University of Groningen Conceptual progress in economics

University of Groningen

Conceptual progress in economics. Abstraction of social kinds versus idealizationRol, Menno Eugen Gerardus Maria

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2007

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Rol, M. E. G. M. (2007). Conceptual progress in economics. Abstraction of social kinds versus idealization.s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-amendment.

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 08-02-2022

CONCEPTUAL PROGRESS IN ECONOMICS

Abstraction of social kinds versus idealization

Doctoral Dissertation for Philosophy Conceptual Progress in Economics

Menno Rol

University of Groningen Faculty of Philosophy

Oude Boteringestraat 52 9712 GL Groningen

Research supported by University of Groningen and BCN

Printed in the netherlands by Ridderprint Offsetdrukkerij BV, Ridderkerk

ISBN 978-90-367-3169-0

RIJKSUNIVERSITEIT GRONINGEN

CONCEPTUAL PROGRESS IN ECONOMICS

ABSTRACTION OF SOCIAL KINDS VERSUS IDEALIZATION

Proefschrift

ter verkrijging van het doctoraat in de

Wijsbegeerte

aan de Rijksuniversiteit Groningen

op gezag van de

Rector Magnificus, dr. F. Zwarts,

in het openbaar te verdedigen op

donderdag 27 september 2007

om 14.45 uur

Menno Eugen Gerardus Maria Rol

geboren op 10 juni 1960

te Zaandam

Promotores: Prof. dr. T. A. F. Kuipers Prof. dr. U. Mäki Prof. dr. A. Nentjes Beoordelingscommissie: Prof. dr. N. L. Cartwright Prof. dr. E. C. W. Krabbe Prof. dr. E. M. Sent

ISBN 978-90-367-3169-0

Acknowledgements

Ideally, a PhD student is never solitary. Inquiries by the Groningen Association of

PhD students, GrASP, show that many who are writing their doctorate thesis have

few contact hours with the supervisors and are not adequately helped when in

trouble or in doubt. As a result of this, they give up. Their situation is far from

ideal. I am surprised to learn about this situation. I do not need to idealise to de-

scribe my own situation as ideal. It has been nothing less than perfect all along,

thanks to the wonderful organization of and great atmosphere in the Theoretical

Philosophy group in Groningen. This is a determining factor of its quality.

Theo Kuipers commented on complex and lengthy texts in a very short

amount of time, and meticulously so. We agreed to meet on a very regular basis at

first, but soon I saw that I had no desire to talk this frequently. The reason was that

Theo is so accessible that I have felt he was always around the corner if I needed

his thorough knowledge and wide experience. I would advise any student to

graduate under his supervision if possible, even if that means changing the topic of

the thesis. But then again, it is probably not necessary to pay such a price, for Theo

is acquainted with so many areas in natural and social science, philosophy, and

mathematics, that it is hard to choose a theme outside his grand vision. Theo de-

serves my gratitude for so many other things, some of which include his encour-

agement given when I needed it and his pleasant receptions at his house.

But there has been more to contribute to a successful completion of the task.

The PCCP is a wonderful institution enabling the in-house pre-testing of our ideas

and papers. It is a forum where everybody is taken seriously, and where nobody

fears to be mocked for a mistake. It is a safe place. All participants, student or pro-

fessor, go to the meetings to learn something. Papers are criticised before publica-

tion, presentations tried out before leaving for the congress. If possible, I shall try

and introduce a similar forum in the places where I will be working.

The faculty offered me a room with a view, more specifically to the academic

garden and into the hall where a zillion promotion celebrations went on before

mine. It is a pleasant place, as Jan-Willem now discovers. Hauke de Vries was al-

ways agreeable if I needed him, even if it was for my own laptop. Daan Franken

was an inexhaustible oracle for questions about English and for keeping my emo-

tional household in order. Barteld Kooi was the oracle for questions on logic. They

made the corridor breathe.

David Atkinson and Jeanne Peijnenburg have had a special role in my de-

velopment. David’s unusually systematic answer to any question one might ask

him give a sort of background solidity to our group from which all PhD students

draw confidence: that no matter the kind of puzzle, everything will eventually turn

out all right, really. Jeanne taught me that not immediately getting published,

when still green in science, need not establish my worthlessness. She made me see

that one can submit a first paper to a sub-top journal also and aim for top journals

later. Jeanne serves as the glue of our group in many ways too.

Outside the faculty, I encountered all the help I needed with my other two

supervisors. It was a fortunate pick to ask Andries Nentjes. He took all the avail-

able time to expose my mistakes in the history of economics in order to protect me.

He had a lot of time available, when not given a deadline; his comments were

prompt, when given a deadline. Hence the planning was in my own hands. It was

also my own choice to ask Uskali Mäki for supervision. He jumped at my proposal

to be involved with my project and never hesitated to fire his deep criticisms onto

me when so needed. He gave me the feeling that if I could get past his thorough

critique, I really had something valuable on paper. His advanced course on the

philosophy of economics counts among the few very best I was ever taught. The

Erasmus Institute for the Philosophy of Economics (EIPE) was my second home

faculty in the first three years; a bastion of internationalization and xenophilia, a

stronghold of forbearance, thus challenging the counterproductive ‘immigration’

policies as I stood facing how these damaged my country so much during the po-

litically grim years when I was writing my dissertation.

But there was more on the cheery side. Jan-Willem Romeyn has grown into

an intimate friend. Over the years, I shared many thoughts about life in its broad-

est sense. Sometimes we were young boys, who talk of girls and cars; often we

were serious, contemplating matters in love and happiness, all in the same gasp of

air with philosophical issues. Pieter Sjoerd Hasper has become a friend of me and

my wife and children. He manages to maintain a multifocus on mathematical is-

sues, on classical languages, Frisian, Dutch, Italian, French, German, and English,

on etymology, philology, and history, on chess, on literature, on my son; it is easy

to forget he also knows his philosophy. Moreover, he is a gracious guest at the

dinner table. In addition, he enabled me to considerably improve the paper for the

Journal of Economic Methodology and, with this, chapter III. I hope the end of my stay

at the faculty will not reduce the frequency of my contact with Jan-Willem and

Pieter Sjoerd. I want to thank these warm people.

There are others who deserve my special gratitude. First are Sheila Dow and

Eric Schliesser who commented on a long chapter that I decided to discard thanks

to them. Next, I am very grateful that Erik Krabbe took an enormous effort and

worked his way through all of the text so scrupulously. His comments have been

unusually detailed; he has helped improve both the philosophical quality and the

readability of the entire text. I am also obliged for his proof in chapter III. Further, I

much appreciate Nancy Cartwright’s invitation to give a talk at the London School of

Economics and her swift reading of the dissertation when the time schedule became

tight. Lastly, I thank Esther-Mirjam Sent for her willingness to read it during the

time when she was on parental leave.

I wish to further express my appreciation to Gerda Bosma for being inter-

ested both in my person and in my project; to Boudewijn de Bruin who made me

think once again (and again) about my chapter III; to John Davis for willing to

share my table for the Thai dinner in Amsterdam and for his many encourage-

ments at several conferences; to John Dupré for another Thai dinner, in Ghent; to

Ben Gales for supervising my masters thesis and for keeping contact afterwards; to

Martin van Hees for his incessant encouragement and his criticism to the first draft

of the paper on abstraction and idealization; to Harold Kincaid for discussing my

comments on one of his most interesting books; to Hans Radder for always taking

my questions on his work seriously; to Don Ross and Nelleke for being so sweet

and for organizing a very nice party at their house in Alabama; to Evert Schoorl for

dragging me into the economics faculty in Groningen; to Allard Tamminga for

always showing interest in my work; to David Teira for the dinner in San Se-

bastián; to Job Zinkstok for taking the time to criticise the Dutch summary.

The whole project has been made possible by the ‘parelsubsidie’ grant offered

by the executive board of the University of Groningen as a prize to the Theoretical

Philosophy group for the top score on all items the research visitation committee

had judged over the years 1996 to 2000. I am honoured that Theo spent it on me

and I want to thank the University for providing this grant.

The research school Behavioural, Cognitive, and Neurosciences (BCN) fi-

nanced courses, part of a visit to Don Ross’ and Harold Kincaid’s congress at the

UAB, and part of the printing costs. Some of the BCN courses taught me a lot of

neurology – as I once thought, perfectly irrelevant for economists. But Don Ross

has convincingly shown in his (2005) that explaining the brain is urgent for eco-

nomics too.

Both Louwke and me dislike the subservient role wives implicitly get when

thanked for the support and the endurance. Such a qualification would underrate

what her presence in my life means. From the start, we decided together that the

project was worth while. It has turned out that we made the right decision.

Preface

Here I comment on some choices I have had to make when writing this thesis.

On ‘model’.

Throughout the thesis I often use the terms ‘model’ in a loose sense, except in

chapter III. Economists tend to use ‘model’ and ‘theory’ interchangeably. In the

context of problems concerning truth and falsity, this is problematic. A theory has

truth value, models have not. The fact of the matter is this. A theory – if conceived

of as a set of propositions – can be true or false depending on the truth value of its

explicit and implicit propositions. Insofar as a theory describes (and explains) a

reality which differs from the actual world (in its past, present, and future states) it

is false simpliciter. But this theory is of course necessarily true of the world it hap-

pens to describe. Chapter III makes use of some concepts from model theory,

which sees possible worlds as models of propositions about possible states of af-

fairs. The set of models of which a proposition (theory) is true belongs to the exten-

sion of the proposition (theory). In this sense it is a category mistake to talk of the

truth or falsity of models. In other contexts, however, when economists use the

term ‘model’ in order to denote a theoretical hypothesis, one can safely attach truth

values to models.

I do not want to speak of truth or falsity as a property of models in any con-

text. But I do also not strictly distinguish between ‘theory’ and ‘model’ if the con-

text does not require so. For this dissertation this means that I use the terms rather

loosely in the first two chapters, only to upgrade the stringency of the distinction

when it becomes necessary, viz. after the intermezzo.

On ‘description’ and ‘explanation’.

A sometimes useful distinction of scientific practices is that between descriptive

and explanatory sciences. Anatomy is an example of a descriptive discipline. So is

a merely quantitative sociology that tries to acquire data on suicide in a number of

countries. Chemistry is an example of an explanatory science. So is economics.

Notwithstanding the rigour suggested by this distinction, there is a subjective as-

pect to it. Suppose nobody knew, at a certain point in time, that metals conduct

electricity, and that copper is a metal; the explanation that metals conduct electric-

ity and that copper conducts electricity because it is a metal considerably increases

understanding. A critic of Hempelian deductive-nomological explanatory schemes

would say that this clarification is hardly elucidating. But under the circumstances

given here it is. The point is that descriptions may strike the learner as explanatory

if his background knowledge is sufficiently poor. Initial descriptions may be in-

structive while these are felt unsatisfactory at a later point in the development of

knowledge. The distinction between description and explanation is perhaps pretty

sharp analytically, the difference is fluid in practice. In this thesis the distinction is

not very important. Especially where it discusses (social) kinds, the idea is that

describing essential objects (properties, relations) is an explanatory undertaking.

I cling to documenting this because economists sometimes make the distinc-

tion in order to set apart rich descriptions of reality from slender theoretical hy-

potheses. Here the concept of abstraction turns up again. Theorists claim that ab-

straction is the best way to understand (economic) reality, while historicist econo-

mists are said to distrust theory and favour the collection of reliable data.

This view of things is, firstly, a caricature of the possible ways to do social

research1. But, secondly, it misses an important point as well. There is no object of

research, natural or social, of which anyone can make a complete description. What

would completeness amount to? The molecular make-up of the Central Bank

president? And would more abstraction also imply more explanation, as the carica-

ture suggests? What counts as explanatory is what increases understanding. As

stated, this is a contextual matter, and the contexts that codetermine it include the

research interests and the particular capabilities of the scientist: clearly subjective

determinants.

What I can say is this. If we accept the analytical distinction, it is clear that an

explanans is always in need of an explanandum, and the explanandum cannot be

denoted without describing it (at least partially). Explanatory sciences are descrip-

tive by their very nature; descriptive sciences need not be explanatory. As I do not

need it to make my points in this thesis I largely ignore the distinction.

On gender.

I noticed that male scientists use the female form to refer to a person in their texts

more often than female scientists do. But this impression is strongly biased, for I

know the work of more male than female economists and philosophers. This phe-

nomenon, then, is precisely what drives men to sometimes talk of scientists in gen-

eral by using the word ‘she’. They see it as fair to women in a time when sex dis-

crimination in the labour market (and elsewhere) is only starting to cease, if at all.

Better, they feel that they thus partake in the fight against such discrimination,

however modestly. Especially in the Netherlands this seems to be appropriate as

for the Dutch a repair of this waste of talent is long overdue. The proportion of

1 Precisely this is Böhm-Bawerk’s central claim in his (1890b) review of Schmoller’s Festgabe for the

founder of the historical school, Wilhelm Roscher.

women in higher positions is scandalously low in this country, and emancipation

is slow.

However, I believe, somewhat hesitantly, that an occasional ‘she’ and ‘her’

in academic texts is not going to change anything about it. This is a contested field,

and I am perchance wrong. I am on the other hand certain that, from a linguistic

point of view, it is an ugly use. I do not endorse it. When I use the words ‘he’ and

‘his’ in general, I refer to men and women alike.

On orthography.

For non-native speakers of the English language, there is the choice between Brit-

ish and American English pronunciation and spelling. Logically, the options in

orthography is available for an Englishman or a US citizen too, but for me the

choice has some stakes. My education is firmly grounded in Dutch practices in the

nineteen seventies. That is the time that secondary school pupils were not sup-

posed to have a choice. Decent English, I was taught, is European. Looking back,

this attitude now strikes me as silly, and nowadays education is, I believe, more

liberal in these matters. Still, a British orientation toward the English language has

been chiselled into my personality. For me no option for American custom exists

without a high cost. So I use British spelling in this dissertation.

I would not have dwelled on this seemingly unimportant point if it had not

confronted me with practical problems. I thought that the only thing I had to ob-

serve was to be consistent, but that turned out less easy as I had thought. ‘Idealisa-

tion’ is British, but hardly a Briton uses it. Yet, the verb ‘realise’ is in normal use in

the Kingdom. What to do, given that no language is used consistently anywhere

outside the classroom? I noted that all ‘—zations’ and other ‘—za’s’ are habitually

written with a ‘z’ by most Britons, even though the use of the ‘s’ instead is not

wrong. I have decided to follow this practice, not in the least because David Atkin-

son told me he does it too. To me, he is the ultimate authority in many Indo-

European languages, in mathematics, and in Indian cooking.

Finally, some apparent German spelling mistakes in quotes are not such, but

Böhm-Bawerk’s nineteenth century Austrian idiosyncratic way of writing.

On the philosophy of science as ‘methodology’

‘Methodology’ is a word often used in the philosophy of economics. Possibly this is

to immunise the discipline against the charge that it is practised by failed econo-

mists. ‘Methodology’ sounds more creditworthy to people who do not profession-

ally reflect on economics in the way philosophers do. It makes sense to distinguish

clearly between the study of rules of method of research and philosophy. But in

this thesis, the distinction is not important. I do nowhere discuss method. In con-

sequence, I use the words interchangeably.

