables into
account. This meant that economic agents could improve their forecasts
by using regressions such as carried out in Sargent's (1973b) test. How-
ever, Sargent did not reject the theory, even though it was discorrobora-
ted. He discerned three types of reasons why such a rejection would be
inappropriate. The first reason implicitly built on the Duhem-Quine thesis
which holds that a (social) scientist cannot test an isolated hypothesis.49
Any discorroboration may then be caused by a 'false' supplementary
hypothesis. That is, the initial conditions (assumptions) may not hold, or
the ceteris-paribus clause may be violated. In particular, Sargent observed
that the assumption about the absence of serial correlation in the disturb-
ance term may not hold. Secondly, Sargent 'retreated' to the methodologi-
cal position of sophisticated falsificationism, which holds that a theory
49
Duhem (1954 (1976), p. 8) had argued that " .. .the physicist (or, more generally, the
scientist] can never subject an isolated hypothesis to experimental test, but only a whole
group of hypotheses; when the experiment is in disagreement with his predictions, what he
learns is that at least one of the hypotheses constituting this group is unacceptable and
ought to be modified; but the experiment does not designate which one should be
changed." Quine (1951 (1964)) provided an even stronger thesis. He extended Duhem's
argument to include the 'laws of logic' and all laws of science, and claimed that "[t]he unit
of empirical significance is the whole of science", and not merely a theory (p. 59).
181
should not be rejected if no better alternative is available. Sargent
(1973b) set up a comparative test ('horse race') and its results indicated
that there was no such alternative. The third reason involved prior beliefs.
It was not used separately but rather in conjunction with the other two
types of argument. Sargent's second test had provided "... some evidence
for rejecting the natural rate hypothesis, although not at an unusually high
confidence level." This indicates that prior beliefs play an important role
in the New Classical assessment of whether a discorroborated theory (and
in particular the NRH) should be rejected. In fact, they insulate the
theory or hypothesis from rejection by demanding discorroboration at
'unusually high' confidence levels.
New Classical theory thus did not appear to be standing on firm
empirical ground. The adoption of the immunizing strategies raises the
question whether the 1973-tests were indeed intended to falsify the joint
NR/RE hypothesis, or whether they 'merely' provided attempts to corrob-
orate it. As Lucas (1981a, p. 2) indicated, the rise of New Classical
business cycle theory was mainly theory-driven, in the sense that it arose
from his and Rapping's dissatisfaction with the lack of microfoundations
of the then prevailing Keynesian orthodoxy. Stated differently, the a priori
beliefs consisted of the view that a competitive-equilibrium explanation of
cyclical fluctuations in output and unemployment was feasible and desir-
able. This view leaves little room for questioning the empirical validity of
the assumption of continuously clearing markets. In turn, this assumption
is closely related to the NRH. Furthermore, New Classicals unequivocally
adopted the rationality postulate, which implies that they considered the
REH to be superior to the AEH. Both the NRH and the REH were thus
defended on a priori (as opposed to empirical) grounds. As a corollary,
the joint NR/RE hypothesis was hardly open to doubt.
182
9. PERSISTENCE, CAPITAL AND
GLOBAL INFORMATION
9.1. INTRODUCTION
The previous chapter showed that Lucas and Rapping, on the one hand,
and, on the other, Sargent developed the joint NR/RE hypothesis. Its
constituent parts were considered to reflect individual optimizing behav-
iour, whereby frictions to the economic system (in the form of adjustment
costs, including those with respect to the agents' information sets) were
assumed to be absent. Mter its deduction from the principle of economic
rationality, the joint hypothesis was confronted with economic data.
Lucas's (1973) empirical test provided some evidence for the empirical
validity of the hypothesis, but this evidence was not conclusive, because it
hinged on only two extreme cases. Sargent's (1973b) test results can even
be considered discorroborative as regards the joint hypothesis, but his
'sophisticated-falsificationist' position prevented the rejection of the latter.
This weak empirical foundation of the joint hypothesis reflects that the
joint NR/RE hypothesis was mainly upheld on a priori (non-empirical)
grounds. This does not mean that empirical considerations did not play
any role at all. Empirical anomalies in particular played a role in the later
development ('growth') of New Classical business cycle theory. This
becomes clear in the discussion on persistence of aggregate output and
unemployment, held in the mid-1970's. This discussion centered on a
criticism which had already been levied by Rees (1970), who argued that
equilibrium business cycle models cannot account for serial correlation
(persistence) in unemployment rates. Lucas (1973) responded to this criti-
cism by including lagged output in his supply curve, so that the effects of
random disturbances in previous periods persist in the present period.
However, this inclusion was not based on economic-theoretical consider-
ations. As a corollary, persistence still had to be explained as the outcome
of the behaviour of individually optimizing economic agents.
This chapter is organized as follows. Section 9.2 discusses two
problems which are involved in accounting for persistence in a general-
equilibrium context, namely the 'impulse problem' and the 'propagation
problem'. The effects of random shocks (impulses) can only be persistent
if they are propagated to the future by a so-called 'propagation mecha-
183
nism'. The discussion briefly touches on Lucas and Prescott's (1974)
analysis of search behaviour and its consequences for persistence, and
Rees's (1970) and Hall's (1975) criticism of the New Classical failure to
explain this phenomenon. Section 9.3 describes Lucas's (1975) example of
a business cycle model which explicitly solved the propagation problem.
This model led other economists to build models which use different
propagation mechanisms. These models are also (albeit briefly) described
in this section. Lucas's (1975) model, however, only included local capital
markets. He himself criticized this feature, and suggested that his model
should be extended by a global capital market. Such a modification should
make it clear whether the global interest rate does not give economic
agents 'too much' global information so as to make expectational errors.
Barro ( 1980} showed that this need not be the case if the information sets
of economic agents only include a global nominal rate of interest and
their respective local commodity prices. Furthermore, the shocks to the
economic system must make it impossible for economic agents to solve
the signal extraction problem. Barro's (1980) model is discussed in section
9.4. Section 9.5 gives a methodological assessment of what may be called
the Lucas programme, which aims to solve the 'Lucas Problem'. In its
broadest formulation this problem concerns the rationalization of the
correlation between money and output in terms of the rationality postu-
late. Its solution is subject to some situational constraints, which form the
requirements to be fulfilled if the solution is to be considered acceptable.
Furthermore, the programme can be characterized by its method of dis-
covery, which is a version of Popper's method of rational model-construc-
tion. Section 9.6 contains some conclusions.
9.2. THE PROBLEM OF PERSISTENCE
General-equilibrium analysis with complete information is difficult to

reconcile with cyclical movements in the variables, because the economic
system will 'merely' react to exogenous shocks by moving to a new equili-
brium position (see Chapter 4). When expectations are formed rationally,
this movement will be even more direct and rapid than under any other
expectations formation mechanism. However, as Rees (1970} had already
pointed out, in reality economic time series on unemployment and output
show important persistence, which means that ex-post non-optimal actions
are not immediately improved, so that past unemployment rates are very
184
helpful in predicting current rates. This appeared to be inconsistent with
Lucas's (1972a) description of a model economy in which global informa-
tion becomes available at the end of the period under consideration, and
in which expectations are rational in the sense of Muth. After all, such
availability of information means that economic agents can discover
whether they have made expectational errors, and that they can immedi-
ately improve their ex-post non-optimal actions. In other words, New
Classicals were confronted with the problem of how to explain persistence
in economic time series. Lucas (1973) had circumvented this problem by
including lagged output in his aggregate supply curve, but this could
hardly be called an explanation of persistence. He appears to have
recognized this, because in their 1974 article 'Equilibrium search and
unemployment' Lucas and Prescott tried to build a business cycle model
which would provide a theoretical explanation of persistence as the
outcome of individual optimizing behaviour. They adopted a distinction
which had already been made by Slutzky (1937) and Frisch (1933).1 The
former had argued that the summation of random disturbances alone
could generate a cyclical pattern in the data, but his analysis could not
make it clear whether business cycles are actually generated by such a
summation; he merely showed that this is possible. Frisch (1933 and 1939)
built on Slutzky's work by showing how the random shocks could become
summed. 2 He used the 'rocking horse' metaphor, which interprets cyclical
movements in economic time series as the outcome of erratic, random
'hits by a club'. 3 This metaphor indicates that there are two problems.
Firstly, the impulse problem addresses the question how such random 'hits'
or shocks come about. As was shown in the previous chapter, New Classi-
cal business cycle theory solved this problem by distinguishing real and
monetary aggregate shocks. The former reflect non-systematic changes in
real variables, such as the distribution of economic agents across markets,
1 As Morgan (1990, p. 80) made clear, Slutzky's 1937 article was already published in
1927, namely in the theoretical journal of the Moscow Institute for Business Cycle
Research. In 1937 it was reprinted (in a revised and expanded form) at the instigation of
Schultz and Frisch.
2 Frisch (1933, p. 99, italics in original) argued that Slutzky "... established that some
sort of swings will be produced by the accumulation of erratic influences, but the exact and
general law telling us what sort of cycles that a given kind of accumulation will create was
not discovered."
3 Frisch (1933, p. 198n) credited Wicksell as the first economist to distinguish between
impulse and propagation, thereby referring to Wicksetrs address to the Norwegian
Economic Society, published in 1907 in the Statsekonomik 1idsskrift, pp. 255- 86.
185
and the latter consist of random additions to the money supply.4 Second-
ly, the effects of the impulses must be propagated to subsequent periods.
This propagation problem concerns the specification of some propagation
mechanism, which explains how random shocks are summed and how they
bring about persistence in the deviations from the mean values of econ-
omic variables.5
From a methodological point of view, as Hirsch and De Marchi
(1990) suggested, the addition of random shocks to an otherwise exact
model can be regarded as an attempt to increase the correspondence of
such a model to an inexact world. Morgan (1990, p. 194) explained this
addition as an attempt to account for either measurement errors (errors-
in-variables models) or omitted variables (errors-in-equations models).
This means that the model deduced from a priori principles (in particular,
the rationality postulate and some auxiliary assumptions) is modified by
adding some error terms, which are assumed to be normally distributed
with zero mean and finite variance. The resulting stochastic model is
subsequently tested, whereby any deviations of its predictions (or retro-
dictions) from observed phenomena are attributed to the shocks. The
inclusion of such shocks thus aims to bring about a closer correspondence
between its exact description of reality, on the one hand, and, on the
other, inexactly observable reality itself. Hirsch and De Marchi (1990, p.
127) therefore concluded that in such a stochastic model "... the realism of
the assumptions of theory or the plausibility of motives is taken as a given
and testing comprises only identifying the size and nature of the error
4
Mullineux (1990, p. 20) argued that "Frisch (1933) ... provides two possible solutions
to the impulse problem: the Frisch I hypothesis that exogenous, purely random, shocks
provide energy to a system {propagation model), with a damped cyclical solution, to
produce the cycles observed in the economy; and the Frisch II hypothesis that the shocks
are provided by the movements of the economic system and these shocks supply the
necessary energy to keep the otherwise damped oscillations from dying out. These shocks
are released systematically, but whether they are regarded as exogenous or endogenous
depends on whether or not a theory of innovations is included in the model." The analysis
in Chapter 8 indicated that early New Classical business cycle theory opts for the Frisch I
hypothesis, because it assumes that the shocks are random and must not be attributed to
changes in technology. In the 1980s New Classicals would tum to real business cycle
theory, whereby the real shocks could be brought about by changes in technology as well.
This appears to be a shift from the Frisch I to the Frisch II hypothesis.
5
As Frisch (1933, pp. 171- 72) stated, the propagation problem is "... the problem of
explaining by the structural properties of the swinging system, what the character of the
swings would be in case the system was started in some initial situation." For a more
detailed analysis of Slutzky's and Frisch's work, see Kim (1988, pp. 59 - 62), Morgan (1990,
pp. 79 - 100), and Mullineux (1990, pp. 19 - 29).
186
term ..." This suggests that New Classical testing procedures are directed
towards corroboration of models, not their falsification.
As was described in the previous chapter, Rees (1970) had critici-
zed Lucas and Rapping (1969a) for not being able to account for persist-
ence in unemployment rates. Lucas ( 1973) had reacted by incorporating
lagged unemployment in his models. However, this 'solution' was not very
satisfying because it did not formulate a propagation mechanism. The
same applies to Lucas and Prescott (1974), who tried to develop a theor-
etical framework within which non-zero equilibrium unemployment can be
discussed as the outcome of individual search processes (p. 174).6 In this
framework unemployment is 'voluntary', given the available information.7
Lucas and Prescott specified an 'island model' with spatially separated
markets, in which economic agents must choose between two mutually
exclusive actions: either they remain in their local market and work at the
market-clearing wage rate on that market, or they leave their market and
search for a market in which they can earn a higher real wage rate. In this
latter case they earn nothing during the entire search period, so that job
search involves costs. Lucas and Prescott (1974, p. 160) assumed that
agents base their decisions on rational expectations, and that they there-
fore know (by definition) the relevant probability distributions, including
that of the workforce over markets (which is assumed to have 'settled
down'). In this setting persistence in unemployment may arise if agents
misperceive their decision situations by overestimating the present value
of job search. Given this estimation, they are induced to switch from one
market to another, so that they enter a pool of unemployed workers.
Their misperception about the present value of job search will lead them
to turn the first job offer down, because they expect a better job offer. As
a result, persistence in unemployment may arise. This persistence is thus
the result of mistaken expectations about the present value of job search.
However, as Lucas (1981, p. 14) remarked, the model of Lucas and Pres-
cott (1974) was difficult to reconcile with business cycles because of a
simplifying assumption. This assumption holds that the job search parame-
ter A. and hence the expected present value of job search is constant,
6
Page numbers refer to the 1981 reprint of Lucas and Prescott (1974).
7
The term 'voluntary' has led to sharp - often normatively based - criticisms on New
Classicism, because it suggests that this type of unemployment is Pareto-optimal. However,
the addition 'given the available information' is crucial in this regard, because it allows for
the claim that under imperfect information the choices may be non-optimal from a full-
information point of view.
187
which clearly does not apply over the business cycle. New Classical
business cycle theory therefore remained open to criticism for not having
explained persistence in unemployment, investment and output over
business cycles.
Hall (1975) started from the observation that variations in the
actual unemployment rate can be explained by variations in its natural
rate and/or by expectational errors about the general price level, i.e. by
shifts in and/or along the (short-term) Phillips Curve. He tried to estab-
lish the relative importance of these components and regressed unemploy-
ment on its own value lagged twice (so that ·both Unt_ 1 and Un •.2 were
considered), using quarterly data from the U.S.A. from 1954 to 1974. He
found that "[o]nly a trivial fraction, less than 1.7 percent, of the variation
of unemployment is attributable to the unemployment-inflation tradeoff in
the rational-expectations model. The remaining 98.3 percent or more is
attributable to unexplained shifts of the Phillips curve" (p. 314). 8 That is,
in Hall's view the variance in unemployment can almost completely be
explained in terms of shifts in the NRU, with deviations from this rate
playing only a minor part. Implying that random disturbances can only
generate random effects on unemployment, he rejected Lucas's (1973)
view that unemployment behaviour can be explained as the result of
serially uncorrelated shocks (p. 312). Furthermore, he considered Sar-
gent's 'explanation' of such correlation in terms of serially correlated
shocks unacceptable, because it provided "... no more than a relabeling of
the same phenomenon" as it did not explain persistence as the outcome of
individual optimizing behaviour. Hall therefore concluded that the models
of Lucas and Sargent did not explain persistence and in fact were discor-
roborated by the data (pp. 312 - 13). Tobin (1977) and Modigliani (1977)
concurred with this criticism. 9 However, as Lucas and Sargent (1978
(1981), p. 313) pointed out, these criticisms failed to distinguish between
the impulse problem and the propagation problem, as discussed above. In
1975 Lucas built a model in which he successfully included a propagation
mechanism. 10
8
In the original the first sentence of this quotation was printed in italics.
9 Modigliani (1977) even claimed that the New Classical explanation of business cycles
implied that "... what happened to the United States in the 1930'swas a severe attack of
contagious laziness." It will be shown below that this conclusion does not follow from New
Classical analysis and hence is mistaken.
10
Apparently, Tobin and Modigliani were completely unaware of this model.
188
9.3. PERSISTENCE AND LOCAL CAPITAL MARKETS
Lucas (1975) adopted a modified neoclassical monetary growth model in

which a constant population of identical economic agents (households)
produces a single output. This output can be divided between consump-
tion and capital accumulation. All markets are competitive and there are
two assets, money (m.) and physical capital goods (k.). 11 The demand for
capital in the next period (kt+ 1) depends on the current period's real rates
of return on money and capital (rmt and rn respectively), and on the cur-
rent period's state of the household k. and m. - p •. The money supply is
assumed to expand at rate p., so that mt+l - m. = p.. Lucas pointed out that
in his model money is neutral, in the sense that "... a once-and-for-all
change in the level of money balances leads to a proportional change in
the price level in the current and all future periods. There are no real
effects." However, money is not superneutral because "... changes in the
rate of increase of money, p., will have real consequences: the higher p. is ...
the larger capital is along its time path and at its stationary point" (p. 185,
italics in original). 12 This 'non-superneutrality' is brought about by a
version of an effect discussed by Tobin (1965). This 'Tobin effect' holds
that a rise in the growth rate of money lowers the real yield on money
and therefore induces economic agents to change the composition of their
portfolios by substituting capital for money.
Lucas subsequently transformed his growth model into a business-
cycle model by adopting the 'islands approach', which he had already used
in his (1972a) and (1973) articles. Again, he specified two types of shocks,
namely real and monetary shocks. The former consist of redistributions of
agents over markets. More specifically, Lucas assumed that at the begin-
ning of each period the agents are randomly distributed over the imper-
fectly linked markets. The monetary shocks take the form of random
changes in the money supply, due to government consumption. These
monetary disturbances can be subdivided into a temporal and a spatial
component, in the sense that they may vary in time and over markets (at
a given time). The former can then be expressed as fluctuations in the
average (over markets) rate of monetary expansion Xu which is assumed to
be normally distributed with mean p. and variance if. The spatial com-
11 The relationships in Lucas's model are log linear, so that the 'variables' are actually
'logs of variables'.
12 Page numbers refer to the 1981 reprint of Lucas (1975).
189
ponent consists of variations over markets (at a given time). These vari-
ations are assumed to differ from xt by the percentage 8t(z) = p()t-l (z) +
&.(z), where z is an index of location, and with 0 < p < 1. They are
normally distributed with mean zero and variance fil6• Both types of
monetary shocks xt and 8t(z) cannot be observed by economic agents, but
their (normal) distributions are assumed to be constant and known.
Jointly with the real shock, these monetary shocks solve the impulse
problem: they all lead to fluctuations in effective demand, creating a
signal extraction problem. Lucas (1975, p. 189) summarized the agents'
expectations, and assumed that the resulting expectations of the represen-
tative agent are Muth-rational. The expectations are then unsystematic, so
that the phenomenon of persistence was yet to be explained (p. 190).
Lucas introduced physical capital goods in order to solve the
propagation mechanism. He assumed that these goods are market-specific,
in the sense that they must remain on the market in which they have been
installed. Stated differently, there are no economy-wide capital markets.
In this setting the periodic redistribution of agents across markets has two
implications for the real rate of return on capital. Firstly, it means that
the capital goods must be rented by agents other than the owners. The
nominal rental of capital goods thus depends on the expectations of these
renting agents about next period's local price. Secondly, the redistribution
implies that the dividends on capital goods will generally be spent else-
where. The appropriate deflator of the nominal rental must then also take
the price movements on the other markets into account. The real rate of
return on capital thus equals the nominal rental of capital goods, deflated
by the expected average price level. The real rate of return on money
equals the observed current local price, also deflated by this price expec-
tation. Lucas inserted both rates of return into the asset demand func-
tions, which yielded a series of equilibrium conditions (pp. 194 - 95). The
mathematical solution of his model specified two equations which des-
cribe the effects of an unanticipated change in the money supply (xt- p.).
The phenomenon of persistence may now be explained as follows.
Suppose that agents are confronted with an unanticipated money shock.
They will then make expectational errors about the nature of the ensuing
rise in their local prices. At least part of this rise will be attributed to real
causes, so that the agents will expand productive capacity. This expansion
has two effects. Firstly, when the agents discover their expectational
errors, they will try to undo this expansion. Lucas assumed that the over-
expanded capacity can only be reduced at the rate of depreciation, so that
190
it takes time. This sluggish adjustment (partly) explains the phenomenon
of persistence in output and unemployment (Mullineux (1990, p. 23)). The
second reason why this phenomenon may occur follows from the fact that
the increase in productive capacity brings about a rise in output. This
retards the upward adjustment of the local prices (and hence the general
price level) to the money shock. These prices thus do not reflect the
shock completely. Since agents must base their expectations about the
state of the economy on these local prices only, they cannot correctly
estimate the magnitude of the shock. This means that they will not
correctly adjust their expectations about this state. According to Lucas
(1975, p. 203), this relatively slow adjustment of expectations reflects a
lagged accelerator effect. Persistence is thus brought about by sluggish
capacity adjustment and a lagged accelerator effect.
Lucas's (1975) model has given rise to several other general-
equilibrium business cycle models, each of which incorporated different
propagation mechanisms. As Barro (198lb, p. 48) observed, Lucas's
modelling approach is not restricted to productive capacity, but in fact
applies to any other variable that responds initially to monetary shocks
and then has a durable aspect that implies a change in future initial
conditions. 13 In general, these mechanisms are instances of what Barro
(1981b, p. 48) called 'adjustment-cost explanations' for the persistence of
the real effects of the shocks. He observed that these explanations imply
that investment will initially rise sharply after the unexpected disturbance,
but would thereafter decline gradually. However, this does not correspond
with empirical evidence, which indicates that investment and output rise
during several periods before declining. This empirical phenomenon can
be explained in two ways. The first explanation adopts Lucas's (1975)
assumption that information lags may be longer than one period, so that
they prevent even the past values of the relevant variables from being
known. This explanation of serial correlation in expectational errors may
be criticized on the account that information about economy-wide monet-
ary variables can often be gathered at short notice and without incurring
many costs. It is therefore not likely to be empirically valid. Secondly,
persistence may also be explained in terms of slow adjustment of the
13
Barro (1981b, p. 48) gives the following examples: Feldstein and Auerbach (1976)
analyze the behaviour of inventories of goods-in-process, which show considerable cyclical
fluctuations. Blinder and Fischer (1981) use the gradual adjustment of inventory stocks of
finished goods. And Sargent (1979, Chapter 16; 1987, Chapter 18) treats labour as a
capital-like input by assuming adjustment costs for variations in employment.
191
capital stock. This explanation combines Lucas's assumption of lagged
downward adjustment of the physical capital goods stock with the notion
of 'time to build'. Kydland and Prescott (1980, pp. 175ff) argued that
investment projects involve long-term planning. In their view, increases in
real output are lagged because "... there are long lags from the time when
changes in its determinants call for an increase in the capital stock until
the time when the new capital starts yielding services." 14 However, given
the length of these lags, one may question whether investment decisions
are taken on the basis of short-term considerations. It appears more
plausible that the length of the lags will induce investors to take long-term
considerations into account, so that it is questionable whether fluctuations
in productive capacity can explain cyclical (short-run) fluctuations in
aggregate output.
As was stated above, Lucas (1975) assumed that investment is
completely internally financed, ·i.e., that there is no economy-wide capital
market. This is an important restriction, because its removal may invali-
date his explanation of cyclical fluctuations. Lucas himself already recog-
nized that the inclusion of such a market could present considerable
difficulties, because it may give economic agents 'too much' information
about economy-wide variables. After all, such a market and the ensuing
global interest rates may convey information about global price changes,
thus providing economic agents with relevant information about the
nature of random shocks. Agents then do not make expectational errors,
so that serial correlation in the deviations from the natural rate of output
cannot be explained in these terms. As a corollary, it must be brought
about by serial correlation in the disturbance terms (i.e., in the shocks).
Hall (1975) had argued that this does not explain but merely relabels the
phenomenon of persistence in unemployment and output. The inclusion of
an economy-wide capital market and the ensuing currently observable
g(obal interest rate may thus poses a serious problem for the New Classi-
cal explanation of cyclical fluctuations. Barro (1980) specified a model in
which economic agents know such a global price. 15
14
Cf. also Hall (1977). It should be noted that Kydland and Prescott are concerned
with the efficiency of fiscal instead of monetary policy, so that they do not study the
influence of monetary shocks on investment decisions. For some criticisms on the Kydland
and Prescott paper, cf. Taylor (1980).
15
It should be noted that Barro (1980) was written well before 1980, namely some-
where between 1976 and 1978. This indication follows from the fact that Barro (1976, p.
3n2) announced to address the problem of an observable global price, as raised by Lucas
192
9.4. CURRENT GLOBAL INFORMATION AND THE SIG-
NAL EXTRACTION PROBLEM
Barro (1980) postulated an island economy in which economic agents

trade in two markets, a local commodity market and an economy-wide
(global) capital ('loanable-funds') market. The latter deals in homo-
geneous, riskless, one-period loans (p. 113). 16 Observation of the global
nominal rate of return on the capital market conveys current global
information to the agents. The signal extraction problem, discussed in the
previous chapter, can only be maintained if the agents' information sets
are restricted in such a way that the agents do not know whether and to
what extent a change in the global nominal rate of return on assets is
caused by real or nominal factors. In order to bring about this restriction,
Barro specified three types of shocks, which concern aggregate money
supply, aggregate money demand, and local excess demand for output. 17
The former two (normally and independently distributed) shocks are
denoted by mt and 4>t respectively (pp. 113 - 14)!8 The specification of
the local excess demand shock was derived from the local demand and
supply functions. These functions contained demand and supply in terms
of the anticipated real rate of return on market z, rt(z), which is defined as
the current price in market z, Pt(z ), discounted at the global nominal rate
of return Rb minus the anticipated future price in a randomly selected
market from the perspective of market z, E?t+l (p. 114}!9 It thus com-
pares current transaction opportunities in market z to anticipated future
opportunities in a randomly selected market, and reflects the importance
of intertemporal reallocation of economic activities in New Classical
business cycle theory. Local commodity supply is positively, and local
(1975). King's (1981, p. 205) list of references shows that Barro (1980) had already been
written in 1978.
16 Page numbers refer to the 1981 reprint of Barro (1980).
17
All variables in this section are expressed in logarithms.
18
The money demand function can be formulated as:
~Y,
4
M, = P, - y R, + + ,,
19
As Barro (1980, p. 114) indicated, "... the earlier analysis in Lucas (1973) and Barro
(1976) amounts to treating Rt = 0, which is appropriate in a model where the only store of
value is money that bears a zero nominal rate of return."
193
commodity demand is negatively related to r,(z). Furthermore, Barro
accounted for wealth effects by deriving a money-wealth variable, which
also influences local commodity demand and supply. This variable only
depends on the unexpected component in the rate of money growth (m, -
E.pr,). 20 In contrast, anticipated changes in this growth rate do not affect
local demand and supply, so that money is superneutral. Barro recognized
that this feature rendered his analysis somewhat different from that of
Lucas (1975), since it excluded the Tobin effect (p. 111).
Barro's specification of the local commodity supply and demand
functions led to two market-specific random disturbance terms. The first,
8 1(z) concerns local commodity supply, and the second, ed1(z), local
1
commodity demand? 1 The mathematical solution of the model yielded

expressions of the local variables r,(z) and y,(z), with which Barro analyzed
the effects of unexpected shocks to the economic system. The case of full
information thereby served as a benchmark. It implies that economic
agents do not make expectational errors, so that m, - E.pr 1 = 0. Both the
real rate of return r,(z) and the level of aggregate output y,(z) are then at
their 'natural' value, which is affected by neither the rate of money growth
(!l) nor random deviations from this rate (m,). In the benchmark, money
is thus not only superneutral, but also neutral. Nonneutrality results from
expectational errors about shocks in ( 1) aggregate money supply (m 1), (2)
aggregate money demand (c/>1), and local excess demand for output (e,(z) ).
There are thus two types of aggregate shocks and one type of relative
shock. Jointly, they solve the impulse problem.
Barro (1980) thus extended Lucas's (1975) model by analyzing an
economy with a global capital market on which a nominal interest rate is
specified. His model yielded similar conclusions to those of Lucas (1972b)
and (1973). It should be noted that it is crucial that the global capital
20
In the derivation of this variable Barro assumed that economic agents are identical
with respect to their expected nominal cash demands in period t+i, and that they correctly
anticipate portfolio balance to obtain in every (future) period (p. 116).
21 These log-linear functions can be formulated as:
y,d(z) = kd(z) - a.;,(z) + Pj.m,-E/") + <<z)
where 't(Z) gives the local real rate of return, (m 1 - E.pr 1) represents the unanticipated
change in the money growth rate, ~(z) and kd(z) represent systematic and invariant supply
and demand conditions, and £11(z) and t\(z) are local random supply and demand shocks.
The a's are relative price elasticities and the {J's are wealth elasticities.
194
market specifies a nominal and not a real interest rate, because "[i]n the
context of an economy-wide real interest mtes, the movements in antici-
pated and realized real rates of return would be coincident ...", so that
monetary shocks cannot induce expectational errors about the actual real
rate of return (pp. 131 - 32). There must be a signal extraction problem,
otherwise economic agents can correctly infer the source of the shocks
from the changes in the global real rate of interest. Furthermore, Barro
noted that the variance of monetary disturbances is inversely related with
the effects on real output, because its rise reduces the agents' confusion
about the source of the price fluctuations. Such a rise indicates that
monetary disturbances become more important. Given the fact that econ-
omic agents form Muth-rational expectations and hence know the 'true'
probability distributions, they will be more in~lined to expect that the
price fluctuations are caused by monetary factors. Output then becomes
less responsive to monetarily induced price fluctuations. As discussed in
the previous chapter, Lucas (1973) had already used this effect in his
search for international evidence for the NRH. Barro showed that it
applies even in the context of an economy with a global capital market, in
which agents observe a global nominal interest rate.
Barro's model, though, abstracts from some important features in
economic reality. Following Lucas's modelling strategy, Barro assumed
that markets continuously clear, so that coordination is ensured. This
reflects a particular methodological role of the notion of general equilib-
rium, as will be shown in the next section. Furthermore, the model could
not account for persistence, because money is assumed to be superneutral.
In this regard Barro (1980) narrowed Lucas's (1975) analysis by elimi-
nating the Tobin effect, although the inclusion of a propagation mecha-
nism does not appear to be problematical. However, Barro would not
undertake such an inclusion, but instead opined that" ... empirical research
would potentially constitute the most fruitful complement to the present
theoretical analysis" (p. 133). This research should estimate the effects of
monetary disturbances on the agents' expectations about real rates of
return, and the importance of such rates as channels of transmission of
such effects (p. 133). In the 1980s he concentrated on such empirical
research, which eventually led him to abandon monetary explanations of
business cycles. The empirical tests of the joint NR/RE hypothesis will be
discussed in the next chapter. The remainder of the present chapter will
concentrate on the methodology of the NCE, which can be specified in
the form of the so-called Lucas programme.
195
9.5. THE LUCAS PROGRAMME
As was shown in Chapter 4, the Hayek Problem sought to explain the

existence of a tendency towards coordination (or intertemporal general
equilibrium), with 'methodological individualism' and 'subjectivism' as its
situational constraints. The method of discovery adopted was the so-called
'compositive method', which deduces the state of the economy as a whole
from introspectively derived knowledge about the decision situations as
perceived by individual economic agents. Empirical testing of this deduced
state plays a minor role, since it is limited to checking whether the ceteris-
paribus clause holds, or whether the initial conditions are met.
New Classical business cycle theory can be analyzed in analogous
terms, in the sense that a so-called 'Lucas Problem' can be formulated,
which is also subject to situational constraints. Verbally, New Classicals
argue that their situational constraints also consist of 'methodological
individualism' and 'subjectivism'. It should be noted, however, that their
heuristic prescription of mathematical formalization necessitates the
inclusion of simplifying assumptions in order to deal with aggregation
problems. It will be shown that these assumptions affect the situational
constraint of 'individualism' in such a way as to alter its meaning, as com-
pared with the meaning it was given in Chapter 4. Furthermore, this
section also contains a description of the New Classical method of dis-
covery, which resembles Hayek's compositive method.
9.5.1. The Lucas Problem

Lucas (1981, p. 3) stated that New Classical business cycle theory aims at
'rationalizing' observed cyclical fluctuations in output and employment.
Such a rationalization means that these fluctuations should be reconciled
with rational behaviour on the part of the economic agents. New Classical
analysis, as almost all economic theory, thus starts from the rationality
postulate. This postulate holds that economic agents 'prefer a greater gain
to a smaller', to speak with Mill (1843, VI, ix, 3).22 Since business cycles
are usually associated with welfare-losses, they will not reflect the inten-
tions of rational economic agents. The Lucas Problem (in a broad sense)
seeks to explain how the actions of rational agents result in an overall
pattern of economic activity which exhibits cyclical fluctuations. New
22 This is not to say that New Classicals adhere to Mill's view that the postulate
reflects a psychological 'law'.
196
Classicals solve this broad Lucas Problem by adopting Hayek's position
that the fluctuations are the 'unintended outcomes' of such behaviour.
This raises the question why the agents' intentions are not fulfilled. That
is, some reasons must be identified why a Natural Rate Equilibrium
(NRE) does not arise.
The rationality postulate implies that rational economic agents do
not suffer from money illusion. The gain which they try to grasp must
therefore be real in nature. However, economic time-series show that
money and output are correlated, suggesting that agents respond to
monetary changes as if they were real. This suggests that economic agents
misperceive their decision situations, and make expectational errors about
the outcomes of their actions. New Classicals base their explanation of the
Lucas Problem on this misperception. This explanation holds that for
some reason economic agents mistake nominal changes for real changes.
In general, agents make such expectational errors because of (1) short-
comings in their expectations formation mechanisms, and/or (2) defi-
ciencies in their information sets. Given the REH, these errors must be
caused by the latter. These deficiencies may concern (a) the structure of
the economy, and/ or (b) the values of relevant economic variables. Since
Muth-rationality implies that the agents' subjective probability distribu-
tions are identical to the objective distributions, these agents do not make
systematic expectational errors and hence know the structure of the
economy. This structure is a competitive market economy, in which the
agents are price takers and in which general equilibrium is ensured by
continuous and instantaneous price adjustment. This means that Hayek's
coordination problem is assumed away, so that there are no endogenous
reasons why economic agents are confronted with unintended outcomes of
their actions. That is, unlike Hayek, the NCE assumes away coordination
failures as a reason for the existence of unintended outcomes of rational
behaviour. In turn, this implies that the NCE does not attempt to explain
these general equilibria (REE's). The broad Lucas Problem can only be
explained exogenously, namely as the result of unanticipated unsystematic
(random) expectational errors which are generated by a lack of informa-
tion about the current values of the relevant global variables. The prob-
lem of providing such an explanation can be called the Lucas Problem in
a narrow sense. It is a narrowed down version of the broad Lucas Prob-
lem, in the sense that it assumes away coordination failures.
Lucas (1972a) partially solved the narrow Lucas Problem by
specifying a stochastic model economy in which the relevant local vari-
197
abies are subject to exogenous random shocks ('impulses'), and in which
global variables were unobservable. Agents were thus confronted with a
signal extraction problem as regards movements in their local prices.
Cyclical fluctuations in output, employment and investment were subse-
quently explained in terms of repeatedly misperceived decision situations.
Lucas (1975) accounted for the phenomenon of persistence by incorporat-
ing a propagation mechanism. However, the resulting model precluded
the existence of observable current global nominal prices. This feature
was 'repaired' by Barra (1980).
These attempts to solve the Lucas Problem were undertaken
against a background consisting of situational constraints. These con-
straints provide criteria in order to assess which solutions are acceptable.
As shown in Chapter 4, Austrians adopted three such constraints, namely
methodological individualism, subjectivism, and methodological dualism.
The solutions to the Lucas Problem are subject to analogous (though not
identical) constraints, which can be subsumed under the headings metho-
dological individualism, subjectivism, and methodological monism. As
stated above, New Classicals thereby adhere to the heuristic prescription
that economic theories should be formalized in the form of mathematical
models. This prescription will prove to have some consequences for the
meaning of the constraints.
9.5.2. Methodological individualism

According to Lucas (1981, p. 2), he and Rapping (1969a) tried to contrib-
ute to the Keynesian orthodoxy by constructing "... a 'microeconomic
foundation' for the wage-price sector of macroeconometric models." This
means that they considered the individual economic agent to be the
appropriate level of reduction, and hence adopted the situational con-
straint of methodological individualism. This concept, however, is not an
unambiguous one, as already shown in Chapter 4. It is therefore useful to
examine the specific ways in which New Classical economists have
provided their macroeconomics with microfoundations. The methodologi-
cal prescription of mathematical formalization plays an important role in
this regard, since it presupposes some sort of aggregation process. In
general, New Classicals adopt two modelling strategies which aim to solve
the aggregation problem, namely the representative-agent approach and the
islands approach.
The representative-agent approach defines the representative
agent (whether individual, household, firm, or otherwise) as the statistical
198
mean of the (sub)system as a whole. The representative agent is thus a
hypostatisation of this (sub )system. Deviations from this mean are
accounted for by adding a probability distribution. The expected value of
the deviations from this mean equals zero, so that there are no a priori
reasons why expectations and actions differ between individuals. The
assumption that the agents do not differ in any systematic way will be
called the homogeneity postulate. It restricts the domain of the NCE, since
it neglects coordination failures and redistribution effects.
The islands approach can be seen as a more sophisticated version
of the representative-agent approach, in the sense that it introduces
several representative agents, one for each local market. 23 These econ-
omic agents only differ with respect to their local information, so that
information is homogeneous across agents in a given market, but hetero-
geneous across markets. As a corollary, 'intra-market' distributional effects
are disregarded, although the analysis of such effects between markets
remains possible. However, the New Classical assumption of Muth-
rationality eliminates even these effects. After all, this assumption holds
that agents know the 'true' (objective) probability distributions of all
relevant variables, and particularly of the difference between local and
global price changes. As a corollary, the agents' information sets do not
differ in any systematic way. Pesaran (1989, p. 57) concluded that this
circumvents the problem of heterogeneity of information across markets is
also circumvented. As Machlup (1983) and Frydman (1983) pointed out, it
is not clear why all economic agents would use the same model, and why
this model would be New Classical in nature.
In conclusion, both the 'islands approach' and the 'representative-
agent approach' adopt a probabilistic version of the homogeneity postu-
late for economic agents. That is, economic agents are assumed not to
differ systematically (with respect to their utility functions, their abilities
to gather information (and hence their relevant information sets), their
expectations formation functions (and, given the information sets, their
expectations) and their productive activities). The claim that the NCE has
provided macroeconomics with microfoundations must therefore be regar-
ded with considerable scepticism. The reformulation of macroeconomics
in terms of the behaviour of a 'representative agent' does not appear to
transform a holistic explanation into an individualistic one, given the
23
In particular, the 'islands parable' was adopted by Lucas (1972, 1973, 1975), Lucas
and Prescott (1974), and Barro (1976, 1980).
199
presumption that it would be more appropriate to reserve the term
'methodological individualism' for the methodological principle which
rejects the homogeneity postulate and instead allows for the analysis of
coordination failures and redistribution effects.
9.5.3. Subjectivism
The emphasis on expectations suggests that the NCE considers the data of
economics to be subjective, in the sense that they are what economic
agents think that they are. According to Lucas (1981, p. 224), an econom-
ist cannot deduce from his observations how a particular agent will act,
unless the latter's perception of the decision situation is specified. This
perception is determined by the agent's information set. Like Hayek, New
Classical economists distinguish between the role of this set in the deci-
sion-making process and its truth status. Business cycles arise as the result
of the misperceived decision situations. For reasons of analytical conveni-
ence, New Classicals adopt the simplifying assumption that the agents'
perceptions are Muth-rational, which holds that the 'subjective' probability
distributions are identical to the 'objective' distributions?4 This simplify-
ing assumption makes it possible to specify precisely the subjective deci-
sion situations. At the same time, however, it presents a substitution of
objective (scientific) knowledge for subjective (agent-specific) knowledge.
This makes it perhaps more appropriate to label New Classical business
cycle theory as 'objectivistic' instead of 'subjectivistic'. It should be noted,
though, that any complete explanation of business cycles in terms of mis-
perception implies some 'objectively correct' perception of the relevant
decision situation, known by the theorist. In providing such an explana-
tion, New Classicals adopt Popper's method of rational model-construc-
tion as their method of explanation.
9.5.4. The method of rational model-construction

Popper (1957) opined that there are no reasons to presume that the social
sciences deal with more complex phenomena than the natural sciences.
He argued that the differences between these types of sciences "... are
24 New Classicals argue that they introduce this simplifying assumption only for 'prac-
tical' reasons, implying that it may be possible, at least in principle, to formulate a model
in which the subjective probability distributions differ from the objective distribution.
Expectations are then not Muth-rational and may even diverge. However, the resulting
model will be analytically very complicated, so that it will be difficult to use for purposes of
prediction and policy-evaluation.
200
differences of degree rather than of kind" (p. 141}. This means that in his
view there are no reasons to presume that the methods of discovery in
both the social and the natural sciences should differ. He nevertheless
identified a method which appears more feasible in the former than in the
latter, namely the method of rational reconstruction, or rational model-
construction.25 Popper (1957, p. 141} described this method as "... the
method of constructing a model on the assumption of the possession of
complete rationality (and perhaps also on the assumption of complete
information) on the part of all the individuals concerned, and of estimat-
ing the deviation of the actual behaviour of people from the model
behaviour, using the latter as a kind of zero co-ordinate." New Classicals
sided with Popper on his methodological monism. In their attempts to
solve the narrow Lucas Problem, they adopted his method of rational
model-construction.26
New Classicals explain the narrow Lucas Problem in terms of
expectational errors, which caused economic agents to misperceive their
decision situation. This implies that they must specify the manner in which
the agents' perceptions are incorrect, that is, New Classicals must specify
some benchmark. This benchmark decision situation is the situation of full
information, in which agents do not make expectational errors. It is used
as Popper's 'zero coordinate'. Deviations from this benchmark are accoun-
ted for by restricting the agents' information sets. These restrictions lead
agents to respond to monetary shocks as if these were real ones. The
responses and the underlying decisions are nevertheless considered to be
rational, in the sense that they follow unambiguously from the logic of the
decision-maker's perception of his decision situation, mistaken as this may
be. This rationality means that the decisions are uniquely determined, or,
in Latsis's (1976) terms, that they are made in single-exit decision situ-
ations. As Huussen (1985, p. 161} observed, rationality in this sense is a
feature of the decision situation rather than a characteristic of the deci-
sion-maker. The New Classical version of Popper's method of rational
model-construction thus holds that economic reality must be interpreted
25 Popper (1957, p. 141n2) noted, though, that this method " ...has some vague parallel
in the natural sciences, especially in thermodynamics and in biology (the construction of
mechanical models, and of physiological models of processes and organs)."
26
This interpretation was already suggested by Huussen (1985, p. 161), who refers to
the NCE as the avant-gardistic disequilibrium approach of Friedman and Phelps (whereby
'disequilibrium' refers to the REE).
201
as the outcome of a rational, single-exit decision-making process. 27 Devi-
ations of actual from model behaviour are explained in terms of the
difference between the subjective decision situations of the agents and the
theorist's conjecture about the 'objective' decision situation. This differ-
ence may result from the agents' expectations formation mechanism
and/ or from the information used in such expectations formation.
Initially, Lucas and Rapping (1969a) specified a partial-equilib-
rium model, attributing expectational errors to lagged adjustment in
expectations revision. However, the substitution of the REH for the AEH
removed this source of expectational errors, so that incomplete informa-
tion sets remained the only possible source of expectational errors. As
Lucas (1981, p. 7) indicated, "[a]t this point, it became clear to me why
Phelps had imagined an island economy, with traders scattered and short
on useful, system-wide information." Lucas's (1972a) model adopted this
'islands approach', and became the modelling style of New Classical
business cycle theory. 28 It thereby considered the construct of general
equilibrium a heuristic principle, not a situation to be explained.
The assumption of perfect price flexibility, implied by Lucas's
general-equilibrium framework, thus eliminates coordination failures. It
explains the narrow Lucas Problem as the result of expectational errors
which arise because of incomplete information only. This incompleteness
may be due to a lack of information about the structure of the economy,
or about the past and present values of relevant variables. The assumption
of Muth-rationality and the implied absence of systematic expectational
errors holds that in the aggregate agents know the 'correct' structure of
the economy. As a result, expectational errors arise only because of
incomplete information about relevant variables.
The New Classical method of discovery was supplemented by a
method of 'proof, which concerns the empirical validity of the logically
derived mathematical models. The first empirical tests was that of Lucas
(1973), who aimed to corroborate the joint NR/RE hypothesis. This
conclusion about its aim follows from two observations. Firstly, Lucas
(1973) included past deviations from the natural rate of output as an
explanatory variable of the persistence in the temporary (cyclical) compo-
27 Koertge (1975, p. 441) and Huussen (1985, pp. 129 - 30) note that Popper's method
of rational model-construction may explain in terms of multiple-exit situations, although
the latter is somewhat sceptical about the feasibility of such explanations.
28 As was shown in Section 8.7, Sargent (l973b) provided his aggregate models with
microfoundations by referring to Lucas's work.
202
nent of real output. This inclusion was not (yet) based on individual
rational behaviour, so that it violated the New Classical situational
constraint of methodological individualism. Instead, it constituted an 'ad-
hoc' stratagem, designed to bring the model into accordance with the
data. The second observation is that Lucas (1973) interpreted his test
results rather favourably, in the sense that his interpretation depended
heavily on two extreme cases. Both observations indicate that Lucas
(1973) sought to corroborate the joint hypothesis, not to falsify it. This
conclusion also applies to Sargent's (1973b) tests. Its results were rather
unfavourably for the joint hypothesis, but Sargent did not reject it.
Instead, he distinguished four reasons why such a rejection would be
premature. As was shown in Chapter 8, these reasons refer to violations
of either the initial conditions of the test or the ceteris-paribus clause. He
thus 'explained away' the discorroborations, which indicates that his tests
tried to corroborate the joint hypothesis.
To conclude, New Classicals adhere to the method of rational
model-construction as their method of discovery, and interpret rationality
in terms of single-exit decision situations. These decision situations were
formalized as general-equilibrium models, whereby the notion of general
equilibrium serves as a heuristic principle. The assumption of Muth-
rationality ensured that the agents' perceptions of these situations differed
only randomly from the objective ('true') situations. Lucas's (1973) and
Sargent's (1973b) empirical tests of these models aimed at corroboration.
The models were subsequently revised in order to eliminate some theor-
etical and empirical deficiencies. The Slutzky-Frisch distinction between
impulses and propagation mechanisms thereby proved very helpful.
9.6. CONCLUSIONS
This chapter described two theoretical developments in New Classical

business cycle theory. The first concerns Lucas's (1975) response to Rees's
(1970) and Hall's (1975) criticism that competitive-equilibrium models
cannot account for persistence in aggregate output and unemployment.
This response showed that the effects of unanticipated and unperceived
random shocks are propagated to subsequent periods by a propagation
mechanism. In particular, Lucas (1975) assumed that capital was market-
specific, so that it can only be transacted on the market in which it had
been invested. Barro (1980) subsequently showed that the introduction of
203
a global capital market would not invalidate Lucas's (1975) analysis if a
third shock was added as well.
New Classical business cycle theory was subsequently recon-
structed in the form of the Lucas programme. This programme consists of
the Lucas Problem and some situational constraints. The broad Lucas
Problem seeks to explain how the actions of rational agents result in an
overall pattern of economic activity which exhibits cyclical fluctuations.
New Classicals narrowed it down by eliminating the coordination problem.
That is, they use the concept of general equilibrium as a heuristic prin-
ciple in terms of which to solve the narrow Lucas Problem. Additionally,
any acceptable solution of this problem should be subject to three con-
straints, namely methodological individualism, subjectivism, and methodo-
logical monism.
New Classicals adopt a probabilistic version of the homogeneity
postulate for economic agents, which holds that economic agents do not
differ from each other in any systematical way. They reformulate macro-
economics in terms of the behaviour of a 'representative agent'. Such a
reformulation, however, does not appear to transform a holistic explana-
tion into an individualistic one, because it does not allow for the analysis
of coordination failures and redistribution effects. As a result, the NCE
adopts a somewhat misleading terminology, in the sense that it attaches
an unusual connotation to the notion of 'individualism'.
The second situational constraint of the NCE concerns the
subjective nature of the data in economics, in the sense that it stresses the
importance of the agents' perceptions of their own decision situations.
Like Hayek, New Classical economists thereby distinguish between the
role of knowledge in the decision-making process and its truth status.
Business cycles are explained as the unintended result of the misperceived
decision situations. For reasons of analytical convenience, New Classicals
adopt the simplifying assumption that the agents' perceptions are Muth-
rational, which holds that the 'subjective' probability distributions are
identical to the 'objective' distributions. This suggests that New Classical
methodology is better interpreted as objectivistic instead of subjectivistic.
New Classicals use a version of Popper's method of rational
model-construction as their method of discovery. The NRE is used as a
benchmark, and deviations from this benchmark are explained in terms of
incomplete information sets only. This means that the NCE constructs
models in which differences between the subjective decision situations of
the agents and the 'objective' decision situation explain social phenomena
204
(such as business cycles) as the unintended outcomes of the former.
Furthermore, the resulting REEs are restricted to single-exit decision
situations, in which the agents' decisions are the determinate outcomes of
their decision situations. This Popperian method of discovery is supple-
mented by a method of proof, which tries to establish whether the logical-
ly derived mathematical models are empirically valid. Early New Classical
econometric practice showed that it was mainly concerned with corrobor-
ation of these models. However, the role of testing in New Classical
Economics would change in time. This becomes clear in the attempts to
test the so-called neutrality proposition. This proposition and its tests are
the subject of the next chapter.
205
10. THE EFFECTIVENESS OF
MONETARY POLICY
10.1. INTRODUCTION
Almost simultaneously with Lucas's (1975) explanation of persistence,

Sargent and Wallace (1975) and Barro (1976) addressed a rather different
issue, which concerns the theory of economic policy. The present chapter
is devoted to their respective analyses of the neutrality (or policy-ineffec-
tiveness) proposition. Section 10.2 describes this proposition, which holds
that active monetary stabilization policy cannot lead to better results than
a fixed-percent money growth rule a la Friedman, so that any such
stabilization policy is ineffective. It also includes the debate between New
Classicals and their opponents on the logical validity of this proposition.
The debate was not only concerned with the logical but also with
the empirical validity of the neutrality proposition. New Classicals tried to
provide some empirical evidence in favour of the proposition. Section 10.3
distinguishes between two test approaches, namely one following Sargent
and Wallace (1975) and the other adopting Barro's (1976) framework.
The discussion of the former also includes Sargent's (1976b) notion of
'observational equivalence', which holds that empirical tests can only
establish the empirical validity of the policy-ineffectiveness proposition
over an estimation period in which a break in policy rule has taken place.
Neftci and Sargent (1978) carried out such a test, which in their view
provided a little bit of evidence of the natural rate hypothesis (and the
policy-ineffectiveness proposition). Barro, on the other hand, took a more
direct route by estimating unanticipated changes in the supply of money.
Initially, his test results favoured the joint NR/RE hypothesis, but later
tests led him to conclude that the general price level could not serve as
the principal channel for the transmission of the effects of money sur-
prises, because output responded faster to such surprises than prices.
The policy-neutrality debate resulted in the insight that the policy-
ineffectiveness proposition crucially depends on the specification of the
agents' information sets. The question arose whether the proposition
remains valid if agents observe a common price and if they hold different
(i.e., agent-specific) information. King (1982) addressed these issues by
extending Barro's (1976) model. He showed that the inclusion of agent-
207
specific information and of common price observations invalidated the
proposition, as will be discussed in section 10.4.
New Classicals explain the relationship between money and
output as the result of expectational errors. This implies that the non-
neutralities of money arise because of unanticipated, or rather unper-
ceived, changes in the supply of money. Barro and Hercowitz (1980) tried
to test whether such unperceived changes (Granger-)cause fluctuations in
output and unemployment. Their (negative) test results will be discussed
in section 10.5. They meant a serious blow for the monetary version of
New Classical business cycle theory. Jointly with the results of Barro
(1978) and Barro and Rush (1980) as regards the empirically insignificant
role of changes in the general-price level, they eventually led to a switch
from monetary to real business cycle theory. This type of business cycle
theory, however, falls outside the realm of this study. Its discussion in sec-
tion 10.6 will therefore be very brief.
The policy-ineffectiveness debate thus led to several empirical
attempts to establish its (in)validity. This raises the methodological
question as regards the role of testing in New Classical Economics.
Section 10.7 provides a methodological assessment of this role. It will be
argued that New Classical econometric practice was mainly concerned
with establishing whether models were satisfactory. Following Morgan
(1988), five criteria of satisfactoriness will be distinguished. Section 10.8
offers some conclusions.
10.2. THE POLICY-INEFFECTIVENESS PROPOSITION
According to Lucas (1980, p. 271), theoretical economics should provide

artificial economic systems (models) in which the effectiveness of econ-
omic policy can be studied. New Classicals first concentrated on the
policy-ineffectiveness proposition. This is not to say that they actually
maintained that monetary policy was ineffective in stabilizing economic
activity. Instead, the proposition was 'merely' used as a point of reference.
It was derived in two ways, namely by Sargent and Wallace (1975) and
Barro (1976). These derivations differed from each other, in the sense
that they adopted different policy aims. Sargent and Wallace (1975)
assumed that the monetary authority aims to minimize the gap between
actual and capacity output, whereby the latter is defined as "... the value
that y. [i.e. aggregate output] would assume if there were no expectational
208
errors and if output equalled capacity in the previous period" (McCallum
(1980, p. 727)). Barro (1976), on the other hand, defined his benchmark
in terms of the absence of expectational errors only. That is, his bench-
mark depends on the absence of expectational errors and on actual
(instead of capacity) output in the previous period. The respective 'bench-
marks' will be called capacity output and full-information output.
As will be shown below, the difference in benchmark does not
lead to different results as regards the effectiveness of monetary policy.
Instead, these results mainly depend on the informational assumptions
adopted. This is not to say, though, that the difference is irrelevant,
because it led to different test approaches.
10.2.1. Minimizing the gap between actual and capacity output

Sargent and Wallace (1975) specified a model which resembles that of
Sargent (1973b) and which consists of an aggregate supply function, an
aggregate demand function, a portfolio balance schedule, and a function
for the determination of productive capacity. Aggregate supply (yt) is
assumed to depend on the previous period's productive capacity (kt_ 1),
expectational errors about the current price level (pt_ 1 - t-tP ·t), and a
random disturbance term utt. 1 The aggregate demand function contains
four terms, namely the previous period's productive capacity, the real rate
of interest (defined as the nominal interest rt minus the anticipated rate of
inflation in period t, formed on t-1 (t_tP.t+ 1 - t-tP·t)), some exogenous varia-
bles (Zt), and a random disturbance term u 2t. The variable Zt is a summa-
tion of exogenous variables (such as government expenditures and taxes)
plus a random error term ~t· Portfolio balance is maintained when econo-
mic agents are satisfied with the division of their assets between money
and financial assets (bonds and equities, which are assumed to be perfect
substitutes). The demand for real money balances is a function of real
output (yt), the real rate of interest (as defined above), the exogenous
variables Zu and a random disturbance term u 3t. The random error terms
are serially uncorrelated. Their conjunction solves the impulse problem.
Unlike Lucas (1975), Sargent and Wallace (1975) did not aim to
explain persistence. Following Poole (1970), they tried to study the effecti-
veness of monetary policy for stabilization purposes. As a result, they did
not adopt Lucas's assumption that capital consists of physical market-
1
All variables in this section are expressed in natural logarithms. All page numbers
refer to the 1981 reprint of Sargent and Wallace (1975).
209
specific goods, but instead concentrated on capital in portfolio, assuming
that it consists of bonds and equities. The respective stocks of these assets
can be adjusted rather easily (disregarding transaction and other adjust-
ment costs), which means that capital is not market-specific. As a corol-
lary, Sargent and Wallace did not adopt the 'islands approach'. Their
model is an aggregate model, with an economy-wide capital market and
without rigid stock variables which may serve as a propagation mech-
anism. Stated differently, this absence of rigidities in their economic
system presupposes that there are no adjustments costs. This assumption
may be called the no-frictions assumption in a narrow sense. This no-
frictions assumption can also be applied to the price system, in the sense
that there are no price-adjustment and transaction costs. This broad no-
frictions assumption then ensures perfect price flexibility.
The analysis of policy-effectiveness presupposes some criterion of
effectiveness. Sargent and Wallace (1975, p. 218) specified an aggregate
'ad-hoc loss function', which provided an indication of the difference
between actual and capacity output? The smaller the 'losses' in aggregate
output (i.e. the deviation of actual from capacity output, as defined
above), the more effective the policy under consideration. The monetary
authority was supposed to minimize this loss function by applying either of
two policies. The first consists of a money supply feedback rule, and the
second pegs the nominal rate of interest by means of a deterministic
linear feedback rule. Sargent and Wallace (1975) analyzed the effec-
tiveness of these rules under the REH. The outcome of this expectations
formation process depends on the agents' information sets, so that some
informational assumptions must be adopted. Sargent and Wallace first
assumed that both the authority's and the agents' sets included the current
and past values of all endogenous and exogenous variables (p. 221). Given
Muth-rationality, a stable money supply rule then cannot 'fool' people
systematically. Expectational errors are therefore random and non-system-
atic, so that in the absence of a propagation mechanism aggregate output
differs only randomly from its natural (capacity) rate. This means that any
deterministic money supply rule is equally ineffective in stabilizing
aggregate output around its capacity level (p. 221). The different rules
2 The term 'ad hoc' was meant to indicate that the function ".. .is not derived from a
consistent set of assumptions about individuals' and firms' objective functions and the
information available to them" (p. 215).
210
only affect the general price level differently.3 This proposition is Sargent
and Wallace's version of the neutrality proposition.
The second policy considered by Sargent and Wallace (1975) aims
to peg the nominal interest rate (as defined above). This policy requires
the monetary authority to accommodate whatever quantity of money is
demanded at the pegged interest rate. This rate then cannot transmit
information about changes in the economy, so that any price expectation
will be as rational as any other. Sargent and Wallace concluded that
"[t]here is nothing to anchor the expected price level. And this is not
simply a matter of choosing the 'wrong' level or rule for the interest rate.
There is no interest rate rule that is associated with a determinate price
level" (p. 224 ). Such a policy is therefore unsuited as a stabilization policy.
The third and final exercise concerned the analysis of a situation
in which the monetary authority follows a deterministic monetary feed-
back rule and in which it has informational advantages over the economic
agents. Sargent and Wallace showed that this informational assumption
may render the neutrality proposition invalid, because "... to exploit the
information discrepancy, the monetary authority must know what it is" (p.
227).
According to Sargent and Wallace (1976, p. 210), these policy-
neutrality results showed that the crucial question in economic analysis
becomes whether policy changes are anticipated.4 They concluded that if
the answer to this question is affirmative, deterministic monetary feedback
policy does not minimize the gap between actual and capacity output.
However, one may criticize their argument by claiming that it takes time
to learn a policy change. As a result, a policy rule with feedback may be
optimal, because during this learning time the new money supply rule can
affect real variables. Sargent and Wallace (1976) rejected this argument
on the grounds that such learning behaviour cannot be exploited, since "...
new rules are not adopted in a vacuum. Something would cause the
change ... " (p. 211). As a result, rational economic agents will not treat the
parameters of these rules as fixed, and will form expectations about the
factors which cause such changes. Given the REH, they will not make
systematic expectational errors about the changes in the policy rule, so
that the neutrality proposition still applies, albeit at the underlying level
of the political decision-making process. Thus, even if it takes time to
3
This result does not hold for non-linear rules. Cf. Shiller (1978) and Snower (1984).
4
Page numbers refer to the 1981 reprint of Sargent and Wallace (1976).
211
learn a change in money supply rule, such learning behaviour cannot be
exploited.
Sargent and Wallace (1975, p. 228) recognized that the neutrality
conclusion crucially depends on their theoretical structure. Changes in this
structure may lead to different conclusions.5 However, both authors
opined that their policy conclusions were rather robust to alterations in
the aggregate demand and portfolio balance schedules, so that they
crucially depend on the joint NR/RE hypothesis. That is, the neutrality
proposition is implied by the conjunction of the joint NR/RE hypothesis
and the assumption that the monetary authority does not have an
informational advantage over the economic agents. Sargent and Wallace
(1976, p. 208) recommended the joint hypothesis on two grounds. Firstly,
they claimed that it was "... consistent with the theoretical core of econ-
omics - optimizing behavior with a coherent general equilibrium frame-
work ... " (i.e. with a 'natural-rate' framework), and secondly, that it was
not discorroborated by the facts. This indicates that the absence of
discorroboration is a criterion for acceptance of the joint hypothesis.
Furthermore, they stressed the need for microfoundations of macroecon-
omics. In their view, the 'natural-rate' models "... seem to be the only
existing ones consistent with individuals' maximizing behavior that are
capable of rationalizing certain important correlations, such as the Phillips
curve, that exist in the data and are summarized by the reduced forms of
macroeconometric models" (1976, pp. 208 - 9). Methodological individual-
ism is thus not only a situational constraint of the Lucas programme, but
also of New Classical work on the theory of economic policy.
To summarize, Sargent and Wallace (1975) and (1976) showed
that the joint NR/RE hypothesis, combined with some particular informa-
tional assumption, has serious consequences for the feasibility of active
stabilization policy. Such a policy will be anticipated and hence incapable
of affecting real variables. This neutrality proposition led to the con-
clusion that Friedman's fixed-percent money growth rule without feedback
is optimal in terms of an 'ad-hoc' quadratic loss function. The monetary
authority can only systematically 'fool' rational economic agents if there is
a discrepancy between its information set and that of the agents, and if it
precisely knows the information discrepancy. However, these results
crucially depend on the theoretical structure. Sargent and Wallace
5 The neutrality proposition need not hold in the case of an open economy (cf. Montiel
(1987)), or if the assets are imperfect substitutes (cf. Minford (1986)).
212
advanced two reasons why this structure was satisfactory. Firstly, it is
consistent with methodological individualism, and secondly, the joint
NR/RE hypothesis was considered to be corroborated by the facts. They
subsequently adopted the sophisticated-falsificationist position that they
knew of no superior alternative models.
Despite the fact that they considered the crucial elements in their
theoretical structure to be satisfactory, Sargent and Wallace (1976, p. 228)
downplayed the importance of the neutrality result.6 The main reason for
this was the fact that this result was derived from an 'ad-hoc' quadratic
loss function which implied that the aim of monetary policy was to
stabilize some measure for aggregate output (or, in a growing economy,
for the rate of growth of aggregate output). Barro (1976) proposed to use
another criterion for policy evaluation.
10.2.2. Minimizing the gap between actual and full-information

output
Barro's (1976) island model contains two types of goods, a non-storable
commodity and money, so that the latter is the only store of value. The
local commodity supply function contains four terms, namely (1) a system-
atic supply term k t(z), (2) the term (Pt(z) - EPt+ 1 l/t(z)), which measures
1
the current local price relative to the expected future price level, and thus
provides an indication of the agents' expectations to gain from inter-
temporal substitution of labour, (3) a wealth variable, which is measured
as next period's expected money supply Mt + .6.Mt+.l/t(z), deflated by the
expected average price level in that period EPt+ 1 I/t(z ), and (4) two
random error terms.7 The first error term concerns shifts in aggregate
output u:, and is assumed to follow a random walk, so that ut = ut_ 1 + v"
where vt is normally distributed with mean zero and variance u; . The
second error term concerns shifts in relative supply, t:(z)) (pp. 3 - 5). The
specification of the local demand function is analogous to that of local
supply, thus also containing two types of shocks. This means that there are
two global and two local shocks. Furthermore, the money supply Mt is
assumed to fluctuate only randomly, so that flMt = Mt - Mt_ 1 = m" which
is assumed to be serially independent and uncorrelated with the other dis-
6
See also Sargent (in Klamer (1984, p. 70)).
7
Barro (1976) expressed all variables in natural logarithm.
213
turbances (p. 7).8 Additionally, agents are assumed to have full local
information, whereas global information is lagged one-period. Information
is then homogeneous across agents on the same market, and heterogen-
eous across markets. 9
Barro (1976) performed two simplifying summations of shocks,
yielding one aggregate and one local commodity shock. His model thus
includes three shocks, one local shock (et(z)) and two aggregate (uu mt)
shocks. Jointly with the assumption of lagged information, the three
shocks solve the impulse problem. The local one is transitory and hence
does not have any effects in the next period. The global shocks, though,
follow a random walk, which means that their effects persist in the next
period, so that they affect that period's general price level (p. 12). In this
sense they are permanent. Barro showed that the magnitude of the effect
of the monetary aggregate shock on local output depends on the variances
of all these shocks (p. 11 ). Given the other variances, this effect decreases
with a rise in its own variance. As Lucas (1973) already suggested, agents
will become more inclined to ascribe an increase in their local price level
to monetary disturbances if the rate of money growth fluctuates more
wildly.
Barro (1976) subsequently addressed the gap between the local
price on a randomly selected market z' and the agents' expectations about
the future general price level, based on current local information. This
gap can be broken down into three components. Firstly, it may arise
because a difference between actual local and actual global prices: Pt+ 1(z')
- Pt+I· Secondly, there may be a difference between the actual and the
expected general price level: Pt+ 1 - EPt+II/t (where It denotes current full
information). Thirdly, the expected general price level may differ from the
expected local price level: EPt+ 1 l/t(z) - EPt+ 1 l/t(z). Barro (1976, pp. 13 -
14) expressed these components in terms of the shocks and their vari-
ances, which enabled him to analyze the effects of (changes in) monetary
policy on the predictability of future prices.
8 This eliminates the Tobin effect, as included by Lucas (1975).
9 Barro (1976, p. 4n4) opined that Lucas's (1975) model also allowed for heterogen-
eous information across agents on the same market, so that it is more complicated in this
regard. However, as was shown in the previous section, Lucas assumed that at the begin-
ning of each period agents 'pool' their estimates. He thus averaged the agents' different
expectations about the state of the system, and summarized them in what may be called
the expectations (ket' met) of the 'representative agent'. Given Muth-rationality, this
procedure eliminates systematic expectational differences across agents on a given market.
214
Sargent and Wallace (1975) had argued that monetary policy
should aim to minimize the gap between actual and capacity aggregate
output. Barro criticized this policy aim because it does not take into
account that the distribution of output may vary across markets in
response to local excess demand shocks e1(z ). He proposed to incorporate
such shocks in his model and therefore replaced Sargent and Wallace's
criterion for policy evaluation by the minimization of the expected
squared gap between actual and full-information output in each market
(p. 15). This squared gap was expressed in terms of the variances of the
three shocks (p. 17). Barro showed that the variance of output about its
full information position is minimized if the variance of money ~ equals
zero. Mter all, an increase in any of the underlying variances ( o! ,u; , or
~ ) reduces the predictability of current prices, which thus become less
accurate signals for market participants. This renders it more difficult for
them to get output at its full-information level. Hence, "[t]o the extent
that the variance of money, o! , can be controlled, the smallest possible
value would be optimal" (p. 18), because monetary policy is optimal when
it is as predictable as possible.
It should be noted that the above policy result holds under
Barro's assumption that the supply of money follows a random walk, and
hence that the rate of money growth equals zero. The monetary authority
was assumed not to include feedback effects from observed economic
variables. Barro (1976) also considered the case in which a feedback
policy is pursued. The optimal policy then crucially depends on the
assumptions about the authority's and the agents' respective information
sets. In particular, he discussed three situations, namely ( 1) the monetary
authority and the public have identical information about the economy,
(2) the authority has superior information about the economy, and (3) the
authority has superior information about its (countercyclical) money
supply rule.
In the first situation, the monetary authority is assumed not to
have an informational advantage about the state of the economy. Both the
authority and the public have lagged global information. Barro (1976, p.
20) assumed that the former's countercyclical policy takes the form of a
compensating shock in aggregate excess demand (v1): f1M1 = m. - "YVt-h
with "Y > 0. This policy rule implies that the authority can only influence
aggregate (global) excess demand, and not local excess demand. Barro
subsequently solved his model and showed, analogously to Sargent and
Wallace (1975), that the countercyclical policy only creates effects which
215
are fully perceived. He concluded that "[b]ecause the market participants
know the form of the money rule, and take this behavior into account in
forming expectations of future prices and monetary growth rates, the
feedback from vt-t to Mt has no effect on the entire distribution of output
[nor on its level]" (p. 21). The agents' full information about the autho-
rity's policy rule renders countercyclical policy ineffective. Thus, in this
situation the substitution of Barro's (1976) criterion of policy evaluation
for that of Sargent and Wallace (1975) does not alter the latter's result.
In the second case, the monetary authority has superior informa-
tion about the state of the economy. In particular, the authority is
assumed to have full current information, whilst as before economic
agents have only lagged global information (p. 22). The authority can then
base its monetary policy on current shocks to aggregate excess demand, v"
so that the appropriate money supply rule becomes: il.Mt = mt - ovu with o
> 0 describing a countercyclical policy (p. 22). The agents' information
sets do not include information about current aggregate shocks, so that
they will not be able to solve the signal extraction problem correctly. The
monetary authority, on the other hand, knows the nature of the shocks.
Given the policy aim of minimizing the variance of aggregate output
around its full information level, it must set o at such a level as to mini-
mize the variance of the aggregate excess demand shock, ol. This vari-
ance is a weighted sum of the respective variances of changes in the
money supply (a;. ) and real shift in aggregate excess demand ( civ ). The
variance in aggregate shocks is minimized if the monetary authority
offsets the effects of independent money shocks mt by changes in aggre-
gate excess demand vt. This prevents a confusion between absolute and
relative price changes. Obviously, the same result may be brought about
by providing the agents with the relevant global information.
In the third case, the monetary authority has superior information
about its own money supply rule according to which it conducts countercy-
clical policy. Barro (1976, p. 24) argued that under these informational
conditions the monetary authority can systematically 'fool' the agents. 10
However, he noted that in his model "... there appears to be no basis for
policy deception as long as the policymaker's objective is based on
minimizing the gap between actual and full-information output" (p. 24 ).
Mter all, given this criterion for policy evaluation, "... the best monetary
policy is always the policy that is most predictable" (p. 25). By assumption,
10 This had already been shown by Taylor (1975).
216
the monetary authority does not have an incentive to 'fool' the agents.
10.2.3. Rules versus discretion

The neutrality proposition implies that fixed-percent money growth rules
are as (in)effective as active monetary stabilization policies which follow
deterministic feedback rules. According to Kydland and Prescott (1977),
the former are even superior to the latter. They explained this superiority
in terms of the so-called time-inconsistency of optimal policy. Optimal
control theory aims to select the best decision or policy at each point,
given the current situation and similar future decision-selecting methods
(p. 619).n Such a best decision is called an optimal policy. Kydland and
Prescott realized that the REH and the ensuing Lucas critique may render
such optimal policies sub-optimal when considered over a longer planning
horizon. Mter all, rational economic agents will form expectations about
the policy-selecting process (i.e. the feedback rule). The Lucas critique
already showed that "... changes in policy induce changes in [economic]
structure, which in turn necessitate reestimation and future changes in
policy, and so on" (p. 620). Kydland and Prescott (1977, p. 620) recog-
nized that this iterative process need not converge. However, given such
convergence, "... the resulting policy was [time-]consistent but sub-
optimal." This sub-optimality can be made clear by the example of active
monetary feedback policy.
Consider a short-term Phillips-Curve relationship, which holds
that the deviation of the unemployment rate from its natural rate (ut - u *)
is a positive function of the agents' underestimation of the rate of infla-
tion (p~ - Pt). Given the government's aim to lower unemployment, it will
be optimal in the current period to increase the rate of money growth, so
that agents underestimate the inflation rate. Rational economic agents
will (eventually) take such increases into account. They will adjust their
inflation expectations and hence their actions. This means that they
correct their errors in the previous period, so that in the long( er) run a
situation of hyperinflation arises without a reduction in unemployment
(which will be at its natural rate). The agents' expectations adjustment
process reflects that their goals are inconsistent with those of the monet-
ary authority, in the sense that what is optimal for the former is sub-
optimal for the latter. According to Kydland and Prescott (1977), this
time-inconsistency of plans favours fixed rules over discretionary policy.
11
Page numbers refer to the 1981 reprint of Kydland and Prescott (1977).
217
The monetary authority can bring about better long-term results if it
adopts a fixed-percent money growth rule instead of a deterministic
feedback rule, at least if agents believe that it will not deviate from the
former. In both cases unemployment will be at its natural rate, but under
fixed-percent money growth rules there will be no hyperinflation. This
fixed-rule policy renders the plans of economic agents and the monetary
authority consistent, but it is sub-optimal for the period in which the
policy-decision was taken. It will be clear that the authority's reputation is
crucial for the agents' responses.
10.2.4. The debate on the neutrality proposition

The neutrality proposition has evoked much criticism. Its plausibility was
questioned by several authors, who particularly criticized the assumption
of perfectly flexible prices. Phelps and Taylor (1977) substituted this
assumption by a form of price rigidity. They followed Gray's (1976)
explanation of price rigidity, which holds that all labour contracts have a
duration of one period and are drawn up at the end of the period previ-
ous to the 'contract period' (i.e. the period in which the contracts
apply}. 12 Economic agents set the nominal wage rate without full infor-
mation about other variables relevant to production decisons. The result-
ing wage rate is therefore 'sticky' over the contract period. Under these
circumstances, the neutrality proposition proves to be invalid, because
agents cannot react to new information.
The second strategy to model price rigidity was adopted by
Fischer (1977). He addressed a situation in which "... economic agents
contract in nominal terms for periods longer than it takes the monetary
authority to react to changing economic circumstances ... " (p. 191}. He
assumed that the duration of contracts is two periods instead of one, and
that at the end of each period new nominal wage-rate contracts are drawn
up for one half of the work force. Such multi-period ('long-term') con-
tracts can be rationalized as the result of the firms' attempts to minimize
their transaction costs of frequent price setting and wage negotiations (p.
194}. Unanticipated (though not necessarily unperceived) shocks to the
economic system then cannot be translated into changes in nominal wage
rates and/ or prices, even if they are fully perceived. This means that the
monetary authority can use its informational advantage (over the contract
12 Gray (1976) herself did not study the relationship between money and output, but
analyzed the effects of indexing the nominal wage rate to the general price level.
218
period) in such a way as to reduce the variability of aggregate output.
Hence, Fischer's long-term contracting approach also renders the neutral-
ity proposition invalid. 13
Barro (1977a) argued that the Gray-Fischer indexation approach
does not explain price rigidity in terms of individual maximizing behav-
iour. In his view, "... the crucial element - and the aspect that accurately
marks this approach as 'non-market clearing analysis' - is the nonexecuti-
on of some perceived mutually advantageous trades (where trades may
include side payments)" (p. 315, italics in original}. It may be profitable
for both firms and workers to execute these trades, thus maximizing the
'total pie available to firms and workers', and subsequently to redistribute
the 'gain in pie' by some payment arrangement (p. 311 ). In this manner
both parties can gain by deciding not to draw up contracts which render
the nominal wage rate 'sticky'. Such contracting behaviour can thus only
be explained in terms of some sort of adjustment (transaction) cost. This
'friction' induces agents to minimize the number of contracts drawn up. 14
New Classicals, however, consider such 'frictions' as 'ad-hoc', in the sense
that their inclusion is not based on economic-theoretical considerations. 15
Sargent and Wallace's (1975) and Barro's (1976) respective ver-
sions of the neutrality proposition led B. Friedman (1979) to argue that
New Classicals have done nothing but to relabel the 'long run' in
informational terms. That is, the NRE (cf. Chapter 8) is simply a full-
information or long-term general-equilibrium situation. He criticized the
NCE on the grounds that they did not explain how such long-term equilib-
rium positions are reached. That is, it is not clear why the probability
distributions of the stochastic shocks have 'settled down', and how agents
come to know them. Other economists have analyzed the circumstances
under which agents will learn these 'objective' distributions and form their
13
Fischer (1977, p. 201) recognized that his model would yield the neutrality proposi-
tion in the case the wage rate is indexed in a way which duplicates the effects of one-
period contracts.
14
Note that this jeopardizes Gray's assumption that contracts are drawn up for each
and every period, at the end of the previous period. Fischer's assumption about the
duration of the contract period appears more plausible in this regard.
15
The above criticisms suggest that the neutrality proposition crucially depends on the
assumption that markets continuously clear. Although there is some truth in this sugges-
tion, McCallum (1977) showed that it overemphasizes the importance of price-stickiness.
219
expectations according to the REH. 16 The results of the studies are rat-
her ambiguous. As Blume, Bray and Easley (1982) have shown, two
approaches exist in economic literature, the one yielding convergence to
rational expectations and the other concluding that the formation process
of rational expectations may not be stable because agents have incorrectly
specified the likelihood functions. The former result crucially depends on
the specific learning process assumed. Bray (1982, pp. 329- 30) concluded
that the REH is at best a long-run phenomenon. Shiller (1978, p. 38) was
more radical in his rejection of the REH. He argued that if a transition
period is allowed during which economic agents may learn to form their
expectations according to the REH, the question arises as to how expecta-
tions are formed during that period and as to how such learning takes
place. Given the non-uniqueness of the solution of New Classical models,
it is not clear which solution economic agents will choose as their starting
pointY McCallum (1983, p. 139) rejected this criticism, using a tu quo-
que argument, which holds that the convergence problem "... is not prop-
erly attributable to the rationality hypothesis but, instead, is a general
feature of dynamic models involving expectations." This suggests that the
use of any expectations formation mechanism cannot be based on econ-
omic theory alone and hence is to some extent 'ad-hoc'.
To summarize, the neutrality proposition was derived in two ways.
Sargent and Wallace (1975) considered the case in which the monetary
authority aims to minimize the gap between actual and capacity output,
whereas Barro (1976) concentrated on the minimization of the gap
between actual and full-information output. They showed that monetary
policy is ineffective for stabilization purposes if the monetary authority
does not have an informational advantage. Given the respective assump-
tions about the policy aims, there are no welfare-theoretic reasons why
the monetary authority should try to 'fool' the agents. Optimal policy
should then be as predictable as possible. This policy conclusion was
criticized by Fischer (1977) and Phelps and Taylor (1977), who both
16
See Shiller (1978), DeCanio (1979), Blume, Bray and Easley (1982), Bray (1982),
and Frydman (1982). Pesaran (1989, Ch. 3) discussed several types of learning models.
17
Furthermore, in the case of initial non-rational expectations formation processes
there is no need for people to use identical expectations formation mechanisms, nor to
revise them in such a manner as to bring about convergence towards the REH. However,
McCallum (1979, p. 287n4) objected that "[a]rguments based on expectational differences
across individual consumers or firms amount to objections to macroeconomics, not rational
expectations." The neglect of such divergent expectations is merely the price to be paid for
'doing' macroeconomics.
220
introduced frictions into New Classical models. These frictions ensured
some form of price rigidity, so that agents cannot respond to new, current
information. The existence of such frictions imply perceived though
willingly foregone profit opportunities. In this sense New Classicals label
them as 'ad hoc'. A second criticism against the neutrality proposition
holds that agents must learn to be Muth-rational. During this learning
process monetary policy will be non-neutral. Lucas and Sargent (1978, p.
315), however, evaded the problem of learning. They adopted the infor-
mational and expectational assumptions merely for analytical convenience,
and not of necessity. 18 This reflects that they do not consider the 'real-
ism' of assumptions to be important as long as their models exhibit
satisfactory analogue properties, that is, as long as these models are
capable of generating the actual economic time-series.
10.3. TESTING FOR NEUTRALITY
New Classicals have not only studied the logical but also the empirical
validity of the neutrality proposition. Again, like the logical derivation of
the neutrality proposition, two approaches may be discerned. The first
adopted Sargent's (and Neftci's) approach, whereas the second built on
that of Barro (and Rush). These approaches will be discussed in that
order.
10.3.1. Sargent's tests