Contents

Introduction 1

I Time as a unifying concept

1 Introduction 7

2 The structure of the Positive Theorie des Kapitales 8

2.1 Judging Böhm-Bawerk 8

2.2 A pyramid-like structure 8

2.3 Dropping assumptions and claiming truth 10

2.4 The key assumptions 12

3 Böhm-Bawerk’s concept of time in economic theory 16

3.1 Capital building: A Robinsonade 16

3.2 A more complex economy 20

4 Subjective Value Theory 22

4.1 Marginal utility as a measure for value 22

4.2 The market 24

4.3 Objective value theory: true as a limiting case, otherwise false 27

5 A new conception of the present and the future 29

5.1 The central idea 30

5.2 The three causes 31

6 Interest and capital yield 34

6.1 Interest 34

6.2 Capital yield: durable goods in general 35

6.3 Capital yield: the labour market and the subsistence market 39

6.4 Positive interest 40

7 Interest and income distribution 42

7.1 The level of interest in the idealised case 42

7.2 The level of interest in the de-idealised case 46

8 Conclusion and summary 49

II Realism and explanation

1 Introduction 51

2 SVT as a unifying tool 52

2.1 The Law of Costs and the Law of the Marginal Agents 52

2.2 Layman’s observation and empirical justification 56

3 Essentialist metaphysics 58

3.1 The ordering principle of the phenomena 59

3.2 A unified order 60

3.3 Essentialism, reduction, abstraction 62

4 Discrete narratives versus continuous mathematics 65

4.1 Böhm-Bawerk’s interest theory in Wicksell’s Wert, Kapital und Rente 66

4.2 The limits of a discrete analysis 69

4.3 The demand curve of capital 72

5 The principle of sufficient complexity 74

5.1 Hypothetical worlds in the theory of the market 75

5.2 Narrativism. A concept from historiography 77

5.3 Highlighting the mechanics from micro to macro. Sufficient complexity 81

6 Conclusion 82

Intermezzo 85

III Abstraction and idealization

1 Introduction 101

2 Isolations: a simple taxonomy 102

2.1 Clauses concerning the ceteri 102

2.2 Horizontal and vertical reasoning 103

2.3 Abstraction and its use in explanatory theories 104

3 Idealization 105

3.1 From Boyle to the van der Waals equation – and back again 105

Three Gas Laws: progressed de-idealization 106

Idealization in the gas laws: the issue of falsity 107

3.2 Idealizational conditionals as conditional implications 110

3.3 Economics and policy 113

Idealizations: truth and external validity 113

External validity of true counterfactuals and economically possible worlds 117

Economically impossible worlds. The case of capital structure 118

4 Abstraction 120

4.1 Abstraction as existential generalisation 120

4.2 A concise antithesis of idealization and abstraction 121

4.3 Abstraction and policy relevance 123

5 Vague qualifications 125

5.1 Hausman: Vague laws as inexact generalisations 126

5.2 Abstraction as focussing in scientific practice 127

5.3 The ‘Social Model’ 128

5.4 Böhm-Bawerk: what closure is it that explains the explanandum? 130

5.5 Abstraction ante explicationem and abstraction post explicationem 132

6 Conclusion 134

IV Four Examples

1 Introduction 137

2 Wicksell’s stationary state 137

2.1 Building on Böhm-Bawerk: rent 138

2.2 Land as capital: the insertion of heterogeneity 139

2.3 Simple interest as a dispensable assumption 143

2.4 Statics versus dynamics, stationary state versus long run growth 145

The stationary state: condition for static analysis or limiting case? 145

Idealization of the stationary state. 146

From de-idealization to abstraction 147

3 Keynes on involuntary unemployment. The case of pure theory. 148

3.1 The neoclassical model 149

3.2 The Keynesian model 151

3.3 Idealization and abstraction in the analyses 152

4 The Dutch market for health, care, and cure. 154

4.1 Inefficiencies in the health care market 155

4.2 Health care and ‘the social model’ 156

5 The economics of environmental governance 158

5.1 The adverse effects of ethical food in your trolley 159

5.2 Idealizational assumptions and their policy implication 161

V Abstraction, vagueness, and social kinds

1 Introduction 163

2.1 Is essentialism tenable? 164

2.1 The theory of extensible concepts: essentialism is not tenable 164

2.2 The theory of redescription: essentialism is there 168

2.3 The theory of exact types: essentialism in economics 169

3 Essentialism is tenable, but not for social science 171

3.1 The theory of recognitional capacities: essentialism is tenable 172

3.2 Our metaphysical appreciation 174

3.3 Exact types and social science 176

4 Essentialism is tenable, also for social science 177

4.1 Kind terms specified de jure 178

4.2 The exaggerated demands on the Causal Theory of Meaning 182

4.3 Vagueness of kind terms 182

Essentialism and pluralism as concordant 183

Minimal concepts, rigid designation de jure, and kinds 185

5 Social kinds as structures of the social 186

5.1 The idea of an actual causal process 187

5.2 The idea of an hypothetical causal mechanism in Austrian economics 190

5.3 Kind terms for social structures 191

Essentialism, not infallibilism 191

Market structures 192

6 Conclusion 194

Conclusion 197

Appendices 201

References 215

Summary 223

Index of names 227

Samenvatting voor groter publiek 229

Introduction

Questions

I got interested in Austrian economics as I taught textbook economic theory to

secondary school children. They would ask me questions I could not answer well

enough, that is, not to my own satisfaction. Why would any entrepreneur try to

engage in risky activities if the end state of perfect markets is zero profit anyway?

The simple answer I gave was that the world is not perfect and that therefore en-

trepreneurs are lucky bastards. Most pupils unconsciously favour socialism by

intuition, so to them it sounded plausible enough. But occasionally a pupil would

ask why a world would be judged ‘ideal’ if no one could earn any profit in its per-

fect markets. How could an ideal state provide for the necessary incentives to in-

novate, to serve customers, and

what is the point of corporate tax without profit? Why, indeed, having corpora-

tions or smaller enterprises anyway if the ideal is this type of perfection?

My own further puzzlement ensued about economic theory. This happened

only after such pupils (who sometimes struck me as brighter than myself) had

posed these deep questions about the economic order. Why do free market econo-

mists describe markets, in consequence, as perfect if the allocation proceeds under

a regime of complete lack of any incentives. The presence of incentives provides a

necessary condition for capitalism; so why qualify that as imperfect?

The Austrian mechanistic approach to economics gives an answer to these

questions. The classical approach made a black box of the mechanism underlying

market phenomena, but the Austrians opened it. The classical theory disregarded,

by ‘assuming away’, as it were, the question of how people deliberate. Ludwig von

Mises and Friedrich Hayek asked questions about the dynamics of decision mak-

ing by individual agents if these entered trade. Their forerunners, Carl Menger and

Eugen von Böhm-Bawerk, had spelled out the subjective basis of the market proc-

ess and applied the subjective value theory to capital and interest, respectively.

Richard Cantillon may have been the first economist to stress the entrepreneurial

aspect of the market process. (Surprisingly, Menger, in his turn, had Irish intellec-

tual descent, as Cantillon had emigrated to France from Ireland.) Further back, the

Scholastic philosophers had asked questions about the role of human action in a

social whole. They were the followers of St. Thomas Aquinas who had discovered

‘competitive price’ (and his own version of the Law of Costs1).

As we shall see, the subjective approach first focuses on what is directly

visible to anyone. It starts with phenomena as these appear familiar to the individ-

ual and with personal intentions, visible by introspection. Next comes an explana-

tory story about the deeper lying mechanism. This raises the question of how a

1 Schumpeter (1954) p.93. See chapter II for an elaboration on Böhm-Bawerk’s criticism of the Law of

Costs.

Introduction

theory about the mechanism can be empirical. The mechanism, which describes

how human action translates into market outcomes, does not seem to be empiri-

cally accessible in any easy way, either to a layman or a scientist. This is different

with the (quasi-)observable results of the mechanism, however, like objective (i.e.

market-) prices, profits, the distribution of wealth and poverty. These can be

known by lay observation. So how can an Austrian theory be at the same time sci-

entific and empirical? My thoughts ran straight into a Methodenstreit.

Related questions focus on explaination and prediction. In ‘neoclassical phi-

losophy of science’ empirical tests are the alpha and omega of theory appraisal. It

seemed to me that a mere empirical approach to comparative theory appraisal

would not help to demarcate the truth content from the falsity content of Austrian

explanatory theories. Does ‘explaining better’ also mean ‘being more truthlike’? Is

it possible that the theory, which explains less satisfactorily, nevertheless is more

truthlike? Is the concept of ‘explanatory progress’ – in contrast to ‘predictive pro-

gress’ – epistemologically fruitful in economics?

The first question this thesis deals with is what makes Austrian theory so dif-

ferent from classical and (much of) neoclassical theory. After all, all three lines of

research seem to bring forth intellectuals with some degree of (neo-)liberal political

orientation. Clearly, attention for the ‘economic mechanism’ is what sets Austrians

apart. Böhm-Bawerk probably was the most polemical defender of the method. He

engaged in lengthy narratives – as I call these – about the reasoning steps an entre-

preneur must follow if his business is to survive. He did so at a price: Knut Wick-

sell and many other later economists looked down on his apparent lack of mathe-

matical dexterity. But I dismiss Wicksell’s criticism. I believe (and I hope to con-

vincingly show) in chapter II that Böhm-Bawerk was able to insert a fair amount of

complexity into his economics in order to highlight an aspect of economic life,

which would otherwise have been ‘assumed away’, that is, remained exogenous.2

Inspired by the Mengerian emphasis on entrepreneurial action, he aimed at a level

of complexity so as to endogenise, for instance, the demand for capital.

My choice for Böhm-Bawerk as protagonist of mechanistic reasoning stems

from my original idea to track down progress in economists’ concept formation.

Böhm-Bawerk had introduced time as a constitutive variable into economic theory,

such that positive interest had to be seen as inherent in all trade and in the invest-

ment period needed to produce something ‘capitalistically’. A stricture of his inter-

est and distribution theory is that his model economy had to be in a stationary

state. But the stricture was helpful, because many had thought that interest was

inherent only in a growing economy (like for instance Jevons – the ‘English

Böhm’). Thanks to Böhm-Bawerk we know that interest is positive even when eco-

nomic development is at a standstill. Wicksell, in turn, had used and praised this

idea, but wanted to make the theory dynamic. The Swedish effort to insert dynam-

2 The expression ‘to assume away’ for disregarding variables by the introduction of a clause is some-

times used in economic methodology. It rightly reveals the deliberateness of the disregard. This is

the reason why I shall copy this use throughout this thesis.

question

ics into the picture is a clear attempt to progress on Böhm-Bawerk. The original

idea, then, was to get an insight into the cognitive patterns involved in such a pro-

gressive step. But it turned out too ambitious for my mainly philosophical thesis,

as I stranded in highly complex Cambridge controversies and capital reversal

problems.

What remains: a scrutiny of Böhm-Bawerk’s magnum opus as a statue of

early Austrian analysis and Wicksell’s interpretation of it; an inquiry into the epis-

temological implications of their interaction in the light of the Methodenstreit; an

evaluation of the apparent essentialist inclinations related to abstraction in Aus-

trian thought; an investigation of how issues of truth and falsity in abstract and in

idealizational theorising may bear on policy issues; and finally, an endeavour to

stretch the feasibility of essentialism in social science up to the point where it is on

the brink of losing plausibility.

The way in which we can resolve issues of the truth of merely explanatory

theories has a bearing on epistemology. In order to cope with the inherent com-

plexity of the socio-economic world scientists exercise two strategies. They (1) ab-

stract from initial descriptions of this world – that is, they theorise – and they (2)

hedge their theorems with clauses. The clauses somehow restrict the domain of

application of the theorems, very often to ideal states of affairs that may never be-

come actual. It seems to me that both the first strategy – abstraction – and the sec-

ond strategy – this use of clauses with the aim to hedge a theoretical proposition –

end up with ideal states or model worlds. It is therefore not clear whether the dis-

tinction between ‘idealization’ and ‘abstraction’ bears philosophical fruit. There is a

lack of clarity is underlined by the fact that I occasionally came to speak with

economists who confuse the two.

An example shows how abstraction and idealization can do similar jobs.

The abstracted perfect circle, for instance, is a mathematical object and the really

existing round objects in the world have properties due to which they are at best

only near-perfect. The conceptualisation of the mathematical object clearly is the

result of abstraction. But it is at first sight unclear whether or not the idea of the

same perfect circle is also produced by the insertion of a (false) clause. If we insert

a clause in our theory, counterfactually saying that all properties, which make the

real life object less than perfectly circular, are absent, that edges are infinitely

smooth, etcetera, are we then not engaged in the same epistemic operation as ab-

straction? In other words, is ignoring well chosen spatio-temporal detail not the

very same as inserting clauses that falsely describe the world as different than it

really is? It seems so. The perfect mathematical object appears to be conceived ei-

ther way.

But the answer depends on how the clause is specified. It will be shown, in

chapter III, that idealization and abstraction are in at least one sense the inverse

(not the opposite) of each other. In what Kuipers calls the internal phase of the

development of scientific research programs, more specifically in the evaluative

phase, scientific improvements often proceed by what is generally coined Idealiza-

Introduction

tion and Concretisation (I&C).3 In seeking increasing success, after a new idea has

broken through in the heuristic phase, specific theories for relatively limited do-

mains are subsequently developed. Every step consists of taking aboard causally

relevant factors that were neglected in the previous theory. The theories share a

core idea that constitutes the scientific dogma. In chapter III I shall judge the label-

ling ‘I&C’ as inadequate due to its inconsistency. Concretisation implies the move

toward a less abstract level of description, so the counterpart of concretisation is

abstraction, not idealization. The terminology stems from Leszek Nowak, in whose

tradition much of current research into isolational practices stands. Nowak com-

bined abstraction with reduction, potentialisation, and transcendentialisation;

terms alternatively categorised as soft and hard deformational procedures.4

But there is a much more interesting point to make about idealization than

internal consistency of labelling. By its very nature, idealization is involved with a

‘move away’ from the actual state of affairs, while abstracted propositions tend to

‘move toward’ the actual world. However, the use of so called vague clauses is very

similar to abstractive reasoning. My conclusion is, first, that these vague clauses

are characterised by the open-endedness of the list of false assumptions and, sec-

ond, that idealization be better reserved for the use of clauses that list a finite num-

ber of false assumptions. The use of well specified clauses of the latter sort is part

of a very different methodology than ignoring detail. I present a model of these

two epistemic orientations that I call idealization and abstraction, respectively. I

shall show that idealization necessarily involves falsity and abstraction does not.

However, both may help constitute true theories, be it in very different ways.

One issue calls attention for the next. Thus, issues of truth and falsity also

bear on a more pressing matter. Social scientists give policy recommendations. As

dominating nature requires knowledge of future possible states of the world, so

does policy in the social realm. In other words, social scientists cannot escape the

pretension to be capable of predicting the future, at least if they give policy advice.

They seem to know what will happen if we intervene and what if we do not. Social

predictions must be true for their policy relevance. As we have to understand how

predictions are assumed to be true, indirectly, the difference between idealization

and abstraction in social theorising turns out to be crucial for questions of policy.

This brings me to my second research question. How can social science lead

to true theories if it is loaded with idealizational practices (especially when predic-

tion comes in)? The answer I give in chapters III and IV makes use of the interest-

ing fact that, although idealizations involve falsity, the propositions in which ide-

alizations are formulated need not be false. Another finding in chapter III is related

to this. Abstract theorising may threaten policy relevance if the complexity of the

theory, relative to the scientific problems to be resolved, is too low.5 But insofar as

3 Kuipers (2001), p.16. 4 Nowak (1989) 5 Siegwart Lindenberg has shown this in his plea for decreasing abstraction. See Lindenberg (1991).

Second

question

policy recommendations must be based on true claims, it is important to see that

the epistemic process of abstraction tends to lead to more truthlike propositions.

The third question is what it is that is abstracted from the superfluous.

Böhm-Bawerk abstracted market mechanisms from the manifold appearances of

economic reality. These were the key explanatory elements required to highlight

the economic process. So what are these mechanisms that translate intentional deci-

sion making into aggregate outcomes? To use Mäki’s terms, which objects make

part of the ‘ontic furniture’ of the world Böhm-Bawerk was redescribing

The fourth question is how the apparent practice of theoretically abstracting

can be squared with the philosophical trouble essentialism is known to lead to.

After all, abstraction is embroiled with the idea of a prior world conception. This

world is taken to deliver (perhaps after some torturing) the kinds that help explain.

More specifically, how can research interests and prior conceptualisations be neu-

tral as regards the objectively existing (social) kinds Böhm-Bawerk, but also pre-

sent-day economists, try to discover? The answer, of course, is that research inter-

ests and prior conceptualisations aren’t neutral at all. But, as I believe, one can

detect essentialist inclinations in economists’ work. Some evidence toward this

claim is given in the intermezzo, which connects the later, more philosophical

chapters to the first two, more economics chapters. The last chapter makes a start

in solving the puzzle (1) how this tacit essentialism can be conceived of as rational

scientific behaviour, and (2) how the importance of ‘subjective’ interests and per-

spectives can be honoured without rejecting a mild essentialist account of science. I

am afraid it is only a start, but I trust it is a worthwhile attempt.

Structure

This thesis can be slimmed in reading by taking the diet version: skip chapter I and

section 4 of chapter II, and appendices 1 and 2. This option may be appealing to

pure philosophers who are bored by economics. The discussion of Böhm-Bawerk’s

economic theory is relevant for the rest insofar as later chapters repeatedly refer to

his work. The conclusions to these first two chapters will give them sufficient am-

munition to understand the points I want to develop later.

The first chapter relatively obediently rewrites the magnum opus of Böhm-

Bawerk, pinpoints crucial concepts, and describes his polemic reasoning strategy.

Much critical analysis is not offered in this opening chapter, but it is in chapter II.

My criticism is directed not only at Böhm-Bawerk, but also at his interpreter and

mathematical translator, Knut Wicksell. In part, Böhm-Bawerk will in fact be de-

fended against his Swedish admirer—critic. The essentialist metaphysics is the

main point of discussion. Böhm-Bawerk’s quest for theoretical unification is rooted

in his metaphysics. Chapter III treats how a strict distinction between abstraction

and idealization can clear up issues concerning policy relevance of theories that are

hedged and that work with – as it is often called – ‘closures’. Chapter IV illustrates

the usefulness of the distinction by four examples, one from Wicksell’s attempts to

insert dynamics into the theory, one about the concept of involuntary unemploy-

question

Fourth

question

Introduction

ment in Keynesian theory, another on policy making aiming to cure inefficient

markets in health care, and yet another on ethical consumption. Finally, chapter V

deals with the sturdy problem of essentialism in social theory.

Appendices 1 and 2 present some further refinements of Böhm-Bawerk’s

theories that would otherwise overload chapter I. Appendix 3 disentangles meta-

concepts of essentialism and holism. It serves to encounter straw man versions of

essentialism, which pop up here and there and make a misconceived attempt to

attack essentialism. There are many good ways to attack strong versions of essen-

tialism, but the bad ones should be discarded. Nevertheless, the third appendix

deals with only one of the misconceptions.

I Time as a unifying concept

1 Introduction

The economic theory developed by Eugen von Böhm-Bawerk is packed in the first

of two volumes of his Positive Theorie des Kapitales (PTK).1 He presented it as a the-

ory of productivity up to the last chapter, in which, as suddenly as unexpectedly, it

evolves into a theory of distribution: Der Kapitalsmarkt in voller Entfaltung. This last

chapter is an apotheosis of the opus. For the tone in which it has been written the

title could be translated as ‘the market of capital in full swing’.

I shall first present the general structure of PTK in order to depict how

Böhm-Bawerk’s theories relate (section 2 of the current chapter). As I shall show,

he did not manage to coherently combine them. For instance, the capital theory

(treated in section 3) turns out to be rather disconnected from the rest. Menger’s

subjective value theory (section 4) is presupposed throughout the work. The inter-

est theory (sections 5 and 6) precedes the distribution theory (section 7) both logi-

cally and in this chapter.

Chapter II can be appreciated in full after reading the current chapter, but

may also be understood without it. The key issues of chapters III to V do not de-

pend on this first chapter; it can therefore be skipped in reading. Nevertheless,

much of what is treated in the current chapter returns in one way or another in

later chapters. Böhm-Bawerk’s critique of the Law of Costs, and of the use theory

of credit, his discrete analysis, and his revolutionary interest theory return in sec-

tions 2, 3, 4, and 5 of chapter II, respectively. The way he abstracts from time is

used in chapter III. Chapter IV extracts an example from his distribution theory.

The law of costs is considered once more in chapter V.

1 It was also published as the second volume of his three-volume Kapital und Kapitalzins. The first vol-

ume of this latter work is a mere critique of such theories up to the end of the nineteenth century and

does not yet propose an original theory. The third volume contains responses to criticisms Böhm-

Bawerk had to cope with in the course of the period running from the first edition of the first vol-

ume, in 1888, to the third edition of the entire work, in 1907. I mainly consulted the German (1921)

fourth edition, and occasionally used the English translation by Huncke and Sennholz (1959).

Chapter I

2 The structure of the Positive Theorie des Kapitales

2.1 Judging Böhm-Bawerk

Schumpeter judges Böhm-Bawerk’s theory above all as an income distribution

theory and he discards common interpretations that it was a mere interest theory.2

I believe it is both an income and distribution theory. Perhaps the best way to in-

terpret Böhm-Bawerk’s contribution to economics is that it is a productivity theory

and an interest theory; though the former disconnected from, the latter leading to a

distribution theory in which interest plays the most important role.

Wicksell has undertaken to conceptualise Böhm-Bawerk’s system in mathe-

matical form offering a simultaneous solution of fourteen equations. In his Über Wert, Kapital und Rente (WKR 1893) he remarked that the result looked very much

like that of Leon Walras (in his Éléments d’économie politique pure), in that equations

of production and those of exchange are combined. But he also noted an important

difference. Walras’ concept of capital comprised only durable goods, whereas

Böhm-Bawerk had rightly understood that the non-durable goods available as

subsistence means for workers, entrepreneurs and landlords are in fact what we

really have to see as social capital.3 This implies that, contrary to what Walras had

said, a dynamic view of a progressive economy is not needed to determine interest.

This then, writes Wicksell, is the great merit of Böhm-Bawerk’s contribution.

2.2 A pyramid structure

The value-, capital-, and interest theories Böhm-Bawerk proposes are represented

in the pyramid-like figure 1. The arrows are meant to show that one market result

delivers the initial conditions for the working of another market. Sometimes, how-

ever, Böhm-Bawerk does not consider explicitly how one market outcome in fact

feeds into the process by which another market produces equilibrium. An example

is that, although capital growth results from savings, he does not show how the

quantity of savings enters the conditions for the generation of interest. Sometimes

he does tell the reader where to find the supposed link.

In the figure, a continuous arrow indicates a well established link in the the-

ory; a dotted arrow shows where the reader has to construct this link for himself.

At the left hand corner of the pyramid we find the goods market, explained by the

theory of value, in fact a theory about how market prices come to be, which Böhm-

Bawerk had inherited from Menger. The market for capital is taken to be a token of

the more general kind a market is supposed to represent. But the precise causes

2 Schumpeter (1954), p.846. 3 WKR, p.142.

and effects of capital accumulation are explained by the theory of capital, which

remains rather independent of the other theories. We find it at the right hand cor-

ner. Böhm-Bawerk can treat the capital theory first since the theory of

Figure I.1 Böhm-Bawerk’s theories: the pyramid structure

value is strictly unnecessary for what I shall call the Robinsonade, which so or-

nately enshrines the capital theory. The Robinsonade is a narrative about a ficti-

tious Robinson Crusoe, representing an entire one-person economy, who manages

to accumulate capital without any exchange. The absence of exchange entails the

absence of markets.

The theory of capital has no well established links with either the theory of

value (more specifically: price theory), or the theory of interest, or the distribution

theory. First, the relation between price theory and capital theory is problematic in

Böhm-Bawerk’s system, because he never arrives at a clear idea as to how to con-

ceive of capital: as physical units (either of fixed capital or of subsistence means) or

as something to be referred to in value terms.4 Furthermore, the explanatory theory

of building (social) capital is not necessary for his interest theory. The mere pres-

ence, and not growth, of (private) capital as an endowment suffices to explain the

origin of interest. Hence, Böhm-Bawerk does not need investigate either the causes

or the effects of the long term growth of an economy. Although Wicksell tried to

improve on Böhm-Bawerk’s static model, the former prized the latter for seeing

that interest emerges even without long run growth or capital accumulation. Fi-

nally, Böhm-Bawerk does not develop a model of distribution, as he lacks the

mathematics for such an undertaking. Henceforth, the links between distribution

4 Wicksell criticised this shortcoming although he never really succeeded in improving on the Austrian

with respect to this point.

Distribution

Value Capital

Interest

Chapter I

and capital remain unclear. It can only be noted that prices in goods and factor

markets determine the distribution of wealth, and that this must be an important

constraint on a population’s saving behaviour.

The Austrian Subjective Value Theory functions as the mechanistic underpin-

ning of all theorems about the generation of equilibrium prices and to Böhm-

Bawerk’s light, interest is, first of all, a market price. He wants to conceptually place

the phenomenon of interest within that of the existence of value and create a unify-

ing picture of these phenomena. Next, interest is also a factor price. Hence, it is an

income. As represented by the vertical arrow of figure 1, interest feeds into the dis-

tribution of wealth. This is where the macroeconomic view begins. The relation

between factor incomes is a function of how produced goods are distributed and

vice versa. The prices of the commodities determine the returns for labour and

capital in the different production lines and the allocation of end products depends

in turn on the distribution of factor incomes. Had Böhm-Bawerk managed to em-

bed his theory of capital accumulation, via the concept of interest, into an integrat-

ing theory of distribution, he would have done the tremendous job of presenting a

General Theory of expanding economies of Walrasian scope, however – and this

would make it splendid – with a concept of the mechanistic underpinnings that

lacked in Walras’ model.5 But this is of course too much to ask. There is no arrow

running from capital to distribution. The reverse direction has been indicated by a

dotted arrow, as the link via savings is obvious but not subject to Böhm-Bawerk's

explicit attention.

2.3 Dropping assumptions and claiming truth

Schumpeter noted that Böhm-Bawerk had a reasoning style of an advocate rather

than that of a scientist. Indeed, he strongly contrasts his theory against competing

theories as the only true one. One of the rhetorical means he employs to do so is

that of switching to and fro between, first, descriptions of ‘hidden’, tendency-like

‘rules’ and, second, descriptions of the phenomena we observe on a daily basis. To

Böhm-Bawerk, the former are primary, the latter are secondary.

The observed phenomena are to verify descriptions of the presumed hidden

patterns. This can for instance be seen when he discusses the Law of Costs. This

law says that, in equilibrium, the costs of production equal prices of end goods.

The classical Objective Value Theory postulates the former as being the cause of

the latter. Böhm-Bawerk contrasts this with the Mengerian Subjective Value The-

ory. He explains that it is demand for finished goods that determines costs of in-

puts, not the other way around. He gives an example, which is more than an illus-

5 I shall contend that awareness of this underpinning helps understand many of his obscurities. It makes

him a true Austrian in one respect, quite to the contrary of what Klaus Hamberger claims in his

(2001), note 14, that ‘Böhm-Bawerk certainly was not an ‘Austrian’ as a capital theorist’.

tration: it has the force of a proof. Its direction runs from observed facts to underly-

ing – and in a sense more ‘real’ – rule. Es ist eine bekannte Tatsache, daß lebhafter Eisenbahnbau die Schienenpreise,

und durch diese den Preis des Eisens hebt […]: Fälle, in denen sichtlich die

steigende Preisbewegung von den Schlußprodukten ihrer Ausgang nahm und

von da auf die Kostengüter sich fortpflanzte.6

This means that changes of the value of inputs are causally posterior to changes of

the value of end products. His proof is such that we only have to check whatever is

‘obvious’ to know this. The direction of the causality is taken to force itself to any

attentive eye. The role of lay knowledge in a context of justification is important: Die Namen subjektiver und objektiver Wert […] scheinen mir den einzigen

Fehler zu haben, daß sie etwas „akademisch“ klingen und nicht recht volk-

stümlich werden können.7

However, the alleged obviousness of ‘real’ relationships is not always so

clear given that the Classical Economists apparently did not see the real causal

direction at all. They defended, after all, the ‘false’ Objective Value Theory. Have

they perhaps not been sufficiently attentive?

In chapter II I shall explain how Böhm-Bawerk could add a pretension of

producing scientific knowledge – in contradistinction with lay knowledge – to a

context of justification that rests on lay observations. On the one hand, he dis-

cerned true science from superficial common sense knowledge of the phenomena

(and from scientific but false theories). On the other hand he appealed to phenom-

ena that are supposed to be well known to anyone.

We shall see in this chapter that, in the course of PTK, the assumptions are

inserted and subsequently dropped again just to show the difference between true

and false theories. It is a strategy to keep track of the distance between abstract

model and ‘known phenomena’. Böhm-Bawerk seeks the checks and balances to

his theory in a continuous movement between the general and the specific, the

economic law and the instantiations of that law, but also the exceptions to the law.

It is, I believe, his peculiar way of guarding over the external validity of his theories.

The stepwise procedure, which shapes the theoretical body of the Positive Theorie des Kapitales (and which generates its heart beat for who reads it from beginning to

end) does not merely serve to persuade the reader, but also to keep track of what

the essentials of the economic world are like and what merely makes part of its

appearance. There are hidden parts of reality waiting to be uncovered. To make

these hidden aspects of reality visible, PTK describes a modelled reality, viz. of

aspects of economic reality as they are expected to be under the force of isolating

assumptions8.

6 PTK, p.312. My emphasis. 7 PTK, p.162. 8 Here I use ‘isolation’ as the epistemic activity, which sets apart a field of exclusion, “sealed off” from

the involvement […] of everything else’ (Mäki 1992b, p.321) in a very general sense. Later the con-

cept will be explicated by the distinction between idealization and abstraction and the concept of ex-

ternal validity will also be made more precise.

Chapter I

An important example of such an assumption is that equilibrium in markets

comes into being instantaneously, without the passage of time. It evidently does

not, but under the force of this isolating condition, we shall see interesting patterns

nevertheless. So, contrary to the view that isolating assumptions take away the

‘realisticness’9 of a model, Böhm-Bawerk believes that assumptions may sometimes

highlight certain aspects of (economic) reality that otherwise remain unseen. Let’s

investigate in detail the particular heuristic Böhm-Bawerk used in finding what he

saw as the only true theory.

2.4 The key assumptions

Let us first consider which assumptions play a key role in his system. I shall list the

main twelve assumptions we find in PTK. There are two basic axioms (taken to be

‘true principles’), the first being that increasing the inputs of some productive fac-

tors and leaving one constant leads to rising and eventually falling extra returns.

The second is that subjects maximise utility, a principle that implies some more

familiar principles, such as a complete and transitive ordering of preferences and

the presence of constraints.

Some of the assumptions are implicit in PTK, others are taken into consid-

eration explicitly. Only the explicit assumptions can serve to be manipulated in

Böhm-Bawerk’s rhetorical strategy.

So which assumptions serve which role? The first seven assumptions consti-

tute some of the many aspects of competition. The most ideal image of competition

pictures this regime as free, full, and perfect. Free competition assumes the absolute

absence of government regulation.10 Assumption 1 excludes price fixing and re-

lated surpluses and shortages. This is a more specific formulation of the condition

under which the ‘exact method’ proposed by Carl Menger can proceed: that to no

one the economic freedom (to pursue one’s interests) will be limited.11

Full competition deals with the assumption that monopolistic market traits

are totally absent, so that market niches cannot exist. One obvious condition of full

competition is that, in order to conceive of it, we must consider markets with many

demanders and suppliers. From a dynamic perspective this also means that new

9 The term has been coined by Uskali Mäki. As ‘realism’ refers to the possible attitude of scientists toward

the aim of theorising (is it to represent social reality as proximate as possible or only to develop tools

for intervention?), the term ‘realisticness’ is more appropriate to describe a property owned or

lacked by the theory itself. See Mäki (2002). 10 Given the nineteenth century controversies over the justifiability of interest or Mehrwert, it is perhaps

easy to forget that any ideological preferences of the theorist a priori are irrelevant at this point. It

may well be true that allusions to normativity lie at the forefront whenever this condition remains

implicit in an economic text, because what remains implicit may be seen as allegedly beyond debate

and therefore as an ideological ‘common ground’. But assuming the absence of external powers is

methodologically independent of any supposed merits or demerits of laissez faire policies. 11 [D]ass kein die ökonomische Freiheit derselben (die Verfolgung ihrer ökonomischen Interessen) beeinträchtigen-

der äusserer Zwang auf sie geübt wird. Menger (1883), p.56.

A Axiom: diminishing marginal returns

B Axiom: maximizing behaviour

Free competition assumptions 1 Absence of government regulation

Full competition assumptions 2 Large number of demanders and suppliers

3 Absence of entrance barriers

Perfect competition assumptions

4 Absence of geographical distance

5 Perfect access to all relevant information for all market parties

6 Infinite divisibility of goods and wants

7 Instant adjustment of market equilibrium

(for which 4+5 are necessary but not sufficient conditions)

Conditions of stability

8 Ceteris paribus clauses. For the demand side there are:

D(1) Tastes and preferences do not change

D(2) Prices of other goods do not change

D(3) Number of demanders does not change

For the supply side there are:

S(1) No exogenous technological development

S(2) Prices of inputs do not change

S(3) Number of suppliers does not change

Other assumptions

9 No endogenous technological development

10 No investment in stock

11 Labour is the only demander of subsistence goods

12 Markets are homogeneous

market parties are allowed to join at any time, so that entrance barriers (other than

those installed by legal authorities, which fall under the first condition) are taken

to be absent (assumption 3).

Assumption 2 is therefore unproblematic as a description of reality. It is not

dropped but inserted in the stepwise process of reasoning. In a modern economy

many people do take part in exchange. The only reason why this assumption does

not prevail from the start seems to be didactic. Böhm-Bawerk tells a story about a

single Robinson who can only be the producer of the goods he consumes. Next we

meet one seller and one buyer, who settle the terms of trade according to their sub-

jective preferences. The point then is that if only two traders (in contrast with a

multitude of them) meet the terms of trade, they have a degree of freedom within

certain boundaries. The next step brings in more traders. The reader learns that, if

the number of demanders and suppliers rises, this interval, which sets the degrees

Chapter I

of freedom, shrinks accordingly. Ultimately, if the number of participants in trade

is large enough to prevent any single market party to exercise a noticeable influ-

ence on the equilibrium market price, there is no interval left.

[D]ie Zone, in welche der beiderseitige Wettbewerb die Preisbildung drängt,

und innerhalb welcher der Markt sein momentanes Gleichgewicht finden

könnte, verschmälert sich zu einem Punkt.12

The terms of trade have become equal to the marginal utilities of the last buyer

and the last salesman. In other words, with a number as large as to turn individual

decision makers into price takers the market price becomes fixed, market equilib-

rium becomes automated. It is a double-edged sword: the assumption of many

suppliers and demanders is realistic all right and it makes the determination of the

resulting market outcome turns easier to explain too.

This is noteworthy in the context of the Methodenstreit. The German histori-

cists disliked ‘theory’ for its supposed lack of realisticness. Their method was ho-

list, it was engaged with a particular concept of realism, and they believed in the

fruitfulness only of detailed historical scrutiny. But the Austrian abstract picture of

Robinsonades and single agents entering a deserted market and computing their

marginal utilities could be dressed – concretized – by a reality of jam-packed mar-

kets with complex information transfers; and this stepwise shaping up of a more

realistic picture precisely served to verify the theoretical hypotheses.

Competition is perfect, finally, only if cheaper potatoes in New York imme-

diately induce buyers from New Orleans to travel – at no cost – to the east coast.

But assumption 4 remains implicit in PTK. Assumption 5, on the availability of

information has its reverberation in Menger’s condition dass [die wirtschaftenden Subjekte] die ökonomischen Sachlage, soweit sie auf die Preisbildung von Einfluss ist, nicht unbekannt sei.13 It is also tacitly assumed. Assumption 6 is mentioned, and not dis-

pensable without harming the consistency of the theory.

Perfection comes close, furthermore, to what physicists would understand as

‘no friction’. It leads to ‘instant adjustment’ or a frictionless market process. This is

the meaning of assumption 7. The idea that changes in initial conditions instantly

lead to a new equilibrium is essential in what I shall call the ‘result’ reasoning

characterising the approach of Walrasian style equilibrium model explanations.

Another way of putting it would be to say that once results are given, there is no

time evolving. Most modern textbook economic theory unproblematically lists

assumption 8 in a context of result-reasoning. But as an elaborate discussion of

Menger’s Subjective Value Theory will show in chapter II, Böhm-Bawerk’s mecha-

nistic approach explains non-equilibrium situations that arise from instability. A

result approach is not equipped to do this.

The theory of interest is mechanistic in this sense too, but while treating in-

terest, Böhm-Bawerk implicitly supposes the goods markets to adjust without fur-

12 PTK, p.287. 13 Ibidem. Two more of his conditions deal with ‘cognitive’ and ‘instrumental’ rationality. See Janssen

(1993), chapter 2.

ther ado. In other words, when the focus is on the interest theory specifically, all

markets other than the one under consideration are understood to produce equilib-

ria as results without any further treatment of their adjustment mechanisms. The

theory of interest offers a partial perspective on the mechanism that explains how

the respective choices, which entrepreneurs make, translate into an equilibrium

interest; the responses of rest of the world are ‘suppressed’ by a stability assump-

For the market equilibria to really be equilibria the assumption is not dis-

pensable; hence it is labelled ‘7R’ in table 1 below. But on the other hand, in

mechanistic reasoning time does evolve, and opening the black box makes it possi-

ble to track what happens in an unstable environment. Instability triggers the

mechanism. The theory can explain why the theoretical equilibria often are not such

in practice. The mechanism can also explain the phenomena during periods of

relatively stable initial conditions, when equilibria can be noticed. In the mechanis-

tic approach the assumption labelled ‘7M’ is dispensable.