The starting point of Sargent's (1976a) test of the NRH was the following
reduced-form equation:
(10.1) Un, = L g;Un,_i + u,

i=l
where Un, is the unemployment rate at time t, gi is a parameter, and u, is

a serially uncorrelated disturbance term with mean zero and finite
variance (pp. 528 - 29). 19 Sargent advanced the idea that the joint
NR/RE hypothesis implies that the forecast of u, cannot be improved on
18
They claim that •[b]oth of these assumptions can be abandoned, albeit at a cost in
terms of the simplicity of the model. •
19
Page numbers refer to the 1981 reprint of Sargent (1976a).
221
the basis of information available at time t-1, so that E(utl 8t_ 1) = 0, where
is the information set at t-1. This implies that the forecast of the
(Jt-1
current rate of unemployment Unu based on lagged unemployment rates
(Unt-i) alone, cannot be improved by the introduction of lagged terms
other than lagged unemployment rates. This can be tested by checking
whether some variables Granger-cause the unemployment rate. This test
involved the following equation:
m ra
(10.2) (Jltt = L &iun,_i + L PiYt-i
j=l j=l
The joint NR/RE hypothesis leads to the null hypothesis that all parame-
.ters {jj are zero, i.e. that Yt does not Granger-cause Unt (p. 531). The re-
gressions were carried out for six variables. They included four lagged
values of the dependent variable (Unt) and six lagged values of the other
variable Yt. The test results showed that the coefficients {jj of two vari-
ables were significant at the 5% level, though not at the 1% level (p. 540),
namely the money supply mt and the wage index w. This result is unfavou-
rable for the joint NR/RE hypothesis. However, the Sims test, which also
tests for Granger-causality although with differently prepared data,
showed that none of the coefficients of the six variables were significant at
the 5% level, thus corroborating the joint NR/RE hypothesis?0
Sargent also carried out the reverse Granger and Sims tests,
which aim to establish whether the unemployment rate Granger-causes
the other variables. The null hypothesis advances that Unt does not
Granger-cause the six variables, that is, that the coefficients of the six
variables do not differ significantly from zero. The results of the reverse
Granger tests, however, indicated that at the 5% level the null hypothesis
must be rejected for only two variables, namely real and nominal govern-
ment expenditures. These two variables are thus Granger-caused by the
unemployment rate. The reverse Sims test yielded a similar result, in the
sense that Un Granger-causes real (though not nominal) government
expenditures. The results of both types of reverse tests indicate that the
unemployment rate Un Granger-causes only one or two (depending on
the test and the significance levels) of the six variables considered. These
results are remarkable because there are no theoretical reasons why the
20 Sargent (1976a, pp. 540 - 41) explained the difference in outcome between the
Granger test and the Sims test as the result of the sensitivity of the tests to the desea-
sonalizing process involved.
222
rate of unemployment would not Granger-cause the other variables. This
led Sargent (1976a, p. 549 -50) to conclude that the empirical results pro-
vided "[s]orne evidence for rejecting the model ... but it is far from being
overwhelming and decisive", although in his view his structural (classical)
model "... is not obscenely at variance with the data" (p. 547).21 However,
as Maddock (1979, pp. 180- 81) noted, the results of the reverse Granger
tests suggest that the power of Sargent's test procedures to discriminate is
rather weak. Sargent's test results are therefore difficult to interpret. This
problem of interpretation is even enlarged bearing in mind the possibility
of serial correlation in the disturbance term ut (p. 181). Such serial
correlation means that ut is correlated with the previous period's infor-
mation set (since this set also includes ut_ 1). As a corollary, the null hypo-
thesis E(utl8t_1) = 0 will never apply. The test for Granger-causality is
therefore invalid in case of serial correlation in the error terms.
Sargent (1976a) had restricted his empirical analysis to a reduced-
form equation. He had derived it from a structural model which exhibited
the 'classical' property of policy-ineffectiveness. In his 1976b-article,
however, he showed that the respective reduced-form equations of
'natural-rate' and 'unnatural-rate' models may be difficult to distinguish on
empirical grounds, so that there is observational equivalence. This has
serious implications for the ways in which the neutrality proposition
should be tested empirically. This can be made clear as follows. Sargent
(1976b, p. 555) first considered the following reduced-form equation: 22
(10.3) Y, = L a,mt-i + L P.Yr-i

i=O jzl
+ TJ,
This equation indicates that current aggregate output is determined by its

own lagged values and by the past and present values of the money
supply. Consider also a deterministic monetary policy rule with feedback.
Such a rule takes the following form:
21
According to Sargent (1976a, p. 550), the fact that his test results do not contradict
the policy-ineffectiveness proposition should disconcert the proponents of active stabiliz-
ation policies which are conducted according to feedback rules. In his view, these propo-
nents should provide empirical evidence in favour of their claim that the active policies are
optimal relative to Friedman's fixed-percent money growth rule. As King (1982, p. 249n1)
noted, New Classicals have been very successful ".. .in shifting the initial position taken in
evaluating stabilization policy. "
22
Page numbers refer to the 1981 reprint of Sargent (1976b).
223
(10.4) m, = E Y,.,_; + E aiY,-i + e,
i=l i=l
This rule will effectively stabilize aggregate output if the parameters in

equation (10.3) remain unaltered after a change in those of (10.4) (p.
556). The assumption that this invariance result prevails may be called
invariance assumption 1. Under this assumption the optimal stabilization
policy will be to minimize the variance of Yt around a target y •. This is
achieved by reducing this variance to that of flu so that y· = Yt - fit· From
equation (10.3) it follows that this optimal feedback rule can be written as
a rule of setting the current value of m as a function of lagged values of
m andy:
(10.5) «om, = - Li=l «1"-t-i - Li=l ~iYt-i + Y•
Thus, if the first invariance assumption holds, the money supply can be
manipulated in order to stabilize output, implying that deterministic
monetary policy with feedback is superior to a fixed-percentage money
growth rule. However, the Lucas critique asserts that economic agents will
form rational expectations about the authority's policy changes, so that the
first invariance assumption will not hold. That is, only the unanticipated
component of mt is relevant, so that equation (10.3) must be replaced by:
(10.6) Y, = L K;(mt-i-Et-i-tm,-;) + L 'J.ty,_; + 11,

i~O i=l
This equation holds that current aggregate output depends on its own past
values and on past expectational errors about changes in the money
supply (i.e. on unanticipated changes in this supply). The assumption that
a change in the parameters of the policy rule (10.4) do not affect the
parameters in (10.6) is called invariance assumption 2. Given the joint
NR/RE hypothesis and hence Muth-rationality, economic agents do not
make expectational errors, so that Et_ 1mt = mt. Substitution of this equality
in equation (10.6) yields an autoregressive representation of y, which is
independent of any deterministic policy rule (such as 10.5):
(10.7) Y, = L AiJ't-i + 11,

i=l
This shows that monetary policy following deterministic feedback rules is
224
ineffective for stabilization purposes. The second invariance assumption is
therefore consistent with the neutrality proposition. The problem now is
that equations (10.3) and (10.6) are observationally equivalent, in the sense
that they have identical residuals (71t) and hence fit the data equally well.
This led Sargent (1976b, pp. 557 - 58) to conclude that "... the empirical
evidence from a single estimation period alone, which can be completely
summarized by [(10.3) or (10.6)] and an autoregression for m, can never
settle things between advocates of rules with feedback and advocates of
rules without feedback." This problem of observational equivalence can
only be solved if it can be ascertained which invariance assumption
applies. Until then it remains ambiguous whether rules with feedback
dominate rules without feedback, or vice versa. The problem of observa-
tional equivalence thus prohibited an empirical solution of the neutrality
debate.
The relevant invariance assumption can only be ascertained after
a change in the parameters of the policy rule. Sargent (1976b, p. 559)
therefore proposed to identify such policy breaks by estimating "...
reduced forms for various subperiods or countries across which policy
rules differed systematically ... " Neftci and Sargent (1978) pursued this
line of research. They tried to identify such a change in the coefficients of
equation (10.4), distinguishing between a pre-war and a post-war sub-
period. In the former they suggested a policy break around January 1930,
using Ml (currency plus adjusted demand deposits) as a measure of the
money supply. Their subsequent test of the null hypothesis that the
reduced-form equation (10.3) is stable across this policy break was rejec-
ted at the 1% significance level, whereas the null hypothesis that (10.6)
was invariant could not be rejected at the 5% level. In conformity with the
Lucas critique, the second invariance assumption was therefore more
appropriate than the first, so that the neutrality proposition was sup-
ported. A similar result was found for the policy break in the post-war
sub-period, which was identified in the first quarter of 1964 (1964-I),
although this break could only be confirmed for Ml, and not for M2
(currency plus deposits at commercial banks). Both tests thus favoured the
second invariance assumption and hence the neutrality proposition. Given
the fact that the latter is deduced from the joint NR/RE hypothesis (plus
the no-frictions assumption), Neftci and Sargent (1978) opined that the
empirical results therefore provide 'a little bit of evidence on the natural
rate hypothesis', as the title of their paper already made clear. However,
they did not make clear why the post-war policy break could only be
225
identified for Ml and not for M2.
Neftci and Sargent's (1978) test can be regarded as a sophisti-
cated-falsificationist attempt to distinguish empirically between equations
(10.3) and (10.6). They did not try to falsify either equation separately but
instead compared their statistical significance. This approach is in sharp
contrast to that of Barro, who did not perform such 'horse races'. Instead,
he adopted a more direct attempt to test the neutrality proposition.
10.3.2. Barro's tests

Barro's empirical tests of the joint NR/RE hypothesis and the neutrality
proposition try to derive an estimate of the unanticipated component in
the growth rate of money. 23 Barro defined this unanticipated money
growth rate, or money surprise, as actual minus anticipated money growth,
whereby the latter is taken to be "... the prediction that could have been
obtained by exploiting the systematic relation between money growth and
[a] set of independent variables" (p. 563)?4 Barro's money supply func-
tion included lagged values of the annual average change in the money
stock at time t (mt-l and mt_2), a measure for real government expenditure
relative to its normal level (gt - g*), and the lagged unemployment rate
(Unt_1) (p. 566).25 He subsequently estimated this equation and assumed
that it is an accurate measure for anticipated money growth. This assump-
tion means that unanticipated money growth corresponds to the residuals
of the estimated money growth equation, that is, to the money surprise
mres,t· This money surprise enters the unemployment equation. Barro
carried out several tests of the hypothesis that only unanticipated money
growth affects the unemployment rate. His test results corroborated this
hypothesis (p. 574). McCallum (1979, p. 290) concurred with this con-
clusion, although he questioned the validity of Barro's procedure to
decompose actual money growth rates in an anticipated and an unantici-
23
Barro's empirical work concentrates on the United States. Several unpublished
papers have provided analogous analyses for other countries. For instance, Attfield,
Demery and Duck (1979) concentrated on the United Kingdom, Saidi and Barro (1976)
and Woigin (1980) on Canada, and Blejer and Fernandez (1979) on Mexico.
24 Page numbers refer to the 1981 reprint of Barro (1977b).
25
The real government expenditure variable captures an aspect of what Barro (1977b)
called the revenue motive for money creation. The federal government is assumed to
finance its expenditures either by taxes or by money issue (p. 564). Both methods of
finance are confronted with costs of raising, which the government minimizes, so that an
increase in the government budget will lead to increases in both types of revenue.
226
pated component, since it takes for granted that economic agents know
the government's money supply rule.
Barro (1977b, p. 582) recognized that his results would be more
reliable if they could be replicated for other periods and/ or for other
countries. He himself took the former route and extended his analysis for
the U.S. back to 1890. He argued that the difference between the pre-war
and post-war money growth processes would render a more powerful test
of the neutrality proposition possible. Furthermore, it would enable a test
of Lucas's (1973) hypothesis that shifts in the variance of money growth
alter the response of economic agents (and hence unemployment rates) to
monetary shocks. These extensions were undertaken in Barro (1978). They
yielded similar results as the 1977b-estimation, so that the joint NR/RE
hypothesis was again supported. However, Barro (1978) not only extended
his previous (1977b) analysis, but he also incorporated the general price
level. For this purpose he specified a price level equation, which
expressed this variable in terms of the money supply (MJ, present and
lagged money surprises (mrea,b mrea,t-h mrea,t-2, mrea,t-3), the nominal interest
rate (rt), a time trend (t), a correction for military conscription (milt), and
the shocks et(z) and ut (p. 595). He subsequently tested four hypotheses
with the help of this equation. The conclusions of these tests were:
(1) The coefficient on Mt did not differ significantly from unity, so that
economic agents do not suffer from money illusion;
(2) The coefficients of the money surprises had significant and correct
(negative) signs, although the lag in the response of the price
level to money surprises proved to be very long (2 to 4 years);
(3) Given the money stock (Mt) and the money surprises (mrea), actual
rates of money growth (m) could not improve the predictions of
the price level, whereas the similar hypothesis as regards unan-
ticipated money growth was rejected;
(4) An increase in the expected rate of inflation, corresponding to antici
pated changes in the growth rate of money and reflected in an
increase in the nominal interest rate ru had a positive effect on
the current general price level, given the other explanatory vari-
ables (p. 602).
These test results appear to support the joint NR/RE hypothesis, with the
exception of the results as regards the price-level equation. According to
Barro, the second test result suggests that the price level cannot be the
principal link between money and output (p. 615). Barro and Rush (1980)
modified and extended the 1978-analysis to quarterly data. They reached
227
similar conclusions to those in Barro's (1977b) and (1978), and again
showed that it is not plausible to conceive of the general price level as the
primary link between money and output. Barro's test approach thus con-
fronted New Classical monetary business cycle theory with an anomaly.
10.3.3. Conclusions
New Classicals have tested the neutrality proposition and the underlying
joint NR/RE hypothesis in two ways. Sargent's (1976a) tests yielded some
favourable evidence, although this result was almost immediately ques-
tioned by Sargent's (1976b) notion of observational equivalence. This
notion holds that the reduced-form equation in his {1976a) cannot be dis-
tinguished on empirical grounds from a reduced-form equation which can
be derived from a model favouring active stabilization policy with feed-
back. This observational problem can only be circumvented if the relevant
invariance assumption is known, which is possible only if a policy break is
identified. Neftci and Sargent (1978) identified such a break around 1964,
so that they could test which of the two invariance assumptions was rele-
vant. This test was thus a sophisticated-falsificationist attempt to establish
the empirical validity of the neutrality proposition and, indirectly, the
underlying joint NR/RE hypothesis. Its outcome favoured the invariance
assumption which is consistent with the neutrality proposition. However, it
crucially depends on Neftci and Sargent's (1978) statistical identification
of a policy break, which was only demonstrated for Ml, not for M2.
Barro did not adopt a sophisticated-falsificationist strategy, but
instead concentrated on 'natural rate' models only. He tried to test the
neutrality proposition more directly, namely by estimating the anticipated
and the unanticipated components of the rate of money growth. These
estimations presuppose that economic agents indeed know the monetary
authority's money supply rule. Initially, his results were favourable to the
neutrality proposition, in the sense that only unanticipated money influ-
enced real variables such as aggregate output and unemployment. How-
ever, the inclusion of a price level equation showed that the response of
the price level to money surprises was very weak, indicating that such
responses cannot provide the principal link between money and output.
This result was confirmed by Barro and Rush (1980), so that the New
Classical link between money and output was confronted with a serious
problem. This problem, however, did not (yet) lead to the rejection of the
New Classical explanation of the correlation between money and output.
The inconclusiveness of the empirical tests discussed above has
228
also been addressed by Hoover (1988), who noted that the two test proce-
dures have yielded results. which may be incompatible. Barro had assumed
that economic agents know the policy rule. Despite the implication of the
Lucas critique that the parameters of the aggregate supply (or unemploy-
ment) equation are not invariant to changes in the policy regime (i.e. to
changes in the parameters of the rule), he had estimated the policy rule
as if its parameters were invariant. Neftci and Sargent (1978), however,
find that the money supply rule, over a period which overlaps with Barro's
estimation period, is not invariant. Hoover observed that both cannot be
right (p. 190). He proposed that this incompatibility may be due to the
fact that Barro's tests do not take Neftci and Sargent's policy breaks into
account, while Neftci and Sargent do not include government expenditure
as an independent variable (p. 190).
The ambiguous test results reflect a balance between favourable
and unfavourable results. As will be shown below, however, the latter
would gain in importance because of two developments. Firstly, King
(1982) showed that the inclusion of agent-specific information overturns
the neutrality proposition. This will be discussed in the next section.
Section 10.5 will address the second development, namely Barro and
Hercowitz's (1980) empirical analysis of the effects of unperceived money
on the general price level and aggregate output. Both developments
strengthened the case against the neutrality proposition and the joint
NR/RE hypothesis. 26
10.4. DIFFERENTIAL INFORMATION AND THE

EFFECTNENESS OF MONETARY POLICY
King (1982) started from Barro's (1976) analysis, in the sense that he
incorporated the latter's local demand and supply functions. Barro had
assumed that the monetary authority adheres to the money supply rule
llM. = M. - M._ 1 = m., which is serially independent and uncorrelated with
26
It is not entirely clear which of the two developments was temporally prior, or
whether they have taken place simutaneously. King (1982) referred to an unpublished
version of his paper, dated in August 1979. The year in which Barro and Hercowitz (1980)
was written, however, is more difficult to determine. It was published in the Journal of
Monetary Economics, in which also King (1981) appeared. Barro (1981b, p. 76) dated the
latter paper in 1979, so that the period between the unpublished and the published ver-
sions is some two years. Assuming that this period also applies to Barro and Hercowitz
(1980), this paper was presumably written in 1978.
229
the other disturbances. King, however, assumed that the monetary author-
ity may choose to follow an active stabilization policy. Such a policy could
respond to excess demand shocks (or 'velocity shocks', vt_ 1), a policy vari-
able (gt_1), and a past autonomous monetary shock (mt_ 1). The money
supply rule can therefore be formulated as (p. 254 ):
(10.8) AM, = M, - M,_t = m, + llr,g,_l + 'irvvt-l + 111mmt-l
= m, + EAM,II,~ 1
where all variables are taken to be independent normal random variables

with zero mean and finite variances u~ U: and, u; ,
and where the 1/;'s are
policy parameters. The equation indicates that changes in the supply of
money consist of two components, namely an autonomous change, mb and
a change which is brought about by current feedback on past aggregates, .
EAMt If/_ 1• Both the monetary authority and the economic agents are
assumed to base their actions on exactly the same information set.
Information about aggregate variables is lagged one period, which implies
that the monetary authority cannot compensate the current autonomous
money shock mb at least not in the present period. Under these circum-
stances, money will be neutral in either of two cases. Firstly, as Barro
(1976) had already shown, the neutrality proposition holds if the monetary
authority does not pursue a feedback policy (so that the 1/;'s equal zero).
Secondly, the proposition will also hold if the aggregate shocks are
currently observable up to a common error term (Xt) across agents. The
agents then know (on average) how the monetary authority will respond
to these shocks, so that they can take this response into account. King
(1982) subsequently showed that the inclusion of agent-specific shocks
overturns these results. He restricted this type of shock to the policy
variable Kt· The agents' observations of this variable then consist of three
components, namely its actual realization gb an imperfection common to
all agents Xu and the agent-specific imperfections ~t(s), so that gt(s) = gt +
Xt + ~t(s), where s = 1, ... , S, is an index for agents. The 'Law of Large
Numbers' is presumed to apply to the latter shocks, in the sense that they
sum up to zero over agents on a given market (p. 259). Furthermore, the
shocks are normally distributed, with zero mean and finite variance
(which is identical for all agents), so that economic agents do not differ
from each other in a systematic way. The agents' information sets can now
be written as /t(s) = [pt(z), gt(s), f;_]. This reflects that agents have incom-
plete information about the current state of the economic system and
about the authority's current policy actions. They are therefore faced with
230
a signal extraction problem.
King (1982, p. 259) described the 'solution' of the signal extrac-
tion problem and the ensuing inference problem as a two-stage process.
Firstly, economic agents are assumed to observe gt(s ). They use this
information to form (or revise) expectations about the actual realization
of the policy variable, Egtl/t(s), and about the 'common' disturbance,
Extl/t(s). King argued that since the mean (and hence expected) value of
the agent-specific component equals zero, an observed change in gt(s) will
be interpreted as the result of a common shock. In turn, this will induce
agents to expect that the authority will bring about a change in the supply
of money, so that they will also alter their expectations about the change
in their local price level, E[pt(z)-p~]1/t(s), where p~ is the expectation of
the current general price level, conditional on information set ~~ (p. 259).
In the second stage of the inference problem, economic agents observe
their actual local price Pt(z). This enables them to revise their expec-
tations about the state of the economy, as reflected in m., g., vt and "t·v
The effects of the introduction of agent-specific information on
policy-effectiveness can now be made clear as follows. Suppose that the
monetary authority unexpectedly increases the policy parameter 1/tg, so that
the observed policy variable gt(s) experiences an unanticipated rise.
Agents are then faced with the problem whether this rise reflects an
increase in the common or in the agent-specific component. King (1982, p.
262) postulated that the agents will interpret this rise as the result of
changes in both components, which means that the magnitude of the
movement in the common component of the policy variable is under-
estimated. Given the fact that only this common component matters for
the determination of market equilibrium, King (1982, p. 260) specified the
price function as follows:
(10.9)
where T 6 and T 8 are complex parameters. 28 This equation shows that for
a given actual movement in prices, Pt(z) - p~, the underestimation of the
common component implies an overestimation of the excess commodity
27
King (1982, p. 259) assumed that there will be no iteration from prices to revisions,
revisions to prices, and so on.
28
King (1982, pp. 274- 75) gives the precise formulation of these parameters.
231
demand shock, {3m. + v. + e.(z).29 This overestimation will apply to both
its aggregate ({jm. + v.) and its local (e.(z)) component. The overestima-
tion of the former part means that agents will overreact as regards their
economic decisions. That is, they will underestimate the relative rise in
their local price and hence the profitability of increasing their output. This
results in a relative decrease of the latter, relative to its full-information
value. As King (1982, p. 263) summarized, "[o]utput ... falls, relative to its
full information value, because agents perceive a larger level of aggregate
commodity excess demand than has actually occurred." Hence, output
rises less than it would have done under full information. The fact that
observations and hence information contains an agent-specific component
enables the monetary authority to influence the agents' output decisions.
As Barro (1981b, p. 55) remarked, King demonstrates that the policy-
irrelevance proposition does not hold in the context of "... differential
information across individuals who also have access to a common price."
The monetary authority then can 'fool' the economic agents. From a
welfare-theoretic point of view, however, it is not clear why it should
pursue such a policy.
King's (1982) analysis shows the crucial importance of the agents'
information sets. It makes clear that the empirical validity of the neutral-
ity proposition depends on the perceptions of the economic agents. The
non-neutrality properties only arise because of currently unperceived or,
rather, incorrectly interpreted shocks. Barro and Hercowitz (1980)
attempted to establish empirically whether Lucas and his followers had
been right in assuming that deviations from the natural rate (of either
output or unemployment) are caused by such unperceived shocks. This
attempt will be discussed in the next section.
10.5. PERCEIVED VERSUS UNPERCEIVED MONEY

GROWTH
According to Begg (1984, p. 167), Barro's empirical work does not avoid
the pitfall of Sargent's (1976b) observational equivalence, in the sense
that it does not make clear why 'natural rate' models should be preferred
over 'unnatural rate' models. As Fischer (1980) showed, "Barro's results
29 King (1982, p. 263) noted that this latter influence involves 'an error in average
opinion about average opinion', thereby referring to an unpublished paper of Weiss (1979).
232
are quite consistent with the view that systematic monetary policy can be
used to affect output: the crucial issue for the potential effectiveness of
policy is whether output is affected by expectations that were formed
before the monetary authority had to commit itself to a particular level of
the money stock" (p. 213). That is, the crucial but yet unanswered ques-
tion is whether 'unanticipated money' is also currently unperceived. New
Classicals explain the correlation between money and output as the result
of expectational errors and incomplete information about the current state
of the economic system. Given the assumption of perfect price-flexibility
and the implied absence of 'long-term' contracting, unanticipated money is
currently unperceived. The relevant monetary residue can then be spec-
ified as:
(10.10)
In contrast, adherents to 'unnatural rate' models assume that prices

exhibit some rigidity, which prohibits economic agents to engage in
unanticipated though perceived favourable transactions. The appropriate
money variable must then be specified as:
(10.11)
where j gives the length of the contracting period. Equation (10.10) is

consistent with the neutrality proposition, whereas equation (10.11) is not.
Fischer (1980) showed empirically that two-year-ahead forecast errors (j
= 2) of the money stock affect the behaviour of real output. He thus indi-
cated that unanticipated though not necessarily unperceived changes in
the money stock could bring about the observed relationship between
money stock and aggregate output. Combined with the Fed's capability to
react to events with less than a two-year lag, Fischer's analysis implies that
there is at least some scope for effective systematic monetary policy and
hence countercyclical policy.
Barro and Hercowitz (1980) tried to test directly for effects of
unperceived money growth. They argued that the differences between
initial and final reports on money growth rates may give an indication for
the unperceived component of money growth, thereby using monthly data
(p. 259). Given the assumption that these differences are indeed a reliable
measure for unperceived money growth, they can be used to test which of
the equations (10.10) and (10.11) is relevant in the relationship between
money and output. Barro and Hercowitz (1980) showed that unperceived
233
money growth lacks explanatory value for output and the unemployment
rate, so that equation (10.11) appears to be more relevant than (10.10). It
should be noted, though, that they qualified this result by arguing that "...
a one-to-one identification of unperceived money growth with the data
revisions does not seem reasonable" (p. 259). Nevertheless, this result
posed an important problem for New Classical explanations of the
relationship between money and output, particularly as it was
strengthened by Boschen and Grossman (1982).
Boschen and Grossman (1982) started from the observation that
since 1965 the Fed has published preliminary data with a lag of only eight
days. The assumption that this information is relevant though unperceived
is clearly inconsistent with the rationality postulate. Furthermore, these
preliminary data are revised over a much longer period. Rational econ-
omic agents will also take these revisions into account. Boschen and
Grossman (1982, p. 310) adopted a model which included both the
preliminary data and the process of data revision. They derived two
hypotheses from this model. The first hypothesis concerns the neutrality of
money. Boschen and Grossman (1982, pp. 310 - 11) tested whether
changes in aggregate output and employment are correlated with the anti-
cipated rate of money growth, as measured by the difference between
current and lagged estimates of current and past stocks of money. The
test indicated, however, that "... current and lagged values of the contem-
poraneous measure of money growth have statistically significant effects
on industrial production and employment" (p. 327). This means that per-
ceived money growth influences output and employment, and hence is
non-neutral. The second hypothesis tested concerned the non-neutrality of
errors in the preliminary data on money growth. These errors are
measured as in Barro and Hercowitz (1980), who had taken the dif-
ferences between the initial and final reports on money growth as a
measure of unperceived money growth. 30 Boschen and Grossman (1982)
tested the hypothesis that this unperceived money growth is not correlated
with industrial production and employment. This hypothesis could not be
rejected, so that unperceived money growth does not Granger-cause these
real variables (p. 329).
There were also other discorroborating tests of New Classical
30 Barro and Hercowitz's (1980) procedure implies that the data are correctly revised
in the next period. In contrast, Boschen and Grossman (1982) allowed " ... for positive
correlation in the subsequent revisions of current estimates of current and past money
stocks" (p. 311). This causes some persistence in the real effects of monetary shocks.
234
models. For instance, Gordon (1982) compared the neutrality proposition
with an alternative hypothesis, which holds that prices only gradually
respond to nominal aggregate demand shocks, even when these shocks are
fully anticipated. His test results corroborated the latter hypothesis, thus
showing short-run price inertia (p. 1112). Mishkin (1983, p. 127) also
rejected the neutrality proposition, because he found that unanticipated
changes in monetary policy do not have larger effects on output and
unemployment than anticipated ones. In subsequent years, the number of
discorroborating results would increase even more. 31
King (1981, p. 200) argued that the above tests need not lead to a
rejection of the neutrality proposition and the underlying joint NR/RE
hypothesis, because they presuppose that the econometrician's assumption
about the agents' information sets is empirically valid. However, if agents
have less information, the joint hypothesis can be maintained. The questi-
on then remains to be answered, of course, why agents would choose to
use such incomplete information. Given the joint NR/RE hypothesis, this
question can only be answered by the introduction of information acquisi-
tion and/or processing costs, that is, by the rejection of the 'no-frictions'
assumption.
10.6. THE SWITCH FROM MONETARY TO

REAL BUSINESS CYCLE THEORY
Barro and Hercowitz's (1980) test indicated that deviations from the
natural rates of both output and unemployment cannot be explained as
the result of unperceived monetary shocks. That is, monetary versions of
equilibrium business cycle theory appear to be empirically invalid. Blan-
chard and Fischer (1989, p. 360) observed that these results "... coming as
they did from proponents of the [New Classical] approach, were signi-
ficant nails in the coffin." Barro (1989a) identified no less than five such
nails. In his view, they even suggested that the causality between money
and output should be reversed. Many economists followed this conclusion
and rejected monetary equilibrium business cycle theory, although it
31
For a list of tests which reject the neutrality proposition, see Visser (1989, p. 48).
235
should be noted that not all New Classicals have taken this step. 32 The
criteria used in this decision remain obscure, so that further research in
this regard may prove insightful. Those who rejected monetary equilib-
rium business cycle theory have followed either of two directions of
research. The first adopted the 'Keynesian' approach of Gray (1976),
Fischer (1977) and Phelps and Taylor (1977). This approach argues that
some forms of multi-period contracting and wage-indexation cause prices
and nominal wages to be 'sticky', so that unanticipated though perceived
movements in nominal aggregate demand have real effects. In a sense this
approach searches for microfoundations of macroeconomic phenomena.
As was already shown above, Barro (1977a (1981), p. 223) criticized it on
the account that it does not explain why economic agents sign contracts
which clearly lead to sub-optimal outcomes of the economic process. Such
an explanation could be provided in terms of adjustment (transaction)
costs, but Barro (and other New Classicals) do not appear to appreciate
such an explanation. Instead, he regarded it as an 'ad-hoc' stratagem. He
therefore choose to adhere to competitive-equilibrium explanations of
business cycles, albeit that he reversed causation. He opined that equilib-
rium business cycle theorists had been mistaken in trying to explain cycli-
cal fluctuations in terms of monetary shocks, and that they should have
emphasized real disturbances as sources of business fluctuations (1989a, p.
4). This means that fluctuations in monetary aggregates are interpreted as
the (endogenous) result, instead of the (exogenous) cause, of cyclical
fluctuations in output and unemployment. The switch shows that causation
may be reversed in research traditions. 33 The earliest and most influen-
tial contributions to real business cycle theory have been provided by
Kydland and Prescott (1982), Long and Plosser (1983) and King and
Plosser (1984). 34 Since the present study is restricted to monetary busi-
32 For instance, Lucas (1987) continued to stress the importance of monetary shocks in
business cycle theory, and Sargent (1987, p. 496) opined that the 'causality issue' remains
"...open and controversial. "
33
Admittedly, this is a matter of definition, or rather classification. After all, this result
depends on the definition of the New Classical research tradition. If the direction of cau-
sation forms part of this definition, then no reversal of this causation can take place within
one and the same tradition. The reversal then implies a change of tradition.
34
Kydland and Prescott (1982) adopted a so-called 'time-to-build' or 'multi-period
construction' explanation of persistence in output and unemployment. They assumed that
economic agents cannot immediately adjust their productive capacity, so that the effects of
unexpected shocks are lagged. These shocks are assumed to be autocorrelated and arise
because of changes in technology and productivity. Long and Plosser (1983) argued that
236
ness cycle theory, these real explanations fall outside its realm and hence
will not be discussed.
10.7. METHODOLOGICAL ASSESSMENT
Both Sargent's and Barro's respective test approaches adopted the view
that economics should build 'artificial economic systems' (in the form of
mathematical models) which may serve as laboratory economies. Like all
theorizing, such models are 'unrealistic' in the sense that they abstract
from those aspects of economic reality which are regarded less relevant
for the economic problem under consideration. That is, the abstractions
determine the relevancy and hence the usefulness of the model. Accord-
ing to Lucas (1980, p. 272), however, not all abstractions and models are
equally useful. He stated that "[t]he more dimensions on which the model
mimics the answers actual economies give to simple questions, the more
we trust its answers to harder questions. This is the sense in which more
'realism' in a model is clearly preferred to less." This statement reflects
the fact that the NCE does not attempt to verify or falsify its models, but
rather that it attempts to build empirical models which describe economic
reality satisfactorily. That is, the NCE is not interested in the truth status
of these models but rather in their usefulness. This concurs with the aims
of econometrics in general, as described by Morgan (1988, p. 199). She
argued that econometricians try to find satisfactory models, and do not try
to prove fundamental theories true or untrue. 35 This raises the question
what criteria must be met for a model to be satisfactory. Morgan (1988, p.
205) distinguished five such criteria, which concern:
economic agents spread unanticipated changes in wealth over both time and consumer
goods, which in tum causes the phenomenon of persistence. Their analysis assumed that
agents have full information, so that there is no signal extraction problem in the sense of
Lucas. The unanticipated changes in wealth are due to random shocks, added to the
outputs of numerous commodities. As was already stated above, King and Plosser (1984)
incorporated financial intermediation, so that they can explain cyclical movements in
monetary variables as the result of changes in real shocks. They assumed that these shocks
occur in government expenditures and the tax system.
35
Morgan's argument was endorsed by Darnell and Evans (1990, p. 73, italics in
original), who concluded that "... the purpose of applied econometrics became focused on
the estimation of models rather than the testing of economic hypotheses. The spirit of those
investigations was more in the vein of verification [corroboration] than falsification."
237
(1) the economic structure: models should exhibit a particular economic
structure, which differs between 'research programmes';
(2) empirical validity: models should meet statistical or historical criteria,
that is, they should be corroborated;
(3) the evaluation of economic policy: models should be useful for policy
evaluation, so that they "... must allow the exploration of policy
options or make predictions about future values" (p. 205);
(4) theory exploration or development: models should expose unsuspected
relationships or should develop the detail of known relationships;
(5) theory rejection: models should facilitate the methodological decision
whether or not to reject a particular economic theory.
The question can now be addressed whether and how these criteria apply
to New Classical econometric practice.
(1) Economic structure

As was shown in the two preceding chapters, New Classical models
include the joint NR/RE hypothesis, some informational assumption, and
the 'no-frictions' assumption. Furthermore, they are formulated in a way
which is consistent with the situational constraints of the Lucas pro-
gramme. This is not to say, however, that each and every New Classical
model should explicitly provide its macro-structure with microfoundations.
The structure of a model is 'satisfactory' on economic-theoretic grounds if
it is consistent with the New Classical modelling style, as set by Lucas
(1972a)?6
(2) Empirical validity

The second criterion concerns the empirical validity of the models under
consideration. Satisfactory models must be corroborated, or at least not
discorroborated. As was shown in Chapter 8, the corroboration of the
joint NR/RE hypothesis by Lucas (1973) and Sargent (1973b) was ambi-
guous. The latter's results showed that the joint hypothesis could be
maintained on 'sophisticated-falsificationist' grounds only. The present
chapter indicated that similar problems arose in the case of the neutrality
36
For instance, the macro-model of Sargent and Wallace (1975, p. 216n1) did not have
explicit microfoundations, but it was nevertheless 'satisfactory• because it closely resembled
the model used by Sargent (1973a). In turn, the latter was " ... heavily dependent on the
analysis of the natural rate hypothesis carried out by Lucas in a series of papers", including
Lucas (1972a) (p. 162). The latter model thus indirectly provided Sargent and Wallace's
(1975) model with microfoundations.
238
proposition. There was some disagreement between New Classicals as
regards the appropriate procedures to test this proposition. Sargent's
(1976a) test approach initially concentrated on the question whether
lagged variables other than lagged unemployment rates could improve the
prediction of the current rate of unemployment. His test results suggested
that the power of the test procedures to discriminate was rather weak.
Sargent's (1976b) observational equivalence emphasized the identification
of policy breaks. Neftci and Sargent (1978) identified such a break around
1964. Their subsequent sophisticated-falsificationist test favoured
'invariance assumption 2' over 'invariance assumption 1', which means
that the neutrality proposition was corroborated. In contrast, Barro's
(1977b) test approach attempted to test the proposition more directly. It
focused on the estimation of anticipated and unanticipated money growth.
Initially, the neutrality proposition was corroborated, but the results of
Barro (1978) and Barro and Rush (1980) raised considerable doubt about
this result. They concluded that the price level provided merely a weak
link between money surprises and output. Barro and Hercowitz (1980)
showed that unperceived money growth lacks explanatory value for output
and unemployment. These unfavourable test results were confirmed and
enhanced by the results of Boschen and Grossman (1982), Gordon (1982),
Mishkin (1983), and others. They made the monetary version of New
Classical business cycle theory less satisfactory on empirical grounds.
However, this did not immediately lead to its rejection.
(3) Evaluation of economic policy

The third criterion holds that models should facilitate policy evalution.
This criterion clearly applies for New Classical Economics, because policy
evaluation was one of its principal aims. As Lucas (1980, p. 271) stated,
"[o)ne of the functions of theoretical economics is to provide fully articu-
late, artificial economic systems that can serve as laboratories in which
policies that would be prohibitively expensive to experiment with in actual
economies can be tested out at much lower costs." In practice, New Clas-
sicals have indeed given considerable attention to the theory of economic
policy, as becomes clear from such issues as the Lucas critique and the
policy-effectiveness debate.
(4) Theory exploration or development