Assumptions … …free

competition

…full

competition

…perfect

competition …stability …other

dispensable

D(1), D(2)

S(1), S(2)

dispensable 1

D(1), S(1)

D(2), S(2)

D(3), S(3)

Explicit

assumptions

part of

stepwise

analysis

D(3), S(3)

11, 12

Implicit assumptions 3 4,5 9, 10

table I.1 Explicit and implicit assumptions in Böhm-Bawerk’s work

The ceteris paribus conditions (with a ‘D’ for the demand side and an ‘S’ for

the supply side) are subordinated under assumption 8. These conditions only deal

with the stability of the environment, i.e. of initial conditions, and not also with

aspects of competition. The difference between result reasoning and mechanistic

reasoning particularly bears on these stability, or ceteris paribus assumptions. The

table allows for a separation of 8R and 8M for the same reason as it does for as-

sumption 7. If preferences or non-own-prices change, the Mengerian mechanism

will take care of the necessary adjustments leading to the results hypothesized.

Note, however, that assumptions 8R.D(3) and 8R.S(3) are listed in the table as ‘not

Chapter I

dispensable’, while 8M.D(3) and 8M.S(3) are seen as ‘part of the stepwise analysis’.

In the course of Böhm-Bawerk’s mechanistic explanations

Changes in technological development are subsumed under stability as-

sumptions, viz. the stability on the supply side. Instability of the environment is

always exogenous. Hence, assumption 9 is needed to rule out any technological

development. As to assumption 10, zero stock changes are not discussed explicitly

in PTK. It comes with the stationary state analysis, but if applied to other cases it

should not be too difficult to adapt the models of the theory accordingly.

Two explicit assumptions in the category ‘other assumptions’ are dropped in

Böhm-Bawerk’s final chapter on (de-isolated) distribution theory: that labour is the

only (or main) demander of subsistence goods (assumption 11) and that the labour,

goods, and capital markets are homogeneous (assumption 12). This is the reason

why I listed these also under the heading of ‘stepwise analysis’. The last chapter of

PTK develops the capital market in the de-idealised case.

3 Böhm-Bawerk’s concept of time in economic theory

3.1 Capital building: a Robinsonade

To isolate what he thinks is the core of a true capital theory, Böhm-Bawerk follows

the strategy known from the work of many economists. He asks his readers to

imagine a society of one individual (whose name, for all romantic implications

involved, is of course Robinson). In this way, possible problems of microreduction

(or methodological individualism) are avoided at the start and the focus is on con-

sumption, production and saving by one man. In other words, there is no market.

Robinson lives on a lonely beach without capital.14 This means that he can only

gather and pick fruit. What we see first is production at the most primitive stage.

In order to start with a clear problem Böhm-Bawerk makes conditions as strict as

possible. From here onwards, we can see which conditions cause which phenom-

ena to happen. Robinson needs all of his available time to barely feed himself, 10

hours a day, and it is clear that no immediate improvement of Robinson’s situation

is possible.

A revision of the example grants Robinson a slightly greater wealth so that 9

hours suffice to maintain subsistence while he can still work 10 hours. Now he will

have an option for change, either increasing his immediate standard of living if he

collects food beyond subsistence level, or increasing his future (mediate) standard

14 In Daniel Defoe’s (1719) story, The Life and Strange Surprising Adventures of Robinson Crusoe of York,

Mariner, Written by Himself, Robinson did actually start with capital goods from the ship that had

wrecked. Apparently, Böhm-Bawerk makes Robinson’s life tougher still.

of living if he devotes the extra hour a day for the production of a hunting device.

By making this device, Robinson improves his uninviting life as a result of saving.

Note that to merely save, eating less than what can be produced – the product of 9

hours work instead of 10 hours - is not sufficient. For progress, he will also have to

invest his savings to materialise the savings effort. Otherwise, in a non-monetary

economy it will be lost.15 At this point Böhm-Bawerk contrasts his view with that of

Adam Smith, who had said that ‘[p]arsimony and not industry is the immediate

cause of the increase of capital’.16 The true description is precisely the other way

around. Seen from a social standpoint, if he chooses to save, what he really saves is

not an amount of consumption goods (nor of capital goods, contrary to what

many, from whom Böhm-Bawerk tries to distance himself, think) but of productive

forces. It is the hour a day.

Given his low standard of living in the first instance, the alternative option,

to use the extra hour for the collection of more consumption goods, may also seem

an attractive option. There is an opportunity cost involved in saving. Suppose Rob-

inson gives up the immediate satisfaction of very tempting needs. He invests the

daily hour into making bow and arrows in a month’s time, i.e. in a total of 30

hours. He will use the hunting device for fowling. Clearly, the bow and arrows will

raise his productivity in food collection after a month from now and hence it raises

his future standard of living. Böhm-Bawerk calls the expected appetising fowl fu-ture goods while the bow and arrows make up his capital goods or Zwischengüter.17

Furthermore, saving and investing increase the capital intensity of this economy.

Saving and investment amount to branching off toward roundabout production or to

increasing the level of ‘Kapitalistische Produktion’18.

How can Robinson satisfy even more wants than fruit and fowl? If he

spends part of his day consuming and resting, he will not even be able to replace

his worn out capital goods. Thus, to maintain the stock of capital, Robinson must

devote another 30 hours to replacement production and this has to be repeated

during his lifetime. The 30 hours are the cost of replacement, till new technological

15 In a monetary economy, this sort of ‘saving’, i.e. without investing it, is called hoarding. 16 Böhm-Bawerk quoting Smith in the note on PTK, p.139. 17 Zukunftsgüter (or –ware) are goods now in existence but the expected revenues of which are the object

of valuation. They are not to be confused with goods that do not exist yet. In note 1 on p.323 PTK,

Böhm-Bawerk remarks that that no fine translation exists in either English or French to indicate the

subtle distinction between the revenues of goods that are and of those that are not existent now. But

‘future goods’ must be taken to exist in the present, for otherwise they cannot be exchanged against

present goods. The possibility of physical exchange is crucial to explain the source of interest. I shall

invariably use the term ‘future good’ for Böhm-Bawerk’s Zukunftsgüter. Furthermore, I shall call the

Zwischengüter ‘intermediary goods’, and not, as Huncke did, ‘intermediate goods’. The latter term

merely suggests circuitousness. The word ‘intermediary’ also hints at the property of mediation.

Böhm-Bawerk stresses that intermediary goods are not only Zwischengüter, but also Zwischenursache.

They are both cause and effect of roundabout ways of production. 18 Böhm-Bawerk interprets capitalism very differently from what Marx had taken it to be. Indeed, the

former needs no claims about property rights, as Marx did. He defines it simply as any production

by means of ‘intermediary products’, like Robinson’s tools. See PTK, p.16.

Chapter I

design (either of the bow and arrows or of the production technique) is available.

The explicit assumption is that no exogenous technological development takes

place19. Given the loneliness of our hero, this seems to be a reasonable axiom, at

least for the time being. If he wants to more than maintain the number and quality

of capital, he will have to save more than what is needed for mere constancy of

capital, or, in more weighty terms, for the continuation of the level of capitalistic

production.

To maintain capital, Robinson has to spend the same amount of time on a

new bow and new arrows. As food gathering, due to the increased efficiency, will

cost just a few hours a day now, this should apparently be easy. He has plenty of

time to reinvest in his device.20 But Böhm-Bawerk does not stress this at all. This is

somewhat curious, for in other places he often puts forward descriptions of eco-

nomic processes as an alleged ‘obvious fact from experience’. Elsewhere (p.130)

Böhm-Bawerk emphasises that, once an economy is engaged in capitalism, the

very possibility to accumulate still more capital has become easier. There is no sign

of awareness that the implicit endogenous technological development makes Rob-

inson’s life easier.21

Instead, Böhm-Bawerk gives a graphical representation that has turned fa-

mous. Figures 2a and 2.b below show two alternative possibilities of the organisa-

tion of capital that stems from roundabout production ways of seven years ago.

Concentric rings display segments of capital of different maturity levels, be

these measured in units of labour. The perspective is from outside in and the inte-

gers refer to the number of years to go till maturity. This means that the outer rings

depict the segments closest to maturity. They are due to loose their intermediary

19 This is made explicit in note 1 on p.140 PTK and again in note 1 on p.147 PTK. Another, more unlikely

assumption is that Robinson does not gain dexterity in the production of bow and arrows. That does

however not matter for the very idea expressed here, because if he does, the number of hours will

simply decrease to a particular limit, say 25 hours, leaving the argument impervious. For Böhm-

Bawerk, exogenous technological development and simple improvement of skill blur into one and

the same phenomenon. 20 In a monetary economy, depreciation is the ‘saving’ of means to purchase replacement capital when it

is due. In this Robinsonade, the non-monetary equivalent is the gradual accumulation of hours of

work (over the days in which Robinson’s capital good, i.e. his hunting device, wears out) spent on a

new bow and new arrows, up to the completion of this capital good. 21 The suggestion becomes even stronger on the pages where he attacks the socialist Rodbertus and his

Das Kapital. See PTK, p.154. Rodbertus believes that it is productivity, not thriftiness, that enables the

accumulation of capital. Böhm-Bawerk’s reasoning is that anyone who lives without capital (von der Hand in den Mund) has little opportunity to spare the necessary working effort for the production of

it. He does not consider new qualitative steps in the development of Robinson’s island economy. His

cases in point are numerical examples from here on. New steps would comprise the accumulation of

more intermediate goods, including goods that serve to bring forth such intermediate goods. In

terms of the robinsonade, Robinson could make a knife for the sharpening of the arrows. It would

take two periods for the knife to mature into food. The first period is when the revenues of the knife

help to create the hunting device. The second is when the revenues of the bow and arrows bring the

owner food. In this way, capital accumulation (Kapitalbildung) can be thought of as ever more nu-

merous layers of capital, each phase closer to maturity, i.e. to the moment when the proceeds of that

capital take the form of goods for consumption.

status, i.e. their predicate ‘capital’, and to be turned into consumer goods. These

sections comprise the greatest amount of labour years. They amass all the labour

that has been put into the process of turning natural resources into simple artefacts,

such as ore that has been altered into iron, but also the labour put into the making

of further capital goods of greater sophistication, like iron into steel, or steel into

railways.22 The inner rings, then, depict younger segments of capital.

Figure I.2a Figure I.2b

Capital structure

Figure 2a represents the situation of an economy that has taken the road to

capitalism relatively recently, as few sections of the economy are as yet capital

intensive. There is a clear difference in the perimeter (and hence, the surface area)

of the rings: the inner rings are much thinner. A more primitive economy has thin-

ner inner rings weil hier weitausholende Produktionsumwege, die ihre Genußfrucht erst nach vielen Jahren bringe, nur selten und spärlich eingeschlagen werden23. Such an econ-

omy has few levels of maturity, so the roundaboutness of production is still lim-

22 Böhm-Bawerk does not distinguish between replacement of worn out fixed capital and replacement

of used floating capital, and hence not between gross and net production. The terms ‘gross’ and ‘net’

are reserved for forms of yield in the English translation by Huncke: Roh and Rein in the German of

PTK. See the section about capital yield on durable goods below. 23 PTK, pp.141-142.

Chapter I

ited. More roundabout production methods are more labour productive. Böhm-

Bawerk notes, also, that not all labour of the inner rings matures into the outer

ring. Many production processes take only one or two years to be completed. Also

due to this fact the inner rings must be smaller: the production methods that re-

quire less roundaboutness do not start life in the middle of the figure, but further

outside, in rings with a lower number. In turn, figure 2b represents a more modern

economy, with more rings, symbolising greater capital intensity.

Fixed capital goods have a given technical duration. For an economy as a

whole under this condition the average duration can always be calculated. Further

below in this chapter we shall see that, in the sections of PTK that treat the interest

theory, a very different concept of capital is used: that of a wages fund. Although

in both the capital theory and in the interest theory a stationary state is assumed,

capital is steadily put in over time and steadily put out in the capital theory. Capi-

tal in the interest theory is steadily put in but unleashed at one point in time. In

this case, the average period of investment at the level of an economy can only be

calculated under the condition that there is no ‘looping structure’: that investment

in roundaboutness only renders consumption goods, and no machinery or other

fixed capital stock that is in turn producing other fixed stock, and so on.

3.2 A more complex economy

Böhm-Bawerk notes that the difference between a socialist economy and one indi-

vidualistically organised24 is that, in the former, savings can be forced upon the

people. But his theory of the origin and effects of capital applies to any sort of

economy.

How could an economy of an aggregate of many individuals increase its

capital stock? There are several ways, the first is to consume less. Another way is

by shifting some goods from ring number 1 back to rings more to the centre. Sow-

ing-seed can either be eaten or sown. If it had been meant for consumption (i.e. to

mature now) but is sown instead, there is a re-investment in younger segments of

capital maturing, in effect, later.25 New lines of capital intensify the capitalism of

the economy.

One can use the seeds to distil brandy. Kindling and firewood heat house

ovens but may be used for heating the blast furnace of an iron foundry instead.

Also in this case, the result is a smaller need for labour years, because less effort

goes into replacement. It is shown in appendix 1 how capital is structured over

maturity classes and what the effect is of dissaving, i.e. of overconsumption, as

opposed to saving.

To briefly sum up what we have seen, one aspect of the concept of capital is,

that it encapsulates ‘disobedient’ natural forces (while labour fences in more sub-

24 Note that the term ‘capitalist – as being dichotomous to socialist – is avoided here. See note 18. 25 Böhm-Bawerk mentions investing what has been saved in the past as a third option.

missive forces). Thus, capital amasses not only labour years, but also natural

forces. Another aspect is that, although a higher capital intensity increases labour

productivity, the amount of roundabout production degressively relates to produc-

tivity. Exogenous technological development helps to improve the relation of mar-

ginal revenue to roundaboutness of production. This means that new technology

makes further lengthening of roundabout production worth while. A third aspect is

that new investments require, as we have just seen, saving. The fourth is that in

non-socialist economies decisions to save are made individually, but work out at

aggregate level.

In accordance with the modern view, Böhm-Bawerk appreciates that capital

is an independent source of income like labour or land, but he notes that this does

not make it an independent cause of production. Apart from cause, capital is

symptom – and by implication, effect – of roundabout production.26 Social capital

comprises seven entities.

SOCIAL CAPITAL

Intermediary goods 1 Constructions attached to the soil (but reversibly so)

2 Productive constructions

3 Machinery

4 Livestock in production

5 Auxiliary materials

Other social capital 6 Unsold stock

7 Money

The first of seven items that form part of social capital is ‘constructions at-

tached to the soil’, but reversibly so, because otherwise it belongs to land. A barn, for

example, can be removed easily, a sewerage system cannot and therefore belongs

to land, rather than to capital. So does fertiliser on manured soil. Fertiliser in stock,

alternatively, belongs to item 6, ‘unsold stock’. Note that the first five items are de facto intermediary goods; the sixth, unsold stock, is not, for this is de facto a finished

product.27 The last one, money, is also no intermediary good. But these latter two

are part of social capital nonetheless, for they serve to complete the roundabout

26 His concept of capital diverges from that of his contemporaries in ways that he has written lengthy

exposés about in part one of his Kapital und Kapitalzins and elsewhere. His polemic with critics can be

found in many volumes of Conrads Jahrbücher, Jena: G. Fischer. In PTK Böhm-Bawerk stresses that

his concept of capital is consistent with Adam Smith’s, but diverges from that of Menger, Marshall,

Wicksell, Clark, Fisher and Landry (PTK p.42, note) 27 Today it is certainly customary to count unsold stock as intermediary goods.

Chapter I

production process.28 They make spatio-temporal movements and exchange of

goods possible. As will become clear below the possibility of these spatio-temporal

movements is indispensable for the central idea behind Böhm-Bawerk’s concept of

future goods.

4 Subjective value theory

In the nineteenth century, economists debated over the truth of opposing theories

that used different concepts of value. Classical economists had developed ideas

around objective value. The value of a good, according to them, consisted in the

costs involved in creating the good. One measure of objective value, used for in-

stance by Karl Marx, was ‘units of labour’. The amounts of these that had gone into

production both had to explain and to determine our valuations. Böhm-Bawerk

pounced on the objective value theory saying that it was not the need to put scarce

means into the manufacture of a good, which determined its value, but the scarcity

of the good itself. Menger’s subjective theory defined value as the importance a

good acquires for the prosperity of a person, as a known condition for a utility,

otherwise lacking.29 The objective value of a good is not the decisive Bestimmungs-grund of our valuations. It is instead only of secondary importance. Put more tell-

ingly, objective value is explanandum, subjective value is explanans.30

4.1 Marginal utility as a measure for value

Böhm-Bawerk’s take on the paradoxes of the objective value theory is that some

goods have a very low value on the market, although their consequence for human

life is crucial and some other goods, alternatively, are very expensive even though

they have no bearing on human subsistence. To explain this, the objective value

theory has been led to distinguish abstract from concrete value. The type ‘air’ has

abstract type-value31, for we cannot do without air. One particular unit of air, how-

ever, is free, because of its abundance, and this would then be concrete value.

Böhm-Bawerk exclaimed that this is ‘a completely misguided creation’:

[S]tatt in dieser Verschiedenheit nur eine kasuistische Besonderheit in der

Geltung eines und desselben Prinzips zu erkennen, konstruierte man zwei

verschiedene Arte von Wert;32

28 Money is now seen as productive in a direct sense, while Böhm-Bawerk takes it as indirectly produc-

tive for its being a mere creator of roundabout production. 29 In German: diejenige Bedeutung, die ein Gut oder Güterkomplex als erkannte Bedingung eines sonst zu ent-

behrenden Nutzens für die Wohlfahrtszwecke eines Subjektes erlangt. PTK, p.167. 30 Erklärungsziel and Erklärungswerkzeug, respectively. PTK, p.161. I shall elaborate this further. 31 In German: einen abstrakten Gattungswert. PTK, p.172. 32 Ibidem. The term kasuistisch reappears so as to sharply separate some kind of ‘fundamental truth’ from

‘superficial truth’.

The examples that give rise to the paradox, like fresh water, air, but also jewellery,

all really fall under the same principle, Böhm-Bawerk believed. So the mistake is

that objective value theorists had introduced more than one principle, and had

missed the point of unification. Note that the underlying premise is that there is an

ontological ground for unification. This unifying principle says that all goods, scarce

or abundant, carry a value, which is a function of two exogenous variables: the

particular sort of need they satisfy and the importance of the satisfaction that sub-

jects are longing for. The principle is represented the following simple table33. In

the scheme, the possible rates of utility of a commodity have been projected with

the two exogenous variables as determinants. The types of needs have been hori-

zontally arranged in order of urgency. The cells with a dash account for needs with

discontinuous satisfaction processes, like fishing water: you can’t fish in one pint of

water. The number of needs within a type have been lined up vertically, again

according to Gossen’s principle that the more needs someone satisfies, the less one

is emotionally attached to it. If three units of a good were in use for the satisfaction

of the first urgency sort of need, the subjective value of the good would equal 8,

because if he lost the good, he would sacrifice the satisfaction of the third need of

type I. This is the marginal utility of the good, because it is the utility for an eco-

nomic agent, given by the last unit the agent has of it.

urgency type of need:

per type: I II III IV V VI VII VIII IX X

1 need fulfilled 10 9 8 7 6 5 4 3 2 1

2 needs fulfilled 9 8 7 - 5 4 3 2 1 0

3 needs fulfilled 8 7 6 - 4 3 2 1 0

4 needs fulfilled 7 6 5 4 3 2 1 0

5 needs fulfilled 6 5 4 - 2 1 0

6 needs fulfilled 5 4 3 - 1 0

7 needs fulfilled 4 3 2 1 0

8 needs fulfilled 3 2 1 0

9 needs fulfilled 2 1 0

10 needs fulfilled 1 0

11 needs fulfilled 0

table I.2 Marginal utility distribution

No rational man will of course distribute the goods exclusively over one sort

of possible use. A bushel of cereal may satisfy the need to alleviate one's hunger,

but also that of one’s bird, the meat of which one can eat later. It can also be used

to distil brandy or to feed an entertaining parrot. These are examples of different

sorts of needs. If you are hungry, the first need comes first. An agent disposing of

33 Adapted from PTK, p.181. Huncke, in Böhm-Bawerk (1959), p.140, copied a somewhat confusing

scheme literally from Böhm-Bawerk, who, in turn, took it from p.93 of Menger’s (1871).

Chapter I

several units of a good will fill out the scheme from top to bottom but also from left

to right. If seven units of a good are in his possession, the subjective value of one

unit of the good equals the marginal utility of that good. In this case it is 7, for if

this subject has to abstain from enjoying the last unit, he cannot satisfy the last

need (in the case shown, the need of type IV; he could as well have chosen to sat-

isfy the fourth need of type I 34). After losing one unit the marginal value of the

good will rise to 8. Thus, the scheme shows how a rational agent computes the

subjective values of goods in and out of his possession.

So how does Böhm-Bawerk explain the low market value of air? Breathing is

very urgent. The explanation proceeds by pointing out, firstly, that air is freely

available, so that, despite the fact that it belongs to the pressing needs of urgency

type I, the value equals a zero marginal utility. Secondly, due to the zero marginal

utility, no one demands it on the market and there is no incentive to supply it. Air

is economically irrelevant (though economically interesting – see below). In con-

clusion, there is no such thing as abstract value. There is only concrete value and

the unifying principle of subjective valuation pervades all possible cases. Menger’s

Subjective Value Theory is a tool in explanatory unification.

The objective exchange value of a good of course falls when supply of the

good is intense. But any understanding of this presupposes a theory of subjective

valuation. Whenever more difficult concrete cases of economic reality blur the pic-

ture of valuation, the theorist can use a passepartout. This is Böhm-Bawerk’s term

and has a strong connotation of ‘unifying principle’. This principle says that a good

in possession is valued by the idea of loss. A good out of possession is valued by

the idea of acquiring it. So to see what the value of a good is, just add an extra unit

of it and count the utility change. Next, start again from the original position, sub-

tract one unit and count again. The lower of the two tells you the value.35

4.2 The market

Tracking down common roots to all the types of value is of interest to psycholo-

gists. According to Böhm-Bawerk, economic theories of value are about the eco-

nomic (i.e. economically relevant) concepts in the scheme only, even if some non-

economic concepts may in a way be economically interesting.36 An example of ex-

trinsic objective economically irrelevant (although economically interesting) value is

the number of Joules a piece of firewood maximally delivers. An example of ex-

trinsic subjective economically irrelevant value is water for a thirsty man who is

sitting ashore at a wealthy stream of fresh water. The subjective element is thirst.

The economic irrelevancy is caused by the wealthy supply of the stream, which

34 A refinement of the table would narrow down the ambiguity here. 35 Böhm-Bawerk does not believe that this counting can actually be carried out. He expresses the idea in

this way as a manner of speech. 36 PTK, p.158.

makes the value of water equal to zero. Still, it is clear that fresh water is economi-

cally interesting. Böhm-Bawerk distinguishes Nützlichkeit from Wert; the former is

merely economically interesting, the latter is economically relevant too.

As to economic goods, we can split into subjective and objective value once

more. Economically relevant objective value is the value goods derive from trade on

the market. That trade is, as we shall see, driven by subjective valuations. Eco-

nomically relevant subjective value (subjektiver Tauschwert) is the non-zero value

attached to a good by any individual. I shall translate it as ‘subjective exchange

value’. Being non-zero the subjective exchange value has market consequences. My

scheme below is a comprehensive representation of Böhm-Bawerk’s many con-

cepts of value.

Scheme I.1 Categorisation of value

Subjective value refers to what the good is worth for a single individual. In a

developed society goods can be used for many different purposes and one of these

is sale. So there is a difference between subjective value in use, derived from how

the subject estimates the value of that particular use (subjektiver Gebrauchswert), and

the subjective value estimated in situations of exchange (subjektiver Tauschwert). Whenever someone estimates the subjective use value of a good lower than the

subjective exchange value, he will be prepared to sell. Perhaps a scholar will not

INTRINSIC VALUE EXTRINSIC VALUE

Eigenwert Wirkungswert aus sinnloser Geiz mit außerhalb der Güter liegendem Zweck

(economically uninteresting) (economically interesting)

OBJECTIVE VALUE SUBJECTIVE VALUE

technisch objektiver Tauschwert (non-economic) (economic)

subjektiver Gebrauchswert subjektiver Tauschwert (non-economic) (economic)

Chapter I

sell an important book even at very high prices, but a bookseller certainly will. The

market, next, turns subjective valuations into the objective value of a good as

‘price’. Let us consider how potential sellers and buyers take their decisions when

in contact on the market.

There will be exchange whenever buyers and sellers engage in opposite

valuations of merchandise (Ware) relative to the exchange good (Preisgut). All par-

ties that meet in the market are in competition. The most competitive (tauschfähig) buyer is the one who has the highest valuation of the sought-after good (das zu erwerbende Gut), i.e. the merchandise, in comparison to the exchange good.37

The most competitive seller is the one who finds the lowest marginal utility

in the merchandise, relative to the sought-after exchange good. The least competi-

tive parties will be excluded from exchange. The market price can be determined in

the way now familiar from textbook microeconomics. In table 3, below, the circum-

stances are depicted in which five horses will be exchanged.38

potential buyers potential sellers

A1 assesses a horse to be

worth at most

300 ƒl B1 assesses his horse to be

worth at least

100 ƒl

worth at most

worth at least

110 ƒl

worth at most

worth at least

150 ƒl

worth at most

worth at least

170 ƒl

worth at most

worth at least

200 ƒl

worth at most

worth at least

215 ƒl

worth at most

worth at least

250 ƒl

worth at most

worth at least

260 ƒl

worth at most

170 ƒl

worth at most

150 ƒl table I.3 Competition in the market39

37 The exchange good is mostly money but this need of course not be the case at all. Examples with

monetary units are in Austrian Gulden, or ƒl. throughout the dissertation. 38 This is Böhm-Bawerk’s own somewhat strange example. Horses do not present us with an example of

homogeneous goods, which must however be assumed here. 39 The table is Böhm-Bawerk’s (1886), p.32. Note that equilibrium price will not be between 210 fl and

215 fl, as the marginal buyer B6 wants to pay less than the potential seller wants to receive.

The price of a horse will turn out to be anything in between 200 ƒl and 220 ƒl, because the five horses can be exchanged in such a way, that salesmen B1 to B5 will

sell for a price equal to or above what they intended to sell it for, with a minimum

of 200 ƒl. Likewise, buyers A1 to A5 will buy at a price equal or below what they

intended to pay for it, with a maximum of 220 ƒl. Salesmen B6 to B8 will be excluded

from exchange and so will potential buyers (Kauflustigen) A6 to A10. What about A1

to A5 and B1 to B5? The salesmen B1 to B4 and the potential buyers A1 to A4 have an

indirect influence on the market process. They do each keep, respectively, one par-

ticular buyer and one seller out of the margin. If A1 were not there, A6 would be

engaged in trade and the price would have developed in between 200 ƒl and 210 ƒl. If, alternatively, B1 were not there, the price would come to rest between 200 ƒl and

215 ƒl. In conclusion, market parties A5 and B5 operate in the margin.40

Given particular institutional conditions – which are generally or approxi-

mately satisfied in a developed market economy – many buyers and sellers will

meet on the market. The margin between the upper and lower limits set by the

most competitive market parties will shrink to, as Böhm-Bawerk phrases it, a sin-

gle point.41 We can see that he uses the concept of infinitesimals without the

mathematics generally involved in such conceptions. Note, however, that this does

not prevent the objective value of goods from being well determined.

Given the way subjective valuations of utility are transformed into an objec-

tive price, Böhm-Bawerk lists six determinants of price. I can summarise these as

four items: the amount of demand42, the intensity of demand, the amount of sup-

ply43 and the intensity of supply. The term ‘intensity’ here summarises the relation

of valuations of the merchandise with those of the exchange good. Demand is in-

tense if, for potential buyers, marginal utility of the merchandise is relatively high

compared to the exchange good. Supply is intense if the inverse conditions hold for

salesmen. Demand and supply can both be intense, in case of which they will cause

intensive trade.

4.3 Objective value theory: true as a limiting case, otherwise false

The ultimate explanatory ground of the price of a traded good lies in subjective

value but it appears that goods derive their value from the costs of production. The

price of finished goods is related to the price of the inputs in the production of

40 A1 to A4 attach a higher marginal utility, or personal value, to the good in question than the marginal

buyer. I see this as a theoretical confirmation of Aristotle’s distinction between ‘value-in-exchange’

and ‘value-in-use’. Schumpeter notes that Aristotle already perceived (in Ethics, V, 1133) that the

former derives from the latter, but adds that this is commonplace. I believe it is not, given the puz-

zles objective value theorists got entangled in (for the discussion by Böhm-Bawerk, see chapter II).

See Schumpeter (1954), p.60. 41 See once more the quotation at note 12. 42 Die Zahl der auf die Ware gerichteten Begehrungen, PTK, p.294. 43 [D]ie Zahl, in der die Ware feil ist. PTK, p.294.

Chapter I

these finished goods. Indeed, reproducible goods tend to acquire a value in the

market ever closer to the level of these costs. This is an ‘observable’ regularity44 and

Böhm-Bawerk refers to it as the Law of Costs (Kostengesetz). This law explains

prices of goods as a function of the effort put into the production of those goods

and it appears to be inferred from an objective as well as from a subjective theory.

Given ideal circumstances (viz. circumstances under which assumptions 1 to 8 are

true) the Law of Costs predicts the same regular succession of phenomena as the

Austrian’s Subjective Value Theory because the mechanism, which levels out dif-

ferences between the prices of finished goods and the prices of their semi-

manufacture inputs, will equate exchange value and production costs. But ideal

circumstances constitute a mere limiting case. Our daily experience may some-

times tell that there is a tendency toward equalisation, the differences between cost

and exchange value persist in a constantly changing practice. No stable result (Er-folg) of the market process materialises and this fact makes the dissimilarities be-

tween Austrian value theory and the competing Law of Costs visible. A subjective

approach uncovers the underlying mechanism of the tendency and, hence, the

direction of the causal chain (Kausalkette).

Böhm-Bawerk uses the example of a variety of finished products made of

iron, with several prices anywhere between 1 ƒl and 10 ƒl, while the price of the

raw material needed is 3 ƒl. Human entrepreneurial skills (Geschäftsgeist) trigger

the exploitation of the difference between the price of the cost good (iron) and the

price of the finished good. Supply in branches that sell finished goods at more than

3 ƒl will be raised until the price of these goods equals 3.- and, consequently, buy-

ers who value it at a lower price will be excluded from trade as extra-marginal

demand. Likewise, whenever a finished good, made out of the raw material iron,

tends to be traded at less than 3 ƒl, the suppliers will stop production.45 Prices will

rise in that line of production up to 3 ƒl. The results of all this do appear to us as if

prices direct themselves toward the cost of production. They are resultant phe-

nomena (Folgeerscheinungen) of the process of correction, but what really happens,

is that [e]s vollzieht sich hier einfach das große Gesetz des Grenznutzens.46 All products,

finished or raw, find their way into the most profitable uses and the price is equal

to the price of the last unit.

Even so, the symmetry of prices and costs is being disturbed by changing

initial conditions all the time. In addition, markets do not work perfectly. Böhm-

Bawerk summarises all these disturbances under the category of ‘friction’ (Rei-bungswiderstände). The conclusion is that, though it explains the facts in absence of

these frictions, the Law of Costs takes its character as a contingent (beiläufig geltend)

44 PTK, p.307. Böhm-Bawerk uses the word Tendenz. The alleged perceptibility of it is implied by phras-

ings like in der Literatur sowie in der Lebenserfahrung on p. 307 and Folgeerscheinungen, on p.311. 45 Obviously, in the context of an indeterminate period analysis, details like the calculation of a shut

down point are not taken to be relevant here. 46 PTK, p.310.

phenomenon.47 Only the competing subjective value theory can take account of

real-life examples. It does so with a story about a sequence of occurrences in a self-

correcting process responding to changes elsewhere. The Subjective Value Theory is a true theory, the proof of which is its ability to describe two things at the same time.

It explains causal processes in a way that discloses the institutional structure, the

rules of the market, underlying these processes and it can accommodate for actual

processes that occur, that is, for instability. Böhm-Bawerk’s point is that the Objec-

tive Value Theory is incapable of doing precisely that. It describes only the process

in the ideal frictionless world and assumes some causal direction accordingly, on

the basis of some superficial experience.

The Subjective Value Theory is presented as a leading principle conceiving

of the market as a social phenomenon, the result of which grows out of individual

decisions. Böhm-Bawerk does go into obvious objections as to the apparent circu-

larity that individual agents also need the same market for their orientations on

subjective valuation, which is where we are to look for finding the start of the

causal chain. But he insists that marginal utility always is die endgiltige Richtschnur’48. The following chapter discusses the epistemological and metaphysi-

cal implications of Böhm-Bawerk’s stance with regard to the Objective Value.

5 A new conception of the present and the future

In the context of the nineteenth century writings on the nature and justification of

interest49, to say that interest is a reward for the supply of capital is begging the

question of what is the nature of capital. Böhm-Bawerk is known for some lengthy

discussions about this in the entire first volume of the Kapital und Kapitalzins. He

introduced time as an essential element in capital theory. It was new to conceptual-

ise interest (the income capital earns) as dependent upon the inclination to estimate

the value of future goods lower relative to present goods. In addition, it was idio-

syncratic to model it as dependent upon the period of production50. Below, I shall

treat Böhm-Bawerk’s much debated ‘three causes’ for the relatively lower valua-

tion of future goods.

47 PTK, p.316. The relevant quote to this claim is discussed in the following chapter, section 2.1. 48 PTK p.299. Valuations that are based on the strategic effort to purchase inputs as cheaply as possible

do matter, Böhm-Bawerk admits, but at best as a psychological step-in-between [bestenfalls eine Art psychologischer Zwischenetappe].

49 As J. Klant remarked: ‘Economics is born from ethics and politics.’ See Klant (1987), p.10. Indeed,

economic research originates in this question (among others): how to justify the payment of interest. 50 However, Schumpeter notes that ‘he had been anticipated, in one essential point by Rae [and] much

more definitely, he was anticipated by Jevons’. Schumpeter (1954), p.846. Of course, Böhm-Bawerk

was in many ways the Austrian Jevons. See Hamberger (2001) for how they relate.