The fourth criterion requires that the models should either develop
already known relationships or discover new ones. Several New Classical
239
models have satisfied this requirement. For instance, Lucas (1972b)
discovered the relationship between the variance of the money supply and
the effectiveness of monetary policy. Sargent (1973a) gave an alternative
explanation of the Gibson paradox to that of Fisher (1930). Lucas (1975)
developed the Slutzky-Frisch distinction between impulses and propaga-
tion mechanisms in a way which enabled him to explain persistence in
deviations from the natural rate of output. Sargent and Wallace (1975)
and Barro (1976) gave alternative formulations of the 'classical' property
that monetary policy is neutral. They showed that this neutrality proposi-
tion crucially depends on the agents' and the authority's respective
information sets.
(5) Theory rejection

Morgan's fifth criterion holds that satisfactory models should facilitate the
methodological decision of acceptance or rejection of the model and the
underlying economic theory. The outcome of this decision does not only
depend on the criteria discussed above, but also on the methodological
position adopted. Given their sophisticated-falsificationism, Sargent
(1976a) and Neftci and Sargent (1978) decided not to reject the joint
NR/RE hypothesis. Lucas (1987, p. 102) argued that such a rejection
would be premature, and Sargent (1987, p. 496) considered it to be
controversial, at least. Barro's alternative test approach, however,
neglected Sargent's observational equivalence, so that it could avoid
comparative tests like that of Neftci and Sargent (1978). Instead, it tested
the neutrality proposition more directly. Its discorroborative results
rendered the underlying joint hypothesis less satisfactory from the point of
view of Morgan's second criterion of empirical 'unsatisfactoriness'. These
results were strengthened by further unfavourable evidence, which may
have induced the switch from monetary to real business cycle theory.
It is not clear when or why precisely Barro decided to reject monetary

business cycle theory. 37 Given the fact that the decision followed some
time after the first discorroborative test results were available, Barro did
not adopt a 'naive-falsificationist' position. The question then arises how
37 According to Barro (1989, p. 4), "... over the past five to ten years, most propo-
nents of the new classical approach have moved away from analyses that emphasize
monetary shocks and towards those that rely on real disturbances as sources of business
fluctuations." This statement indicates that the decision to reject monetary New Classical
business cycle theory has been taken in the 1980's.
240
his methodological position should be characterized. The answer to this
question will depend on the reasons why he rejected monetary business
cycle theory. The conjecture that Barro adhered to 'sophisticated-falsifica-
tionism' can be regarded as a useful suggestion for further research. It is
based on the assumption that his ultimate decision of rejection was taken
in 1984. By that time Kydland and Prescott (1982) and Long and Plosser
(1983) had provided an alternative business cycle model. However, these
models did not explain cyclical movements in monetary aggregates. This
latter property means that their models cannot be considered empirically
more satisfactory than their monetary counterparts. King and Plosser's
(1984) reversal of causation between money and output interpreted such
monetary movements as the result of financial intermediation. They rem-
edied the empirical shortcoming of real business cycle theory, and may
have made it possible for sophisticated-falsificationists to switch from
monetary to real business cycle theory.
10.8. CONCLUSIONS
Sargent and Wallace (1975) and Barro (1976) derived the neutrality
proposition from the joint NR/RE hypothesis and the 'no-frictions'
assumption. The propostion was very controversial, and gave rise to a
fierce debate, which particularly concentrated on the assumption of
perfect price-flexibility. Alternative models without this feature led to
non-neutrality results, whereas McCallum showed that the agents' infor-
mation sets are crucial for the logical validity of the neutrality proposition.
As was pointed out in the previous chapter, this view also formed the
starting point of Barro's (1980) study, in which the proposition holds even
if agents observe a global nominal interest rate. Finally, King (1982) con-
cluded that the inclusion of agent-specific shocks invalidate the neutrality
proposition.
The debate on the proposition did not only concern its logical
but also its empirical validity. New Classical econometric practice, how-
ever, is not concerned with the truth status of economic propositions.
Instead, it aims to build so-called 'laboratory models', which are 'satisfac-
tory' if they meet Morgan's (1988) five criteria. The second of these
criteria, concerning the model's empirical validity, proved very difficult to
meet. Initially, Sargent (1976a) provided some corroborative evidence, but
his (1976b) 'observational evidence' rendered this evidence meaningless.
241
Neftci and Sargent (1978) subsequently identified a policy break, so that
they could circumvent this observational equivalence. Their test results
corroborated the neutrality proposition. Lucas ( 1987) and Sargent ( 1987)
therefore continued to adhere to the view that money may cause cyclical
fluctuations in aggregate variables. Barro's rival test approach, however,
led to opposite results. It tested the neutrality proposition more directly
by estimating the relevant money surprises, thereby neglecting the policy
breaks as identified by Neftci and Sargent (1978). As shown by Barro
(1978) and Barro and Rush (1980}, it yielded unfavourable results for the
neutrality proposition. Barro and Hercowitz (1980) claimed that unper-
ceived money could not explain fluctuations in aggregate output, thus
contradicting the implications of the joint NR/RE hypothesis. Lucas-type
models therefore became less and less satisfactory from an empirical
point of view. The availability of King and Plosser's (1984) alternative
may have induced the switch from monetary to real business cycle theory.
242
PART III. A COMPARISON
11. AUSTRIAN VERSUS NEW CLASSICAL
ECONOMICS
11.1. INTRODUCTION
Recently economists have debated the issue whether Austrian and New
Classical economics share the same roots, as implied by Lucas's (1977, p.
216) suggestion that the latter embroiders on Hayekian business cycle
theory. In the subsequent literature this claim has been confirmed by
several authors, such as Kantor (1979), Colander and Guthrie (1980),
Laidler (1982) and Scheide (1982). These authors stress that both
Austrians and New Classicals explain business cycles as unintended
consequences of rational behaviour, whereby prices do not transmit all
relevant information. Scheide (1982), for instance, advanced the view that
the differences between Austrians and New Classicals are merely seman-
tic, arising from the fact that the meaning of the term 'equilibrium' has
changed since Hayek formulated his business cycle theory. He suggested
that sequences of New Classical rational-expectations equilibria (REE's)
constitute what Hayek called disequilibrium dynamics. This suggestion
overlooks, however, that Hayekian business~cycle theory explains cyclical
fluctuations in terms of coordination failures, which New Classicals
eliminate by assuming price-taking economic agents. As Butos (1986, p.
341) pointed out, Hayek increasingly recognized the limitations of gen-
eral-equilibrium analysis.
The present chapter aims to provide a comparison of the Austrian
School of economics and New Classical Economics. This comparison
particularly concerns their respective methodologies, and hence the Hayek
programme and the Lucas programme. It starts with a comparison of the
central problems of these programmes, namely the Hayek Problem and
the (narrow) Lucas Problem, which are addressed in Section 11.2. As was
advanced in Chapters 4 and 9, the solutions of these problems are subject
to constraints, one of which concerns the method of discovery. Section
11.3 compares Hayek's compositive method with the New Classical ver-
sion of Popper's method of rational model-construction. These methods
closely resemble each other, particularly since in the early 1950s Hayek
substituted Popper's methodological monism for his own methodological
dualism. The close resemblance between these metho~s, however, does
245
not mean that the resulting types of explanation are also similar. The
Hayek Problem suggests that economists should concentrate on disequilib-
rium dynamics, whereas the Lucas Problem favours comparative-static
analysis. This results in different types of explanations, in the sense that
Austrians aim to provide 'genetic-causal' explanations, whilst New Classi-
cals adhere to 'mathematical-functional' explanations. This means that
Austrians try to explain the tendency towards intertemporal general-
equilibrium, whereas New Classicals use the notion of 'general-equilib-
rium' as a heuristic principle. The respective types of explanations are
discussed in Section 11.4. Section 11.5 addresses the subjective nature of
the data and the dispersion of knowledge. Austrians and New Classicals
both stress the fact that economic agents have imperfect information.
They both distinguish between different types of knowledge in order to
explain in what sense the information is restricted. However, it will be
shown that these distinctions differ in important respects. As a result, they
attach different meanings to the notion of 'dispersion of knowledge'.
Section 11.6 subsequently points out that the New Classical heuristic
prescription of mathematical formalization affects the phenomena to be
analyzed. In particular, it eliminates fundamental uncertainty and hence
the role of entrepreneurship, so that it only captures Knightean risk. The
heuristic prescription of mathematical formalization thus limits the
domain of New Classical Economics. Section 11.7 discusses a crucial
simplifying assumption with which the NCE simplifies its analysis, namely
the 'homogeneity postulate'. This postulate holds that the elements of a
given class do not differ from each other in relevant (systematic) ways, so
that the class may be regarded as homogeneous. For reasons of simplicity
and analytical convenience, New Classicals formulate such a postulate for
both economic agents and capital goods. Austrians, on the other hand,
allow for the uniqueness of the 'individual', because this uniqueness is a
necessary prerequisite for the process of competition. Furthermore, since
their business cycle theory is a 'malinvestment' theory, capital goods must
be heterogeneous. Austrians thus reject the validity and usefulness of the
homogeneity postulate for agents as well as goods. This rejection has
considerable implications for the feasibility of providing complete explana-
tions of social phenomena, that is, for the 'degree of explanation'. Section
11.8 shows that Austrians merely formulate generic schemes, which are
difficult to test empirically. New Classical models, on the other hand, are
of a higher 'degree' (completeness, precision) of explanation, and hence
have far more empirical content. This reflects the relationship between
246
\
analytical rigour and empirical content. Section 11.9 states some con-
clusions.
11.2. THE HAYEK PROBLEM VERSUS THE LUCAS

PROBLEM
The central problem in the Hayek programme is the Hayek Problem,

which concerns the question how coordination can be achieved in market
economies. According to Hayek (1945), this problem cannot be solved by
some central board because of the dispersion of all relevant knowledge.
Instead, the Hayek Problem studies "... how to secure the best use of
resources known to any of the members of society, for ends whose relative
importance only these individuals know" (p. 78). That is, it addresses the
question how a decentralized price system can bring about a situation of
coordination. In turn, this means that the attempt to solve the Hayek
Problem forms a quest for discoordination dynamics, which explains a
situation of coordination as the end result of a sequence of discoordina-
tion situations. Such a sequence will be brought about if discoordination
prices transmit sufficient information for agents to learn their way into
coordination. Kirzner stressed the role of Misesian entrepreneurship in
this regard. He explained that entrepreneurship aims to discover and
grasp profit opportunities. The process of competition is presumed to
eliminate such opportunities, so that the agents' plans and actions become
coordinated. However, since it is not clear why entrepreneurship will be
successful in discovering all profit opportunities, the Hayek Problem
remains unsolved, in the sense that it is an empirical matter, depending
on the agents' learning behaviour. Latter-day Austrians, such as Garrison,
recognize that the presupposition of its existence rests on 'belief.
The Lucas programme aims to solve the Lucas Problem, which
concerns the correlation between money and output. At first sight, this
correlation is inconsistent with the rationality postulate, which implies that
agents do not suffer from money illusion. Lucas and Rapping, however,
adopted Friedman's interpretation that the correlation is an 'unintended
outcome' of rational behaviour. They suggested that it is brought about by
expectational errors and hence by the agents' misperceptions about their
decision situation. In his later work, Lucas substituted general-equilibrium
analysis for their own earlier, and Friedman's, partial-equilibrium frame-
work. He thereby assumed that agents are price-takers, and that coordina-
247
I
tion is continuously and instantaneously ensured by the price mechanism.

This assumption is the result of the New Classical heuristic prescription
that economic theories should be formulated as mathematical general-
equilibrium models, which leads New Classicals to assume away problems
which cannot (yet) be mathematically formalized, including the coordina-
tion problem. Butos (1986, p. 337) therefore concluded that in the New
Classical view of science, "... techniques (both mathematical and
econometric) define the boundaries and possibilities of economic analy-
sis." The general-equilibrium construct (particularly the REE) is a heuris-
tic principle, and not a situation to be explained. The absence of
coordination failure implies that there are no endogenous reasons why
agents would misperceive their decision situations. The Lucas Problem
can then only be explained as the result of some exogenous source, which
New Classicals introduce in the form of lagged information about econ-
omy-wide variables. The persistence in the effects of this informational lag
is accounted for by the inclusion of some propagation mechanism.
The difference between Austrians and New Classicals in the
methodological status of the general-equilibrium construct is reflected in
their respective 'types' of explanation.
11.3. TYPES OF EXPLANATION
Mayer (1932, p. 148) distinguished between two types of explanation,

mathematical-functional explanation and genetic-causal explanation. 1 The
, 1Hempel (1965 (1970) p. 447) defined a genetic explanation as an explanation which

"... presents the phenomenon under study as the final stage of a developmental sequence,
and accordingly accounts for the phenomenon by describing the successive stages of that
sequence." He argued that "... schematically speaking, a genetic explanation will begin with
a pure description of an initial stage; thence, it will proceed to an account of a second
stage, part of which is nomologically linked to, and explained by, the characteristic features
of the initial stage, while the balance is simply added descriptively because of its relevance
for the explanation of some parts of the third stage, and so forth." The genetic explanation
can then be expressed as follows:
S1 .. s; + D
2 = S2 .. s; + D
3 = S3 ....... s;_1 + Dn_ 1 = Sn_1 .. Sn
in which each arrow indicates a presumptive law-like (nomic) connection between two
successive stages. The arrows presuppose universal or statistical uniformities. S 1, Sf• ·~ S0
express "...all the information that the genetic account gives about the first, second, ... ,ntn
stage." D 1, D 2, ••• , D 0 _1 denote " ... information about further facts which are adduced
without explanation, because of their explanatory significance for the next stage" (pp. 449-
50). It should be noted, as Hempel did, that in practice the stages in genetic explanations
248
former was described as the method which aims to exhibit the conditions
of equilibrium under a given situation, that is, given the values of
exogenous variables. This type of explanation describes the equilibrium
conditions of variables such as prices, wages, interest rates, commodity
supply and demand. 2 It treats these elements of such general-equilibrium
situations as simultaneously determined and mutually interdependent. 3
The resulting description of the equilibrium situations is inherently static,
in the sense that it does not allow for endogenous changes in these situ-
ations. Any changes in the state of the economic system can only be
caused exogenously.4 The NCE therefore introduces exogenous shocks
(impulses). Mathematical-functional analysis does not describe the
resulting adjustment processes, but merely gives an account of the 'new'
equilibrium, so that it is of a comparative-static nature. The major innova-
tion and sophistication of the NCE is its distinction between two types of
general equilibrium. The first type is the NRE, which is defined as an
intertemporal general-equilibrium in which economic agents have full
information. The second and more important type is the REE, in which
cannot be as stringently separated as in this scheme. Furthermore, as will be shown below,

Hayek would have objected to the above representation, because it neglects the dispersion
of knowledge, presupposing 'too much' knowledge on the part of the economist.
2 Mayer argued that "[d]ie Gleichgewichtstheorie setzt sich als Erkenntnisaufgabe die
exakte Beschreibung der quantitativen Beziehungen, die zwischen den Preisen, den
angebotenen und nachgefragten Mengen aller Giiter auf einem einheitlichen Markte
bestehen, wenn der Ruhezustand erreicht ist, d.i. keine weiteren Verschiebungen der Giiter
durch Tauschakte mehr erfolgen." It should be noted that his notion of 'equilibrium' was
restricted to full-information general equilibrium, but it can be broadened to include the
REE as well, without loosing any of its relevance. The crucial issue here is that there are
no endogeneous reasons for change.
3
In Mayer's (1932, p. 151) words, "[a]lle Elemente, die sich gegenseitig im Ruhezu-
stand das Gleichgewicht halten, werden als gleichzeitig 'gegeben', d.i. existierend, wenn
auch nicht gleichzeitig 'bekannt' (im mathematischen Sinne) angenommen, kein Element
ist vor irgendwelchen anderen gegeben, es besteht kein einseitiger Kausalzusammenhang
zwischen ihnen, sondem sie bestimmen sich aile wechselseitig, stehen als variabele
Elemente eines geschlossenen Systems untereinander in allseitiger,reversibler Abhiingigkeit,
im Verhaltnis der 'allgemeinen Interdependenz', derart, da {3, wenn ein Element in seiner
Gro ~ sich iindert, automatisch sich aile anderen sich ins Entsprechungsverhiiltnis setzen."
4
Huussen (1985, p. 161) argued that New Classicals show that a situation of general
equilibrium may itself generate 'endogenous dynamics' as long as economic agents must
act on the basis of incomplete information. However, it appears more appropriate to label
this dynamics 'exogenous' because (1) the assumption of continuous and instantaneous
market-clearing assumes away the coordination problem, so that there are no endogenous
reasons for agents to change their actions, and (2) the lack of information arises from
exogenous impulses.
249
economic agents have imperfect global though full local information. As
was shown in Chapter 9, this distinction means that the agents' knowledge
need not be correct. That is, the agents may base their actions on incor-
rect knowledge. The distinction between the NRE and the REE thus
involves a separation between the 'truth status' of knowledge and its role
in the decision-making process. This separation, in turn, enables the NCE
to interpret deviations from the NRE (such as business cycles) as
sequences of imperfect-information general-equilibria.
In contrast, Austrians do not assume that markets and prices clear
continuously and instantaneously. Instead, the Hayek Problem aims to
explain how a situation of coordination can come about. Hayek pointed
out that the process of competition may be regarded as a learning pro-
cess. Kirzner subsequently tried to explain the role of entrepreneurship in
this process. In Mayer's (1932, p. 148) terminology, these Austrians
attempted to generate genetic-causal theories.
11.4. METHODS OF DISCOVERY
In the early 1940s Hayek attacked what he called 'scientism', which he

defined as the view that the social sciences should adopt the methods of
discovery as employed in the natural sciences. In his view, the 'scientistic
prejudice' led many social scientists to adopt the analytical method, which
explains the behaviour of economic agents in terms of developmental laws
for society as a whole (cf. Chapter 4). Instead, Hayek took a methodologi-
cal-individualist stance, and proposed the so-called compositive method.
As Huussen (1985, p. 126) pointed out, Hayek's compositive
method contains three stages. In the first stage the agents' actions are
observed and interpreted as purposive. The method of Verstehen is used to
reconstruct their decision situations. These situations depend on the
agents' perceptions of so-called 'external stimuli', and hence on their
mental states. Stated differently, Austrians explain the agents' decisions
psychologistically, whereby a decision is considered to depend crucially on
the decision-maker's mental state. In Hayek's (1942c, p. 277) view, the
method of Verstehen is a 'key' which enables the social scientist to trans-
late observed behaviour into statements about the underlying decision
situations of human actions and hence about their mental states. That is,
since social scientists have themselves a mind similar to those of the
agents, they can understand the agents' perceptions from what the latter
250
do and say. He thus advanced that the method of Verstehen can yield
'true' knowledge about the actual decision situations of agents. In the
second stage of the compositive method, the state of the (macroeconomic)
system as a whole is deduced from this 'true' knowledge. In the third
stage of the compositive method the deduced models are confronted with
the actual behaviour of the system under consideration and its constituting
elements. Since Hayek considered the results of the method of Verstehen
infallible, differences between model behaviour and actual behaviour may
only result from logical errors, incorrectly specified initial conditions,
and/or violations of the ceteris-paribus clause. The deduced model is
considered to be a classification scheme.5
Popper ( 1957) modified the compositive method in two respects.
Firstly, he criticized Hayek's interpretation of the method of Verstehen as
a method of justification (proof). Instead, he argued, our intuitions of the
agents' decision situations may be mistaken, which means that introspec-
tion need not lead to infallible results (p. 138). He therefore eliminated
the method of Verstehen from the context of justification, although he
recognized that it may play a role as a method for the derivation of
hypotheses (p. 137). That is, he restricted the method to the context of
discovery. Empirical testing must then establish (at least in principle)
whether the deductions are empirically valid.
Popper's (1957, p. 105n) second criticism holds that Hayek had
misrepresented the method of discovery in the natural sciences. Instead,
he redefined the latter's notion of 'scientism' as the imitation of what
people erroneously believe to be the method of discovery in the natural
sciences. As a corollary, Popper (1957) rejected Hayek's methodological
dualism and took a methodological-monist stance, arguing that the differ-
ences between natural and social sciences "... are differences of degree
rather than of kind" (p. 141). Hayek (1967a, p. viii) subsequently agreed
with this criticism, adopting Popper's methodological-monism. He main-
tained, though, that social phenomena are more complex than natural
5
Hayek (1943b, p. 73) opined that "[a]ll that the theory of the social sciences attempts
is to provide a technique of reasoning which assists us in connecting individual facts, but
which, like logic or mathematics, is not about the facts. It can, therefore, ... never be
verified or falsified by reference to facts. All that we can and must verify is the presence of
our assumptions in the particular case. . .. [T]he theory itself, the mental scheme of inter-
pretation, can never be 'verified' [nor 'falsified'] but only tested for its consistency. It may
be irrelevant because the conditions to which it refers never occur; or it may prove
inadequate because it does not take account of a sufficient number of conditions. But it
can no more be disproved by facts than can logic or mathematics."
251
phenomena, so that the two types of sciences cannot yield the same
'degree' (precision, completeness) of explanation (and prediction). This
issue will be addressed in Section 11.8.
Popper's (1957) criticisms of the compositive method can be used
to formulate an alternative method of discovery, which he called the
method of rational model-construction (or rational reconstruction). He
described this method as "... the method of constructing a model on the
assumption of the possession of complete rationality (and perhaps also on
the assumption of complete information) on the part of all the individuals
concerned, and of estimating the deviation of the actual behaviour of
people from the model behaviour, using the latter as a kind of zero co-
ordinate" (p. 141).6 As was shown in Chapter 9, the New Classical
method of discovery closely resembles this method. It uses a situation of
full-information (the NRE) as a benchmark (zero-coordinate), with which
actual, incomplete-information states of the system (REEs) are compared,
whereby agents decide in single-exit decision situations.
The fact that New Classicals use a version of Popper's method of
rational model-construction implies ~hat their method closely resembles
Hayek's compositive method. The main differences between the respective
methods concern the appropriate context of the method of Verstehen, and
the complexity of the phenomena studied. Hayek's later adoption of
Popper's position as regards the former issue reduced this difference even
more, so that only the complexity of social phenomena remains a source
of disagreement. This alleged complexity results from the subjective
nature of the data in economics.
11.5. THE SUBJECTIVE NATURE OF THE DATA
Austrian economists distinguish two types of knowledge, namely objective

(scientific) and subjective (specific) knowledge. According to Garrison
(1986b, p. 442), the former includes "... an understanding of how the
economic system works - knowledge of the structure of the economy."
6 This description suggests that the method is more readily applicable in the social
than in the natural sciences. However, Popper (1957, p. 141n2) pointed out that this
method " ... has some vague parallel in the natural sciences, especially in thermodynamics
and in biology (the construction of mechanical models, and of physiological models of
processes and organs)." In fact, he noted that it is a version of Hayek's (and Menger's)
compositive method (p. 141nl).
252
This knowledge is held by the economic theorist who provides the analysis
in question. It concerns (conjectural) knowledge about economic theories,
which allows the theorist to explain how some overall order of economic
activity arises. Since it does not depend on the particular theorist holding
it, this type of knowledge will be called objective knowledge.1 However,
agents need not possess this type of knowledge. They are assumed to base
their actions on knowledge about the particular circumstances in which
they find themselves. Such knowledge may be called subjective knowledge,
since it depends on the subject holding it. It includes "... normal market
information coupled with various degrees of entrepreneurial insights" (p.
442). Its content will depend on the specific market under consideration.
Furthermore, this content will vary in time. As a result, subjective knowl-
edge is market- and time-specific, which means that it can only be known
to the participants on a particular market at a particular point in time. In
this sense it is dispersed over the agents in the economy.
It should be noted, though, that in Austrian economics the notion
of 'subjectivism' is somewhat ambiguous. As was shown in Chapter 6, the
Austrian revival changed the meaning of this notion. 'Older' Austrians
(Mises, Hayek) had adopted static subjectivism, assuming that decisions
are fully determinate (i.e. taken in single-exit decision situations). Dyna-
mic subjectivists, on the other hand, claim that this is inconsistent with
man's freedom to choose. They propose instead that there is no determi-
nate relationship between the outcomes of these decisions and the circum-
stances that prevailed before the choice was made. That is, dynamic
subjectivism means that decisions cannot be predicted (and hence are
taken in multiple-exit decision situations). This broadening of the concept
of subjectivism allows Austrians to consider entrepreneurial decision-
making, although at the same time it reduces the determinateness of their
analyses, in the sense that 'too much is possible'. As a result, Austrian
economics may be confronted with the criticism that 'complexity of
analysis' is not a virtue in itself.
Following Muth (1961), New Classicals adopt the assumption of
Muth-rationality for analytical reasons, that is, because it allows them to
mathematically formalize the agents' decision situations. In particular, the
assumption holds that the subjective probability distributions are identical
7
Since this type of knowledge is scientific in nature, it belongs to Popper's 'World
Three'. This 'world' is an unintended consequence of man's purposive activities, in the
sense that its content is man-made but independent from any agent or subject. In this
latter sense it is objective. Cf. Popper (1972, Chapter 4).
253
to the objective distributions, which means that subjective knowledge dif-
fers only randomly from objective knowledge. The random deviations of
subjective from objective knowledge are explained in terms of the differ-
ence between local and global information, which result from the assump-
tion that agents do not know the (current) values of global variables. 8
To summarize, Austrians assume that subjective knowledge may
systematically differ from objective knowledge. The ensuing dispersion of
knowledge means that economic theorists cannot give full explanations of
social phenomena in terms of the agents' decision situations. New Classi-
cals, on the other hand, adopt the simplifying assumption that subjective
knowledge does not differ systematically from objective knowledge, al-
though from the theorist's perspective, the former may be incomplete.
This difference in informational assumptions between Austrians and New
Classicals results from the latter's heuristic prescription of mathematical
formalization. The effects of this prescription may also be expounded in
terms of Knight's (1921) distinction between risk and uncertainty.
11.6. RISK AND UNCERTAINTY
Knight (1921 (1948), p. 233, italics in original) argued that "[t]he practical
difference between the two categories, risk and uncertainty, is that in the
former the distribution of the outcome in a group of instances is known ...
while in the case of uncertainty this is not true, the reason being in
general that it is impossible to form a group of instances, because the
situation dealt with is in a high degree unique." The outcome of a 'risky'
phenomenon may thus be predicted by using the 'true' probability dis-
tribution. Conversely, no information about such a distribution is available
in the case of an uncertain phenomenon, so that it cannot be predicted
using probability calculus. 9
Uncertainty plays an important role in Austrian views on the
process of competition. As Garrison (1986a, p. 94) stated, "[m]arket par-
ticipants must make decisions without knowing what the relevant true
8
This conclusion also holds in the case of King's (1982) introduction of agent-specific
shocks, because agents continue to know the 'true' objective probability distributions.
9 Note, as Knight himself already did, that it is possible for uncertainty to become risk.
This will be the case when over time some instances of the phenomenon under consi-
deration have occurred so that the individual may draw. up a probability distribution.
254
probabilities are and even without knowing what the full range of possible
outcomes are." They face an inherently unpredictable future, and hence
are confronted with fundamental uncertainty. Entrepreneurship attempts
to deal with this fundamental uncertainty. It plays a creative role, in the
sense that it allows economic subjects to form an image about the future
(including the other agents' future preferences and actions). Their subse-
quent actions aim to bring the actual future in closer correspondence with
this image, because the closer this correspondence, the more successful
their actions and hence the larger the profits. These profits need not be
pecuniary but may be 'merely' psychological. In the Austrian view, they
are not only a social reward for eliminating profit opportunities (and
hence discoordinations), but also a remuneration for bearing uncertainty.
.This 'uncertainty-bearing' property of entrepreneurship, however, implies
that the outcome of the market process cannot be predicted in any
reliable sense. As a corollary, entrepreneurial activities cannot be sub-
sumed under a probability distribution. Since Austrians concentrate on the
Hayek Problem and the ensuing process of competition, they stress the
crucial role of entrepreneurship and hence uncertainty.
In contrast, New Classicals disregard fundamental uncertainty.
They interpret business cycles as "... repeated instances of essentially
similar events." As Lucas (1975, p. 187) pointed out, New Classicals res-
trict their analyses to the situation in which "... the relevant distributions
have settled down to stationary values and thus can be 'known' by tra-
ders." Given these stationary distributions, systematic errors (i.e., system-
atic deviations from the mean value of the distribution) are easy to detect.
Rational economic agents will then avoid such expectational errors, so
that in the aggregate their expectations are correct. Lucas (1977, p. 224)
therefore concluded that "... it will be reasonable to treat agents as react-
ing to cyclical changes as 'risk' ... ". 10
Barreto (1989) showed that the elimination of fundamental uncer-
tainty follows from three core assumptions of the modern theory of the
firm. He argued that this theory presupposes some production function,
the logic of rational (situationally determinate) choice, and perfect
information (p. 102). He claimed that these core assumptions were intro-
duced because the ensuing consistent, interlocking theory was considered
to be desirable. In his view, "[c]onsistency represents a professional norm
10
Lucas (1977, p. 224) even claimed that •[i]ncases of uncertainty, economic reasoning
will be of no value. •
255
reflecting a mechanistic conception of real-world phenomena'L (p. 141).
Such a conception assumes away any indeterminateness from the econ-
omic system, and hence eliminates creativity and entreneurship.
Entrepreneurship implies the existence of relevant though
unknown (and unknowable) influences on the future outcome of the econ-
omic process. Such lack of knowledge about relevant influences cannot be
mathematically formalized. New Classicals therefore assume away funda-
mental uncertainty and indeterminateness by adopting the assumption of
Muth-rationality. The stationarity of the relevant probability distributions
ensures that the future does not differ systematically from the present
(and the past), so that any unknown and unknowable influence is elimin-
ated. Paraphrasing Barreto (1989, p. 137), entrepreneurship (as a short-
hand for fundamental uncertainty) has no place in New Classical Econ-
omics because it directly clashes with its heuristic prescription of math-
ematical formalization. The elimination of entrepreneurship is thus the
price to be paid for mathematical formalization. Obviously, New Classi-
cals are prepared to pay this price, whereas Austrians are not.
11.7. THE HOMOGENEITY POSTULATE
Since mathematical models are stylized pictures of reality, they must

abstract from those features of reality which are deemed less relevant for
the problem under consideration. According to Hoover (1988, p. 135), this
means that model-builders are confronted with what he called the Coumot
problem. This problem holds that economic reality is too complex to be
modelled fully, so that mathematical formalization presupposes the
adoption of simplifying assumptions. Macroeconomists usually solve this
problem by aggregating over agents and/or goods. Aggregation over goods
amounts to the assumption that there is one type of good, which may be
viewed as a standard basket of goods. Aggregation over agents takes a
more complex form. As was shown in Chapter 9, New Classicals adopt
two modelling strategies, both of which introduce the homogeneity postu-
late for the class of economic agents, namely the representative-agent
approach and the islands approach. Both approaches imply that agents and
their information sets do not differ systematically. The absence of such
differences between agents implies that they belong to a class of homo-
geneous elements. This allows for the introduction of the concept of a
representative agent, which is defined as the statistical mean of all agents.
256
The unsystematic deviations of this mean are accounted for by adding a
frequency distribution (with mean zero and finite variance), which can be
included in mathematical models.
Austrians start from the dispersion of knowledge, which implies
that the ex post profitability of courses of action can only be determined
according to some decentralized selection mechanism. In economics, this
mechanism is the competitive process, in which capable entrepreneurship
is rewarded and incapable entrepreneurship penalized. Such a selection,
however, is possible only if courses of action differ systematically, which
will be the case only if agents differ in information, skills and/ or abilities.
Hayek (1964, p. 35) therefore concluded that" ... most of the phenomena
in which we are interested, such as competition, could not occur at all
unless the number of distinct elements involved were fairly large, and ...
the overall pattern that will form itself is determined by the significantly
different behaviour of the different individuals ... " Such differences "...
cannot be overcome by treating them [i.e., the individuals] as members of
a statistical collective" (p. 35). 11 As a result, the process of competition
and the alleged complexity of social phenomena rule out the homogeneity
postulate for agents.
Austrians also refuse to accept the homogeneity postulate for
goods. Instead, they emphasize that goods are vertically related in the
structure of production. The ensuing distinction between types of goods
allows them to distinguish between the real-cash-balance effect and the
capital-allocation effect, as discussed in Chapter 3. The latter plays a
crucial part in Austrian explanations of discoordination phenomena. In
contrast, New Classicals adopt the homogeneity postulate for goods, thus
eliminating this effect. More correctly, they assume either that capital is
homogeneous, or that it consists of financial assets only. The latter are
'homogenized' by the 'no-frictions' assumption, which implies that there
are no transaction costs. As a result, the assets are perfect substitutes.
The difference between Austrians and New Classicals as regards
the appropriateness of the homogeneity postulate for agents and goods
leads to a difference in their respective business cycle theories. Austrians
explain such cycles in terms of coordination failures between the agents'
11
Hayek (1964, p. 30) argued that "the statistical method is ... of use only where we
either delibarately ignore, or are ignorant of, the relations between the individual elements
with different attributes, i.e., where we ignore or are ignorant of any structure into which
they are organized." This quotation reflects that the fact that Hayek's rejection of probabi-
lity calculus is not only based on systematic differences between agents ('different attribu-
tes'), but also on the interrelations between them ('the structure').
257
savings and investment decisions. These failures arise from nonneutral
monetary injections in the economy, which disturb the structure of relative
prices. This disturbance bring about rna/investments, culminating in distor-
tions of the structure of production. 12 New Classicals, on the other hand,
assume that there is 'helicopter money', which can be spread immediately
and proportionately over the economy. From a full-information point of
view, money is then neutral in a comparative-static sense. Furthermore,
the homogeneity postulate for goods eliminates malinvestments. Conse-
quently, the NCE explains business cycles in terms of overinvestments.
Austrians thus stress the complexity of social phenomena, whereas
New Classicals reduce this complexity in their endeavours to mathemat-
ically formalize their views. This difference affects what may be called the
respective degrees of explanation.
11.8. THE 'DEGREE OF EXPLANATION'
Popper (1959, pp. 112- 13) argued that the testability of a theory depends
on the class of opportunities for which it is discorroborated by experience,
that is, on the class of potential falsifiers. The larger this class, the more
events are ruled out and hence the more the theory under consideration
says about the world of experience. In turn, the empirical content of the
theory under consideration then increases, as does its testability. The
problem is, however, that it is difficult to measure the degree of test-
ability, because all classes of potential falsifiers are infinite, in the sense
that they contain an infinite number of falsifying basic statements, or
occurrences (pp. 113 - 14). Nevertheless, according to Popper (1959, p.
114), the relative degree of testability of a particular theory may be
established when the class of potential falsifiers of a particular theory a is
a proper subclass of that of theory {3. This is the case if all elements of
the former are potential falsifiers of {3, whilst {3 also has other potential
falsifiers. In this case the respective degrees of testability and hence the
12Leijonhufvud (1981, p. 143) already reached this conclusion when stating that
"[a]mong Wicksell's intellectual descendants, the Austrian business-cycle theorists were
particularly insistent that monetary impulses must disrupt the coordination of saving and
investment decisions and shove the system off its equilibrium growth path. Among the
Monetarists, Friedman has most strongly argued the view that the banking system cannot,
except very transitorily, affect the real rate of interest on which saving and investment
depend. The later Rational Expectations Monetarism [i.e. the NCE] need not invoke the
interest rate as part of the transmission mechanism at all."
258
respective empirical contents can be unambiguously determined. However,
if the two classes of potential falsifiers intersect, without one being
included in the other, no such unambiguous outcome results. Popper
(1959, p. 114) distinguished another, more intuitive measure, namely that
of the degree of composition of permitted occurrences (events, state-
ments).13 He advanced that "... basic statements, combined by conjunction
with other basic statements, again yield basic statements which, however,
are 'more highly composite' than their components." The more permitted
occurrences, the higher the degree of composition, and the greater the
empirical content of the theory under consideration. This implies that the
empirical content of a theory will increase with its precision. Furthermore,
Popper (1959) noted that theories will be more easily falsified if there are
fewer initial conditions to be taken into account. That is, "[ t ]he fewer the
magnitudes which are needed for determining the initial conditions, the
less composite will be the basic statements which suffice for the falsifica-
tion of the theory" (p. 127). These results help to explain Hayek's views
on the role of empirical testing.
Hayek (1964, p. 25) claimed that social phenomena are more
complex (in the sense of composite) than natural phenomena, using the
minimum number of variables necessary to describe the characteristic
features of the phenomenon under consideration as a measure of the
degree of complexity. In particular, this complexity results from the
subjective nature of the data in economics, the ensuing dispersion of
knowledge, and the heterogeneity of human behaviour in the competitive
process. These features do not allow fully articulated and completely
specified explanations. The incomplete and hence imprecise nature of
these explanations led Hayek (1942c, p. 289n1) to propose that econom-
ists should merely provide generic schemes .14 Their imprecision means, as
shown above, that they have little empirical content and hence are diffi-
cult to falsify. Stated differently, the schemes have a relatively low degree
13
Popper (1959, p. 114) pointed out that the composition of forbidden occurrences
cannot be used as a measure, because " ... the events forbidden by a theory can be of any
degree of composition", so that this measure cannot discriminate between theories of
different degrees of composition.
14
Hempel (1965, p. 238) suggested that scientifically acceptable schemes (or explanati-
on sketches, as he called them) should contain suggestions for further research. This
research should lead to a more detailed ('filed-out') explanation. Austrians, however, reject
this suggestion on the grounds that the dispersion of knowledge does not allow for an
increase in detail. In their view, generic schemes are necessarily incomplete.
259
of explanationY Paque (1990) pointed out that Hayek's (1964, p. 24)
suggestion that the social sciences may yield falsifiable pattern predictions
could not be maintained, as shown in Chapter 4. Latter-day Austrians,
such as Garrison and Wainhouse, appear to recognize the merits of this
reinterpretation. They do not dismiss economic testing as entirely useless,
but stress that the complex (composite) nature of economic phenomena
do not allow for rigorous testing.
According to Lucas (1980, p. 271), one of the main objectives of
New Classical model-building is "... to provide fully articulated, artificial
economic systems that can serve as laboratories in which policies that
would be prohibitively expensive to experiment with in actual economies
can be tested out at much lower cost." Since the availability of math-
ematical techniques restricts the domain of the NCE, this domain only
contains those problems which can be fully and determinately analyzed in
terms of mathematical models. The degree of precision of these fully
articulated and mathematically formalized explanations will be relatively
high, as compared with that of generic schemes. As a result, New Classical
models have relatively great empirical content and hence a relatively high
degree of explanation of the problems which they address. This means
that they can be tested relatively easily, and that econometric tests may be
applied for falsification purposes. However, as Lucas's statement expli-
cates, the purpose of policy-evaluation presupposes that the models used
are empirically satisfactory, in the sense that they should mimic the behav-
iour of actual economies (p. 272). New Classical econometric practice
concentrates on corroboration, not falsification. As Barro's test approach
reflected, however, repeated discorroborations may result in the theory's
refutation, particularly when some 'better' (more satisfactory) alternative
is available. Sargent's test approach, on the other hand, places doubts on
Barro's results by showing that the latter's implicit assumption about the
absence of a policy break may not be warranted.
New Classical econometric practice thus shows that some scepti-
cism as regards the scope of empirical testing is appropriate. However,
Austrians appear to overstretch this scepticism by rejecting such testing,
because at least they fail to appreciate its role in suggesting new direc-
tions for the very search.
15 Hayek (1955b, p. 4) adopted the hypothetical-deductive method of explanation,