Chapter I

5.1 The central idea

Future goods are not present goods that will be consumed later, but promised

goods that are not here yet. Take for instance credit (to be rewarded by means of

interest). The idea of the influential writings of David Ricardo was that the creditor

lends goods (inclusive of money) to the debtor only to get back the very same goods

later, in quite a similar way as renting a house. Interest, then, is a payment for the

use of something. Böhm-Bawerk, in contrast, notes that the one who borrows, say

kindling or firewood, will hand back some other samples of firewood to the previ-

ous owner than those he actually got after the agreed period of lending. The origi-

nal firewood will have been burnt. This is one of the many examples Böhm-Bawerk

uses that also illustrates his concept of ‘empirics’. Much of what we can easily

imagine is reliable to serve as proof. Apparently, then, the interest (payable in the

form of some extra amount of the wood) is not a reward for borrowing but some-

thing else. If I borrow money from you for two years, you will suffer a disutility as

you have to abstain from the use of that money for the agreed period of time, be-

cause you prefer money now over and above money after two years of waiting. In

other words, goods in the future are worth less than goods now. I should like to

call this a form of ‘liquidity preference’. After all, the supplier of future goods is

prepared to buy liquidity above par. Thus, a creditor will have to give additional

future goods later in exchange for the present goods given out now. This reasoning

is typical for the Tauschtheorie, as opposed to the Gebrauchstheorie, and undercuts

moral judgements to the effect that interest be unfair. (Böhm-Bawerk makes use of

it to attack Marxist arguments into that direction.)

But this central idea encompasses more. In the labour market the wages are

also present goods. Labourers receive them in exchange for their supply of future

goods. So what do wages offer and why are they categorised as future goods? They

deliver labour, not bringing any yield before the production process will have been

finished. Only after the fruits of that labour have been brought to maturity in the

slowly unfolding production process from ore to iron, from iron to steel and semi-

manufactured produce, and ultimately to finished products that can be sold in the

consumer goods market (or in the business to business market placing it even fur-

ther from maturity), the yield is due. So the return for wages is future goods.51

Böhm-Bawerk applies the same simple principle to all markets and thus

achieves a great deal of theoretical unification. All rewards essentially are interest.

Even if consumers buy durable goods, they essentially exchange present goods in

51 Surplus value, as defined by Marx, is nothing more than a just order that takes into account the rela-

tively lower valuation of future goods. Moreover, [d]ie Kapitalrente, welche heute die Sozialisten als ei-nen Ausbeutungsgewinn, als einen Raub am Arbeitsprodukte schmähen, würde auch im Sozialistenstaate nicht verschwinden, sondern gerade von der sozialistisch organisierten Gemeinschaft selbst gegenüber den Ar-beitern in Kraft gelassen werden […] müssen, (PTK, p.437). In his attacks on Marxist reasoning Böhm-

Bawerk always lies the burden of proof at the Marxist side: no proof has been presented for the un-

justness of this aspect of the capitalist order. The justifiability of surplus earnings over and above

wages is a recurring theme in Böhm-Bawerk’s work. See also note 18.

return for goods in the future. According to Schumpeter ‘interest now entered the

theories of rent and wages in an entirely new manner’. Of course, Irving Fisher is

known for his conceptualisation of interest as unifying all types of income, but

Böhm-Bawerk’s theory of capitalisation proves that this was in fact his view, be it

that Fisher’s terminology brought it out better than Böhm-Bawerk’s.52

5.2 The three causes

It has thus far not been made clear precisely what sets the value of future goods

lower than present goods. As stated, there are three causes for this.

First, men generally expect to be better supplied with sources of subsistence

in the future. This is of course not necessarily the case, but it is quite often so, at

least in progressing economies. In times of progress, better economic prospects

dominate. In addition, the value we attach to an extra unit of a particular good –

the marginal utility – decreases as the amount we dispose of rises. In combination

with the expectation of persistently better economic prospects for the future, the

hypothesis of falling marginal utility logically entails a tendency to estimate future

goods of less value than present goods.

Second, people have an unexceptional inclination to irrationality. They fail to

take account of needs in the future as of needs with the same intensity now. It may

be because of bad estimation practices, weakness of will, or sheer uncertainty

whether one will still be alive by the very time that we should rationally include in

our responsible deliberations, like old age. These two ‘grounds’, as Böhm-Bawerk

calls them, have a subjective character.

In contrast, the third cause has an objective character. A unit of labour ma-

tures over the time involved in the chosen roundabout method. The subjective

value of this unit of labour is not determined by the wage level (which is the objec-tive value) but ultimately by the value the buyers of the consumption good attach

to the final good, i.e. the good, which comes into existence at the end of the round-

about production. The third cause can be represented as shown in table 4.53

The average productivity of a month’s labour (ML) that is put into a one

year production technique as of, say, the first of January 1909 till the 31st of Decem-

ber, yields 100 physical units of product. But it is possible to extend the period of

production. Let the beginning of 1909 be the present. Wages are present goods in

this year and the product financed by the wages are the future goods. If the round-

about production is set at a two year period, the labour bought with these wages is

an investment in a future end product maturing, or coming ready, by the end of

1911. The table shows that in this case the average productivity of an ML is 200. If

52 See Schumpeter (1954) p.932. 53 For a reason unknown to me, Huncke swapped the vertical and horizontal axes of Böhm-Bawerk’s

table, showing the figures ‘from the back’. I stick to Böhm-Bawerk’s own for the ease of representa-

Chapter I

the roundaboutness is now set at a three year period, productivity climbs to 280,

etcetera. Compare also different starting dates with the same maturity date. The

table shows that a present ML (i.e. from January 1909) to mature at December 31st,

1914, will yield 440 units, whilst an ML to be set in use not in 1909 but as of the

beginning of next year (1910) will yield only 400 by the end of the same maturity

year 1914. The same principle shows itself if we look back in time (viz. if we as-

sume that the present is not 1909 but, say, 1916) and consider the yield of older

ML’s against younger ones. In sum, as Böhm-Bawerk puts it, the facts are that ‘in aller Regel gegenwärtige Güter aus technischen Gründen vorzüglichere Mittel für unsere Bedürfnisbefriedigung sind und uns daher auch einen höheren Grenznutzen verbürgen als künftigere.54

…yields an average product

in the economic period…

a month’s labour from the year…

1909 1910 1911 1912

1909 100

1910 200 100

1911 280 200 100

1912 350 280 200 100

1913 400 350 280 200

1914 440 400 350 280

1915 470 440 400 350

1916 500 470 440 400

table I.4 Technical superiority of present over future means

Let us now turn to the table 5. This combined table adds the first and second

cause of relative undervaluation of future goods. It includes possible marginal

valuations to the finished goods that come out of each production line. Four over-

views are given, from 1909 to 1912, each with four columns that refer to one source

of a particular ML. Note that the first of these four columns is a repetition of table

4: ‘average productivity’. This column refers to the third cause. The second column

contains a hypothetical distribution of marginal utilities given in table 2 above.55 It

refers to the first cause. But the third column introduces something we have not yet

seen. It refers to what Böhm-Bawerk counts as his second cause. It is the product of

discounting the results that follow from the first cause.

The third column must be read as a correction of the figures in the second

column, as the second cause is a revision of valuations on the basis of the first

cause. (For a totally rational being we would need no correction, as his perspective

would not deform reality. ‘Perspectival’ (marginal) utility would equal marginal

54 PTK, p.339. 55 The numbers in tables 2, 4 and 5 are Böhm-Bawerk’s. The conception of table 5 is mine.

utility.) The fourth column, finally, shows the result in units of future utility. It lists

the subjective values of a ML in each economic period (indicated vertically).

Labour from the year…

1909 1910 1911 1912

…in the economic period…

1909 100 5 5 500 5 5 5 5 5 5

1910 200 4 3.8 760 100 4 3.8 380 4 3.8 4 3.8

1911 280 3.3 3 840 200 3.3 3 600 100 3.3 3 300 3.3 3

1912 350 2.5 2.2 770 280 2.5 2.2 616 200 2.5 2.2 440 100 2.5 2.2 220

1913 400 2.2 2 800 350 2.2 2 700 280 2.2 2 560 200 2.2 2 400

1914 440 2.1 1.8 792 400 2.1 1.8 720 350 2.1 1.8 630 280 2.1 1.8 504

1915 470 2 1.5 705 440 2 1.5 660 400 2 1.5 600 350 2 1.5 525

1916 500 1.5 1 500 470 1.5 1 470 440 1.5 1 440 400 1.5 1 400

second cause

first cause or wahrer Grenznutzen

third cause as technical datum

table I.5 Technical superiority of present over future means and marginal utilities

There has been much commotion about Böhm-Bawerk’s third ground as de-

pendent on the first two, mentioned above. One of the reasons for this is, I believe,

that he explains his third cause while mixing it with the subjective utility of the

yields of production. The technical superiority of present over future means points

into the same direction as the two other causes and the table clearly shows how

they work in tandem. However, without the first and second cause, the valuation

could not be decided upon and we would not be able to spot an optimum in the

valuations. The third cause has an independent effect on interest, but the underly-

ing subjective value patterns make for an optimum.56

56 It seems to me that indistinctness as to the subjective and the technical status of the respective causes

have played a role too. In some critiques, Böhm-Bawerk’s drei Gründe are erroneously altered into

‘three reasons’. See for instance Boianovsky (1998). Böhm-Bawerk sometimes replaces his ambiguous

Grund with the more precise Ursache. Huncke, in his translation (PTK59), correctly talks of ‘causes’.

Chapter I

6 Interest and capital yield

6.1 Interest

Whoever has an urgent need for present goods or estimates their value relatively

high will act as a buyer of present goods. The sellers of present goods are those

people who judge these to have a lower value. In table 6 the price for present

goods is indexed at 100; for future goods at 100 + x. The price mark-up of present

goods relative to future goods consists of the surplus, x, of future goods that buy-

ers are prepared to pay over and above the amount of present goods. The market

price of present goods is determined precisely like that of horses (compare table 3

above). The upper margin of the movement of the price is set by the buyer’s maxi-

mum readiness to pay, the lower margin by the seller’s minimum readiness to

receive.

In the table, the interest will be around 6½%, because the price of 106½ is in

between the offers of the marginal buyers and sellers. When many parties meet on

the market, also in this case, this has the effect of narrowing down the margin be-

tween the two offers that are closest to equilibrium.

sellers

present

minimum amount of

future goods wanted

in return

buyers

present

maximum amount of

future goods ready to pay

S1 100 108 B1 100 105

S2 100 107 B2 100 106

S3 100 106 B3 100 107

S4 100 104 B4 100 108

table I.6 The market price of present goods

According to the Tauschtheorie demand for credit is demand for present pur-

chasing power in exchange for future purchasing power. The interest is the pre-

mium (Aufgeld) payable relative to the amount borrowed. We can think of positive,

zero, and negative57 interest. To illustrate the latter case Böhm-Bawerk gives the

example of a brewer who needs ice in the future while he currently has ice galore.

Ice melts. The brewer will be happy to give some ice to a colleague in exchange for

a smaller amount of ice later.

57 Note that negative nominal interest, the explanation of which is given by reference to inflation, is not

at issue. Inflation is no topic in Böhm-Bawerk’s theory, or at most in embryonic state only.

6.2 Capital yield: durable goods in general

Why do capitalists earn interest? Because they dispose of the opportunities to

lengthen the roundabout production methods. The rate of interest these capitalists

earn is again fully determined by subjective valuations. The relevant utility func-

tions are, again, those of the ultimate future buyers. But how to determine the

marginal utility of future goods? The leading idea, of subjective valuations being

determinant, is that the capital, used to roundabout produce final goods has the

same marginal utility as these future, final, and mature (genußreife) goods. This is

due to the fact that, as scheme II.2 shows, economically interesting value is extrin-

sic: any good derives its value owing to its instrumentality.

Two qualifications have to be made.

The first is that the consumption of the mature goods will take place only in

the future: the future must be discounted to get current values of the proceeds

because of the difference in utility valuation. This is how the subjective utilities of

end produce translate into valuations of intermediary goods. Just like the subjec-

tive valuations per se, the (positive) discount rate also is a function of individual

agents’ given preferences.

The second qualification is, that an intermediary good can be used for sev-

eral different mature goods.58 Intermediary goods or fixed capital goods (aus-dauernde Kapitalgüter) can be put into service for all sorts of mature consumption

goods. But each and every consumption good requires another length of round-

about production. So, for example, it is possible to distinguish final good A, which

will mature promptly59 (i.e. A has a 0 length of production time left over to ma-

ture), from final good X, with a 5 year period to go before maturity, and this one

subsequently from final good Y, with a 10 year period to go. Under the plausible

assumption that discontinuous combinations of capital help produce the final

goods, the subjective valuations are attributed to these combinations. In other

words, the entrepreneur’s own subjective judgement is irrelevant for this question,

because, by assumption, the subjective valuations by future marginal buyers of

final consumption goods do all the work. Had this not been the case, then Böhm-

Bawerk could not have contrasted Subjective Value Theory so sharply with cost

theories, like the Objective Value Theory.

If there are, at t=060, 500 units of capital available, split up in groups of 100

units, then undiscounted marginal values may be as in table 7. This and the follow-

ing table must be read in a largely similar fashion as table 2 above.

58 Appendix 1 discusses that, during the time that elapses before maturity, it is possible to shift the

intermediary goods into other lines of roundabout production. 59 The capital used for A stands , as he phrases it, in Gegenwartsdienst. 60 The variable ‘t’ is a mere time index, not to be confused with for roundaboutness, viz. the capital

intensity of production. In the calculation of the rate of interest, treated below, ‘t=0’ refers to a zero

capital intensity, or the first-stage Robinson economy.

Chapter I

subjective marginal utility (U’), no time preference

types of final goods: A X Y

years before maturity: 0 5 10

U’ for the 1st group of 100 units of capital 6 9 16

U’ for the 2nd group of 100 units of capital 5 8 12

U’ for the 3rd group of 100 units of capital 4 7 8

table I.7 Subjective valuations of fixed capital under alternative ends

The five shaded areas represent the implication that, without discounting, a

utility-maximising entrepreneur would choose to invest his 500 units of capital in

the production of three units of Y and two units of X. No capital is put into the

production of A.

Suppose, next, that the discount rate z (Zins) is 5%, as in table 8. Clearly,

given time preference, maximisation now leads to producing two units of Y, two

units of X, and one unit of A. So how much should the entrepreneur value the capi-

tal goods? Böhm-Bawerk says this:

Allgemein gefaßt: der Wert der Produktivmitteleinheit richtet sich nach dem Grenznutzen und Werte desjenigen Produktes, welches unter allen, zu deren Erzeu-gung die Produktivmitteleinheit wirtschaftlicherweise hätte verwendet werden dür-fen, den geringsten Grenznutzen besitzt.61

subjective marginal utility (U’) and time preference

types of final goods: A X Y

years before maturity: 0 5 10

time preference term: (1+z)t 1.050 1.055 1.0510

U’ for the 1st group of 100 units of capital, U’/ (1+z)t 6 7.05 9.82

U’ for the 2nd group of 100 units of capital, U’/ (1+z)t 5 6.27 7.37

U’ for the 3rd group of 100 units of capital, U’/ (1+z)t 4 5.48 4.91

table I.8 Discounted subjective valuations of fixed capital under alternative ends

Recall that an agent values a good in possession by the idea of loss and a

good out of possession by the idea of acquiring it, i.e. the utility one unit of the

good bestows on the user when he is on the verge of losing that unit (or of gaining

it – and as noted, if there is any difference between these, the lower of the two).

This, then, means that the value of a unit of capital turns out to be 6.

Over time, the discounting term decreases and, as a consequence, capital

‘grows’, as it were, into a higher value. The capital goods used for X grow into a

final value of 8, because 200 units of it will be used for a production line that im-

poses a delay of 5 years to the goods before maturity. In table 8 it is shown why it

is 200 units; table 7 shows why it is a value of 8. But, given the present value of

61 Böhm-Bawerk (1886), p.69. Italics in the original.

these capital goods, the price an entrepreneur will have to pay for a unit of capital

is only 6, while the value it is expected to obtain is 8. Analogously, capital used for

Y is worth 6 now but will grow into a value of 12. Böhm-Bawerk concludes that the

interest on capital for X is 8/6-1 or 33.3% and capital used for Y earns an interest of

12/6-1 or 100%. The better result for the longer roundabout way of producing is

due to the choice Böhm-Bawerk makes in the examples, i.e. to work with figures

that reflect the assumptions discussed in the first section, such as the higher yields

of roundabout production techniques relative to less capitalist methods. The rates

of 33.3% (over 5 years) and 100% (over 10 years) are supernormal yields in the

sense that 5% is the apparent (yearly) normal rate, given the 5% discount assumed

in the example. Thus, for the X line of production, a zero supernormal yearly yield

would have to be 5% or a value growth from 6 to 7.66, i.e. a total of 27.6% over the

full period. The surplus is 5.9 percentage point (33.3% against 27.6%). For the Y line

of production, a zero supernormal yearly yield62 would have to be 5% or a value

growth from 6 to 9.77, i.e. a total of 62.9% over the full period. The surplus is then

27.1 percentage point (100% against 62.9%). Note however that, in this conception,

even the supernormal yield earned is not an unjustifiable windfall collected by the

entrepreneurs and capitalists. It is a natural consequence of the discount future

goods receive due to our psychological constitution (causes one and two) and due

to the technical superiority of present goods over future goods (the third cause),

which, as we see again, places Böhm-Bawerk in opposition to Marxist theorists. He

makes quite a point of the question of what makes the price of labour lower than

the value of the mature product it helps to bring about. The answer lies in the uni-

fying analysis of the labour market as the locus of trade in present and future

goods.63

Durable goods, then, derive their value from the renditions of service (Nutz-leistungen) they give in the future. This is the case both for capital with respect to

entrepreneur and for consumption goods with respect to consumers. Any material

object has the same value as the sum of the services it offers to the owner. Entre-

preneurs reap the advantages of the productivity of capital goods, thus enabling

man to bring forth future final goods in roundabout ways of production.

But the theory of interest is more sophisticated than that. Böhm-Bawerk dis-

tinguishes Rohzins or Bruttoerträgnis (gross yield) from Reinzins (net yield). Both

originate in the very durability of the goods:

Und die Ursache, welcher der Reinzins seine Existenz verdankt, ist nichts an-

deres als ein Wertschwellen der anfangs minderwertigen, aber während der

Gebrauchsdauer in die Gegenwart ein-, oder wenigstens ihr näher rückenden

künftigen Nutzleistungsraten.64

62 ‘Supernormal yield’ is any income from entrepreneurship over and above what may be counted as

‘normal’. A shopkeeper counts as normal the income he would earn with the same activity but on a

payroll. ‘Normality’ also refers to the current long run yield of AAA rated government bonds. 63 For instance in Exkurs VI’, PTK2, pp.121-6. 64 PTK, p.416.

Chapter I

Böhm-Bawerk’s quantitative example explains the general case of the inter-

est on durable goods. Suppose someone disposes of durable goods that bear fruit

with a present worth of ƒl.100 each year during six years in succession65. Table 9

displays the respective values at the start and at the end of each year66 (column (3)),

the non-discounted yields (column (4)), depreciation, for which we also use the

discounting technique (column (5)) and finally, the net yield (column (6)). The

value of this durable good must be calculated in the way shown, with a discount

on the future fruits (say of 5%). The value of the most remote yield is calculated as

100/(1+0.05)6 = 74.62 because it will take six years before the product has gained its

value of 100. The yield that is second most remote in time is worth 100/(1+0.05)5 = 78.35. The value decreases stepwise as the services are ‘used up’ each year. At the

end of the first year the value decreases by 74.62, the next year by 78.35 and so on,

till after six years these goods are exhausted completely. The total value of the

capital goods under consideration is the sum of the discounted values of these

services: 74.62 +78.35 +82.27 +86.38 +90.70 +95.24 =507.57.

Net yield from durable goods; evenly distributed gross yields; 5% discount;

(1) (2) (3)

Year 1st Jan 31st Dec

Gross yield

(5) = (2) – (3)

Depreciation

(6) = (4) – (5)

Net yield

1 507.57 432.95 100 74.62 25.38

2 432.95 354.60 100 78.35 21.65

3 354.60 272.32 100 82.27 17.73

4 272.32 185.94 100 86.38 13.62

5 185.94 95.24 100 90.70 9.30

6 95.24 0 100 95.24 4.76

TOTAL YIELD: 600 92.43

table I.9 Yield of durable goods and growth into ‘presentness’

As noted, durable goods in general constitute the case for both durable capi-

tal goods and durable consumer goods, like a sewing machine. The garments pro-

duced with the help of it will be used more than once. As to the capital which helps

make the sewing machine, its yields are in time twice dissociated from their role in

the satisfaction of needs, the garments. It takes time, firstly, to technically exhaust

capital, but it also takes time, secondly, for the durable end product to mature in

the sense of bearing its fruit over time. (In appendix 2 it is explained how the sew-

ing machine derives its value.) Another possible view is that the gross yield of

durable capital is the sales price of semi-manufactured or half product. Seen in this

65 For the ease of the presentation, all pecuniary figures must be taken to refer to such guilders. I have

corrected some of Böhm-Bawerk’s minor calculation errors. 66 The assumption is that the raping of the fruit, i.e. the revenues of production with this particular set of

capital goods, will take place at the end of each year. Böhm-Bawerk also considers cases in which

this happens at the start of the year. In such cases, due to the shorter (5 year) interval till exhaustion,

the total discount is smaller. The result is a higher value of capital at the start.

way, capital derives its value from the future services of the end product after hav-

ing been channelled into another line of production. Also in this case the yields are

twice (or perhaps three times) dissociated from the future satisfaction of needs.

Land is a durable good too, so its earning must originate in the same way.

However, the yields are perpetual. This entails that there will be no depreciation

(Abnutzung). The conclusion must be, that land earns rent as the total of the dis-

counted values of the infinite proceeds, or, in other words, as the sum of a geomet-

ric series. If the yearly gross yield is 100, like in the previous examples, the series

adds 95.24 +90.70 +86.38+…=2000. Rent is pure income (reines Einkommen) or net

yield. (As we shall see, Wicksell had some amendments to this treatment.)

6.3 Capital yield: the labour market and the subsistence market

Labour is used to generate – in the future – final goods. As noted, labour is a future

good in roundabout production processes. In pressing need of present goods, a

labourer sells it. The conception of the labour market in terms of the exchange of

future against present goods is in line with the explanatory unification Böhm-

Bawerk aims for all over PTK. The fact that he also takes the revenues of labour,

not labour itself, as Zukunftsgüter is problematic. We have seen that Böhm-Bawerk

makes quite a point of the fact that the German term approaches the intended

meaning better than any word in either English or French: future goods in the

sense intended must be interpreted as goods that render fruit in the future but do exist already in the present. This interpretation is in line with Böhm-Bawerk’s own

suggestions, but it will be clear that this makes labour something completely dif-

ferent in a capitalistic compared to a primitive society.67

Whoever has to live without any subsistence means for the immediate satis-

faction of daily needs can choose from two options, either to produce goods in a

non-roundabout way or to sell labour in return for these subsistence means, i.e. sell

future goods in return for present goods (or, alternatively, to demand credit from

banks). In contrast, agents who do dispose of subsistence means can engage in

roundabout production, for they will be able to wait till the produce matures. Such

agents possess sufficient present goods for the marginal utility of future goods to

exceed that of present goods, or so the application of the first two causes indicates.

It is clear that, in the systematisation by Böhm-Bawerk, the credit market

and the labour market both are subcategories of markets, the common stakes of

which are to put in subsistence goods so as to lengthen roundabout production

ways and increase productivity. He explains the functioning of the market for la-

67 Huncke sometimes translated Zukunftsgüter into ‘intermediate goods’. His translator’s note indicated

that, whenever it seemed harmless, he would use the term ‘future goods’ for Zukunftsgüter, other-

wise ‘intermediate good’. The latter is unfortunate. All intermediary goods are future goods in

Böhm-Bawerk’s sense but, while labour certainly is not an intermediary good, it is a future good

nonetheless.

Chapter I

bour by a treatment of the general market for subsistence goods. The concept of a

market for means of subsistence unifies the description of trade in credit, capital,

labour, and land.

Demand for subsistence goods is exercised by three groups of agents: labourers,

entrepreneurs and consumers.68 Therefore the market for subsistence means is a

central concept in the theory. Supply of subsistence goods is total present social capi-

tal69 minus that part, which is needed for tending to needs of the capitalists them-

selves. The more there is, the longer roundabout production intervals may become.

Due to the staggered70 structure of production, the formula according to which this

quantitative relation holds states that, if x is the production interval, (x+1)/2 is the

time for which subsistence means will be needed. Note that Böhm-Bawerk assumes

the arithmetic mean of the invested labour days’ worth of subsistence fund to indi-

cate the average production interval, not the geometric or indeed any other mean.71

This choice is arbitrary. Note, furthermore, that the average production interval as

calculated here is yet another measure for capital intensity. The importance of the

average period (calculated in whatever way) lies however somewhere else. Wick-

sell tried to use a more elaborate form of Böhm-Bawerk’s average period (based on

compound interest) as a measure to relate interest to: interest as a reward for capi-

talists for the supply of a definite amount of capital-value. The difficulty of estimat-

ing the value of heterogeneous capital would be a source of controversy until well

into the twentieth century – the so called Cambridge controversies.

6.4 Positive interest

Positive interest, we are now ready to understand, exists due to demand for means

of subsistence exceeding supply. Positive interest on the market is the materialisa-

tion of the premium on present goods, or, which amounts to the same thing, of the

68 Land owners can be seen as a fourth group. But rent is to be understood as independent from interest.

See PTK, p.402, note 1. 69 Huncke calls this ‘accumulated wealth’. This is faithful to the original that sometimes speaks of aufge-

sammelter Vermögensstock and, elsewhere, aufgestapeltes Vermögen. Yet the translation I propose is

more in line with Böhm-Bawerk’s categorisation of social and private capital. 70 Böhm-Bawerk uses the term staffelweise, or ‘graduating the interest’. 71 Böhm-Bawerk takes for example an absolute five year interval of roundabout production. Labour put

into the lowest stage of production has to wait five years before new subsistence goods discharge

into the total amount of social capital, labour in the next stage waits four years, etcetera. The time the

amount of subsistence means is needed for is then half of 5+4+3+2+1, or 3 years. Graduation into

months will result in a 30½ months’ period, i.e. half of 5 years plus half a month. A subsistence fund

must suffice für die halbe Produktionsperiode und überdies noch für die halbe Dauer derjenigen Zeitstufe, welche der gesellschaftlichen Staffelung der Produktion zu Grunde gelegt ist.

discount on future goods.72 Furthermore, positive interest does not only have a

cause but also an effect. It is the enabling condition for equilibrium.73

To see this, we need to understand the mechanism of the market for subsis-

tence means. Interest proceeds from the growth of the yields of durable goods into

‘presentness’. Böhm-Bawerk uses a reductio argument to show how a positive rate

of interest will force itself mechanistically. Suppose interest were nil. This means

that future goods would be traded at par against present subsistence goods. On the

one hand entrepreneurs would earn no income apart from the proceeds of their

own direct labour, on the other a further lengthening of roundabout production

would considerably increase productivity. The investment would require more

subsistence goods, which were to be purchased cheaply at par against future goods.

This means that the zero interest level assumed would be too low to reflect the

value present subsistence goods have in a capitalist economy. The returns after

investment in more capital intensive ways of production would rise, whilst wages

would not. Note that the cause of the wage level to remain fixed is twofold. The

exogenous cause is Ricardian: competition on the labour market prevents a wage

rise. But, interestingly, there is an endogenous cause as well. The roundabout pro-

duction method is not yet sufficiently long to finance the necessary wages fund,

which can only be augmented after a capitalism driven greater fruit is realised.74

Profitable investments generate a premium on present goods, due to which a per-

sistent zero interest cannot last: it causes roundabout production routes to lengthen

indefinitely. Much in line with the previous paragraph, it leads to the exhaustion of

the social subsistence fund even before new, capitalistic manners of production

could bear sufficient fruit to augment the fund. Again, demand for the means of

subsistence would bid up the price of present goods. (We can see the mechanism of

low interest and associated inflation already in Böhm-Bawerk’s interest theory in

an embryonic state, awaiting theoretical elaboration by Wicksell.75)

72 It takes a review of the Robinsonade above to see that less developed economies have higher premi-

ums on present goods than economies that have already entered roundabout production long ago.

Rising capital intensity decreases marginal productivity in developing economies. 73 Blaug accuses Böhm-Bawerk of begging the question about positive interest, by assuming it and

deriving it as a conclusion of the assumptions. See Blaug (1997), p.481. This accusation is, I believe,

mistaken. It seems to me that he falls victim to the same misunderstanding as Wicksell does about

the nature of the demand for capital in PTK , viz. the demand for a wages fund. 74 One could highlight the endogenous cause by conceiving of an imaginary case, much in the spirit of

Wicksell: were the entrepreneurs to pay higher wages, for instance due to lack of competition on the

labour market, then competition for subsistence goods would drive up prices of the latter. Real

wages would remain the same, i.e. nominal wages rise with inflation. 75 In Wicksell (1898) and (1907).

Chapter I

7 Interest and income distribution

Present goods satisfy present needs, future goods do so in the future. Interest is the

premium on present goods, payable in an amount of future goods and the level of

this premium is an endogenous outcome of the market process. Supply and de-

mand of present goods in exchange for future goods set a margin within which the

premium on present goods develops.

The discount rate on labour in the case of one entrepreneur and one worker

lies somewhere in the margin between the lowest possible price the entrepreneur

wants to receive for the present goods and the highest price the labourer wants to

pay, viz. the highest amount of work he wants to do for a given wage.76 As the

labourer lacks subsistence means, present goods tend to be expensive (the wage

tends to be low) in relatively immature economies. At later stages of development,

wages will rise due to soaring productivity. The interest rate for a consumer loan

with one demander and one supplier also lies in the margin between what the

supplier wants to receive and the demander wants to pay. Consumers’ decisions

are directly determined by Böhm-Bawerk’s first two causes. The level of interest

for producer loans in case of one demander and one supplier will be between the

same limits. Here, the third cause is the prime determinant of individual decisions.

The above assumed cases of single suppliers and demanders are idealised

limiting cases. With these cases Böhm-Bawerk intends to disclose a supposed gen-

eral pattern only to fill it in later with ever more of the real-life details. The ideali-

zation must be rescinded. In the following subsection, the assumption of one de-

mander and one supplier on the market will be suspended. The concluding subsec-

tion is to show the workings of the ‘the rate of interest under the fully developed

market for capital’ (Kapitalsmarkt in voller Entfaltung).

7.1 The level of interest in the idealised case

Böhm-Bawerk asks two questions with regard to interest: what is its source and

how can we calculate its level. The source of interest lies, as we have seen, in the

productivity of waiting. This means that the concept of ‘waiting’ essentially refers

to an independent productive factor. (Nevertheless he insists that land and labour

form original – independent – productive factors and that capital is mere accumu-

lated land and labour.) The level of interest is the quantitative part of the interest

theory.

76 For a treatment of interest it makes sense to interchange merchandise (labour) and exchange good

(wage): the labourer pays labour for means of subsistence. Interest is the extra amount of future

goods, relative to the number of present means, that he produces: the discount rate.

Total demand for present goods in exchange for future goods faces total77

supply. To see the mechanism which makes interest come into being, Böhm-

Bawerk employs some special assumptions. (1) There is a persistent lack of present

goods, and therefore demand by labourers is strongest; (2) labour supply is wage-

inelastic; (3) present goods are homogeneous; (4) the stock of present goods is exo-

genously given; (5) marginal yields are the same in all directions of production and

in all industries;78 (6) labour is homogeneous; (7) wages are independent of capital

intensity (this is implied by the essentially partial analysis, which shows that only

at a wage of 500 the wages fund and supply of labour ‘fit’ each other); (8) there is

no investment in stock, the goods market is cleared; (9) only consumption goods

are produced – there is no ‘loop structure’ of investment79.

Assumptions (2) and (7) allow Böhm-Bawerk to stick to a partial analysis.

Assumptions (2) to (7) are dropped in the concluding chapter, which generalises

toward the distribution theory. In fact, by dispensing with these assumptions, the

step toward his version of a general equilibrium theory is taken.80 Assumption (8)

is associated with the notion of perfect markets. In his concluding words Böhm-

Bawerk refers to it in passing: Wäre die Freizügigkeit des Kapitales eine vollkommene, so könnten die parti-

kulären Abweichungen von der normalen Zinshöhe weder eine irgend erheb-

liche Stärke, und noch weniger eine erhebliche Dauer gewinnen.81

Finally, as to assumption (9), the change from a discussion of fixed capital goods in

the capital theory to a wages fund doctrine in the interest theory provides for an

important simplification but turns out to be a source of confusion for later readers.

Wicksell will note, later, that the initial distinction between fixed and floating capi-

tal has fainted completely in this set-up.82

I shall copy Böhm-Bawerk’s numerical examples but organise them below

more coherently than he (lacking spreadsheets) could have done and correct some

of his small errors. We first have to suppose that supply of labour amounts to 10

77 I use the somewhat vague term ‘total’ in this case in order to avoid possible confusions. ‘Joint’ de-

mand and supply refers to the aggregation of all markets in one type of good, whereas ‘aggregate’ is

the term economists reserve for macro-analysis, i.e. referring to the stock of all goods and services in

the economy. The present goods in question make up a heterogeneous set, but do not include capital

goods (other than the real wages fund). So neither does the term ‘joint’ capture the set very well, nor

does the term ‘aggregate’. The term ‘aggregate’ seems to be formally correct as, mostly, Böhm-

Bawerk conceives of capital as floating only. But his capital theory, on the other hand, distinguishes

between floating and fixed capital and durable and non-durable capital goods. 78 In mathematical terms, this is the assumption of homogeneity of production functions in different

industries. 79 In reality, present goods are invested to produce durable capital goods, but these in turn help produce

newer capital goods, etcetera. This is called the ‘loop structure of capital’. 80 In what follows, Böhm-Bawerk explores the consequences of different wage levels for interest and

distribution. To that extend his analysis is ‘semi-general’ rather than partial. 81 PTK, p.484. 82 Wicksell, WKR, p.121.

Chapter I

million labour years and that available capital is 15 billion83. Equilibrium, then, is

‘the fit’ of quantities of labour to quantities of capital, viz. that these constitute no

sub-optimal surpluses.

Optimum investment periods for two different exogenously given wage lev-

els are calculated below, in tables 10 and 11. For convenience, the symbols in the

headings of the columns are taken from Wicksell’s WKR, since these will be used

again in the next chapter. The wage level is indicated by ℓ and employment by A,

while K refers to the capital stock; p, z and t refer to the yearly monetary yield of

investment, interest, and the investment period, respectively. The second column

lists (digressively rising) returns of a labour year with increased roundabout peri-

ods and the third column shows the digressive rise in marginal figures.