which considers explanations and predictions as structurally identical (cf. Hempel (1965, p.
367)). This so-called symmetry thesis thus implies that the above conclusions not only
apply to explanations but also for predictions.
260
11.9. CONCLUSIONS
The present chapter has discussed in what respects Austrian economics

and New Classical Economics differ from each other. The first difference
concerned the central problems of the respective research traditions.
Hayek and other Austrians aimed to solve the Hayek Problem, which
concerns the coordination of the agents' plans and activities. Given the
subjective nature of the data in economics and the ensuing dispersion of
knowledge, the problem must be solved in some decentralized way. The
crucial question of how a decentralized price system can bring about a
tendency towards coordination is answered in reference to the process of
competition between economic agents. Since agents are creative decision-
makers, the description of this process must necessarily be indeterminate
and hence incomplete. Furthermore, the competitive process presupposes
significant differences between agents. Austrians therefore stress the
complexity of economic reality. They do not attempt to give full genetic-
causal explanations of the competitive process, but instead provide merely
generic schemes (explanation sketches). These schemes cannot explain
economic phenomena in full detail because of the dispersion of knowl-
edge. Their incompletenes means that they have a low degree of testabi-
lity, in the sense that discorroborations may always be attributed to incor-
rectly stated initial conditions or violations of the ceteris-paribus clause.
New Classicals eliminate the coordination problem by assuming
perfect price-flexibility and price-taking agents. This reflects Butos's
(1986) conclusion that they define the boundaries and possibilities of
economic analysis in terms of techniques available, as the adoption of the
homogeneity postulate for agents and goods shows. General-equilibrium is
considered to be a heuristic principle, not a situation to be explained. The
resulting mathematical-functional analysis assumes away the coordination
problem and the alleged informational advantages of decentralization (cf.
K.lausinger (1989a, p. 179)), as reflected by the assumption of Muth-
rationality. This assumption is based on the idea that the repeated nature
of business cycles allows agents to learn the objective probability distribu-
tions. The ensuing elimination of Knightean uncertainty and entrepreneur-
ial activities make it possible to build fully articulated mathematical
models. Since these models are fully specified, they have relatively high
empirical content, so that empirical testing becomes meaningful. However,
New Classical tests have yielded ambiguous results.
The differences in attitude towards the alleged complexity of
261
social phenomena is reflected in the respective explanations of business
cycles. The Austrians' refusal to accept the homogeneity postulate for
goods allows them to distinguish between the real-cash-balance effect and
the capital-allocation effect. The former concerns the agents' portfolio
holdings, whereas the latter apply to physical capital goods and hence the
vertical structure of production. Business cycles result from monetary
injections, which generate Cantillon effects and coordination failures
between investment and savings decisions. This results in malinvestments
and in distortions of the structure of production. New Classicals, on the
other hand, assume that money is neutral in a comparative-static sense.
The homogeneity postulates for agents and goods respectively ensure that
there are no Cantillon effects and malinvestments. In the New Classical
view, business cycles thus comprise overinvestments only.
262
12. SUMMARY, CONCLUSIONS AND EPILOGUE
As was argued in Chapter 1, the Popperian or Lakatosian criterion of

progressiveness is less relevant for our purposes, because the Austrian
research tradition does not try to attain it. Instead, the present study has
concentrated on what may best be called the respective 'developments' of
both the Austrian and New Classical traditions. The former was discussed
in Part I, and the latter in Part II. This chapter summarizes the main
argument of these parts.
12.1. DEVELOPMENT OF AUSTRIAN ECONOMICS
Austrian business cycle theory started from those continental explanations

of cyclical fluctuations which stressed the importance of the rate of
interest in the agents' savings and investment decisions. Mises elaborated
these explanations by adding his monetary theory and by incorporating
Menger's structure of production. The resulting theory explains business
cycles in terms of monetary injections and the ensuing fall of the money
rate of interest below the Wicksellian natural rate. The additional money
on the money market is assumed to find its way to entrepreneurs (inves-
tors), who increase their investments, thus lengthening the structure of
production. As O'Driscoll (1977, pp. 153 -54) pointed out, this means that
the Austrian business cycle theory stands in the Cantillon tradition,
emphasizing distribution effects. The lengthening of the structure of
production reflect malinvestments, i.e., investments in unprofitable pro-
jects and hence in the 'wrong' type of goods. These unprofitable projects
are abandoned during the depression, so that the economy eventually
returns to a new equilibrium position (which need not be the same as the
initial equilibrium position). Mises thus advanced that there is a tendency
towards general equilibrium. However, this tendency is continuously
hampered by changes in the data.
Mises's analysis suffers from two defects. Firstly, it adopts a rather
static version of subjectivism, which means that decisions are made in
single-exit decision situations, according to some pure logic of choice.
Instead, Hayek would 'improve' on Mises by broadening the latter's
benchmark of the 'evenly rotating economy' to an intertemporal equili-
brium. This may be seen as a theoretical improvement, because it allows
263
the inclusion of expectations without loosing any of the analytical features
of Mises's theory. It appears that Hayek's version of Austrian business
cycle theory encompasses that of Mises, although the theory remained
'sketchy'. In Hayek's (and other Austrians') view, the dispersion of
knowledge prohibits the explication of subjective knowledge and hence of
the agents' decision situations, so that theories cannot be tested in
meaningful ways. Hayek concluded that economists should merely provide
explanation sketches, instead of fully articulated explanations.
The second defect of Mises's theory is that it does not explicitly
distinguish between individual and general equilibrium. This suggests that
in his view both amount to the same thing, so that a movement in the
direction of the former also leads to a movement towards the latter. This
result, as Hayek would later show, crucially depend on the agents' skills
and abilities in gathering and processing information. Mter explicitly
distinguishing between the two equilibrium concepts, Hayek could address
the coordination problem and hence formulate the Hayek Problem. This
problem forms the core of the Hayek programme, which aims to explain
the alleged tendency towards coordination as the unintended consequence
of purposive behaviour. The concomitant programme is in fact a quest for
discoordination dynamics, whereby the process of competition is presumed
to ensure that all relevant subjective knowledge is communicated on
decentralized markets, thus bringing about a tendency towards coordina-
tion. In Austrian explanations of business cycles the coordinating tendency
checks both the boom and the depression, so that in the absence of
exogenous shocks the economy (eventually) moves towards some equilib-
rium position. It should be noted that this new equilibrium position may
differ from the initial one.
Lachmann (1945) followed Hayek's research agenda, in the sense
that he explicitly discussed the nature of knowledge and expectations in
economics. Criticizing Lange's (1944) attempt to obtain an 'objective'
measure for Hicks's (1939) 'elasticity of expectations', he pointed out that
this elasticity should not be interpreted mechanistically, since it depends
on the agents' free choices. His subsequent rejection of the usefulness and
appropriateness of mathematical formalization of this elasticity was based
on the agents' freedom to choose. This freedom means that decisions do
not follow deterministiclly from the decision situation in which it was
taken. That is, Lachmann substituted multiple-exit for single-exit decision
situations, rejecting the pure logic of choice as an appropriate way of
describing the agents' decision-making process. The ensuing substitution
264
of dynamic for static subjectivism meant a broadening of the Austrian
framework, thus transforming the Hayek Problem. After all, agents will
only be free to choose if their decision-making processes are indeter-
minate. This indeterminacy implies that the resulting actions need not
become coordinated, so that a situation of coordination need not arise.
Lachmann therefore abandoned the analysis of the tendency towards
coordination and instead proposed to study sequences of discoordination
situations. However, later Austrian developments failed to appreciate this
programmatic change. Other Austrians continued to direct their research
activities to the tendency towards coordination. The most influential in
this regard have been Rothbard and Kirzner.
Rothbard's main contribution to Austrian economics has been his
(1963) explanation of the crisis of 1929 and the ensuing depression in
terms of the Misesian version of Austrian business cycle theory. This
means that he adopted the latter's static benchmark, to the neglect of
Hayek's and Lachmann's elaborations and improvements. Kirzner (1973),
on the other hand, followed the Hayek Problem more closely. He tried to
explain the original Hayek Problem in terms of Misesian entrepreneurial
decision-making. Such decision-making was interpreted as creative, that is,
as taking place in multiple-exit decision si~uations. This means that it is
indeterminate and unpredictable. In the 1970s this emphasis on the role
of entrepreneurship in decentralized market economies would revive Aus-
trian economics. This revival led O'Driscoll (1977), Garrison (1978), and
Thalenhorst and Wenig (1984) to restate Austrian (macro)economics.
As was shown in Chapter 11, Austrians regard the process of
decision-making as indeterminate. The relevant knowledge in this process
is subjective and hence dispersed throughout the economy. This subjecti-
vity of knowledge means that the relevant knowledge is private and can
only be known by the decision-maker under consideration. Stated differ-
ently, Austrians emphasize the importance of mental states in the
decision-making process. They stress the psychological nature of choice.
The concomitant dispersion of knowledge does not allow economic theor-
ists to explain social phenomena in full detail. As a result, Austrians
downplay the importance of empirical testing, because the inevitably
incomplete explanations of social (hence complex) phenomena cannot be
rigorously tested. As a corollary, empirical testing is considered to be
rather meaningless because it can only yield ambiguous results.
265
12.2. DEVELOPMENT OF NEW CLASSICAL ECONOMICS
Part II discussed the development in New Classical Economics, also

concentrating on its business cycle theory. This theory is an equilibrium
theory, which finds its origin in Friedman's and Phelps's attempt to
rationalize the Phillips Curve. Friedman rejected the implication of this
curve that agents suffer from money illusion. Instead, he tried to
rationalize it with reference to expectational errors. His Natural Rate
Hypothesis (NRH) holds that there is no relationship between money and
output (unemployment), so that the implied Phillips Curve is vertical.
Since he adopted a version of the quantity theory of money, long-term
(anticipated) changes in the money supply only affect the general price
level. This implies that Friedman abstracted from capital-allocation and
Cantillon effects. The short-run non-verticality of the Phillips Curve was
explained as the result of expectational errors. This means that some
expectations formation mechanism had to be formulated. Lucas and Rap-
ping (1969a) adopted the AEH for this purpose. However, given the
absence of frictions in the economic system, this hypothesis is in danger of
being inconsistent with the rationality postulate. Lucas (1972a) therefore
replaced it by Muth's (1961) REH.
Since the REH does not allow expectations to adjust adaptively to
new information, Lucas sought some other source of expectational errors.
He found it in Phelps's 'island parable', which suggested that global
information may be lagged or even absent. The ensuing 'island approach'
became the New Classical modelling standard. 1 From a full-information
position both approaches retain the 'classical' neutrality proposition, so
that they are consistent with Friedman's NRH. The joint NR/RE hypo-
thesis became the hallmark of New Classical Economics, whereby the
NRH was assumed to be path-independent. Its underlying assumption of
continuous and instantaneous market-clearing (i.e. price-taking behaviour)
assumed away the very coordination problem which formed the central
feature of the Hayek programme. The notion of intertemporal general
equilibrium constitutes a heuristic device, not a situation to be explained.
The New Classical version of Popper's method of rational model-construc-
tion thereby explained the agents' decisions in terms of single-exit decision
situations.
1
Sargent and Wallace often adopted the much simpler 'representative-agent approach',
but they did so only with reference to Lucas's 'islands approach'.
266
Lucas (1972a) improved on Friedman (1968), Phelps (1967), and
Lucas and Rapping (1969a), in the sense that he mathematically forma-
lized a general-equilibrium model with imperfect information on the part
of the agents. This model implied that anticipated changes in the supply
of money are neutral, whereas unanticipated changes affect real output
and unemployment. Lucas (1972b) derived another implication from his
joint NR/RE hypothesis, namely that the usual methods of policy-evalua-
tion exhibited serious shortcomings because of the REH. This Lucas Criti-
que holds that changes in the policy parameters may affect the parameters
in the behavioural equations, so that there are so-called cross-equational
restrictions. These restrictions were logically derived from the REH and
the definition of policy in terms of rules. They implied that past tests of
the joint NR/RE hypothesis had been invalid, as was simultaneously
though independently shown by Sargent (1971). In turn, serious testing of
the joint NR/RE hypothesis was yet to be undertaken. Lucas (1972b) sug-
gested an alternative test procedure. However, he himself did not follow
this suggestion, although the Lucas critique provided a valuable insight
which was used in Lucas's (1973) test. This critique had suggested a
relationship between the variance of monetary policy and the slope
parameter of the short-term Phillips Curve. This relationship implies that
systematic monetary policy becomes less effective the more it has been
used, because agents will attribute a larger proportion of a given price
change to monetary causes. Lucas's (1973) test results corroborated this
suggestion and hence the underlying joint NR/RE hypothesis. However, it
should be noted that this corroborative result depends on the ad-hoc
inclusion of lagged values of real output, required in order to deal with
persistence.
Sargent (1973b) tested the joint NR/RE hypothesis along diffe-
rent lines. The results of this test were rather unfavourable. Nevertheless,
Sargent did not reject the theory, but instead listed three types of reasons
why such a rejection would be inappropriate. The first type concerns
incorrectly specified initial conditions and/ or violations of the ceteris-
paribus clause. Secondly, Sargent 'retreated' to the methodological
position of sophisticated falsificationism, which holds that a theory should
not be rejected if no better alternative is available. Sargent's (1973b)
comparative test indicated that there was no such alternative. The third
type involved prior beliefs, and was used in conjunction with the other two
types. In other words, the joint NR/RE hypothesis was defended on a
priori grounds.
267
The early New Classical models, discussed in Chapter 8, could not
account for persistence in the deviations in real output (unemployment)
from its natural rate. Lucas (1975) solved this problem by adopting the
Slutzky-Frisch distinction between impulses and propagation mechanisms.
This solution ensured that random shocks could generate cyclical move-
ments in aggregate variables. Lucas's model included local capital markets
only. He himself recognized that the introduction of a global capital
market could jeopardize his (1975) non-neutrality result. Barro (1980)
generalized Lucas's model by showing that this result would also hold in
the presence of such a global capital market. The necessary prerequisite
for this result was the introduction of a third shock, namely an aggregate
money demand shock. It should be noted, though, that Barro's model does
not represent an unambiguous generalization of Lucas's model, since it
did not include a propagation mechanism. On the other hand (as Barro
himself claimed), it appears that its extension with such a mechanism does
not pose any theoretical problem.
Sargent and Wallace (1975) and Barro (1976) shifted attention to
the neutrality proposition, and away from business cycle theory. This shift
was accomplished by combining the joint NR/RE hypothesis with the 'no-
frictions' assumption, which holds that there are no transaction and
adjustment costs, so that information can be acquired costlessly and stock
variables can be adjusted immediately and perfectly. Agents then do not
make systematical expectational errors. King (1982) subsequently showed
that this neutrality proposition does not hold in an economy in which
perceived nominal prices are subject to agents-specific shocks.
There have been two approaches to test the neutrality proposition
and the underlying joint NR/RE hypothesis. The first approach started
with Sargent's (1976a) test, which provided some corroborative evidence.
However, his (1976b) 'observational equivalence' almost immediately ren-
dered this evidence meaningless. Neftci and Sargent (1978) subsequently
identified a policy break, so that they could circumvent this observational
equivalence. Their test results corroborated the neutrality proposition.
Lucas (1987) and Sargent (1987) therefore maintained the view that cycli-
cal fluctuations in aggregate variables may result from monetary distur-
bances. Barro (1978) and Barro and Rush (1980) adopted a more direct
test approach, which yielded opposite and hence unfavourable results for
the neutrality proposition. Additionally, Barro and Hercowitz (1980)
claimed that unperceived money could not explain fluctuations in aggre-
gate output, thus contradicting the implications of the joint NR/RE
268
hypothesis. This discorroborative result, however, was based on Barro's
neglect of Neftci and Sargent's policy breaks. Nevertheless, Barro and
many other New Classical economists concluded that monetary explana-
tions of business cycles were empirically invalid. They therefore reversed
causation, suggesting that fluctuations in monetary variables reflect
endogenous changes, caused by (exogenous) changes in real variables.
That is, they switched from monetary to real business cycle theory. The
availability of King and Plosser's (1984) empirically more satisfactory
alternative may have induced this switch.
12.3. EPILOGUE
The above summaries show that the Austrian school and New Classical
Economics study different problems. The former aims to explain the
tendency towards coordination in a genetic-causal way, whereas the latter
eliminates this problem from the domain of economics. That is, Austrians
consider coordination a phenomenon to be explained, whereas New Clas-
sicals uses this concept as a heuristic principle. Furthermore, the different
problems are studied in different ways. The coordination problem arises
because of the complexity of social reality. In addressing this problem,
Austrians stress this complexity. Their dynamic subjectivism only allows
for so-called explanation sketches, which explain social phenomena in
generic terms. The New Classical heuristic prescription of mathematical
formalization does not allow for such complexity. Therefore, the NCE
(implicitly) introduces the homogeneity postulate for agents and goods.
Furthermore, the prescription leads to a removal of indeterminateness
from its analysis. In particular, this means that agents are presumed to
decide in single-exit decision situations. As a corollary, New Classicals
disregard fundamental uncertainty and hence entrepreneurship.
Austrian and New Classical economics thus differ considerably.
These differences do not only concern the context of discovery, but also
the context of justification. As was shown above, Austrians use the
method of Verstehen in this context, and hence do not consider empirical
testing meaningful. This conclusion is enhanced by the so-called Duhem-
Quine thesis, which holds that a hypothesis must be tested in conjunction
with many supporting assumptions. A discorroborative test result then
need not lead to the rejection of the hypothesis under consideration.
However, the use of the method of Verstehen in the context of justification
269
may be subjected to serious criticism. As Popper pointed out, the results
of this method are not infallible and hence cannot be used as a criterion
for theory selection. New Classicals follow Popper's rejection of the use of
the method of Verstehen in the context of justification, and hence must lay
down different criteria in order to be able to choose between rival
theories. As was shown in Chapter 10, they attach an important role to
corroborations, implying that such corroborations increase the plausibility
of the tested theory. However, New Classical econometric practice also
shows that there may be rival test approaches which yield opposite results.
This ambiguity of test results reflects that empirical testing need not and
will not establish unambiguously the empirical (in)validity of economic
theories. It thus appears that there is at least some (albeit little) merit in
Austrian scepticism as regards the meaningfulness of empirical testing.
270
BIBLIOGRAPHY
Agassi, J. (1969) 'The novelty of Popper's philosophy of science', International Philosophi

cal Quarterly 8, pp. 442 -63.
Agassi, J. (1979) 'The legacy of Lakatos', Philosophy of the Social Sciences 9, 316- 26.
Andersen, L.C.,and J.L. Jordan (1968) 'Monetary and fiscal actions: a test of their relative
importance in economic stabilization', Review of the Federal Reserve Bank of St.
Louis, vol. 50, no. 1. Reprinted in H.R. Williams and W.R. Woudenberg (eds.)
(1970) Money, Banking, and Monetary Policy, New York: Harper & Row.
Andersen, T.M. (1989) 'Comment on B.T. McCallum, "New Classical Macroeconomics: a
sympathetic account"', Scandinavian Journal of Economics 91,252- 58.
Aschheim, J., and C.-Y. Hsieh (1969) Macroeconomics, Income and Monetary Theory,
Columbus, Ohio: Charles E. Merril.
Attfield, C.L.F.,D. Demery, and N.W. Duck (1979) 'A quarterly model of unanticipated
monetary growth in the U.K., 1963 - 1978', unpublished, University of Bristol,
September.
Attfield, C.L.F.,D. Demery, and N.W. Duck (1985) Rational Expectations in Macroecono
mics, Oxford: Blackwell.
Backhouse, R. (1987) A History of Modern Economic Analysis, Oxford: Blackwell.
Barreto, H. (1989) The Entrepreneur in Microeconomic Theory, London: Routledge.
Barro, R.J. (1976) 'Rational expectations and the role of monetary policy', Journal of
Monetary Economics 2, 1 - 32.
Barro, R.J. (1977a) 'Long term contracting, sticky prices, and monetary policy' ,Journal of
Monetary Economics 3, 305- 16. Reprinted in R.J. Barro (1981a) pp. 213- 24.
Barro, R.J. (1977b) 'Unanticipated money growth and unemployment in the United States',
American Economic Review 61, 101 - 15. Reprinted in R.E. Lucas, Jr., and T.J.
Sargent (eds.) (1981).
Barro, R.J. (1978a) 'Unanticipated money, output, and the price level in the United States',
Journal of Political Economy 86, 549- 80. Reprinted in R.E. Lucas, Jr., and T.J.
Sargent (eds.) (1981).
Barro, R.J. (1980) 'A capital market in an equilibrium business cycle model', Econometrica
48, 1393- 1417.
Barro, R.J. (1981a) Money, Expectations, and Business Cycles, New York: Academic Press.
Barro, R.J. (1981b) 'The equilibrium approach to business cycles', in R.J. Barro (1981a)
pp. 41- 78.
Barro, R.J. (ed.) (1989) Modern Business Cycle Theory, Oxford: Blackwell.
Barro, R.J., and R.G. King (1984) 'Time-separable preferences and intertemporal-
substitution models of business cycles', Quarterly Journal of Economics 98, 817-
39.
Barro, R.J., and S. Fischer (1981) 'Recent developments in monetary theory', in R.J. Barro
(1981a) pp. 3 - 40.
Barro, R.J., and Z. Hercowitz (1980) 'Money stock revisions and unanticipated money
growth', Journal of Monetary Economics 6, 257 - 67.
Barro, R.J., and M. Rush (1980) 'Unanticipated money and economic activity', in S.
Fischer (ed.) (1980) pp. 23 - 74.
Barry, N.P. (1979) Hayek's Social and Political Philosophy, London: MacMillan.
Begg, D.K.H. (1980) 'Rational expectations and the non-neutrality of systematic monetary
policy' ,Review of Economic Studies 44,293- 303.
Begg, D.K.H. (1982) 'Rational expectations, wage rigidity and involuntary unemployment',
Oxford Economic Papers 34, 23 - 48.
Begg, D.K.H. (1984) The Rational Expectations Revolution in Macroeconomics, Oxford:
Philip Allan.
Bellante, D., and R. Garrison (1988) 'Phillips curves and Hayekian triangles: two perspecti
271
ves on monetary dynamics', History of Political Economy 20, 207 - 34.
Binmore, K. (1990) Essays on the Foundations of Game Theory, Oxford: Blackwell.
Blanchard, O.J. (1989) 'Comment on B.T. McCallum, "New Classical Macroeconomics: a
sympathetic account"', Scandinavian Journal of Economics 91, 259- 64.
Blanchard, 0., and S. Fischer (1989) Lectures on Macroeconomics, Cambridge (Mass.):
MIT Press.
Blaug, M. (1962) Economic Theory in Retrospect, Cambridge: Cambridge University Press,
1990.
Blejer, M.l., and R.B. Fernandez (1979) 'The effects of unanticipated money growth on
prices, output, and its composition in a fixed exchange-rate open economy',
unpublished, Hebrew University of Jerusalem, April.
Blinder, A.S., and S. Fischer (1981) 'Inventories, rational expectations, and the business
cycle' ,Journal of Monetary Economics 8, 277- 304.
Block, W. (1980) 'On Robert Nozick's "On Austrian Methodology"', Inquiry 23, 397 - 444.
Block, W. (1988) 'On Yeager's "Why Subjectivism?"' ,Review of Austrian Economics 2, 199
-208.
Blume, L.E., M.M. Bray, and D. Easley (1982) 'Introduction to the stability of rational
expectations equilibrium', Journal of Economic Theory 26, 313 - 17.
Blume, L.E., and D. Easley (1982) 'Learning to be rational', Journal of Economic Theory
26,340- 51.
Bohm, S. (1982) 'The ambiguous notion of subjectivism: comment on Lachmann', in I.M.
Kirzner (ed.) (1982) pp. 41 -52.
Bohm-Bawerk, E. von (1889) Kapital und Kapitalzins, 3 vols. (1. Positive Theorie des
Kapitales, II. Positive Theorie des Kapitales, III. Geschichte und Kritik der
Kapitalzins-Theorien), Jena: Fischer, 1921.
Boland, L.A. (1978) 'Time in economics vs. economics in time: the "Hayek Problem"',
Canadian Journal of Economics 11, 240 - 62.
Boland, L.A. (1982) The Foundations of Economic Method, London: Allen & Unwin.
Boland, L.A. (1984) 'Methodology: reply' ,American Economic Review 74,795- 97.
Boland, L.A. (1986) Methodology for a New Microeconomics, Boston: Allen & Unwin.
Boland, L.A. (1987) 'On the relevance of neo-Walrasian economic theory: a review',
History of Political Economy 19,659- 66.
Bordo, M.D. (1989) 'Equation of Exchange', in J. Eatwell et al. (eds.) (1989c), pp. 151 -
56.
Boschen, J.F., and H.l. Grossman (1982) 'Test of equilibrium macroeconomics using
contemporaneous monetary data', Journal of Monetary Economics 10,309- 33.
Boulding, K.E. (1966) 'The economics of knowledge and the knowledge of economics',
American Economic Review 56,3- 13.
Braithwaite, R.B. (1953) Scientific Explanation, Cambridge: Cambridge University Press.
Bray, M.M. (1982) 'Learning, estimation, and the stability of rational expectations', Journal
of Economic Theory 26, 318- 39.
Brechling, F. (1968) 'The trade-off between inflation and unemployment', Journal of
Political Economy 76, 712 - 37.
Brock, W.A. (1972) 'On models of expectations that arise from maximizing behavior of
economic agents over time', Journal of Economic Theory 5, 348- 76.
Brock, W.A. (1974) 'Money and growth: the case of long run perfect foresight', Inter
national Economic Review 15,750- 77.
Bronfenbrenner, M. (ed.) (1969) Is the Business Cycle Obsolete?, New York: John Wiley.
Brunner, K. (1968) 'The role of money and monetary policy', Federal Reserve Bank of St.
Louis Review 50, 8- 24.
Brunner, K., and A.H. Meltzer (eds.) (1976) The Phillips Curve and Labor Markets,
Carnegie-Rochester Conference Series on Public Policy, vol. 1, Amsterdam:
North-Holland.
Brunner, K., and A.H. Meltzer (eds.) (1981) The Costs and Consequences of Inflation,
Carnegie-Rochester Conference Series on Public Policy, vol. 15, Amsterdam:
272
North-Holland.
Buiter, W. (1980) 'The macroeconomics of dr. Pangloss: a critical survey of the new
classical macroeconomics', Economic Journal90, 34- SO.
Burmeister, E. (1980) 'On some conceptual issues in rational expectations modelling',
Journal of Money, Credit and Banking 12, 800- 16.
Butler, E. (1983) Hayek: his contributions to the political and economic thought of our time,
London: Temple Smith.
Butler, E. (1988) Ludwig von Mises:fountainhead of the modern microeconomics revolution,
London: Gower.
Butos, W. (1986) 'Hayek and general equilibrium analysis', Southern Economic Journal 52,
332- 43.
Cagan, P. (1956) 'The monetary dynamics of hyperinflation', in M. Friedman (ed.) (19S6a).
Cagan, P. (1965) Determinants and Effects of Changes in the Supply of Money, New York:
NBER.
Cagan, P. (1989) 'Monetarism', in J. Eatwell et al. (eds.) (1989c) pp. 195- 205.
Cagan, P., and A. Gandolfi (1969) 'The lag in monetary policy as implied by the time
pattern of monetary effects on interest rates', American Economic Review 59,277
- 84.
Cairnes, J.E. (1965) The Character and Logical Method of Political Economy, New York.
Caldwell, B. (1980) 'Positivist philosophy of science and the methodology of economics',
Journal of Economic Issues 14, 53 - 76.
Caldwell, B.J. (1982) Beyond Positivism, London: Allen & Unwin.
Caldwell, B.J. (1984) 'Praxeology and its critics: an appraisal', History of Political Economy
16,363- 79.
Caldwell, B.J. (1986) 'Towards a broader conception of criticism', History of Political Econ
omy 18, 675 - 82.
Caldwell, B.J. (1988) 'Hayek's transformation', History of Political Economy 20,513- 41.
Caldwell, B.J. (1991) 'Has formalization gone too far in economics: a comment', Methodus
3 (1), 27 - 29.
Caldwell, B.J. (ed.) (1984) Appraisal and Criticism in Economics, Boston: Allen & Unwin.
Camus, A. (1975) De Myte van Sisyfus, Amsterdam: De Bezige Bij, 1985.
Cantillon, R. (1755) Essai sur Ia Nature du Commerce en General (edited and translated
into English by H. Higgs), 1931, London: MacMillan.
Cassel, G. (1918) Theoretische Sozialokonomie, Book IV (1918, written in 1914), Leipzig:
Scholl, 1932. English translation, The Theory of Social Economy (1923), New
York: Kelley.
Chipman, J.S. (1954) 'Empirical testing and mathematical models', The Review of Econo
mics and Statistics 36, 363 - 65.
Clower, R. ( 1967) 'A reconsideration of the microfoundations of monetary theory', Western
Economic Journal6 (1), 1 - 8.
Clower, R. (1980) Money and Markets, Cambridge: Cambridge University Press.
Cohen, M.R. (1931) Reason and Nature: An Essay on the Meaning of Scientific Method,
New York.
Cohen, R.S.,et al. (eds.) (1976) Essays in Memory of Imre Lakatos, Dordrecht: Reidel.
Colander, D. and R. Guthrie (1980) 'Great expectations: what the Dickens do rational
expectations mean ?',Journal of Post-Keynesian Economics 3, 219- 34.
Copleston, F. (1964) A History of Philosophy, London: Burns and Oates, 1985, Vol. VI,
'Wolff to Kant', Part IV, 'Kant', pp. 180- 440.
Cournot, A.A. (1838) Researches into the Mathematical Principles of the Theory of Wealth,
(transl. Nathaniel T. Bacon), New York: MacMillan, 1927.
Cournot, A.A. (1843) Exposition de Ia Theorie des Chances et des Probabili~s. Oeuvres
Completes, vol. I, ed. Bernard Bru, Paris: Librairie Philosophique J. Vrin, Ch. II,
'Des chances et de Ia probabilite mathematique', pp. 21 - 35.
Craver, E. (1986) 'The emigration of the Austrian economists', History of Political
Economy 18 (1), 1 - 32.
273
Cross, R. (1984a) 'Monetarism and Duhem's thesis', in P. Wiles and G. Routh (eds.)
(1984) pp. 78 - 99.
Cross, R. (1984b) 'Reply to comments', in P. Wiles and G. Routh (eds.) (1984) pp. 111 -
15.
Cross, R. (ed.) (1988) Unemployment, Hysteresis and the Natural Rate of Unemployment,
Oxford: Blackwell.
Cross, R. (1991) 'Alternative accounts of equilibrium unemployment', inN. de Marchi and
M. Blaug (eds.) (1991) pp. 294- 323.
Cukierman, A. (1979) 'Rational expectations and the role of monetary policy: a generali
zation', Journal of Monetary Economics 5, 213- 30.
Cukierman, A. (1981) 'Interest rates during the trade cycle, inventories and monetary
policy - a theoretical analysis', in K. Brunner and A. H. Meltzer (eds.) (1981) pp.
87- 144.
Darby, M.R. (1976) 'Rational expectations under conditions of costly information', Journal
of Finance 31, 889- 95.
Darnell, A.C.,and J.L. Evans (1990) The Limits of Econometrics, Aldershot: Edgar Elgar.
Debreu, G. (1984) 'Economic theory in the mathematical mode', American Economic
Review 74, 267 - 78.
DeCanio, S.J. (1979) 'Rational expectations and learning from experience', Quarterly
Journal of Economics 93, 47 - 57.
Desai, M. (1975) 'The Phillips curve: a revisionist interpretation', Economica 42, 1 - 19.
Desai, M. (1984) 'Wages, prices and employment a quarter century after the Phillips
curve', in D.F. Hendry and K.F. Wallis (eds.) (1984) Econometrics and Quantita-
tive Economics, Oxford: Blackwell, pp. 253- 73.
Diamond, P., and Fudenberg ( 1989) 'Rational expectations business cycles in search
equilibrium', Journal of Political Economy 97, 606- 19.
Dolan, E.G. (ed.) (1976) The Foundations of Modern Austrian Economics, Kansas City:
Sheed & Ward.
Dorfman, R. (1954) 'A catechism: mathematics in social science', The Review of Economics
and Statistics 36, 374- 77.
Dotsey, M., and R.G. King (1983) Monetary Instruments and Political Rules in a Rational
Expectations Environment, NBER Working Paper Series, Cambridge.
Duesenberry, J.S. (1954) 'The methodological basis of economic theory', The Review of
Economics and Statistics 36, 361- 63.
Duhem, P. (1954) 'Physical theory and experiment', Chapter VI in The Aim and Structure
of Physical Theory, Princeton: Princeton University Press. Originally published in
French in 1906. Reprinted in S.G. Harding (ed.) (1976), pp. 1 - 40.
Eatwell, J., M. Milgate and P. Newman (eds.) (1989a) General Equilibrium, The New
Palgrave, London: MacMillan.
Eatwell, J., M. Milgate and P. Newman (eds.) (1989b) The Invisible Hand, The New
Eatwell, J., M. Milgate and P. Newman (eds.) (1989c) Money, The New Palgrave, London:
MacMillan.
Eatwell, J., M. Milgate and P. Newman (eds.) (1989d) Utility and Probability, The New
Eatwell, J., M. Milgate and P. Newman (eds.) (1990) Capital Theory, The New Palgrave,
London: MacMillan.
Egger, J.B. (1978) 'The Austrian method', in L.M. Spadaro (ed.) pp. 19- 39.
Egger, J.B. (1986) 'A sympathetic critic of the Austrian business-cycle theory', in I.M.
Kirmer (ed.) (1986) pp. 56- 71.
Endres, A.M. (1987) 'The origins of Bohm-Bawerk's "greatest error": theoretical points of
separation from Menger', Journal of Institutional and Theoretical Economics 143,
291- 309.
Endres, A.M. (1991) 'Menger, Wieser, Bohm-Bawerk, and the analysis of economizing
behavior', History of Political Economy 23,279- 99.
274
Estey, J.A. (1956) Business Cycles, Englewood Cliffs, N.J.: Prentice-Hall.
Faber, M. (ed.) (1986a) Studies in Austrian Capital Theory, Investment and 1ime, Berlin:
Springer Verlag.
Faber, M. (1986b) 'On the development of Austrian capital theory', in M. Faber (1986a).
Fair, R.C. (1979) 'An analysis of the accuracy of four macroeconometric models', Journal
of Political Economy 87, 701 - 18.
Fama, E. (1980) 'Banking in the theory of finance', Journal of Monetary Economics 6 (1),
39- 57.
Feldstein, M.S., and A. Auerbach (1976) 'Inventory behavior in durable goods manufactu
ring: the target-adjustment model', Brookings Paper on Economic Activity 2, 351 -
96.
Fellner, W. (1980) 'The valid core of rationality hypotheses in the theory of expectations',
Journal of Money, Credit and Banking 12, 763 - 87.
Ferguson, A. (1767) An Essay on the History of Civil Society, Edinburgh: Edinburgh Uni
versity Press, 1978.
Fetter, F.A. (1977) Capital, Interest, and Rent, Mission (Kansas).
Fischer, S. (1977) 'Long term contracts, rational expectations and the optimal money
supply rule', Journal of Political Economy 85, 191 - 205.
Fischer, S. (1979) 'Anticipations and the nonneutrality of money', Journal of Political
Economy 87, 225 - 52.
Fischer, S. (1980) 'On activist monetary policy with rational expectations', in S. Fischer
(ed.) (1980) pp. 211 - 48.
Fischer, S. (1981a) 'Is there a real-balance effect in equilibrium?', Journal of Monetary
Economics 8, 25- 39.
Fischer, S. (1986a) 'Monetary rules and commodity money schemes under uncertainty',
Journal of Monetary Economics 17,21- 35.
Fischer, S. (1986b) 'Friedman versus Hayek on private money, review essay', Journal of
Fischer, S. (1988) 'Recent developments in macroeconomics', Economic Journal 98, 294 -
339.
Fischer, S. (ed.) (1980) Rational Expectations and Economic Policy, Chicago: University of
Chicago Press.
Fisher, F. (1972) 'Comment on paper by R. Lucas, in 0. Eckstein (ed.) (1972) The Econo
metrics of Price Determination, Washington, D.C.. ·
Fisher, I. (1922) The Purchasing Power of Money, New York: MacMillan.
Fisher, I. (1925) 'Our unstable dollar and the so-called business cycle', Journal of the
American Statistical Association 20.
Fisher, I. (1926) 'A statistical relationship between unemployment and price changes',
International Labour Review 13.
Fisher, I. (1930) The Theory of Interest, New York: MacMillan, reprint by August M. Kelly,
1961.
Foss, N.J. (1989) 'Hayek's transformation: some addenda to the standard account',
unpublished paper, Copenhagen: Center for Educational Research.
Foss, N.J. (1990) 'A note on the lag-structure in Austrian business cycle theory', unpublis
hed paper, Copenhagen: Center for Educational Research.
Friedman, B. (1979) 'Optimal expectations and the extreme information assumptions of
"Rational Expectations Macromodels"', Journal of Monetary Economics 5, 23- 41.
Friedman, M. (1948) 'A monetary and fiscal framework for economic stability', American
Economic Review 38, 245 - 64.
Friedman, M. (1953) Essays in Positive Economics, Chicago.
Friedman, M. (1956) 'The quantity theory of money: a restatement', in M. Friedman (ed.)
(1969).
Friedman, M. (1957) A Theory of the Consumption Function, Chicago: Aldine.
Friedman, M. (1958) 'The supply of money and changes in prices and output', in Joint
Economic Committee Print, The Relationship of Prices to Economic Stability and
275
Growth 85th Congress, 2nd session, Washington, D.C.: US Government Printing
Office, pp. 241- 56. Reprinted in Friedman (1969), pp. 171- 87.
Friedman, M. (1968) 'The role of monetary policy' ,American Economic Review 58, 1- 17.
Friedman, M. (1969) The Optimal Quantity of Money and Other Essays, London: MacMil
Ian.
Friedman, M. (1971) 'A theoretical framework for monetary analysis', NBER Occasional
Paper 112, New York: Columbia· University Press.
Friedman, M. (ed.) (1956) Studies in the Quantity Theory of Money, Chicago.
Friedman, M., and A.J. Schwartz (1963a) A Monetary History of the United States, 1867-
1960, Princeton, N.J.: Princeton University Press, 1970.
Friedman, M., and A.J. Schwartz (1963b) 'Money and business cycles', in M. Friedman
(1969).
Friedman, M., and A.J. Schwartz (1982) Monetary Trends in the United States and the
United Kingdom, Chicago: University of Chicago Press.
Friedman, M., and A.J. Schwartz (1986) 'Has government any role in money?', Journal of
Frisch, R.A. (1933) 'Propagation problems and impulses problems in dynamic economics',
in Economic Essays in Honour of Gustav Cassel London: Allen & Unwin.
Reprinted by London: Cass and Co., 1967,pp. 171-205.
Frisch, R.A. (1939) 'A note on errors in time series', QuarterlyJournal of Economics 53,
639-40.
Frydman, R. (1982) 'Towards an understanding of market processes: individual expec
tations, learning, and convergence to rational expectations equilibrium', American
Frydman, R. (1983) 'Individual rationality, decentralization, and the rational expectations
hypothesis', in R. Frydman and E.S. Phelps (eds.) (1983).
Frydman, R., and E.S. Phelps (eds.) (1983) Individual Forecasting and AggregateOutcomes,
Cambridge.
Garrison, R.W. (1978) 'Austrian economics: a diagrammatical exposition', in L.M. Spadaro
(ed.) (1978) pp. 167- 204.
Garrison, R.W. (1981) 'The Austrian-Neoclassical Relation: a study in monetary dynamics',
unpublished Ph.D. -dissertation, University of Virginia.
Garrison, R.W. (1982) 'Austrian economics as the middle ground: comment on Loasby', in
I.M. Kirzner (ed.) (1982) pp. 131- 38.
Garrison, R.W. (1984) 'Time and money: the universals of macroeconomic theorizing',
Journal of Macroeconomics 6, 197- 213.
Garrison, R.W. (1985a) 'A subjectivist view of a capital-using economy', in G.P. O'Driscoll,
Jr., and M.J. Rizzo (1985) pp. 160- 87.
Garrison, R.W. (1985b) 'lntertemporal coordination and the invisible hand: an Austrian
perspective on the Keynesian vision', History of Political Economy 17, 309 - 21.
Garrison, R.W. (1986a) 'From Lachmann to Lucas: on institutions, expectations, and
equilibrating tendencies', in I.M. Kirzner (ed.) (1986) pp. 87 - 101.
Garrison, R.W. (1986b) 'Hayekian trade cycle theory: a reappraisal', Cato Journa/6, 437 -
53.
Garrison, R.W. (1987) 'Full employment and intertemporal coordination: a rejoinder',
History of Political Economy 19, 335- 41.
Garrison, R. W. ( 1989) 'The Austrian theory of the business cycle in the light of modem
macroeconomics', Review of Austrian Economics 3, 3- 29.
Garrison, R.W. (1991) 'New Classical and Old Austrian economics: equilibrium business
cycle theory in perspective', Review of Austrian Economics 5, 91 - 103.
Gibson, A.H. (1923) 'The future course of high class investment values', The Bankers'
Magazine, January 1923.
Gordon, R.J. (1976a) 'Recent developments in the theory of inflation and unemployment',
Journal of Monetary Economics 2, 185- 219.
Gordon, R.J. (1976b) 'Can econometric policy evaluation be salvaged?: a comment', in
276
'The Phillips Curve and Labor Markets', pp. 47- 61, supplement to the Journal
of Monetary Economics 2, ed. by K. Brunner and A. Meltzer.
Gordon, R.J. (1981) 'Output fluctuations and gradual price adjustment', Journal of
Economic Literature 19,493- 530.
Gordon, R.J. (1983) 'A century of evidence on wage and price stickiness in the United
States, the United Kingdom, and Japan', in J. Tobin (ed.) Macroeconomics,
Prices, and Quantities; Essays in Memory of Arthur M. Okun, Oxford: Blackwell.
Granger, C. W .J. ( 1969) 'Investing causal relations by econometric models and cross-
spectral methods', Econometrica 37, 424 - 38. Reprinted in R.E. Lucas, Jr., and
T.E. Sargent (eds.) (1981) pp. 371 - 86.
Granger, C.W.J., and P. Newbold (1977) Forecasting Economic Time Series, New York:
Academic Press.
Gray, J.A. (1976) 'Wage indexation: a macroeconomic approach', Journal of Monetary
Economics 2.
Greenwald, B., and J.E. Stiglitz (1987) 'Keynesian, New Keynesian and New Classical
Economics', Oxford Economic Papers 39, 119 - 33.
Grossman, H. I. (1980) 'Rational expectations, business eyeless, and government behavior',
in S. Fischer (ed.) (1980) pp. 5 - 22.
Grossman, H.I. (1983) 'The natural-rate hypothesis, the rational-expectations hypothesis,
and the remarkable survival of non-market-clearing assumptions', Carnegie-
Rochester Conference Series 19,225- 46.
Grossman, S.J. (1976) 'On the efficiency of competitive stock markets where trades have
diverse information', Journal of Finance 31, 573 - 85.
Grossman, S.J. (1977) 'The existence of future markets, noisy rational expectations and
informational externalities', Review of Economic Studies 44,431- 49.
Grossman, S.J. (1979) 'Rational expectations and the allocation of resources under
asymmetric information: a survey', unpublished, University of Pennsylvania.
Grossman, S.J. (1981) 'An introduction to the theory of rational expectations under
asymmetric information', Review of Economic Studies 48,541- 59.
Gurley, J., and E. Shaw (1960) Money in a Theory of Finance, Washington DC: Brookings
Institute.
Haberler, G. (1937) Prosperity and Depression, New York: United Nations, Lake Success,
1946.
Haberler, G. (1976) 'The world economy, money, and the great depression 1919-1939',
Washington, D.C.: American Enterprise Institute for Public Policy Research.
Haberler, G. (1980) 'Critical notes on rational expectations', Journal of Money, Credit and
Banking 12, 833 - 36.
Hacking, I. (1975) The Emergence of Probability, Cambridge: Cambridge University Press.
Hacking, I. (1989) 'Probability',in J. Eatwell et al. (eds.) (1989d) pp. 163-77.
Hacking, I. (1990) The Taming of Chance, Cambridge: Cambridge University Press.
Hahn, F.H. (1965) 'On some problems of proving the existence of equilibrium in a mone
tary equilibrium', in F. Hahn and F.B.R. Brechling (eds.) (1965) pp. 126- 35.
Hahn, F.H. (1980a) 'Monetarism and economic theory',Economica 47 (185), 1- 17.
Hahn, F.H. (1980b) 'Unemployment from a theoretical viewpoint',Economica 47,285-98.
Hahn, F. H., and F.P.R. Brechling (eds.) (1965) The Theory of Interest Rates, London:
MacMillan.
Hall, R.E. (1975) 'The rigidity of wages and the persistence of unemployment', Brookings
Papers on Economic Activity 2, 301 - 35.
Haltiwanger, J.C., and M. Waldman (1989) 'Rational expectations in the aggregate',
Economic Inquiry 27,619- 36.
Hamminga, B. (1991) 'Comment on Hands', in N. de Marchi and M. Blaug (eds.) pp. 76 -
84.
Hands, D.W. (1991) 'The problem of excess content: economics, novelty and a long
Popperian tale', inN. de Marchi and M. Blaug (eds.) pp. 58- 75.
Hansen, A.H. (1927) Business-Cycle Theory, Boston: Ginn. Reprinted by Westport: Hype
277
rion Press, 1979.
Hansen, A.H. (1932) 'Book review of F.A. von Hayek's Prices and Production', American
Economic Review 22, 332- 35.
Harding, S.G. (ed.) (1976) Can Theories Be Refuted?, Dordrecht: D. Reidel.
Hausman, D.M. (ed.) (1988) The Philosophy of Economics: an anthology, Cambridge:
Cambridge University Press.
Hawtrey, R.G. (1913) Good and Bad Trade, New York: Kelley, 1970.
Hawtrey, R.G. (1932) 'Book review of F.A. von Hayek's Prices and Production, Economica,
119- 25.
Hawtrey, R.G. (1940) 'The trade cycle and capital intensity', Economica VII, 1 - 22.
Hayek, F.A. von (1928) 'lntertemporal price equilibrium and movements in the value of
money', in F.A. von Hayek (1984) ed. by R. McCloughry.
Hayek, F.A. von (1929) Geldtheorie und Konjunkturtheorie, Wien: Holder-Pichler-Tempski
(reprint Salzburg: Wolfgang Neugebauer, 1976); English translation Monetary
Theory and the Trade Cycle (1933a}, Clifton, N.J., 1975.
Hayek, F.A. von (1931a) Prices and Production, London: Routledge & Kegan Paul, 1935,
2nd. ed.; German translation Preise und Produktion (1931b), Wien: Julius
Springer.
Hayek, F.A. von (1932) 'Money and capital: a reply' ,Economic Joumal42, 237- 49.
Hayek, F.A. von (1933b) 'The trend of economic thinking', Economica No. 40, 121- 37.
Hayek, F. A. von ( 1933c) 'Price expectations, monetary disturbances and malinvestments',
in F.A. von Hayek (1939) pp. 135- 56.
Hayek, F.A. von (1933d) 'The present state and immediate prospects of the study of
industrial fluctuations', in F.A. von Hayek (1939) pp. 171- 82.
Hayek, F.A. von (1933e) 'Uber "neutrales Geld"',Zeitschrijtjar Nationaliikonomie 4, 659-
61. Translated and reprinted as 'On "neutral money"',in F.A. von Hayek (1984).
Hayek, F.A. von (1936a) 'Technical progress and excess capacity', in F.A. von Hayek
(1984). First published as 'Technischer Fortschritt und Uberkapazitiit', Oster-
reichische Zeitschrift for Bankwesen , 1, 9 - 23.
Hayek, F.A. von (1936b) 'Utility Analysis and Interest', Economic Journal46, 44- 60.
Hayek, F.A. von (1937) 'Economics and knowledge', Economica 4, 33 - 54. Reprinted in
F.A. von Hayek (1949) pp. 33- 56.
Hayek, F.A. von (1939) Profits, Interest and lnvestmellt, London: Routledge & Kegan Paul,
1950.
Hayek, F.A. von (1941) The Pure Theory of Capital, London: Routledge & Kegan Paul,
1950.
Hayek, F.A. von (1942a) 'The Ricardo effect', Economica IX, 127- 52. Reprinted in F.A.
von Hayek (1949).
Hayek, F.A. von (1942b) 'A comment', Economica IX, 383- 85.
Hayek, F.A. von (1942c) 'Scientism and the study of society', Part I (sections i - iv),
Economica IX, 267- 91. Reprinted in F.A. von Hayek (1955a).
Hayek, F.A. von (1943a) 'Scientism and the study of society', Part II (sections v - x),
Economica X, 34- 63. Reprinted in F.A. von Hayek (1955a).
Hayek, F.A. von (1943b) 'The facts of the social sciences', Ethics LIV, 1 - 13. Reprinted in
F.A. von Hayek (1949) pp. 57- 76.
Hayek, F.A. von (1944a) The Road to Serfdom, London: Routledge & Sons, 1946.
Hayek, F.A. von (1944b) 'Scientism and the study of society', Part III (sections xi- xiv),
Economica XI, 27- 39. Reprinted in F.A. von Hayek (1955a).
Hayek, F.A. von (1945a) 'Time-preference and productivity: a reconsideration', Economica
12,22- 25.
Hayek, F.A. von (1945b) 'The use of knowledge in society' ,American Economic Review 35,
519- 30. Reprinted in F.A. von Hayek (1949) pp. 77- 91.
Hayek, F.A. von (1946a) 'Individualism: True and false', twelfth Finlay Lecture, Dublin:
Hodges, Figgis & Co., Oxford: Blackwell. Reprinted in F.A. von Hayek (1949)
pp. 1- 32.
278
Hayek, F.A. von (1946b) 'The meaning of competition', in F.A. von Hayek (1949) pp. 92 -
106.
Hayek, F.A. von (1947) "'Free" competition and competitive order', in F.A. von Hayek
(1949) pp. 107 - 18.
Hayek, F.A. von (1949) Individualism and Economic Order, London: Routledge & Kegan
Paul, 1976.
Hayek, F .A. von (1952) The Sensory Order, London: Routledge.
Hayek, F.A. von (1955a) The Counter-Revolution of Science, London: Free Press of
Glencoe (Collier - MacMillan), 1964.
Hayek, F.A. von (1955b) 'Degrees of explanation', in F.A. von Hayek (1967a) pp. 3- 21.
Hayek, F.A. von (1961) The Constitution of Liberty, London: Routledge & Kegan Paul.
Hayek, F.A. von (1964) 'The theory of complex phenomena', in F.A. von Hayek (1967a)
pp. 22-42.
Hayek, F.A. von (1965a) 'Kinds of rationalism', in F.A. von Hayek (1967a) pp. 82- 95.
Hayek, F.A. von (1965b) 'Wiener Schule', Handworterbuch der Sozialwissenschaften 12,68-
71.
Hayek, F.A. von (1967a) Studies in Philosophy, Politics and Economics, Chicago: Universi
ty of Chicago Press.
Hayek, F.A. von (1967b) 'The results of human action but not of human design', in F.A.
von Hayek (1967a) pp. 96- 105.
Hayek, F.A. von (1968) 'Der Wettbewerh als Entdeckungsverfahren', in F.A. von Hayek
(1969a), pp. 249 - 65. Translated as 'Competition as a discovery procedure', in
F.A. von Hayek (1978), pp. 179- 90.
Hayek, F.A. von (1969a) Freiburger Studien, Walter Eucken Institut, Tiibingen: J.C.B.
Mohr (Paul Siebeck.
Hayek, F.A. von (1969b) 'Three Elucidations of the Ricardo Effect', Journal of Political
Economy 77,274- 85. Reprinted in F.A. von Hayek (1978) pp. 165- 78.
Hayek, F .A. von (1972) A Tiger by the Tail, London: Institute of Economic Affairs.
Hayek, F.A. von (1974) 'The pretence of knowledge', in F.A. von Hayek (1978) pp. 23- 34.
Hayek, F.A. von (1976) Denationalization of Money, London: Institute of Economic Affairs.
Hayek, F.A. von (1978) New Studies in Philosophy, Politics, Economics and the History of
Ideas, London: Routledge.
Hayek, F.A. von (1979) Law, Legislation and Liberty, London: Routledge & Kegan Paul.
Hayek, F.A. von (1984) Money, Capital & Fluctuations, ed. by Roy McCloughry, London:
Routledge.
Hayek, F.A. von (ed.) (1935) Collectivist Economic Planning, London: Routledge & Sons.
Hayek, F.A. von (ed.) (1973) Toward Liberty: Essays in honor of Ludwig von Mises, Menlo
Park: Institute for Humane Studies.
Heiner, R.A. (1985-86) 'Rational expectations when agents imperfectly use information',
Journal of Post-Keynesian Economics 8, 201- 7.
Hempel, C.G. (1965) Aspects of Scientific Explanation, New York: Free Press, 1970.
Hicks, J.R. (1932) The Theory of Wages, London: MacMillan, 1966.
Hicks, J.R. (1935) 'Wages and interest', Economic Journal45, 456- 68.
Hicks, J.R. (1937) 'Mr. Keynes and the Classics: a suggested interpretation', Econometrica
5, 147- 59.
Hicks, J.R. (1939) Value and Capital, Oxford: Clarendon Press, 1941.
Hicks, J.R. (1950) A Colllribution th the Theory of the Trade Cycle, Oxford: Clarendon
Press.
Hicks, J.R. (1965) Capital and Growth, Oxford: Oxford University Press.
Hicks, J.R. (1967) 'The Hayek Story', in J.R. Hicks (1967) Critical Essays in Monetary
Theory, Oxford: Clarendon Press, pp. 203- 15.
Hicks, J.R. (1973) Capital and Time, a Neo-Austrian Theory, Oxford: Oxford University
Press.
Hicks, J.R., and W. Weber (eds.) (1974) Carl Menger and the Austrian School of Econo
mics, Oxford.
279
High, J. (1982) 'Alertness and judgment: comment on Kirmer', in I.M. Kirmer (ed.)
(1982) pp. 161 - 68.
High, J. (1986) 'Equilibration and disequilibration in the market process', in I.M. Kirmer
(ed.) (1986) pp. 111- 21.
Hirsch, A. (1986) 'Caldwell on praxeology and its critics: a reappraisal', History of Political
Economy 18, 661 - 68.
Hirsch, A., and N. de Marchi (1990) Milton Friedman, Hemel Hempstead: Wheatsheaf.
Homer, The Odyssey, London: Penguin Books, 1991.
Hoover, K.D. (1984a) 'Two Types of Monetarism', Journal of Economic Literature 22, 58 -
76.
Hoover, K.D. (1984b) 'Methodology: A Comment on Frazer and Boland, II', American
Hoover, K.D. (1988) The New Classical Macroeconomics, Oxford: Blackwell.
Hoover, K.D. (1991) 'Scientific research program or tribe? A joint appraisal of Lakatos
and the New Classical Macroeconomics', inN. de Marchi and M. Blaug (eds.)
pp. 364-94.
Howitt, P. (1981) 'Activist monetary policy under rational expectations', Journal of Political
Economy 89, 249 - 69.
Hume, D. (1748) An Inquiry Concerning Human Understanding, Indianapolis, 1977.
Humphrey, Th.M. (1991) 'Nonneutrality of money in classical monetary thought', Econ
omic Review Federal Reserve Bank of Richmond, 77, no. 2.
Hurwicz, L. (1973) 'The design of mechanisms for resource allocation', American Econ
omic Review, 63, 1 - 30.
Hutchison, T. W. (1937) 'Expectation and rational conduct', Zeitschrift fur Nationaloko
nomie 8, 636 - 53.
Hutchison, T.W. (1938) The Significance and Basic Postulates of Economic Theory, New
York: Kelley, 1965.
Hutchison, T.W. (1977) Knowledge and Ignorance in Economics, Chicago: University of
Chicago Press.
Hutchison, T.W. (1981) The Politics and Philosophy of Economics, New York: New York
University Press.
Huussen, G.M. (1985) Economisch Evenwicht, Amsterdam: VU-uitgeverij, dissertation.
Huussen, G.M. (1988) De Keerzijde van de Munt, Amsterdam: VU-uitgeverij.
Huussen, G.M. (1989) 'Mises and the praxeological point of view', Journal of Economic
Studies 15, 121 - 33.
Huussen, G.M. (1990) 'Maakte Hayek een methodologische ommezwaai?', Maandschrift
Economie 54, no. 2, 115 - 27.
Johre, W.A. (1977) 'Note on Professor Hayek's 'True theory of unemployment', Kyklos 30,
713- 23.
Johnson, H. G. (1971) 'The Keynesian revolution and the Monetarist counter-revolution',
American Economic Review 61, 1 - 14. Reprinted in H.G. Johnson (1972) Further
Essays in Monetary Economics, London: Allen & Unwin, chapter 2, pp. 50- 69.
Johnston, W.M. (1972) The Austrian Mind, Berkeley: University of California Press.
Jordan, J.S., and R. Radner (1982) 'Rational expectations in microeconomic models: an
overview' ,Journal of Economic Theory 26,201- 23.
Kaldor, N. (1934) 'A classificatory note on the determinateness of equilibrium', Review of
Economic Studies 1, 122 - 36.
Kaldor, N. (1937) 'Annual survey of economic theory: the recent controversy on the theory
of capital', Econometrica, July.
Kaldor, N. (1938) 'On the theory of capital: a rejoinder to professor Knight', Econometrica,
April.
Kaldor, N. (1939) 'Capital intensity and the trade cycle' ,Economica VI, 40- 66.
Kaldor, N. (1942) 'Professor Hayek and the concertina effect', Economica IX, 359- 82.
Kamath, S.J. (1989) 'A rational reconstruction of the rational expectations revolution', in
D.A. Walker (ed.) pp. 217- 44.
280
Kantor, B. (1979) 'Rational Expectations and Economic Thought', Journal of Economic
Literature 17, 1422- 41.
Katouzian, H. (1980) Ideology and Method in Economics, London: MacMillan; chapter 7:
'Economic theory and mathematical economics: abstraction and generalisation in
economic science', pp. 157- 83.
Katmer, D.W. (1991) 'In defense of formalization in economics', Methodus 3 (1), 17- 24.
Kauder, E. (1965) A History of Marginal Utility Theory, Princeton.
Kauder, E. (1957) 'Intellectual and Political Roots of the Older Austrian School', Zeit
schriftfor Nationalokonomie XVII, 411- 25.
Keizer, W. (1986) 'The concept of 'consumer' sovereignty',Research Memorandum 1986-28,
Amsterdam: Free University.
Keynes, J.M. (1923) A Tract on Monetary Reform, London: MacMillan, 1971.
Keynes, J.M. (1930) A Treatise on Money, London: MacMillan, 1965.
Keynes, J.M. (1936) The General Theory of Employment, Interest and Money, The Collected
Writings of John Maynard Keynes, vol. VII, London: MacMillan, 1983.
Keynes, J.M. (1937) 'The "ex-ante" theory of the rate of interest', Economic Journal41, 663
-69.
Keynes, J.M. (1938) 'Letter to R.F. Harrod, July 4, 1938', in Collected Writings v. 14, D.
Moggridge (ed.), London: MacMillan, 1973.
Kim, J., N. de Marchi, and M. Morgan (1991) 'Batch characteristics and belief in the
natural rate hypothesis', Paper presented at the conference on 'The Significance
of Testing in Econometrics', Tilburg, December 15- 17, 1991.
Kim, K. (1990) Equilibrium Business Cycle Theory in Historical Perspective, Cambridge:
King, R.G. (1978) 'Asset markets and the neutrality of money: an economy-wide bond
market', unpublished, University of Rochester.
King, R.G. (1981) 'Monetary information and monetary neutrality', Journal of Monetary
Economics 1, 195- 206.
King, R.G. (1982) 'Monetary policy and the information content of prices', Journal of
Political Economy 90, 247- 79.
King, R.G., and C.l. Plosser (1984) 'Money, credit and prices in a real business cycle',
American Economic Review 14, 363 - 80.
King, R.G., and C.l. Plosser (1986) 'Money as the mechanism of exchange', Journal of
Monetary Economics 17,93- 115.
Kirmer, I.M. (1960) 1he Economic Point of View, Kansas: Sheed & Ward, 1976.
Kirmer, I.M. (1962) 'Rational action and economic theory', Journal of Political Economy
70,380- 85.
Kirmer, I.M. (1963) Market Theory and the Price System, Princeston: D. van Nostrand.
Kirmer, I.M. (1966) An Essay on Capital, New York.
Kirmer, I.M. (1967) 'Methodological individualism, market equilibrium, and market
process', II Politico 32, 787- 99.
Kirmer, I.M. (1969) 'Entrepreneurship and the market approach to development', in I.M.
Kirmer (1979), pp. 107 - 19.
Kirmer, I.M. (1973) Competition and Eturepreneurship, Chicago: University of Chicago
Press.
Kirmer, I.M. (1975) 'Hayek, knowledge, and market process', in I.M. Kirmer (1979), pp.
13- 33.
Kirmer, I.M. (1976a) 'Equilibrium versus market process', in E.G. Dolan (ed.) (1976) pp.
115 - 25. Reprinted in I.M. Kirmer (1979) pp. 3 - 12.
Kirmer, I.M. (1976b) 'Knowing about knowledge: a subjective view of the role of informa
tion', in I.M. Kirmer (1979), pp. 137 - 53.
Kirmer, I.M. (1978a) 'Economics and error', in L.M. Spadaro (ed.) (1978) pp. 57 - 76.
Reprinted in I.M. Kirmer (1979) pp. 120- 36.
Kirmer, I.M. (1978b) 'Alertness, luck, and entrepreneurial profit', in I.M. Kirmer (1979),
pp. 154- 81.
281
Kirmer, I.M. (1979) Perception, Opportunity and Profit. Studies in the Theory of Entrepre
neurship, Chicago: University of Chicago Press.
Kirmer, I.M. (1984) 'Prices, the communication of knowledge, and the discovery process',
in K.R. Leube and A.H. Zlabinger (ed.) (1984) pp. 193 - 206.
Kirmer, I.M. (1988) 'The economic calculation debate: lessons for Austrians', Review of
Austrian Economics 2, 1 - 18.
Kirmer, I.M. (1989) Discovery, Capitalism, and Distributive Justice, Oxford: Blackwell.
Kirmer, I.M. (ed.) (1982) Method, Process, and Austrian Economics, Lexington: Lexington
Books.
Kirmer, I.M. (ed.) (1986) Subjectivism, Intelligibilityand Economic Understanding, London:
MacMillan.
Klamer, A. (1984) 'Levels of discourse in new classical economics', History of Political
Economy 16, 263 - 90.
Klamer, A. (1985) The New Classical Macroeconomics, Brighton (Sussex): Wheatsheaf.
Klant, J .J. ( 1978) Spelregels voor Economen , Leiden: Stenfert Kroese.
Klant, J.J. (1991) 'Plausibility' ,Methodus 3 (1), 64- 66.
Klausinger, H. (1989a) 'Hayek and New Classical Economics on equilibrium analysis',
Jahrbuchfor Sozialwissenschaft 40, 171- 86.
Klausinger, H. (1989b) 'Equilibrium methodology as seen from a Hayekian perspective',
paper presented at the 'Troisieme Colloque de I' Association Charles Gide pour
l'Etude de Ia Pensee Economique', Strasbourg, September 21 - 22.
Klausinger, H. (1989c) 'On the history of neutral money', in D.A. Walker (ed.) pp. 171 -
86.
Klausinger, H. (1991) Theorien der Geldwirtschaft, Berlin: Duncker & Humblot.
Klein, L.R. (1954) 'The contributions of mathematics in economics', The Review of
Economics and Statistics 36, 359- 61.
Knight, F. (1937) Risk, Uncertainty and Profit, London.
Knight, F. (1938) 'On the theory of capital: in reply to Mr. Kaldor', Econometrica, January.
Knorr-Cetina, K. (1981) 'The micro-sociological challenge of macro-sociology: towards a
reconstruction of social theory and methodology', in K. Knorr-Cetina and A. V.
Cicourel (eds.) Advances in Social Theory and Methodology, Boston, pp. 1 - 47.
Koertge, N. (1975) 'Popper's metaphysical research program for the human sciences',
Inquiry 18,437- 62.
Koningsveld, H. (1987) Het Verschijnsel Wetenschap, Meppel/Amsterdam: Boom.
Koopmans, T. (1954) 'On the use of mathematics in economics', The Review of Economics
and Statistics 36, 377 - 80.
Koyck, L.M. (1954) Distributed Lags and Investment Analysis, Amsterdam: North-Holland.
Krabbe, J., et al. (1989) Austrian Economics: Roots and Ramifications Reconsidered,
Bradford: MCB University Press.
Kromphardt, J. (1989) 'Konjunkturtheorie heute: ein Oberblick', Zeitschrift for Wirtschafts-
und Sozialwissenschaften 109, 173- 231.
Kuhn, T.S. (1962) The Structure of Scientific Revolutions, Chicago: University of Chicago
Press, 1970.
Kuhn, T.S. (1970a) 'Logic of discovery or psychology of research?', in I. Lakatos and A.
Musgrave (eds.) (1970) pp. 1 - 23.
Kuhn, T.S. (1970b) 'Reflections on my critics', in I. Lakatos and A. Musgrave (eds.) (1970)
pp. 231- 78.
Kydland, F.E., and E.C. Prescott (1977) 'Rules rather than discretion: the inconsistency of
optimal plans', Journal of Political Economy 85, 473 - 91.
Kydland, F.E., and E.C. Prescott (1980) 'A competitive theory of fluctuations and the
feasibility and desirability of stabilization policy' ,inS. Fischer (ed.) (1980).
Kydland, F.E., and E. C. Prescott (1982) 'Time to build and aggregate fluctuations',
Econometrica NS 10, 1345- 69.
Lachmann, L.M. (1937) 'Uncertainty and liquidity preference', Economica 4, 295- 308.
Lachmann, L.M. (1939) 'On crisis and adjustment', Review of Economics and Statistics 21,
282
62-68.
Lachmann, L.M. (1940) 'A reconsideration of the Austrian theory of industrial fluctua
tions',Economica 1, 179-96. Reprinted in L.M. Lachmann (1977), pp. 267- 86.
Lachmann, L.M. (1943) 'The role of expectations in economics as a social science',
Economica X, 12- 23.
Lachmann, L.M. (1945) 'A note on the elasticity of expectations', Economica XII, 248- 53.
Lachmann, L.M. (1951) 'The science of human action', Economica 18,412- 27.
Lachmann, L.M. (1956) Capital and its Structure, Kansas City, 1978.
Lachmann, L.M. (1969) 'Methodological individualism and the market economy' in E.
Streissler, et al. (eds.) (1969) pp. 89- 105.
Lachmann, L.M. (1970) The Legacy of Max Weber, London: Heinemann.
Lachmann, L.M. (1973a) 'Ludwig von Mises and the Market Process', in F.A. von Hayek
(ed.) (1973) pp. 38 - 52.
Lachmann, L.M. (1973b) 'Sir John Hicks as a Neo-Austrian' (Review Article), The South
African Journal of Economics 41, 195- 207.
Lachmann, L.M. (1975) Makrookonomischer Formalismus und die Marktwirtschaft,
Tiibingen.
Lachmann, L.M. (1976) 'From Mises to Shackle: An essay on Austrian Economics and the
Kaleidic society' ,Journal of Economic Literature 14, 54- 62.
Lachmann, L.M. (1977) Capital, Expectations, and the Market Process, ed. by W. Grinder,
Kansas City: Sheed, Andrews & McMeel.
Lachmann, L.M. (1978) 'Interview with Ludwig M. Lachmann', Austrian Economics
Newsletter 1, nr. 3.
Lachmann, L.M. (1982a) 'Ludwig von Mises and the extension of subjectivism', in I.M.
Kirmer (ed.) (1982) pp. 31 - 40.
Lachmann, L.M. (1982b) 'The salvage of ideas, problems of the revival of Austrian
economic thought', Journal of Institutional and Theoretical Economics 138, 629 -
45.
Lachmann, L.M.,and F. Snapper (1938) 'Commodity stocks in the trade cycle',Economica
5, 435-54.
Laidler, D.E.W. (1982) Monetarist Perspectives, Oxford: Philip Allan.
Lakatos, I. (1970) 'Falsification and the methodology of scientific research programmes', in
I. Lakatos and A. Musgrave (eds.) (1970) pp. 91 - 196.
Lakatos, I. (1971) 'History of science and its rational reconstructions', in I. Lakatos (1978)
pp. 102- 38.
Lakatos, I. (1978a) Philosophical Papers, vol. I: The Methodology of Scientific Research
Programmes, Cambridge: Cambridge University Press, 1986.
Lakatos, I. (1978b) Philosophical Papers, vol. II: Mathematics, Science and Epistemology,
Cambridge: Cambridge University Press.
Lakatos, I., and A. Musgrave (eds.) (1970) Criticism and the Growth of Knowledge,
Cambridge: Cambridge University Press.
Langlois, R.N. (1982) 'Austrian economics as affirmative science: comment on Rizzo', in
I.M. Kirmer (ed.) (1982) pp. 75- 84.
Latsis, S.J. (1976) 'A research programme in economics', in S.J. Latsis (ed.) (1976) pp. 1 -
41.
Latsis, S.J. (ed.) (1976) Method and Appraisal in Economics, Cambridge: Cambridge
University Press.
Lavoie, D. (1981) 'A critique of the standard account of the socialist calculation debate',
Journal of Libertarian Studies 5, No. 1, 41- 87.
Lavoie, D. (1985) Rivalry and Central Planning, The socialist calculation debate reconside
red, Cambridge: Cambridge University Press.
Lavoie, D. (1991) 'The progress of subjectivism', inN. de Marchi and M. Blaug (eds.) pp.
470- 86.
Leamer, E.E. (1983) 'Let's take the con out of econometrics', American Economic Review
73, 31 - 43.
283
Leamer, E. E. (1991) 'Comment: has formalization gone too far?', Methodus 3 (1), 25 - 26.
Lerner, A.P. (1972) 'The economics and politics of consumer sovereignty', American
Economic Review 62-II, 258 - 78.
Leijonhufvud, A. (1968) On Keynesian Economics and the Economics of Keynes Oxford:
Oxford University Press.
Leijonhufvud, A. (1968) 'Comment: Is there a meaningful trade-off between inflation and
unemployment?', Journal of Political Economy 16, 738- 43.
Leijonhufvud, A. (1981) Information and Coordination, Oxford: Oxford University Press.
Leijonhufvud, A. (1984) 'What would Keynes have thought about rational expectations', in
D. Worswick and J. Trevithick (eds.) (1984) Keynes and the Modern World,
Cambridge: Cambridge University Press, pp. 179- 221.
Lindahl, E. (1939) Studies in the Theory of Money and Capital, London: Allen & Unwin.
Lipsey, R.G. (1960) 'The relation between unemployment and the rate of change of money
wage rates in the United Kingdom, 1862 - 1957: a further analysis', Economica
27' 1 - 31.
Lipsey, R.G. (1963) An Introduction to Positive Economics, London: Weidenfeld and
Nicolson.
Littlechild, S.C. (1982) 'Equilibrium and the market process', in I.M. Kirzner (ed.) (1982)
pp. 85- 102.
Loasby, B.J. (1976) Choice, Complexity, and Ignorance, Cambridge.
Loasby, B.J. (1982) 'Economics of dispersed and incomplete information', in I.M. Kirzner
(ed.) (1982) pp. 111- 30.
Long, J .B., Jr., and C.I. Plosser (1983) 'Real business cycles' ,Journal of Political Economy
91,39- 69.
Lucas, R.E., Jr. (1965) 'Distributed lags and optimal investment policy', in R.E Lucas, Jr.,
and T.J. Sargent (eds.) (1981) pp. 39- 54, previously unpublished.
Lucas, R.E., Jr. (1966) 'Optimal investment with rational expectations', in R.E. Lucas, Jr.,
and T.J. Sargent (eds.) (1981) pp. 55- 66, previously unpublished.
Lucas, R.E., Jr. (1972a) 'Expectations and the neutrality of money', Journal of Political
Economy 83, 1113- 44. Reprinted in R.E. Lucas, Jr. (1981a) pp. 66- 89.
Lucas, R.E., Jr. (1972b) 'Econometric testing of the Natural Rate Hypothesis', in 0.
Echstein (ed.) The Econometrics of Price Determination Conference, Washington
D.C.: Board of Governors of the Federal Reserve System, 50- 59. Reprinted in
R.E. Lucas, Jr. (1981a) pp. 90- 103.
Lucas, R.E., Jr. (1973) 'Some international evidence on output-inflation trad.eoffs',
American Economic Review 63, 326 - 34. Reprinted in R.E. Lucas, Jr. (1981a)
pp. 131- 45.
Lucas, R.E., Jr. (1975) 'An equilibrium model of the business cycle, Journal of Political
Economy 83, 1113-44. Reprinted in R.E. Lucas, Jr. (1981a) pp. 179- 214.
Lucas, R.E., Jr. (1976) 'Econometric policy evaluation: a critique', in K. Brunner and A.H.
Meltzer (eds.) The Phillips Curve and Labor Markets, Carnegie-Rochester Confer-
ence Series on Public Policy, vol. 1, Amsterdam: North-Holland, pp. 19 - 46.
Reprinted in R.E. Lucas, Jr. (1981a) pp. 104- 30.
Lucas, R.E., Jr. (1977) 'Understanding business cycles', Journal of Monetary Economics'
supplement, 7 - 30. Reprinted in R.E. Lucas, Jr. (1981a) pp. 215 - 39.
Lucas, R.E., Jr. (1980a) 'Methods and problems in business cycle theory', Journal of
Money, Credit and Banking 12, 696 - 715. Reprinted in R.E. Lucas, Jr. (1981a)
pp. 271-96.
Lucas, R.E., Jr. (1980b) 'Equilibrium in a pure currency economy', Economic Inquiry 18,
203-20.
Lucas, R.E., Jr. (1980c) 'Rules, discretion, and the role of the economic advisor', in S.
Fischer (ed.) (1980) pp. 199- 210.
Lucas, R.E., Jr. (1981a) Studies in Business-Cycle Theory, Oxford: Blackwell.
Lucas, R.E., Jr. (1981b) 'Discussion of: Stanley Fischer, "Towards an understanding of the
costs of inflation: II", inK. Brunner and A. Meltzer (eds.) (1981) pp. 43- 52.
284
Lucas, R.E., Jr. (1987) Models of Business Cycles, Oxford: Blackwell.
Lucas, R.E., Jr., and E.C. Prescott (1971) 'Investment under uncertainty', Econometrica 39,
659- 81.
Lucas, R.E., Jr., and E.C. Prescott (1974) 'Equilibrium search and unemployment', Journal
of Economic Theory 7, 188- 209. Reprinted in R.E. Lucas, Jr. (1981a) pp. 156-
78.
Lucas, R.E., Jr., and L.E. Rapping (l969a) 'Real wages, employment, and inflation',
Journal of Political Economy 71, 721 -54. Reprinted in R.E. Lucas, Jr. (1981a)
pp. 19- 58.
Lucas, R.E., Jr., and L.E. Rapping (1969b) 'Price expectations and the Phillips curve',
American Economic Review 59,342- 50.
Lucas, R.E., Jr., and L.E. Rapping (1972) 'Unemployment in the Great Depression: is
there a full explanation?', Journal of Political Economy 80, 186- 91. Reprinted in
R.E. Lucas, Jr. (l981a) 59 - 65.
Lucas, R.E., Jr., and T.J. Sargent (1978) 'After Keynesian macroeconomics', Federal
Reserve Bank of Boston, After the Phillips Curve: Persistence of High Inflation
and High Unemployment. Reprinted in R.E. Lucas, Jr., and T.J. Sargent (eds.)
(1981) pp. 295- 319.
Lucas, R.E., Jr., and T.J. Sargent (eds.) (1981) Rational Expectations and Econometric
Practice, London: Allen & Unwin.
Lukes, S. (1968) 'Methodological individualism reconsidered', in A. Ryan (ed.) (1973) The
Philosophy of Social Explanation, Oxford: Oxford University Press.
Lundberg, E. (1937) Studies in the Theory of Economic Expansion, Stockholm and London:
Stockholm Economic Series.
Lutz, F.A.,and V. Lutz (1951) 1he Theory of Investment of the Firm, Chapter XI, 'Capital
Intensity and the Trade Cycle', Princeton: Princeton University Press, pp. 137 -
42.
Machlup, F. (1963) Essays on Economic Semantics, Englewood Cliffs, N.J.: Prentice Hall.
Machlup, F. (1978) Methodology of Economics and Other Social Sciences, New York.
Machlup, F. (1983) 'The rationality of "rational expectations"', Kredit und Kapital16, 172-
83.
Machlup, F. (ed.) (1977) Essays on Hayek, London.
Mackay, D.I., and R.A. Hart (1974) 'Wage inflation and the Phillips Relationship',
Manchester School 42.
Maddock, R. (1979) Rational Expectations, Political Business Cycles and the Course of
Macroeconomic Theory, Ph.D.-dissertation, Duke University, Durham (N.C.).
Maddock, R. (1984) 'Rational expectations macrotheory: a Lakatosian case study in
program adjustment', History of Political Economy 16,291- 309.
Maddock, R. (1991) 'The development of New Classical Macroeconomics: Lessons for
Lakatos', inN. de Marchi and M. Blaug (eds.) pp. 335- 59.
Maddock, R., and M. Carter (1982) 'A child's guide to rational expectations', Journal of
Economic Literature 20, 39 - 51.
Malinvaud, E. (1977) The Theory of Unemployment Reconsidered, Oxford: Blackwell.
Mandeville, B. (1714) The Fable of the Bees: or, Private Vices, Publick Benefits, London:
Penguin Classics, 1989.
Mandeville, B. (1723) 'A search into the nature of society', essay added to B. Mandeville
(1714), London: Penguin Classics, 1989.
de Marchi, N. (1976) 'Anomaly and the development of economics: the case of the
Leontief paradox', in S.J. Latsis (ed.) (1976) pp. 109- 27.
de Marchi, N. (1990) '"Discovery"and the unimportance of Lucas' test of the natural rate
Hypothesis', unpublished paper, draft version October 30, 1990.
de Marchi, N. (1991) 'Introduction: rethinking Lakatos', in N. de Marchi and M. Blaug
(eds.) pp. 1 - 30.
de Marchi, N. (ed.) (1988) The Popperian Legacy in Economics, Cambridge: Cambridge
University Press.
285
de Marchi, N., and C. Gilbert (1989) 'Introduction', Oxford Economic Papers 41, 1 - 11.
de Marchi, N., and M. Blaug (eds.) (1991) Appraising Economic Theories: Studies in the
Methodology of Research Programs, Aldershot: Elgar.
Manne, C., and K.A. McDermott (1989) 'The quantity theory of money of J.M. Keynes:
from Indian Currency to the General Theory', in D.A. Walker (ed.) pp. 148 - 68.
Marshall, A. (1890) Principles of Economics , London: MacMillan, 1947.
Marshall, A. (1923) Money, Credit and Commerce, London: MacMillan.
Mayer, H. (1932) 'Der Erkenntniswert der funktionellen Preistheorien', in H. Mayer (ed.)
Die Wirtschaftstheorieder Gegenwart, vol. 2, Wien, pp. 147- 239.
McCallum, B. T. (1977) 'Price-level stickiness and the feasibility of monetary stabilization
policy with rational expectations', Journal of Political Economy 85 (1977), 627 -
34. Reprinted in Lucas and Sargent (eds.) (1981) pp. 277- 84.
McCallum, B.T. (1978) 'Price level adjustments and the rational expectations approach to
macroeconomic stabilization policy', Journal of Money, Credit and Banking 10,
418 -36.
McCallum, B.T. (1979) 'The current state of the policy-ineffectiveness debate', American
Economic Review 69, 240- 45. Reprinted in Lucas and Sargent (eds.) (1981) pp.
285- 92.
McCallum, B.T. (1980a) 'Introduction', Journal of Money, Creditand Banking 12,691-95.
McCallum, B.T. (1980b) 'Rational expectations and macroeconomic stabilisation policy',
Journal of Money, Credit and Banking 12,716 -46.
McCallum, B.T. (1989a) Monetary Economics; Theory and Practice, New York: MacMillan.
McCallum, B.T. (1989b) 'Real business cycles models', in R.J. Barro (ed.) (1989) pp. 16-
50.
McCallum, B.T. (1989c) 'New Classical Macroeconomics: a sympathetic account', Scandi
navian Journal of Economics 91,223- 52.
McCallum, B.T., and J.K. Whitaker (1979) 'The effectiveness of fiscal feedback rules and
automatic stabilizers under rational expectations', Journal of Monetary Economics
5, 171- 86.
McCloskey, D.N. (1985) The Rhetoric of Economics , Brighton (Sussex): Wheatsheaf.
McCloskey, D.N. (1991) 'Economic science: a search through the hyperspace of assumpti
ons?', Methodus 3 (1), 6 - 16.
McCormick, B. (1984) 'Comment 1' (on Cross (1984a)), in P. Wiles and G. Routh (eds.)
(1984) pp. 100- 4.
Menger, C. (1871) Grundsiitze der Volkswirtschaftslehre, Gesammelte Werke, Band I,
Tiibingen: J.C.B. Mohr (Paul Siebeck), 1968.
Menger, C. (1883) Untersuchungen uber die Methode der Sozialwissenschaften und der
politische Okonomie insbesondere, Gesammelte Werke, Band II, Tiibingen: J.C.B.
Mohr (Paul Siebeck), 1969.
Menger, C. (1888) Zur Theorie des Kapitals, Gesammelte Werke, Band III, Tiibingen:
J.C.B. Mohr (Paul Siebeck), 1970.
Menger, C. (1889) Grundzuge einer Klassifikation der Wirtschaftswissenschaften, Gesam
melte Werke, Band III, Tiibingen: J.C.B. Mohr (Paul Siebeck), 1970.
Milford, K. (1988) 'Menger's solution of the problem of induction: on the history of
methodological thought in economics', Working Paper No. 8806, University of
Vienna, Dep. of Economics.
Milgate, M. (1979) 'On the origin of the notion of "intertemporal equilibrium"' ,Economica
46, 1 - 10.
Mill, J.S. (1843) A System of Logic, Ratiocinative and Inductive, 2 vols., London: John W.
Parker, 1856. Reprint of Book VI, 'The Logic of the Moral Sciences', London:
Gerald Duckworth & Co., 1987.
Minford, P. (1986) 'Rational expectations and monetary policy' ,Scottish Journal of Political
Economy 33,317- 33.
Minford, P., and D. Peel (1981) 'The role of monetary stabilization policy under rational
expectations', The Manchester School 49, 39 -50.
286
Mirowski, P. (1989) 'The probabilistic counter-revolution, or how stochastic concepts came
to neoclassical economic theory', Oxford Economic Papers 41,217- 35.
Mises, L. von (1920) 'Die Wirtschaftsrechnung im sozialistischen Gemeinwesen', Archiv Jar
Sozialwissenschaften und Sozialpolitik 41, No. 1. Translated (by S. Adler) as
'Economic calculation in the socialist commonwealth', in F.A. von Hayek (ed.)
(1933) Collectivist Economic Planning, Clifton: Kelley, 1975.
Mises, L. von (1912) Theorie des Geldes und der Umlaufsmittel, Miinchen: Verlag Von
Duncker & Humblot, 1924.
Mises, L. von (1928) Geldwertstabilisierungund Konjunkturpolitik, Jena.
Mises, L. von (1929) Kritik des lnterventionismus; Untersuchungen zur Wirtschaftspolitik und
Wirtschaftsideologieder Gegenwart, Jena.
Mises, L. von (1933) Grundprobleme der Nationalokonomie, Jena.
Mises, L. von (1943) '"Elastic Expectations" and the Austrian Theory of the Trade Cycle',
Economica X, 251- 52.
Mises, L. von (1949) Human Action: A Treatise on Economics, Yale University Press.
Reprinted by Chicago: Contemporary Books, Inc., 1966.
Mises, L. von (1953) 'Bemerkungen iiber die mathematische Behandlung nationalokonomi
scher Probleme', Studium Generale VI, No. 2, Berlin-Gottingen-Heidelberg:
Springer Verlag. Translated as 'Comments about the mathematical treatment of
economic problems', Journal of Libertarian Studies 1 (1977}, 97- 100.
Mises, L. von (1960) Epistemological Problems of Economics, New York: New York
University Press.
Mises, L. von (1962) The Ultimate Foundation of Economic Science, Kansas City, 1978.
Mishkin, F.S. (1983) 'The remarkable survival of non-market-clearing assumptions:
comment on the Grossman paper', Carnegie-Rochester Conference Series 19, 247 -
52.
Modigliani, F. (1977) 'The monetary controversy, or should we forsake stabilisation
policies',American Economic Review 61,1- 19.
Montiel, P.J. (1987) 'Output and unanticipated money in the dependent economy model',
l.M.F. Staff Papers 34, no. 2.
Morgan, B. (1978) Monetarists and Keynesians, London: MacMillan.
Morgan, M. (1988) 'Finding a satisfactory empirical model', in N. de Marchi (ed.) (1988)
pp. 199 - 211.
Morgan, M. (1990) The History of Econometric Ideas, Cambridge: Cambridge University
Press.
Morishima, M. (1984) 'The good and bad use of mathematics', in P. Wiles and G. Routh
(eds.) (1984) pp. 51 - 73.
Moss, L.S., and K.I. Vaughn (1986) 'Hayek's Ricardo Effect: A Second Look', History of
Political Economy 18,545- 65.
Moss, L.S. (ed.) (1976) The Economics of Ludwig von Mises, Kansas City: Sheed and
Ward.
Mullineux, A. (1990) Business Cycles and Financial Crises, Hemel Hempstead: Harvester
Wheatsheaf.
Muth, J. (1961) 'Rational expectations and the theory of price movements', Econometrica
29,315-35.
Myrdal, G. (1933) 'Der Gleichgewichtsbegriff als Instrument der geldtheoretischen
Analyse', in F.A. von Hayek (ed.) Beitriige zur Geldtheorie, Vienna: Julius
Springer.
Nash, J. (1950) 'The bargaining problem', Econometrica 18, 155- 62.
Nash, J. (1953) 'Two-person cooperative games', Econometrica 21, 128- 40.
Neftci, S., and T.J. Sargent (1978) 'A little bit of evidence on the natural rate hypothesis
from the U.S.' ,Journal of Monetary Economics 4, 315- 19.
Niehans, J. (1987) 'Classical monetary theory, new and old', Journal of Money, Credit and
Banking 19,409- 24.
Nordhaus, W.D. (1969) Invemion Growth and Welfare: A Theoretical Treatment of
287
Technological Change, Cambridge (Mass.): M.I.T.Press.
Novick, D. (1954) 'Mathematics: logic, quantity, and method', 1he Review of Economics
and Statistics 36,357- 58.
Nozick, R. (1974) Anarchy, State, and Utopia, New York: Basic Books.
Nozick, R. (1977) 'On Austrian methodology', Synthese 36,353- 92.
O'Driscoll, G.P., Jr. (1977a) Economics as a Coordination Problem, Kansas City: Sheed
Andrews and Mcmeel.
O'Driscoll, G.P., Jr. (1977b) 'Spontaneous Order and the Coordination of Economic
Activities' ,Journal of Libertarian Studies 1, No.2, 137- 51.
O'Driscoll, G.P., Jr. (1982) 'Rational expectations, politics, and stagflation', in M.Rizzo
(ed.) (1982) pp. 153 - 76.
O'Driscoll, G.P., Jr., and S.R. Shenoy (1976) 'Inflation, recession, and stagflation', in E.G.
Dolan (ed.) (1976) pp. 185- 211.
O'Driscoll, G.P.,Jr., and M. Rizzo (1985) 1he Economics of 1ime and Ignorance, Oxford.
Okun, A.M. (1980) 'Rational-expectations-with-misperceptions as a theory of the business
cycle' ,Journal of Money, Credit and Banking 12,817- 25.
Okun, A.M. (1981) Prices and Quantities: A Macroeconomic Approach, Oxford: Blackwell.
Pagenstecher, U. (1987) Verstehen und Erkliiren in der Nationalokonomie, Methodenkon
troversen 1930-1985, Niimberg: Preuszler-Verlag.
Paque, K.H. (1985) 'How far is Vienna from Chicago', Kyklos 38,412- 34.
Paque, K.-H. (1990) 'Pattern prediction in economics: Hayek's methodology of the social
sciences revisited', History of Political Economy 22, 281 - 94.
Parsons, S. (1990) 'The philosophical roots of modem Austrian economics: past problems
and future prospects', History of Political Economy 22,295- 319.
Patinkin, D. (1948) 'Price flexibility and full employment', American Economic Review 38,
543-64.
Patinkin, D. (1956) Money, Interest and Prices, New York: Harper and Row, 1965.
Patinkin, D. (1965) 'An indirect-utility approach to the theory of money, assets, and
savings',in F.H. Hahn and F.P.R.Brechling (eds.) (1965) pp. 52-79.
Patinkin, D. (1972) 'Keynesian monetary theory and the Cambridge School', Quarterly
Review, Banco Nazionale del Lavoro, no. 101.
Patinkin, D. (1989) 'Neutrality of money', in J. Eatwell, et al. (eds.) (1989c).
Patinkin, D., and 0. Steiger (1989) 'In search of the "veil of money" and the "neutrality of
money": a note on the origin of terms', Scandinavian Journal of Economics 91,
no. 1.
Pellengahr, I. (1986) 'Austrians versus Austrians', in M. Faber (ed.) (1986a).
Pesaran, M.H. (1989) 1he Limits of Rational Expectations, Oxford: Blackwell.
Phelps, E.S. (1967) 'Phillips curves, expectations of inflation and optimal unemployment
over time', Economica 34,254- 81.
Phelps, E.S. (1968) 'Money-wage dynamics and labor-market equilibrium', Journal of
Phelps, E.S. (1970) 'The new microeconomics in employment and inflation theory', in E.S.
Phelps (ed.) (1970) Microeconomic Foundations of Employment and Inflation
Theory, New York: Norton.
Phelps, E.S., and J.B. Taylor (1977) 'Stabilizing powers of monetary policy under rational
expectations', Journal of Political Economy 85 (1977), 163- 90.
Phelps, E.S. (1990) Seven Schools of Macroeconomic Thought, Oxford: Oxford University
Press.
Phillips, A.W. (1958) 'The relation between unemployment and the rate of change of
money wage rates in the United Kingdom, 1861- 1957' ,Economica 25,283- 99.
Phillips, R.P. (1934- 35) Modern Thomistic Philosophy, Westminster, Md., vol. 2.
Pigou, A. C. (1917) 'The exchange value of legal tender', Quarterly Journal of Economics 32.
Pigou, A. C. (1943) 'The classical stationary state', Economic Journal 53, 343- 51.
Poole, W. (1970) 'Optimal choice of monetary policy instruments in a simple stochastic
macro model', Quarterly Journal of Economics 84, 197- 216.
288
Popper, K.R. (1945) 'The autonomy of sociology', Chapter 14 in K.R. Popper (1945) The
Open Society and its Enemies , London: Routledge & Kegan Paul.
Popper, K.R. (1957) The Poverty of Historicism, London: Routledge & Kegan Paul, 1976.
Reprinted from Economica 11 (no. 42 and 43, 1944) and 12 (no. 46, 1945).
Popper, K. (1959) The Logic of Scientific Discovery, London: Unwin Hyman, 1990, first
published as Logik der Forschung, Vienna: Springer 1934.
Popper, K.R. (1963) Conjectures and Refutations, London: Routledge & Kegan Paul, 1989.
Popper, K.R. (1968) 'On the theory of the objective mind', in K.R. Popper (1972), pp.
153 - 90, first published in a German version in the Akten des XIV. lnternatio-
nalen Kongressftir Philosophie, vol. 1, Vienna, pp. 25- 53.
Popper, K.R. (1972) Objective Knowledge, Oxford: Oxford University Press, 1989.
Popper, K.R. (1983) Realism and the Aim of Science, Totowa, NJ: Rowman and Littlefield.
Popper, K.R., and J.C. Eccles (1977) The Self and Its Brain, Berlin: Springer International.
Reprinted by London: Routledge & Kegan Paul, 1983.
Quine, W. (1948) 'On what there is', Review of Metaphysics 2, 21 - 38. Reprinted in W.
Quine (1953).
Quine, W. (1951) 'Two dogmas of empiricism', Philosophical Review 51,20- 43. Reprinted
in W. Quine (1953).
Quine, W. (1953) From a Logical Point of View, Cambridge (Mass.): Harvard University
Press, 1964.
Radner, R. (1972) 'Existence of equilibrium of plans, prices, and price expectations in a
sequence of markets', Econometrica 40,289- 303.
Radner, R. ( 1982) 'Equilibrium under uncertainty', in Arrow and Intriligator (eds.)
Handbook of Mathematical Economics, Amsterdam: North-Holland, Vol. II, Ch.
20, pp. 923 - 1006.
Radner, R. (1989) 'Uncertainty and general equilibrium', in J. Eatwell, et al. (eds.)
(1989a), pp. 305 - 23.
Reekie, W.D. (1984) Markets, Elltrepreneurs and Liberty, an Austrian view of Capitalism,
Brighton (Sussex).
Rees, A. (1970) 'On equilibrium in labor markets', Journal of Political Economy 78, 306 -
10.
Ricardo, D. (1817) The Principles of Political Economy and Taxation, London. Reprinted
by London: Everyman's Library, 1978.
Rizzo, M.J. (1982) 'Mises and Lakatos: a reformulation of Austrian methodology', in I.M.
Kirnter (ed.) (1982) pp. 53 - 74.
Rizzo, M. (ed.) (1979) Time, Uncertainty and Disequilibrium: Explorations of Austrian
Themes , Lexington: Lexington Books.
Robbins, L. (1932) An Essay on the Nature and Significance of Economic Science, London:
MacMillan, 2nd ed., 1935.
Robbins, L. (1961) 'Hayek on Liberty', Economica, February. Reprinted in L. Robbins
(1961) Politics and Economics, London.
Robinson, J. (1954) 'The impossibility of competition', in E. Chamberlain (ed.) (1954)
Monopoly and Competition and Their Regulation, London.
Rothbard, M.N. (1951a) 'Mises' "Human Action": Comment', American Economic Review
41, 181 - 85.
Rothbard, M.N. (1951b) 'Praxeology: Reply to Mr. Schuller', American Economic Review
41,943- 46.
Rothbard, M.N. (1957) 'In Defense of "Extreme Apriorism"', Southern Economic Journal
23, 315- 20.
Rothbard, M.N. (1962) Man, Economy and State, Princeton: D. van Nostrand, 1970.
Rothbard, M.N. (1963) America's Great Depression, Kansas City: Sheed and Ward, 1975.
Rothbard, M.N. (1973) 'Praxeology as the method of economics', in M. Natanson (ed.)
(1973a) pp. 311- 39.
Rothbard, M.N. (1976) 'Praxeology: The methodology of Austrian economics', in E.G.
Dolan (ed.) (1976) pp. 19- 39.
289
Rothbard, M.N. (1990) 'Interview with Murray N. Rothbard', Austrian Economics
Newsletter, summer 1990.
Rotwein, E. (1959) 'On the methodology of positive economics', Quarterly Journal of
Economics 73,554- 75.
Runde, J., and C. Torr (1985) 'Divergent expectations and rational expectations', South
African Journal of Economics 53, 217- 25.
Saidi, N.H., and R.J. Barco (1976) 'Unanticipated money, output, and unemployment in
Canada', unpublished, University of Rochester, July.
Samuelson, P. (1954) 'Introduction: mathematics in economics - no, no or yes, yes, yes?',
The Review of Economics and Statistics 36,359.
Samuelson, P. (1954) 'Some psychological aspects of mathematics and economics', The
Review of Economics and Statistics 36, 380 - 82.
Samuelson, P.A., and R.M. Solow, (1960) 'Analytical aspects of anti-inflation policy',
American Economic Review 50, 177 - 94.
Sargent, T.J. (1969) 'Commodity price expectations and the rate of interest', Quarterly
Journal of Economics 83, 127- 40.
Sargent, T.J. (1971) 'A note on the 'accelerationist' controversy', Journal of Money, Credit
and Banking 3, 721 - 25. Reprinted in R.E. Lucas, Jr., and T.J. Sargent (eds.)
(1981) pp. 33 - 38.
Sargent, T.J. (1972) 'Anticipated inflation and the nominal rate of interest', Quarterly
Journal of Economics 86, 212 - 25.
Sargent, T.J. (1973a) 'Interest rates and prices in the long run', Journal of Money, Credit
and Banking 5, 385 - 449.
Sargent, T.J. (1973b) 'Rational expectations, the real rate of interest, and the natural rate
of unemployment', Brookings Papers on Economic Activity 2, 429 - 72. Reprinted
in R.E. Lucas, Jr., and T.J. Sargent (eds.) (1981) pp. 159- 98.
Sargent, T.J. (1973c) 'The fundamental determinants of the interest rate: a comment',
Review of Economics and Statistics 55, 391- 393.
Sargent, T.J. (1976a) 'A classical macroeconometric model for the United States', Journal
of Political Economy 84, 207- 38. Reprinted in R.E. Lucas, Jr., and T.J. Sargent
(eds.) (1981) pp. 521 - 51.
Sargent, T.J. (1976b) 'The observational equivalence of natural and unnatural rate theories
of macroeconomics', Journal of Political Economy 84. Reprinted in R.E. Lucas,
Jr., and T.J. Sargent (eds.) (1981) pp. 553- 62.
Sargent, T.J. (1987) Macroeconomic Theory, New York: Academic Press.
Sargent, T .J., and N. Wallace ( 1973) 'Rational expectations and the dynamics of hyperinfla
tion', International Economic Review 14, 328- 50.
Sargent, T.J., and N. Wallace (1975) '"Rational" expectations, the optimal monetary
instrument and the optimal money supply rule', Journal of Political Economy 83,
241 -54. Reprinted in R.E. Lucas, Jr., and T.J. Sargent (eds.) (1981) pp. 215-
28.
Sargent, T.J., and N. Wallace (1976) 'Rational expectations and the theory of economic
policy' ,Journal of Monetary Economics 2, 169- 83. Reprinted in R.E. Lucas, Jr.,
and T.J. Sargent (eds.) (1981) pp. 199- 213.
Scheide, J. (1986) 'New Classical and Austrian business cycle theory: Is there a differen
ce?', Weltwinschaftliches Archiv 122, 575 - 98.
Schuller, G.J. (1950) 'Book review of "Human Action: A Treatise on Economics" by
Ludwig von Mises', American Economic Review 40,418- 22.
Schuller, G.J. (1951) 'Rejoinder', American Economic Review 41, 185-90.
Schumpeter, J.A. (1908) Das Wesen und der Hauptinhalt der theoretischen Nationaloko
nomie, Leipzig.
Schumpeter, J.A. (1926) 1heorie der winschaftlichen Entwicklung, Miinchen und Leipzig;
English translation, The Theory of Economic Development, Cambridge, 1934.
Schumpeter, J.A. (1939) Business Cycles, vol. I and II, New York and London.
Schumpeter, J.A. (1954) History of Economic Analysis, London: Allen & Unwin, 1986.
290
Selgin, G.A. (1988) 'Praxeology and Understanding: An Analysis of the Controversy in
Austrian Economics', Review of Austrian Economics 2, 19- 58.
Senior, N.W. (1852) 'Four introductory lectures on political economy', London: Longman,
Brown, Green, and Longmans. Reprinted 1966, New York: August M. Kelley.
Shackle, G.L.S. (1952) 'On the meaning and measure of uncertainty: I', in G.L.S. Shackle
(1955).
Shackle, G .L.S. (1955) Uncertainty in Economics and Other Reflections, Cambridge:
Shackle, G.L.S. (1966) 1he Nature of Economic Thought, New York.
Shackle, G.L.S. (1972) Epistemics and Economics, Cambridge: Cambridge University Press.
Shackle, G.L.S. (1981) 'F.A. Hayek', in D.P. O'Brien and J.R. Presley (eds.) Pioneers of
Modern Economics in Britain, London: MacMillan.
Shearmur, J. (1991) 'Popper, Lakatos and theoretical progress in economics', in N. de
Marchi and M. Blaug (eds.) pp. 35 -52.
Shiller, R.J. (1978) 'Rational expectations and the dynamic structure of macroeconomic
models', Journal of Monetary Economy 4, 1- 44.
Shove, G.F. (1933) 'Review article on The Theory of Wages by J.R. Hicks', Economic
Journal43, 460- 72.
Shreffin, S.M. (1983) Rational Expectations, Cambridge: Cambridge University Press.
Sims, C.A. (1972) 'Money, income, and causality',American Economic Review 62,540-52.
Sims, C. A. (1977) 'Exogeneity and causal ordering in macroeconomic models', in C. A.
Sims (ed.) New Methods in Business Cycle Research: Proceedings from a Confer-
ence, Minneapolis: Federal Reserve Bank of Minneapolis.
Skousen, M. (1990) 1he Structure of Production, New York: New York University Press.
Slutsky, E. (1937) 'The summation of random causes as the source of cyclical processes',
Econometrica 5, 105 - 46.
Smith, A. (1776) An Inquiry into the Nature and Causes of the Wealth of Nations, 2 vols.,
ed. E. Cannan, reprint, Chicago: University of Chicago Press, 1976.
Snippe, J. (1985) 'Non-stationarity and the collapse of Panglossian economics: a note on
the microfoundations of new classical models', unpublished paper, Groningen,
Rijksuniversiteit.
Snippe, J. (1986) 'Even modest pretence cannot be justified: on the negative implications
of Hayek's "Economics and knowledge"', Research Memorandum nr. 201, Insti-
tute of Economic Research, Faculty of Economics, University of Groningen.
Snippe, J. (1986 - 87) 'Varieties of rational expectations: their differences and relations',
Journal of Post-Keynesian Economics 9, 424- 37.
Snippe, J. (1987) 'lntertemporal coordination and the economics of Keynes: comment on
Garrison', History of Political Economy 19, 329- 34.
Snower, D.J. (1984) 'Rational expectations, nonlinearities, and the effectiveness of
monetary policy', Oxford Economic Papers 36, 177 - 99.
Solow, R.M. (1954) 'The survival of mathematical economics', The Review of Economics
and Statistics 36, 372 - 74.
Solow, R.M. (1978) 'Down the Phillips Curve with gun and camera', in R.L. Teigen (ed.)
Readings in Money, National Income, and Stabilization Policy, Homewood, Ill.:
Irwin. Published originally in D.A. Belsley, E.J. Kane, P.A. Samuelson, and R.M.
Solow (eds.) (1976) Inflation, Trade, and Taxes, Ohio State University Press.
Solow, R.M. (1991) 'Discussion notes on "Formalization"' ,Methodus 3 (1), 30- 31.
Spadaro, L.M. (ed.) (1978) New Directions in Austrian Economics, Kansas City.
Spiethoff, A. (1902) 'Vorbemerkungen zu einer Theorie der Uberproduktion', Jahrbuchjar
Gesetzgebung, Verwaltung und Volkswirtschaft
Spiethoff, A. (1903) 'Die Krisentbeorie von M. v. Tougan-Baranowsky und L. Pohle',
Jahrbuchjar Gesetzgebung, Verwaltung und Volkswirtschaft
Spiethoff, A. ( 1925) 'Krisen', Handworterbuch der Staatswissenschaften
Spiethoff, A. (1933) Der Stand und die niichste Zukunft der Konjunktuiforschung, Festschrift
ffir Arthur Spiethoff, Miinchen: Duncker & Humblot.
291
Sraffa, P. (1932) 'Book review of F.A. von Hayek's Prices and Production', Economic
Journal42, 42 - 53.
Stigler, G.J. (1961) 'The economics of information', Journal of Political Economy 64, 213 -
25.
Stiglitz, J.E. (1988) 'Money, credit, and business fluctuations', Economic Record 64, 307-
22.
Streissler, E., et al. (eds.) (1969) Roads to Freedom. Essays in Honour of Friedrich A. von
Hayek, London.
Streissler, E. (1972) 'To what extent was the Austrian school marginalist?', History of
Streissler, E., and W. Weber (1974) 'The Menger tradition', in J.R. Hicks and W. Weber
(eds.) (1974).
Suppes, P. (1968) 'The desirability of formalization in science', Journal of Philosophy 65,
651- 64.
Taylor, J.B. (1975) 'Monetary policy during a transition to rational expectations', Journal of
Taylor, J.B. (1980) 'Comment' on Kydland and Prescott (1980), in S. Fischer (ed.) (1980)
pp. 191- 94.
Thalenhorst, J., and A. Wenig (1984) 'F.A. Hayek's "Prices and Production" re-analyzed',
Jahrbiicher fUr Nationalokonomie und Statistik 199, pp. 213- 36.
Thornton, H. (1978) An Enquiry into the Nature and Effects of the Paper Credit of Great
Britain, Fairfield, N.J.: A.M. Kelley. First printed 1802 by J. Hatchard and F. and
C. Rivington, London. Reprinted 1939 by Allen & Unwin, London.
Tinbergen, J. (1954) 'The functions of mathematical treatment', The Review of Economics
and Statistics 36, 365- 69.
Tintner, G. (1938) 'A note on economic aspects of the theory of errors in time series',
QuarterlyJournal of Economics 53, 141- 49.
Tisdale, C. (1975) 'Concepts of rationality in economics', Philosophy of the Social Sciences
5, 259- 72.
Tobin, J. (1965) 'Money and economic growth',Econometrica 33,671-84.
Tobin, J. (1977) 'How dead is Keynes?', Economic Inquiry 15,459- 68.
Tobin, J. (1980a) Asset Accumulation and Economic Activity, Oxford: Blackwell.
Tobin, J. ( 1980b) 'Are new classical models plausible enough to guide policy?', Journal of
Money, Credit and Banking 12,788- 99.
Townshend, H. (1940) 'Review article on Profits, Interest and Investment by F.A. von
Hayek', Economic Journal 50,99- 103.
Townshend, R.M. (1983) 'Forecasting the forecasts of others', Journal of Political Economy
91,546- 88.
Tullock, G. (1988) 'Why the Austrians Are Wrong About Depressions', Review of Austrian
Economics 2, 73 - 78.
Tversky, A., and D. Kahneman (1986) 'Rational choice and the framing of decisions',
Journal of Business 59, S251 - 78.
Uhr, C.G. (1960) Economic Doctrines of Knut Wicksell, Berkeley: University of California
Press.
Vicarelli, F. (1985-86) 'Natural laws and economic policy: some considerations on the
theoretical foundations of the new classical macroeconomics', Journal of Post-
Keynesian Economics 8, 298- 314.
Visser, H. (1971) 'Neutraal geld, een overzicht',De Economist 119,393-439.
Visser, H. (1980) Monetaire 1heorie, Leiden: Stenfert Kroese.
Visser, H. (1984) New Classical Macroeconomics as seen by an impressed non-believer or,
Keynes and the Classics all over again, SUERF Series 49A, Tilburg: Societe
Universitaire Europeenne de Recherches Financieres.
Visser, H. (1989) 'The monetary order', Research Memorandum no. 1989-3, Amsterdam:
Free University.
Visser, H. (1991) Modern Monetary Theory, Aldershot: Edward Elgar.
292
Wainhouse, C.E. (1982) Hayek's Theory of the Trade Cycle: The Evidence from the Time
Series, Ph.D. dissertation, New York University.
Wainhouse, C.E. (1984) 'Empirical evidence for Hayek's theory of economic fluctuations',
in B. Siegel (ed.) Money in Crisis, San Francisco: Pacific Institute, pp. 37- 71.
Wartofsky, M.W. (1976) 'The relation between philosophy of science and history of
science', in R.S. Cohen, et al. (eds.) (1976) pp. 717-37.
Weintraub, E.R. (1983) 'On the existence of a competitive equilibrium: 1930 - 1954',
Journal of Economic Literature 21, 1 - 39.
Weintraub, E.R. (1985) General Equilibrium Analysis: studies in appraisa~ Cambridge:
Weintraub, E.R. (1989) 'Methodology doesn't matter, but the history of thought might',
Scandinavian Journal of Economics 91, 477- 93.
Weintraub, E.R. (1991) 'Stabilizing dynamics', in N. de Marchi and M. Blaug (eds.) pp.
273-90.
Weintraub, S. (1978) Keynes, Keynesians, and Monetarists, New Jersey: University of
Pennsylvania Press.
Weiss, L. (1980) 'The role for active monetary policy in a rational expectations model',
Journal of Political Economy 88, 221 - 33.
White, L.H. (1976) 'Entrepreneurship, imagination, and the question of equilibrium',
unpublished paper.
White, L.H. (1977) 'The Methodology of the Austrian School Economists', Occasional
Paper Series, Center for Libertarian Studies, No. 1, New York.
White, L.H. (1982) 'Mises, Hayek, Hahn, and the market process: comment on Littlechild',
in I.M. Kirmer (ed.) (1982) pp. 103 - 10.
Wicksell, K. (1898) Geldzins und Gaterpreise. Jena: Gustav Fischer.
Wicksell, K. (1919) 'Professor Cassel's System of Economics', Ekonomisk Tidskrift, no. 9.
Reprinted in English in K. Wicksell (1934 (1949), pp. 219- 57).
Wicksell, K. (1934) Lectures on Political Economy, vol. 1: General Theory, London:
Routledge & Kegan Paul, 1949.
Wicksell, K. (1935) Lectures on Political Economy, vol. II: Money, London: Routledge &
Kegan Paul, 1950.
Wicksell, K. (1958) Selected Papers on Economic Theory, London: Allen & Unwin.
Reprinted by New York: Augustus M. Kelley, 1969.
Wiles, P., and G. Routh (eds.) (1984) Economics in Disa"ay, Oxford: Blackwell.
Wilson, T. ( 1940) 'Capital theory and the trade cycle', Review of Economic Studies 7, 169 -
79.
Windsperger, J. (1983) 'Gleicbgewicht und Wissen: Analyse des Hayek'scben Konzeptes',
Jahrbuchjar Sozialwissenschaft 34,232- 45.
Woigin, G. (1980) 'Unemployment and monetary policy under rational expectations: some
Canadian evidence', Journal of Monetary Economics 6, 59 - 68.
Wong, S. (1978) The Foundations of Paul Samuelson's Revealed Preference Theory,
London: Routledge & Kegan Paul.
Wulwick, N.J. (1987) 'The Phillips Curve: which? whose? to do what? bow?', Southern
Economic Journal 53, 834- 57.
Wulwick, N.J. (1989) 'Phillips' approximate regression', Oxford Economic Papers (special
issue on The History and Methodology of Econometrics), 170- 88.
Yeager, L.B. (1987) 'Why Subjectivism?', Review of Austrian Economics 1, 5- 32.
Yeager, L.B. (1988) 'Reply to Comment by Walter Block', Review of Austrian Economics
2, 209- 10.
Yeager, L.B. (1989) 'The Austrian School on Money and Gold', in J.J. Krabbe, et al. (eds.)
(1989) pp. 92 - 105.
Zamowitz, V. (1985) 'Recent work on business cycles in historical perspective: a review of
theories and evidence', Journal of Economic Literature 23,523- 80.
Zuidema, R.P. (1989) 'On the Austrian contribution to capital theory', in J.J. Krabbe, et al.
(eds). (1989) pp. 64- 78.
293
Zuidema, R.P. (1990) 'lntrest en de hogere produktiviteit van omwegproduktie', Maand
schrift Economie 54, 93 - 104.
Zijp, R.W. van (1990a) 'Hayek en Lucas: een vergelijking', Maandschrift Economie 54,
128- 39.
Zijp, R.W. van (1990b) 'Why Lucas is not a Hayekian', paper presented at ISINI (Paris,
August 1990), Research Memorandum 1990-27,Free University Amsterdam.
Zijp, R.W. van (1991a) 'The methodology of the Neo-Austrian research programme',
Research Memorandum 1991-34,Free University Amsterdam.
Zijp, R. W. van ( 1991 b) 'Tussen micro en macro: de produktiestructuur', Economisch-
Statistische Berichten 16, nr. 3820,pp. 802- 3.
Zijp, R.W. van, and H. Visser (1992) 'Mathematical formalization and the analysis of
Cantillon effects', Research Memorandum 1992-2, Free University Amsterdam.
294
NAME INDEX
Arrow, K.J. 135 Colander, D. 2, 245

Aschheim, J. 29n13 Cross, R. 143n18
Attfield, C.L.F. 226n23
Auerbach, A. 191n13 Darnell, A. C. 87n10, 237n35
Davidson, D. 19n16
Backhouse, R. 14 Debreu, G. 119n25, 135
Barone, E. 23 DeCanio, S.J. 220n16
Barreto, H. 225-226 Demery, D. 226n23
Barro, R.J. 150n2, 184, 191-195,198, Dolan, E.G. 83
199n23,203,207-209,213-216, Duck, N. W. 226n23
219-221,226-230,232-237,239- Duhem, P. 181n49
-242,260,268-269
Barry, N.P. 49n8 Easley, D. 220
Begg,D.K.H. 145n22,147n23,232 Egger, J.B. 94n23, 95n26, 124n33
Bellante, D. 117 Endres, A.M. 15n4, 93n19
Bernholz, P. 17n10 Evans, J.L. 87n10,237n35
Blanchard, O.J. 235
Blaug, M. 16n6, 60n23, 63 Faber, M. 17n10
Blejer, M.I. 226n23 Fehl, U. 111
Blinder, A.S. 191n13 Feldstein, M.S. 191n13
Blume, L.E. 220 Ferguson, A. 67n30
Bohm, s. 26, 39, 40, 79n45 Fernandez, R.B. 226n23
Bohm-Bawerk, E. von 6, 11, 15, 16, Fischer, S. 191n13, 218-220,232-233,
54,92-93,114n16,115n20 235-236,240
Boland, L.A. 52, 53n16, 69, 76, 90, Fisher, I. 135-137,147,151,169-176,
94n24,149, 162n22 181
Bordo, M.D. 135n5 Foss, N.J. 50, 125
Boschen, J.F. 234,239 Friedman, B. 219
Boulding, K.E. 107 Friedman, M. 7, 97n30, 117, 135-137,
Bray, M.M. 220 138-140,142, 143-144,147-148,
Bronfenbrenner, M. 135n1 149,151,153-154,155n9,176,
Brunner, K. 135n2 180,201n26,247,258n12,266-
Buiter, W. 158n16 -267
Butos, W.N. 2, 51n13, 53, 245,248, Frisch, R.A. 7, 135, 148, 185, 186n4,
261 186n5
Butler, E. 28n11, 30, 32n21 Frydman, R. 199, 220n16
Cagan, P. 135n2, 137n7, 145,170 Garrison, R.W. 2, 15n4, 19n16, 20, 21,
Caldwell, B.J. 51n14, 53n16, 64, 30n15,33n23,93n19,112,115-
81, 101, 116n21 -117, 120-121,247,252,254,
Cantillon, R. 29n13 260,265
Cassel, G. 11, 14 Gibson, A.H. 168
295
Gordon, D. 235,239 Kamath, S.J. 158n16
Granger, C.W.J. 126 Kantor, B. 2, 245
Gray, J.N. 218-219,236 Katmer, D.W. 119n24
Grossman, H. I. 157n14, 234,239 Keizer, W. 37
Guthrie, R. 2, 245 Keynes, J.M. 61, 83-85, 101, 114n16,
135,138-139,147,168-169,181
Haberler, G. 11, 14, 19n14,36, 124 Kim, K. 2, 186n5
Hacking, I. 113 King, R.G. 192n15,207, 223n21,229-
Hahn, F.H. 28n11, 147n24 -232,235-236,241-242,254n8,
Hall, R.E. 184,188,192,203 268-269
Haltiwanger, J.C. 157n15 Kirmer, I.M. 7, 33-34,40, 84, 93,
Hansen, A.H. 56n20, 81 96,98-100,101-109, 110n13,
Hart, R.A. 141n13 112,130,247,250,265
Hawtrey, R.G. 49n7, 62n27 Klamer, A. 149, 163n23, 176n45, 213n6
Hayek, F.A. von 2, 7-8, 16, 19n16, Klausinger, H. 2, 261
20n17,26,34,35n26,40,42, Knight, F. 54, 83, 109nl1,254
43-81,83-86,87n12,88-93, Knorr-Cetina, K. 68n31
94n23,95n26,96-97,99, 101- Koertge, N. 202n27
-103, 110n13,114-115,117-118, Koyck, L.M. 145
121n28,122-124,125-129,135, Kydland, F.E. 192,217,236,241
196,200,204,245,247,250-
-253,257,259-260,263-264 Lachmann, L.M. 7, 15n3, 26, 39, 40,
Hempel, C.G. 73n40, 74, 248n1, 68, 70n32, 79n45,83-96,99-100,
259n14,260n15 101-102,110,114, 124n33,130,
Herbener, J. 120n26 264
Hercowitz, Z. 208,229,232-235,239, Laidler, D.E.W. 2, 245
242,268 Lakatos, I. 5n2
Hicks, J.R. 7, 58,61-62,81, 84n3, Lange, 0. 90,264
86,88n14,90n17,93n21,95, Langlois, R.N. 118
101, 114n17, 264 Latsis, S.J. 106n9, 201
High, J. 110-111,130 Lavoie, D. 23,34n24
Hines, A.G. 141n13 Leamer, E.E. 120n26
Hirsch, A. 142n16,144,149,186 Lerner, A.P. 84n3
Homer 1 Leijonhufvud, A. 19n16, 114n16,
Hoover, K.D. 2, 141-142,149,229, 116n22,258n12
256 Lindahl, E. 93n21
Hoppe, H.-H. 120n26 Lipsey, R.G. 136, 141
Hsieh, C.-Y. 29n13 Loasby, B.J. 104n5, 111-112,130
Hutchison, T.W. 121n28 Long, J.B. 236,241
Huussen, G.M. 24n2, 72n35, 72n37, Lucas, R.E. 2, 7-8, 148, 150, 151-168,
201,202n27,249n4,250 175-176,180,181-182,183-192,
193n19,194-195,196-198,
Jevons, W.S. 12,114n16 199n23,200,202-204,207-209,
Johnston, W. 35n26 214,221,232,236n32,237-238,
240,242,245,247,255,260,
Kaldor, N. 55,58-61,64,84n3,88 266-268
296
Lukes, S. 67 Orosel, G.O. 17n10
Lundberg, E. 93n21
Paque, K.-H. 75, 260
Mach, E. 72n35 Parsons, S. 24n2
Machlup, F. 49n9, 199 Patinkin, D. 28n11, 116
Mackay, D.I. 141nl3 Peel, D. 152n2
Maddock, R. 149, 154,223 Pellengahr, I. 17n10
Marchi, N. de 142n16, 144, 149, Pesaran, M.H. 147n23, 147n25, 199,
159n17,162,167, 186 220n16
Marshall, A. 18n12, 135, 137 Phelps, E.S. 7, 136, 143n19, 144-
Mayer, H. 88n13, 248, 249n2, 249n3, -148, 150-151,154-156,180,
250 201n26,218,220,236,266-267
McCallum, B. T. 157n14, 209, 219n15, Phillips, A.W. 136, 140-142
220,226,241 Pigou, A.C. 135, 137
McCloskey, D.N. 121n29 Plosser, C.l. 236,241-242,269
Menger, C. 11, 14-17,27n8, 71n34, Poole, W. 209
73n39,81,88n13,92-93, Popper, K.R. 2-5, 53n16, 65, 70n32,
114n16,252n6,263 71n33, 74,94n25,200-201,
Mill, J.S. 72n34, 196 202n27,204,245,251-252,
Minford, P. 152n2, 212n5 253n7,258-259,266,270
Mises, L. von 6, 14-16,21,23-45, Prescott, E.C. 184, 187, 192, 199n23,
48-49,69, 72n7, 76,78,80-81, 217,236,241
85n5,89,91-93,94n23,95n26,
95n27, 96-97,99, 101-102, Quine, W. 181n49
103n3, 104,105n7,109n11,
110n13,116,119,121n29,130, Radner, R. 160n19
144n21,253,263 Rapping, L.E. 150-156,158, 166,
Mishkin, F.S. 235,239 176n46,180, 182,183,187,
Modigliani, F. 188 198,202,247,266-267
Montiel, P.J. 212n5 Rees, A. 155, 166, 183-184,187,203
Moore, H.L. 12 Ricardo, D. 19n14, 56
Morgan, M. 12n2, 87n10, 185n1, 186, Rizzo, M. 40, 101, 105-106,118, 130,
208,237-241 Rosenstein-Rodan, D.N. 88n13
Moss, L.S. 28n11, 55, 57,63-64 Rothbard, M.N. 72n37, 84, 95n26, 96-
Mullineux, A. 186n4, 186n5, 191 -97,99-100,109, 110n13,265
Muth, J. 148, 156-158, 160n19, 162, Runde, J. 157n14
173,181,253,266 Rush, M. 208,221,227-228,239,242,
268
Neftci, S. 207,221,225-226,228-
-229, 239-240,242, 268-269 Saidi, N.H. 226n23
Newbold, P. 126n37 Samuelson, P.A. 136, 142
Nozick, R. 36, 67 Sargent, T.J. 6, 150-151,163n23, 164-
-165, 168-180,181-182,183,
O'Driscoll, G.P. 25n5, 40, 55, 63, 191n13,202n28,203,207-213,
89n16, 101-102,105-106,115, 215-216,219-226,228-229,232,
118, 130, 263,265 236n32,237-242,260,266n1,
297
267-269 Uhr, C.G. 19n14
Scheide, J. 2, 245
Schmoller, G. von 71n34 Vaughn, K.I. 55, 57,63-64
Schultz, H. 185nl Visser, H. 29n13, 31n20, 50n11, 137n8,
Schumpeter, J.A. 16n8,29n13,44n2, 169n32,235n31
85, 105n7
Schwartz, A.J. 137n7, 144,151 Wainhouse, C.E. 124, 126-130,260
Selgin, G.A. 25n5 Waldman, M. 157n15
Shackle, G.L.S. 18n12,40,90n17, Wallace, N. 176n44,207-213,215-216,
94n24,104n5,105n7,107 219-220,238n36,240-241,
Shiller, R.J. 21ln3, 220 266n1,268
Sims, C.A. 126n37 Walras, L. 114n16
Sisyphus Weber, W. 16
Skousen, M. 18 Weintraub, E.R. 120
Slutzky, E. 7, 135,148,185,186n5 Weizsicker, C.C. von 17n10
Smith, A. 19n14 Wenig, A. 121-124,265
Snapper, F. 84n2 White, L.H. 39, 109
Snippe, J. 157n13 Wicksell, K. 6, 11, 15,19-21,27,
Snower, D.J. 211n3 30n15,43n1,44,48n5, 114nl6,
Solow, R.M. 136, 142 116,143n17,170n35,185n3,
Spiethoff, A. 11, 13-14,48n5 258n12
Sraffa, P. 18n12, 62n27 Wieser, F. von 16
Stackelberg, H. von 17n10 Wilson, T. 55,57-61,88
Stigler, G.J. 18n12, 158 Woigin, G. 226n23
Streissler, E. 16 Wong, S. 3-4
Suppes, P. 119n25 Wulwick, N.J. 140n12
Taylor, J.B. 192n14, 216n10, 218, 220, Yeager, L.B. 28nll

236
Thalenhorst, J. 121-124,265 Zuidema, R.P. 15n4, 17n10, 93n19
Thornton, H. 19n14 Zijp, R.W. van 4n1, 24n2, 72n37
Tobin, J. 188
Torr, C. 157n14
Tugan-Baranowsky, M. 11-13
298
SUBJECT INDEX
accelerator effect, 191 dualism, methodological, 41,251

agent-specific shocks, 207, 229-232, Duhem-Quine thesis, 269
241,268 dynamics, situational, 76,91-92,94,96,
alertness, see entrepreneurship 123,264
analytic method, 71
entrepreneurship, 103-105,108-112,247,
Cantillon effects, see distribution 255-257,265,269
effects equilibrium,
capacity output, see output individual, 51,264
capital allocation effect, 30, 32, 116, 257, intertemporal, 43-45,263
266 general, 51, 61, 159-161,247,
capital structure, see structure of produc 263-264,267
tion Natural Rate (NRE), 159-160,
choice, pure logic of, 26, 39, 65, 69,255, 197,204,219,249-
264 -250,252
circularity, problem of, 18, 129 partial, 158-159,161,247
comparative-statics, 55,61-64, 123 Rational Expectations (REE),
competition, as a process, 78-80, 102, 160,197,205,248-
110-112 -250, 252, 256
complex phenomena, 71,200,251-252, evenly rotating economy, 25, 31,96-97,
259,265,269 263
compositive method, 71-76,250-252 expectations,
constraints, situational, 3-4, 7, 196, 203 adaptive (AEH), 145-147,154,
coordination (problem), 43, 50-55,65-66, 156, 172-173,266
80-81,83,94,96,98-99,102- elasticity of, 88-90, 264
-105,114,120,153,197,204, rational (REH), 150, 156-158,
247-248,250,257,264-266,269 162-165,167, 172,
corroboration, 212,239,260,268,270 195,204,219-220,
costs, transaction and/or adjustment, 222,227,238,241,
see no-frictions assumption 266-268
currency competition, see free banking explanation,
Currency School, 31 deductive-nomological, 74
degree of, 258-260
decision-situation, 68 genetic-causal, 73, 248, 250,
single-exit, 69,202,252-253, 269
263,266,269 hypothetical-deductive, 72
multiple-exit, 253,264-265 mathematical-functional, 248-
discoordination dynamics, see dynamics -250
dispersion of knowledge, 39, 52,253,257, explanation sketch, 74,259,264,269
259, 264-265
distnbution effects, 63, 117, 153,204, falsificationism,
263,266 naive, 240
299
sophisticated, 179,213,228, -162,266
238,240,267 objective vs. subjective, 69,
Fisher's indirect effect, 169-170 252-254, 264-265
forced savings, doctrine of, 34, 48, 117,
129 liquidity trap, 138
free banking, 49 logic, situational, see reconstruction
frictions, see no-frictions assumption Lucas critique, 163-166,229,239,267
full-information output, see output Lucas Problem, 196-198,201-202,247-
-248
generic scheme, see explanation sketch Lucas programme, 196-203,238, 247
Gibson paradox, 168, 173 Lucas supply function, 150, 152
Granger causality, 126, 174,222
maladjustment, 11-17
Hayek Problem, see coordination model-construction, method of rational,
Hayek programme, 64-81,247,264, 266 see reconstruction
homogeneity postulate, 112-113,199-200, money, (non-)neutrality of,
203,256-258,269 29-30,58,117,123,
household, representative, 127-128,136, 161,
see representative agent 189,194,266,268
(non-)supemeutrality of,
impulse problem, 185, 190, 194, 198,203, 189,194
209,214,240,268 surprise, 226-228
individualism, methodological, 36-37, 67, monism, methodological, 201,251
198-200,203,212,250 Muth-rationality, see expectations
information, see knowledge
innovation, 85-86 Natural Rate Hypothesis (NRH),
interest, 143-148,150,154,162,167,
market (money) rate of, 14, 17, 176,179,188,221,227,238,
19-21,33,46-47,128 241, 266-268
natural rate of, 18-21,46-47, natural rate of interest, see
nominal rate of, 169, 171-174 interest
real rate of, 169, 171, 175, 195 neutrality proposition, see money
intertemporal inconsistency of plans, see no-frictions assumption, 210, 219, 238,
time-inconsistency of plans 241,257,268
invariance assumption, 224-225,228
investment, observational equivalence, 223-225,228,
malinvestments, 17-18,21,32, 240-241' 268
89, 114, 116, 123, output,
129,258,263 capacity, 209-213,220
overinvestments, 17, 21, 32, full-information, 213-216,220
190,258
islands approach, 155-156,189, 198-200, pattern predict!on, 75,260
202,210,256 Peel's Bank Charter Act, 31
perceived vs. unperceived money growth,
knowledge, 232-235
global vs.local, 155-156,161- period of production, 15-16,31,49
300
Permanent Income Hypothesis (PIH), satisfactoriness of models, 237-242
139,153 scientism, 64, 70, 250
persistence, 155, 166, 184-192,268 Shackle-Boulding paradox, 107, 158
Phillips Curve, 140-142,146-147,151-152, signal extraction problem, 48, 89, 146,
154,167-168,175,188,217,266 193, 195
policy, singularism, methodological, 38
(in)effectiveness, 35, 49, 97, structure of production, 15-21,54, 120,
207-236,238-242,267 122,257,263
rules versus discretion, 217-218 subjectivism,
praxeology, 23-27 dynamic, 39-40,69,79, 102,
problem situation, 2-4, 7 105-106,117-118,
process analysis, 91-95 253,264-265
production time, aggregate, 115-116 static, 39-40,69, 105-106,200,
progress, scientific, 5-6 253,263
propagation problem, 186, 190-191,195, supemeutrality of money, see money
198,203,240,268 symmetry thesis, 74
quantity theory of money, 136-140,266 technological change, see innovation

theoretical aims, 2-3
Rational Expectations Hypothesis, time-inconsistency of plans, 217-218
see expectations Tobin effect, 189, 194-195
real business cycle theory, 235-236,240-
-242, 257. 268 understanding, 2-3, 5-6, 70, 72, 87, 250-
real cash balance effect, 29, 116 -252, 269-270
reconstruction, method of rational,
3-5, 7, 200-203,266 verisimilitude, 5
representative agent, 152, 198-199,204, Verstehen, see understanding
256
Ricardo effect, 55-64, 83 waiting time, average, 15-16
risk vs. uncertainty, 109,254-256,269
301
NEDERLANDSE SAMENVATTING (DUTCH SUMMARY)
Gedurende de laatste twee decennia heeft de economische wetenschap

een opmerkelijke verandering in aandachtsveld ondergaan. Sinds de
Tweede Wereldoorlog werd zij gedomineerd door de Hicksiaanse versie
van de Keynesiaanse macro-economie. Ondanks de monetaristische
aanvallen duurde deze situatie voort tot bet begin van de jaren '70. De
theoretische problemen van de Keynesiaanse traditie met bet verklaren
van stagflatie bood twee andere economisch-liberale onderzoeksprogram-
ma's de gelegenheid om zich te profileren. Deze programma's, de Oosten-
rijkse School en de Nieuw-Klassieke Economie (NCE) benadrukken de
allocatieve efficientie van bet marktmechanisme. Hun nadruk op conjunc-
tuurtheorie deed dit onderzoeksveld weer opleven.
De gezamenlijke opkomst van beide 'scholen' en beider nadruk
op de informatie-doorgevende functie van prijzen heeft geleid tot de claim
tot ze eenzelfde achtergrond hebben en wellicht tot dezelfde onderzoeks-
traditie behoren. Deze suggestie werd versterkt door Lucas' (1977, p. 216)
claim dat de Nieuw-Klassieke Economie voortborduurt op bet werk van
de Oostenrijkse econoom Hayek. In de loop der tijd heeft deze claim
zowel voor- als tegenstanders gekregen. Helaas hebben deze voor-en
tegenstanders beide onderzoeksprogramma's slechts bekeken in hun
volledig ontwikkelde vorm, zodoende beider ontwikkeling negerend. Het
gevolg hiervan is geweest dat de theoretische problemen van Oostenrij-
kers en Nieuw-Klassieken niet voldoende duidelijk zijn geworden om de
zgn. common-roots claim op haar merites te beoordelen. De onderhavige
studie poogt een analyse te geven van de respectievelijke historische
ontwikkelingen om daarmee zo'n verantwoorde beoordeling mogelijk te
maken. Aangezien de studie dus een poging omvat tot begrijpen wordt
Popper's versie van de begrijpende methode als analyse-methode gebruikt.
Deze methode, ook wel de methode van rationele reconstructie genoemd,
vat een theorie op als een poging om een theoretisch probleem (theoreti-
cal aim) op te lossen. De voor de bestudeerde wetenschapper accpetabele
oplossingen moeten aan bepaalde voorwaarden (situational constraints)
voldoen. Tezamen vormen theoretische probleem en voorwaarden de pro-
bleemsituatie van de wetenschapper. Het begrijpen van een theorie
betekent dat de wetenschapshistoricus de probleemsituatie reconstrueert.
Gegeven Popper's kritisch-rationalistische wetenschapsfilosofie is zo'n
reconstructie 'slechts' een gissing, waarvan de waarheidsstatus nooit met
volledige zekerheid kan worden gekend.
Wetenschapsgeschiedenis houdt zich bezig met een opeenvolging
van theorieen. Gelet op Popper's interpretatie van een theorie als een
tentatieve oplossing van een probleemsituatie, betekent dit dat een
specifieke wetenschapshistorische geschiedenis moet worden gekarakteri-
seerd als een opeenvolging van probleemsituaties. Het proces van weten-
schap doorloopt dan vier fasen. De eerste betreft de formulering van een
probleemsituatie. De tweede geeft een tentatieve oplossing voor die
situatie. In een derde wordt die oplossing onderworpen aan (zowel logi-
sche als empirische) bekritisering, opdat duidelijk wordt in welke zin zij
tekortschiet. Volgens Popper leidt deze confrontatie tussen theorie en
kritiek in een vierde fase tot een volgende probleemsituatie, die door een
volgende theorie moet worden opgelost. In deel I van bet proefschrift
wordt een dergelijke opeenvolging voor de Oostenrijkse School gegeven,
en in deel II voor de Nieuw-Klassieke Economie. De methodologische
opvattingen van beide benaderingen worden in hoofdstuk 11 met elkaar
geconfronteerd.
In hoofdstuk 2 worden de theorieen uiteengezet die relevant zijn voor de

Oostenrijkse theoretische doelstellingen. Deze betreffen vooral de conti-
nentale conjunctuurtheorieen van Tugan-Baranowski, Spiethoff en Cassel,
die bet belang van de interestvoet voor de besparings- en investeringsbe-
slissingen van agenten benadrukken. Bovendien worden Wicksells onder-
scheid tussen de natuurlijke en marktinterestvoet en Mengers produktie-
structuur uiteengezet. Hoofdstuk 3 beschrijft Mises' samenvoeging van bet
Wickselliaanse onderscheid, Mengers kapitaaltheorie, en zijn eigen mone-
taire theorie. De resulterende conjunctuurtheorie verklaart cycli als de
reele effecten van monetaire verstoringen, waarbij die effecten vooral
bestaan uit investeringen in de verkeerde produktiegoederen en dus in
verliesgevende projecten (rna/investments). De onjuiste investeringen
verlengen de produktiestructuur, creeren daardoor op termijn een tekort
aan consumptiegoederen. Dit tekort zal ondernemers doen realiseren dat
zij onjuiste investeringen hebben gepleegd. Zij zullen hun investeringen
ombuigen van 'hogere' naar 'lagere' kapitaalgoederen en vooral consump-
tiegoederen. Deze ombuiging leidt volgens Mises tot een omslag in de
conjunctuurcyclus, tijdens welke de verliesgevende investeringsprojecten
worden geelimineerd.
In hoofdstuk 4 wordt Hayeks verbetering van Mises' theorie
gegeven. Deze verbetering betreft vooral de dynamisering van de analyse,
in de zin dat een intertemporeel raamwerk wordt aangenomen, waarbin-
nen verwachtingen en dus kennis een fundamentele rol spelen. Bovendien
maakte Hayek een expliciet onderscheid tussen individueel en algemeen
evenwicht. Dit maakte bet voor hem mogelijk om expliciet bet coordina-
tieprobleem te bekijken. De verklaring van de tendentie naar intertempo-
reel algemeen evenwicht, en daarmee coordinatie, werd bet centrale
probleem van de Oostenrijkse School. Acceptabele oplossingen van dit
Hayek Problem moeten voldoen aan drie methodologische voorwaarden,
te weten methodologisch individualisme, subjectivisme, en methodologisch
dualisme. Tezamen met bet Hayek Probleem vormen zij bet Hayek Pro-
gramma, dat in wezen een zoektocht naar onevenwichtigheidsdynamica is.
De volgende twee hoofdstukken beschrijven hoe volgende genera-
ties hebben geprobeerd bet Hayek programma uit te voeren dan wel te
wijzigen. In hoofdstuk 5 wordt vooral aandacht besteed aan Lachmanns
bestudering van de rol van verwachtingen. Hij verving Mises' en Hayeks
statisch subjectivisme door dynamisch subjectivisme, met als gevolg dat
menselijk handelen onbepaald wordt en niet een logica van de situatie
volgt. De resulterende onbepaaldheid van bet menselijk handelen impli-
ceert dat de uitkomst van bet marktproces onbepaald is. Ten gevolge
hiervan kan dan ook niet worden geconcludeerd dat het proces een
tendentie naar coordinatie teweeg zal brengen. Volgens Lachmann
bestond bet proces uit zowel coordinerende als discoordinerende tenden-
ties. Hij stelde daarom voor dat economen opeenvolgingen van discoordi-
natie-situatie zouden moeten bestuderen. Dit voorstel tot proces-analyse is
echter niet door andere Oostenrijkse economen overgenomen.
In hoofdstuk 6 wordt duidelijk gemaakt hoe de Oostenrijkse
revival in de jaren '70 leidde tot een analyse van bet voor bet marktproces
zo cruciale ondernemerschap. Kirzner combineerde daarbij dit Misesiaan-
se begrip met de Hayekiaanse visie op bet marktproces als een leerproces,
tijdens welke ondernemers discoordinaties in de vorm van winstmoge-
lijkheden ontdekken. De inherente fundamentele onzekerheid van onder-
nemerschap maakt bet echter onmogelijk bet concurrentieproces in een
deterministisch of stochastisch wiskundig model te incorporeren. Dit
verklaart dan ook waarom Oostenrijkers de Hayekiaanse conjunctuurthe-
orie niet in een dergelijk model hebben geformuleerd, al hebben niet-
Oostenrijkers als Thalenhorst en Wenig wel een aanzet in die ricbting
gegeven. Een dergelijke formalisering betekent echter dat Hayeks psycho-
logistische benadering van keuze-gedrag wordt vervangen door een bena-
dering die de nadruk legt op de logische aspecten van zulk gedrag, en
tevens een vervanging van zijn methodologische dualisme door een
methodologisch monisme.
Deel II bevat een rationele reconstructie van het Nieuw-Klassieke denken.

In hoofdstuk 7 wordt de oorsprong daarvan gevonden in de monetaristi-
sche verklaring van de Phillips Curve. Het incorporeert dus Friedmans
versie van de kwantiteitstheorie en diens 'Natural Rate'-hypothese. Deze
versie impliceert dat economische subjecten niet aan geldillusie lijden,
zodat de lange termijn Phillips Curve verticaal is. De negative belling van
de korte-termijn curve wordt verklaard door een vertraagde aanpassing
van verwachtingen. Aangezien de Nieuw-Klassieke Economie (NCE) de
Rationele Verwachtingen Hypothese (REH) aanvaardt, moet zij een
alternative verklaring voor die negatieve belling geven.
Hoofdstuk 8 beschrijft de ontwikkeling van dit Nieuw-Klassieke
alternatief. Lucas en Rapping (1969) formaliseerde Friedmans NRH in
hun 'Lucas supply function', die bet verscbil tussen natuurlijke en werkelij-
ke werkloosheid uitdrukt in termen van verwachtingsfouten. Lucas (1972a)
combineerde deze functie met de REH, zodat deze verwachtingsfouten
niet systematisch zijn. De 'random' fouten worden daarentegen verklaard
uit een gebrek aan informatie aan de zijde van de subjecten. Lucas
formaliseerde dit tekort met behulp van Phelps' eilanden-parabel, die stelt
dat informatie over de economie als geheel niet beschikbaar is. Hierdoor
was hij genoodzaakt bet partieel-evenwichtsmodel van Lucas and Rapping
(1969a) te vervangen door een algemeen-evenwichtsmodel, waarin
markten continu in evenwicht zijn door volledig flexibele prijzen.
De REH impliceert dat economische agenten alle bescbikbare
informatie zullen gebruiken in hun verwachtingvormingsproces, inclusief
die ten aanzien van bet door de monetaire autoriteiten gevoerde beleid.
Dit beleid wordt geformuleerd als een stochastische geldgroeiregel. Actief
stabilisatiebeleid betekent dan dat die regel feedback van huidige econo-
mische variabelen incorporeert. Economische subjecten zullen echter met
veranderingen in bet beleid rekening houden. Daar de gebruikelijke
methoden van beleidsevaluatie geen gebruik maken van de REH, leveren
zij een vertekend beeld op. Deze Lucas critique suggereert een relatie
tussen de variantie van monetair beleid en de belling van de korte-termijn
(onvolledige-informatie) Phillips curve, in de zin dat dit beleid minder
effectief wordt naarmate bet meer wordt aangewend. Lucas' (1973) test
maakte gebruik van deze implicatie. De Lucas critique werd ongeveer
terzelfdertijd en onafhankelijk van Lucas ontdekt door Sargent (1971).
Deze laatste auteur volgde echter een andere route dan Lucas, in de zin
dat hij zich concentreerde of wat Keynes de Gibson paradox had ge-
noemd. Hij verwierp Irving Fishers verklaring, die luidde in termen van
een vertraagde aanpassing van verwachtingen, daarbij de REH als een
criterium voor plausibiliteit gebruikend. Net als Lucas verklaarde ook
Sargent (1973a) verwachtingsfouten als bet resultaat van onvolledige
informatie. Zijn analyse leidde tot de conclusie dat Friedmans NRH en
Fishers verklaring van de Gibson paradox op hetzelfde neerkomen.
De NR/RE hypothese is zowel door Lucas (1973) als Sargent
(1973b) getest. Lucas' test leverde een gunstig resultaat op, met dien
verstande dat de persistentie in deviaties van de 'natural rate of output'
werden verklaard door een vertraagde waarde van reeel inkomen op te
nemen. De test-resultaten van Sargent (1973b) waren aanzienlijk minder
gunstig. Zij leidden echter niet tot de verwerping van de hypothese, maar
werden 'wegverklaard' door te wijzen op mogelijke problemen ten aanzien
van de initiele omstandigheden en/ of de ceteris paribus clausule. Boven-
dien concludeerde Sargent (1973b) op basis van een comparatieve test dat
er geen enkel ander model beter presteerde dan bet 'natural rate' -model.
Hoofdstuk 9 behandelt de Nieuw-Klassieke verklaring van bet
fenomeen 'persistentie'. Die persistentie betreft bet verschil tussen de
natuurlijke en werkelijke waarde van reele output (inkomen, werkloos-
heid). De Nieuw-Klassieke modellen uit hoofdstuk 8 namen een vertraag-
de waarde van die variabele op teneinde rekening te houden met persis-
tentie. Dit kan echter niet als een verklaring van bet fenomeen worden
gezien. Die verklaring werd gegeven door Lucas (1975) en diens incorpo-
rering van bet verschil tussen schokken en doorgeef-mechanismen, zoals
in de jaren '30 al onderkend door Slutzky en Frisch. Die mechanismen
bestaan uit voorraad-variabelen die zich niet direct kunnen aanpassen,
zodat zij de effecten van 'random'-verstoringen doorgeven naar volgende
perioden. Lucas (1975) model bevatte echter alleen lokale kapitaalmark-
ten, zodat kapitaalgoederen alleen van nut zijn op de markt waarin zij zijn
geinvesteerd. Barro (1980) veralgemeniseerde dit model door ook een
kapitaalmarkt voor de economie als geheel toe te staan. Monetaire
schokken hebben dan nog steeds tot reele effecten, mits economische
subjecten alleen een nominale interestvoet kunnen waarnemen.
Hoofdstuk 9 vat ook bet voorgaande samen in bet zgn. Lucas pro-
gramme. Dit bestaat uit bet Lucas Problem en de bijbehorende situatione-
le voorwaarden. Het eerstgenoemde poogt de cyclische relatie tussen
veranderingen in de geldhoeveelheid en deviaties van de 'natural rate of
output' te verklaren. Lucas (1972a} verklaarde dit verband in termen van
exogene schokken en doorgeefmechanismen. Zijn situationele voorwaar-
den kunnen net als die van Hayek in termen van methodologisch individu-
alisme en subjectivisme duidelijk worden gemaakt. De betekenis van deze
termen verschilt echter, in de zin dat de NCE 'slechts' bet gedrag van
representatieve individuen bestudeert, waarbij de REH systematische
verschillen tussen die individuen elimineert. Dit laatste betekent dat de
subjectivistische data van de economie worden 'geobjectiveerd'.
Hoofdstuk 10 behandelt een verschuiving in bet centrale pro-
bleem van de NCE, teweeggebracht door Sargent and Wallace (1975) en
Barro (1976). Zij toonden aan dat de NR/RE hypothese impliceert dat
systematisch monetair beleid geen reele effecten heeft, mits er geen
fricties (in de vorm van transactie- en andere aanpassingskosten) in bet
economisch systeem zijn, en mits de waargenomen nominale. prijzen niet
onderhevig zijn aan voor elke agent specifieke schokken. Het belang van
de propositie is echter vooral dat bet tot talrijke testen van de NR/RE
hypothese heeft geleid. Er zijn in deze twee benaderingen te onderschei-
den. De eerste volgt Sargents (1976a) test en leverde in bet begin gunstig
resultaat op. Sargents (1976b) 'observational equivalence'-resultaat gaf
echter meteen aan dat deze gunstige uitkomst was verkregen met een test
die weinig tot geen discriminerend vermogen bezat. Dit zou wel bet geval
zijn indien de test-periode een verandering in beleid zou omvatten. Neftci
en Sargent (1978) onderscheidden vervolgens zo'n verandering. Hun test
leverde weer een gunstig resultaat voor de NR/RE hypothese. De tweede
test-benadering, die van Barro (1978), gaf echter een tegengestelde
uitkomst te zien. Deze uitkomst werd versterkt door additioneel onder-
zoek, o.a. van Barro en Rush (1980) en Barro and Hercowitz (1980).
Deze discorroboraties waren echter gebaseerd op bet negeren van Neftci
and Sargents (1978) identificatie van een beleidsverandering. Barro en
andere Nieuw-Klassieken concludeerden desalniettemin dat empirisch
gezien monetaire verstoringen niet de oorzaak van conjunctuurcycli
konden zijn, vooral omdat de prijsrelatie een te zwakke schakel vormde
tussen onverwachte veranderingen in de geldhoeveelheid en bet reele-
outputniveau. King en Plosser (1984) gaven een alternatieve, reele
verklaring van conjunctuurcycli, die de causaliteit tussen geld en output
omdraaide. De meeste Nieuw-Klassieken zouden Barro volgen in zijn
overstap van monetaire naar reele conjunctuurtheorie.
Hoofdstuk 11 geeft een vergelijking van bet Hayek Programme en
bet Lucas Programme. Beide programma's blijken verschillende centrale
problemen te bestuderen. Het eerste richt zich op bet probleem hoe een
tendentie naar coordinatie tot stand komt, terwijl bet tweede aile coordi-
natieproblemen wegverondersteld. Dit verschil komt tot uitdrukking in
methodologische verschillen, die met name betrekking hebben op de ver-
klaringswijzen (genetisch-causaal versus mathematisch-functioneel) en de
in de analyses geincorporeerde typen kennis (objectief/subjectief versus
globaal/lokaal). Oostenrijkers volgen bovendien Hayeks 'compositieve
methode', met een psychologistische benadering van keuze-gedrag. De
verspreiding van psychologisch-subjectieve kennis over de samenleving als
geheel betekent dat economische theorieen de sociale werkelijkheid niet
volledig kunnen verklaren. Oostenrijkers concluderen daarom dat sociale
wetenschappers slechts verklaringsschetsen van sociale verschijnselen
kunnen geven. Nieuw-Klassieken volgen daarentegen Poppers methode
van rationele model-constructie, die de logische aspecten van keuze-
gedrag benadrukt. Aangezien de gissingen omtrent de dit keuze-gedrag
niet empirisch geldig hoeven te zijn, spelen empirische testen een rol. De
methode betekent bovendien dat Nieuw-Klassieken geen verschillen zien
tussen de complexiteit van natuur- en sociale verschijnselen. Als een
gevolg hiervan is er geen reden waarom verklaringen van de laatste niet
net zo volledig kunnen zijn dan die van de eerste. Oostenrijkers en
Nieuw-Klassieken richten zich dus niet aileen op verschillende sociale
verschijnselen, maar pogen die ook op een andere manier te verklaren.
De verschillen in hun respectievelijke centrale problemen en methoden
van ontdekking en rechtvaardiging Iaten zien dat de 'common-roots claim'
niet kan worden gehandhaafd.
The Tinbergen Institute is the Netherlands Research Institute and Graduate School for General and
Business Economics founded by the Faculties of Economics (and Econometrics) of the Erasmus
University in Rotterdam, the University of Amsterdam and the Free University in Amsterdam. The
Tinbergen Institute, named after the Nobel prize laureate professor Jan Tinbergen, is responsible for the
PhD-program of the three faculties mentioned. Since January 1991 also the Economic Institute of the
University of Leiden participates in the Tinbergen Institute.
Copies of the books which are published in the Tinbergen Institute Research Series can be ordered
through Thesis Publishers, P.O. Box 14791, 1001 LG Amsterdam, The Netherlands, phone: +3120
6255429; fax: + 3120 6203396.
The following books already appeared in this series:
Subseries A. General Economics
no. 1 Otto H. Swank, "Policy Makers, Voters and Optimal Control, Estimation of the Preferences
behind Monetary and Fiscal Policy in the United States". ·
no.2 Jan van der Borg, "Tourism and Urban Development. The impact of tourism on urban
development: towards a theory of urban tourism, and its application to the case of Venice,
Italy".
no. 3 Albert Jolink, "Liberte, Egalite, Rarete. The Evolutionary Economics of Leon Walras".
no. 5 Rudi M. Verburg, "The Two Faces of Interest. The problem of order and the origins of
political economy and sociology as distinctive fields of inquiry in the Scottish Enlightenment".
no.6 Harry P. van Dalen, "Economic Policy in a Demographically Divided World".
no. 8 Matjan Hofkes, "Modelling and Computation of General Equilibrium".
no. 12 Kwame Nimako, "Economic Change and Political Conflict in Ghana 1600-1990".
no. 13 Ans Vollering, "Care Services for the Elderly in the Netherlands. The PACKAGE model".
no. 15 Cees Gorter, "The dynamics of unemployment and vacancies on regional labour markets".
no. 16 Paul Kofman, "Managing primary commodity trade (on the use of futures markets)".
no. 18 Philip Hans Franses, "Model selection and seasonality in time series".
no. 19 Peter van Wijck, "Inkomensverdelingsbeleid in Nederland. Over individuele voorkeuren en

distributieve effecten".
no. 20 Angela van Heerwaarden, "Ordering of risks. Theory and actuarial applications".
no. 21 Jeroen C.J.M. van den Bergh, "Dynamic Models for Sustainable Development".
no. 22 Huang Xin, "Statistics of Bivariate Extreme Values".
no. 23 Cees van Beers, "Exports of Developing Countries. Differences between South-South and
South-North trade and their implications for economic development".
no. 24 Lourens Broersma, "The Relation Between Unemployment and Interest Rate. Empirical
evidence and theoretical justification".
no. 26 Michel de Lange, "Essays on the Theory of Financial Intermediation".

no. 27 Siem-Jan Koopman, "Diagnostic checking and intra-daily effects in time series models".
no. 28 Richard Boucherie, "Product-form in queueing networks".
no. 29 Frank A. G. Windmeijer, "Goodness of Fit in Linear and Qualitative-Choice Models".
no. 30 M. Lindeboom, "Empirical duration models for the labour market".
no. 31 S. T.H. Storm, "Macro-economic Considerations in the Choice of an Agricultural Policy".
no. 33 R.W. van Zijp, "Austrian and New Classical Business Cycle Theories".
Subseries B. Business Economics
no.4 Rob Buitendijk, "Towards an Effective Use of Relational Database Management Systems".
no. 7 P .J. Verbeek, "Two Case Studies on Manpower Planning in an Airline".
no.9 T.C.R. van Someren, "Innovatie, emulatie en tijd. De rol van de organisatorische vernieuwin-
gen in bet economisehe proces".
no. 10 M. van Vliet, "Optimization of manufacturing system design".
no. 11 R.M.C. van Waes, "Architectures for Information Management. A pragmatic approach on
architectural concepts and their application in dynamic environments".
no. 14 Shuzhong Zhang, "Stochastic Queue Location Problems".
no. 17 Paul Th. van de Laar, "Financieringsgedrag in de Rotterdamse maritieme sector, 1945-1960".
no. 25 E. Smeitink, "Stochastic Models for Repairable Systems".
no. 32 H.E. Romeijn, "Global Optimization by Random Walk Sampling Methods".

Rudy Van Zijp - Austrian and New Classical Business Cycle Theories - A Comparative Study Through The Method of Rational Reconstruction-Edward Elgar Pub (1993)

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Rudy Van Zijp - Austrian and New Classical Business Cycle Theories - A Comparative Study Through The Method of Rational Reconstruction-Edward Elgar Pub (1993)

Hochgeladen von

Copyright:

Verfügbare Formate

AUSTRIAN AND NEW CLASSICAL

BUSINESS CYCLE THEORIES

AUSTRIAN AND NEW CLASSICAL

ter verkrijging van de graad van doctor aan

Rudolf Willem van Zijp

Thesis Publishers Amsterdam

PART I. AUSTRIAN BUSINESS CYCLE THEORY

2. EARLY VERTICAL MALADJUSTMENT THEORIES 11

PART II. NEW CLASSICAL ECONOMICS

7. MONETARIST ROOTS OF NEW CLASSICISM 135

PART Ill. A COMPARISON

11. AUSTRIAN VERSUS NEW CLASSICAL ECONOMICS 245

NAME INDEX 295

SUBJECT INDEX 299

1.1. THE AIM OF THE STUDY

In the last two decades economics has experienced a remarkable shift in

1.2. THE METHOD OF THE STUDY

Following Popper (1972, p. 164), a theory is a tentative solution to some

1.3. THE STRUCTURE OF THE STUDY

The study is organized as follows. Part I consists of five chapters, which

One of the earliest 'vertical-maladjustment business cycle' theorists was

Arthur Spiethoff was strongly influenced by Tugan-Baranowski. His

Cassel (1918 (1923)) built on Spiethoffs over-investment theory. He

2.5. THE STRUCTURE OF PRODUCTION4

The early business cycle theories under consideration distinguished

4 The purpose of this section is not to give a complete account of controversies in

5 It should be noted that Bohm-Bawerk (1889 (1921}, pp. 39 - 43) distinguished

6 Blaug (1962 (1990), p. 508) formulated Bohm-Bawerk's (1889 (1921), p. 118)

Knut Wicksell ( 1898) distinguished between two ·different notions of inte-

17 It may be argued that Hayek's (1928) notion of intertemporal equilibrium is an

Ludwig von Mises provided three main contributions to Austrian econ-

3.1. PRAXEOLOGY AND ECONOMICS

3.2. MONETARY THEORY

3.3. BUSINESS CYCLE THEORY

Mises's contribution to the Austrian theory of the business cycle is that he

16 This effect was discussed at the end of the previous chapter.

3.4. MISESIAN METHODOLOGY

3 .4.1. Methodological individualism

3.4.2. Methodological singularism

3.4.3. Radical subjectivism

4.2. INTERTEMPORAL EQUILIBRIUM

4.3. HAYEK'S EARLY BUSINESS CYCLE THEORY

4.4. THE COORDINATION PROBLEM

4.5. CAPITAL, INVESTMENT, AND DISCOORDINATION

4.6. THE DEBATE ON THE RICARDO EFFECT

In 1939 Hayek published his Profits, Interest and Investment, a collection of

4.6.2. The debate on the Ricardo effect

4.6.3. The re-opening of the debate: Hicks's 'The Hayek story'

27 For instance, see Sraffa (1932) and Hawtrey (1932).

To summarize, Hayek's business cycle theory is a theory of the effects of

4.7 THE HAYEK PROGRAMME

In the early 1940s, Hayek tried to convince his fellow-economists of the

4.7.1. The Hayek Problem

31 Therefore, it is perhaps better to speak of methodological situationalism. This

4.7.4. The compositive method

36 'Understanding' aims to obtain knowledge about other minds, whereas 'explaining'

concerns social interaction.

41 Such an explanation subsumes an explanandum (phenomenon to be explained)

4.8. COMPETITION, LEARNING AND THE HAYEK

Interpreting social phenomena as unintended consequences of individual

Hayek elaborated Mises's business cycle theory by replacing its bench-

The formulation of the methodological constraints of the Hayek pro-

5.2. INNOVATION AND THE AUSTRIAN BUSINESS