Capital stock: 15 billion (15 b.) Share of capital needed: ½ Wage 300

Labour stock: 10 million (10 m.)

period

pecuniary

marginal

interest

payable*

surplus value

per worker

labour

demand

surplus

interest

‘z’

t p ∆p ∆p ⁄ ½ℓ p-ℓ K ⁄½ℓ t K(p-ℓ) ⁄ ½ℓt (p-ℓ) ⁄½ℓt 0 150 ** ** ** ** ** **

1 350 200 1.333 50 100.0 m. 5.0 b. 0,3333

2 450 100 0.667 150 50.0 m. 7.5 b. 0,5000

3 530 80 0.533 230 33.3 m. 7.7 b. 0,5111

4 580 50 0.333 280 25.0 m. 7.0 b. 0,4667

5 620 40 0.267 320 20.0 m. 6.4 b. 0,4267

6 650 30 0.200 350 16.7 m. 5.8 b. 0,3889

7 670 20 0.133 370 14.3 m. 5.3 b. 0,3524

8 685 15 0.100 385 12.5 m. 4.8 b. 0,3208

9 695 10 0.067 395 11.1 m. 4.4 b. 0,2926

10 700 5 0.033 400 10.0 m. 4.0 b. 0,2667

*Interest payable for ∆p=1 (or the rate of return)

= marginal pecuniary yield ÷ year’s subsistence income.

**Rows xi of columns (3) to (8) refer to rows xi – xi-1 of columns (1) and (2) .

table I.10 Surplus demand on the labour market

The maximum interest payable is what the entrepreneur is willing to maxi-

mally pay for a further lengthening of the production period. A lengthening from 0

to 1 years (the first step into capitalist production) delivers a yearly marginal yield

of 200 per unit of labour that has to be hired. Thus, the maximum amount of inter-

est the entrepreneur could pay, without deteriorating his economic situation, is

83 In contrast to the examples with physical quantities in earlier sections, capital is counted here in

monetary terms. Böhm-Bawerk gives an explicit defence of this practice in PTK, p.441. An assump-

tion is, of course, that real and nominal quantities do not differ: no inflation. See also note 57.

200. As the production loan to finance such an extra unit of labour equals the

yearly wage of 300, the maximum interest rate payable amounts to 66.7%. Assumptions

(2) and (7) allow that marginal wage is equal to average wage. Column four shows

that the maximum interest payable by the maximising entrepreneur decreases with

rising capital intensity, hence the more developed an economy is, the lower the

general level of interest can be. This column can be interpreted as representing the

demand side of the market for capital, the figures referring to the subjective mar-

ginal prices demanders are willing to pay for the product.

The average surplus value of labour comes next. After lengthening round-

about production from 0 to 1 years, the yearly yield per unit of labour is 350. The

surplus value therefore equals 50. In case of a two year production period, the av-

erage surplus amounts to 450-300=150, and so on.

But how much of labour is demanded? Available capital, worth 15 billion,

provides the means to nourish the labourers during the time needed for round-

about production. Note that for each extra year of labour done, the entrepreneur

needs to invest an initial capital of only 150 per labourer, because the wage of 300

is paid out, not entirely at the beginning of the year, but gradually over time, e.g.

monthly or weekly. As a result the average time of investment is half the total time:

the ‘share’ of capital needed amounts to ½. So 15 billion guilders hires 100 million

labourers for one year, or 50 million labourers for two years (they earn 600 each in

these two years, for which an average capital of 300 is needed), or 33⅓ million la-

bourers for three years, and so on. Columns (7) and (8) show that the highest total

surplus is earned with a three year production period.84 Hence, in the optimum of

the entrepreneur there is an excess demand of 23⅓ million labourers. The individ-

ual optimum does not coincide with the social optimum: the wage is too low.

This is different in table 11 below, which shows that, in a 6 year roundabout

production period, the equilibrium wage level of 500 takes an investment of

½(6×500) =1500. Hence, a capital stock of 15 billion hires and nourishes 10 million

labourers during the 6 years of production before the maturity of new subsistence

means. Of course, other social capital stock and other marginal productivity levels

result in different optima. It is easy to see from table 11 that the individual opti-

mum for entrepreneurs coincides with the social optimum. To compare: in case the

wage level is 300, like in table 10, the interest earned amounts to 7⅔ billion relative

to the stock of 15 billion, or approximately 51%. In the social optimum interest is

1½ billion relative to 15 billion, or 10%. Both tables show that higher wages in-

crease capital intensity (roundaboutness of production), coined the ‘Ricardo ef-

fect’.85

84 In chapter II the meaning of column (8) will be subject to further scrutiny. Note that columns (4) and

(8) both represent some form of interest, but that they result in identical figures only in equilibrium.

The ‘interest payable’ is payable only from the point of view of an individual entrepreneur. In the

next chapter we shall also discuss the fact that the tables tend to cause puzzlement due to the double

perspective of a partial analysis (column (4)) and a general equilibrium analysis (column (8)). 85 In The Netherlands this has recently become known as the ‘Kleinknecht’ thesis (Kleinknecht (1994)).

Chapter I

Stock of capital K: 15 billion (15 b.) Share of capital needed: ½

Wage 500

Stock of labour: 10 million (10 m.) prod.

period

pecuniary

marginal

interest

payable*

surplus value

per worker

labour

demand

surplus

interest

‘z’

t p ∆p ∆p ⁄ ½ℓ p-ℓ K ⁄½ℓt (p-ℓ)K ⁄½ℓt (p-ℓ) ⁄½ℓt 0 150 ** ** ** ** ** **

1 350 200 0.800 -150 60.0 m. -9.0 b. loss

2 450 100 0.400 -50 30.0 m. -1.5 b. loss

3 530 80 0.320 30 20.0 m. 0.6 b. 0,0400

4 580 50 0.200 80 15.0 m. 1.2 b. 0,0800

5 620 40 0.160 120 12.0 m. 1.44 b. 0,0960

6 650 30 0.120 150 10.0 m. 1.5. b. 0,1000

7 670 20 0.080 170 8.6 m. 1.46 b. 0,0971

8 685 15 0.060 185 7.5 m. 1.39 b. 0,0925

9 695 10 0.040 195 6.7 m. 1.3 b. 0,0867

10 700 5 0.020 200 6.0 m. 1.2 b. 0.0800

*Interest payable for the lengthening of roundabout production by 1 year

= marginal pecuniary yield ÷ year’s subsistence income.

**Rows xi of columns (3) to (8) refer to rows xi – xi-1 of columns (1) and (2) .

table I.11 Labour market clearance

The wage levels of 300 and 500 assumed in the tables are supposed to come

into being by market forces, but the numerical examples are tacit about this equi-

librium generating mechanism. The Subjective Value Theory Böhm-Bawerk inher-

ited from Menger is believed to take explanatory care of this piece of ‘underlying

reality’. The mechanism, which translates individual decisions into market out-

comes, is supposed to be ubiquitous as far as markets are concerned and, hence, to

really exist and drive equilibrium seeking forces in the capital market just as well

as in the market for horses. Equilibrium interest, then, will rise with ceteris paribus falling social capital, and so it does in case of a ceteris paribus growing labour force

or a ceteris paribus rise in productivity.

7.2 The level of interest in the de-idealised case

I shall now describe what may be called ‘the close of his system’, the last chapter in

which Böhm-Bawerk arrives at his distribution theory, which is meant to answer

the question what brings the whole system to equilibrium when some of the sim-

plifying assumptions are dropped (or, rather, in which he proposed to drop them)

but did not calculate the consequences for the level of interest.86 Giving a partial

86 He could not have done so. This was the job performed by Wicksell who was the one, coming at the

right time, disposing of the required mathematical skills to deal with the resulting complexities. The

next chapter treats the difference mathematics made for Wicksell in comparison to Böhm-Bawerk.

analysis, he postponed the treatment of distribution till what you may call the final strike: ‘The rate of interest under the fully developed market for capital’. It turned

the semi-general analysis into a general ‘model’. Interest theory and distribution

theory merged in a way that then waited to be forged into a system of equations by

Wicksell. From an explanatory point of view, the use of partial analysis while there

are numerous markets is unproblematic. Markets communicate and arbitrage will

produce a converging tendency. The forces triggered by differences in general

supply and general demand have more noticeable macro-effects than the forces

triggered by local sub-markets.

Let us consider the effect of dropping the assumptions one by one. I do this

in a particular order that is convenient for the discussion. First, dropping assump-

tion (11), we can see that different industries will be faced with different produc-

tion functions (or, as Böhm-Bawerk phrased it, different marginal product patterns

or technical relations). This entails the existence of a whole number of entrepre-

neurs looking for present goods, each of whom attach another marginal utility to

the acquisition of these goods. But the degrees of freedom for the price to material-

ize shrinks as more market parties gather. Böhm-Bawerk reasons that dropping

this assumption does not put the idealised theory into jeopardy at all. Thus, as a

consequence of market processes, present (subsistence) goods will first flow into

the industries with the highest productivity rates. Dropping assumption (12) has

the same sort of effect: heterogeneous goods allow for more markets and as long as

many demanders and suppliers (assumption (2)) gather to exchange these and,

given the prevalence of full, free and perfect competition, some determinate equi-

librium will be attained.

Assumption (1) has been mentioned in passing already in the earlier chap-

ters. Government interventions alter market results, either toward sub-optimality

or toward optimality, the latter for instance in the case of antitrust policy. Assump-

tion (2) loses its idealising character as now an entire society is in focus. Likewise,

assumptions (3) to (8) are unproblematic in the last chapter. Assumptions (7) and

(8) are now seen from their ‘result perspective’. All considerations about the

mechanism by which tendencies toward equilibria develop are set aside. Assump-

tions (3) to (6) may not be as harmless, but Böhm-Bawerk sees no trouble.

Assumption (9) is and remains the big problem. Wicksell would later try to

drop it and allow that capital has a loop structure87, turning out to cause many

difficulties and ending up in the infamous Cambridge controversies.

What about the more general assumption that the number of present goods

is exogenously given? This is another weakness, related to the disconnectedness of

the capital and distribution theory. Thus, the partial analysis has it that wages are

exogenous and, in a stationary state economy, constant. Böhm-Bawerk notes that

his theory of the relation between wage and subsistence fund appears to have

many similarities with the English wage fund theory. He stresses that wages tend

87 See note 79.

Chapter I

to rise if longer roundabout periods are taken (e.g. if the supply of capital rises)

and judges the inclusion of this phenomenon into the possible set of instantiations

of his model as more advanced than the English classical wage fund theory, be-

cause it endogenises the wage level: Die englische Theorie will, daß die Lohnhöhe der Arbeiter einfach aus einer

Division des Lohnfonds durch die Zahl der vorhandenen Arbeiter resultieren

soll. Das ist ganz falsch.88

The wages-fund theory is indeterminate as to what causes the size of that fund.

Böhm-Bawerk’s theory is in fact equally indeterminate as to the size of the subsis-

tence fund, but at least to his own understanding he clarifies some relations be-

tween wages and degree of ‘capitalism’. While the English theory allows that an

increase in the wages-fund ceteris paribus raises the wages too, the Austrian theory

explains how: an increase in the number of subsistence goods will flow into the

project of lengthening the roundaboutness of production first. Due to falling mar-

ginal productivity the distributive share of interest, i.e. in gross domestic product,

will fall. The relative wage earnings will rise, be it not proportional to the initial

rise of the fund itself. As Böhm-Bawerk puts it, [d]ie englische Lohnfondstheorie ent-hält somit ein Körnchen Wahrheit eingehüllt in eine ganz überwiegende Masse des Irr-tums.89

The pretension implied by Böhm-Bawerk seems to be that he offers a general

equilibrium analysis with the same success. Indeed, he elaborates on relationships

into which the subsistence fund is interwoven with the capital intensity of the

economy, but the size of it remains undetermined. He lacks the mathematics to

really endogenise wages and this is a dead end in his theory. So precisely in the

chapter, which had to be the final strike, interest is indeterminate. One of the de-

terminants of interest is, anyhow, the supply side of the capital market: the subsis-

tence fund of present goods. In his rephrasing of Böhm-Bawerk’s system, Wicksell

tried to give the distribution theory and some of the key aspects, like the nature of

the subsistence fund, a more determinate treatment. As Böhm-Bawerk had worked

out the interest theory with not much more than hints toward a distribution the-

ory, some of his work as we read it now bears the character of a partial analysis,

and some of it refers to a more general analysis. Wicksell’s endeavour would not

be easy.

88 PTK, p.481. 89 PTK, p.481. The ‘grain of truth’ presumably lies in the fact that the Classical Economists also used a

wages fund concept of capital; the mass of error, then, lies in the disconnectedness of this concept

with an interest theory.

8 Conclusion and summary

Böhm-Bawerk pretended to develop a general theory of production and distribu-

tion with capital not as given, but as itself subject to decision making by economic

agents. However, his capital theory remained disconnected from his interest and

distribution theories, as the possibility to save and dissave played no role in the

rest of his work. Overall his model was one of a stationary state, that is of an econ-

omy with constant capital over time. The fact that positive interest existed in this

economy nevertheless was revolutionary. In the following chapter I shall spend a

word on Wicksell’s admiration for Böhm-Bawerk’s important insight.

He used a peculiar rhetoric. He produced economic models subjected to a

number of assumptions, some of which would be dropped successively. When

seeing assumptions dropped the reader is struck with the impression that the the-

ory is capable of explaining model situations and real situations alike. This comes

to the fore best when Böhm-Bawerk explains Menger’s theory of value. The Subjec-

tive Value Theory (SVT) is foundational for the entire work, the Positive Theorie des Kapitales, or PTK. He was able to show that the theory can explain a market process

that leads to (stable) equilibria and one that does not lead to any stability, but re-

mains in constant flux due to ever changing initial conditions. The Classical Theo-

rist’s Objective Value Theory (OVT) ignores transformation processes and cannot

explain the absence of equilibrium other than by a lot of ad hoc mending. The Aus-

trian approach opens the black boxes that markets appear to be in the classical line

of thinking. The theorists in the Austrian tradition all found a kind of mechanism,

even though they did not all find the same mechanism.

The mechanistic approach gave time an important role in the unravelling of

complex market processes. But in Böhm-Bawerk’s political economy time has two

different (and complementary) meanings. The first meaning is important in the

value theory and in the parts of PTK where the value theory is presupposed. It is a

concept that comes closest to the common sense interpretation of the word. Time

evolves and, often, before a tendency toward some equilibrium state has devel-

oped itself to a point where one can observe the tendency in the succession of phe-

nomena it has been counteracted by other tendencies. Thus, if initial conditions

remain in flux, a particular equilibrium state, which may be expected to develop on

the basis of the theory, simply will not be achieved. This aspect of life fits well into

a social theory that focuses on human action as being of a foundational explana-

tory kind. But in later parts of PTK the current of time is absent from his explana-

tions. The interest theory helps calculating equilibria that seem to arise under the

force of stability assumptions. The current of time is ‘isolated’ and put out of sight.

The second meaning of the crucial word ‘time’ allows it to have a constitu-

tive role in capital formation and in the dynamics of individual utility maximiza-

tion. Time is also a productive factor, much in the sense in which money is produc-

tive: without it fewer transactions take place. In the course of capital building ever

more productive ways of bringing forth goods are made possible. Capital intensive

Chapter I

production is ‘roundabout production’. To fish with a fishing rod is more efficient

than to do it barehandedly, but for the rod time must be invested during which no

fish can be caught. Hence, the development towards more roundabout production

methods requires abstinence. Furthermore, labour is a future good in the sense that

it will yield mature goods only after the production process has been completed.

One can run a business with labour, but present goods – the goods the labourers

are going to consume during the years of labour – have to be supplied immedi-

ately. Present goods are worth more than future goods.

As any market can be subsumed under the description of the exchange of

present against future goods, the concept of time in this second meaning enabled

Böhm-Bawerk to unify many economic phenomena. Unification is here taken to

mean the subsumption of what, superficially seen, is a diversity of phenomena

under few explanatory hypotheses and with the help of a slim conceptual appara-

tus. Thus, for Böhm-Bawerk, good theories grow out of the combination of ex-

planatory and conceptual unification.

At the end of his Magnum Opus Böhm-Bawerk generalises many hypothe-

ses elaborated in previous chapters in order to bolster the distribution theory.

Many of the assumptions that limited the domain of application of his model are

now dropped and he achieves a form of ‘de-isolation’. It is not straightforward

whether this amounts to concretization – leading to a decreasing level of abstrac-

tion – or to dropping restrictive clauses. The question which is what will be made

very important in the coming chapters.

I shall first investigate the reception by Knut Wicksell of Böhm-Bawerk’s

work and draw conclusions about the specific type of realism that comes with

Böhm-Bawerk’s ways to theorise. At least in his theoretical work the economist

and minister from Vienna has some essentialist inclinations. It is difficult to ascer-

tain whether his essentialism explains his antagonistic ways of expressing himself

in the Methodenstreit, or the other way around.

II Realism and explanation

1 Introduction

Eugen von Böhm-Bawerk was convinced that his theories approached the eco-

nomic truth closer than competing theories of income and distribution. Although

in method he was opposed to the German historicist – non-theoretical – conception

of good social science, he took above all on the British Classical Economists who

had studied Political Economy theoretically.

Section 2 shows how the quest for conceptual unification was the main en-

gine that directed the development of the Böhm-Bawerkian position opposing the

British objective value theories. Section 3 discusses the essentialist epistemology

that is involved in his idea of the existence of a leading principle that tells the sci-

entist what aspect of economic reality is primary and what secondary. The primary

aspects are the really explanatory ones, but they are often ‘underlying’, in the sense

that these do not easily come to the phenomenological fore. Deduction and theo-

retical research is needed.

The Swedish economist Knut Wicksell is known as the economist’s econo-

mist as he reformulated the theories of others. His 1892 Kaptalzins und Arbeitslohn

(K&A) summarised Böhm-Bawerk’s system. The 1893 Über Wert, Kapital und Rente (WKR) condensed the interest and distribution theory of Böhm-Bawerk into a

mathematical form. Section 4 summarises this mathematical model and assesses

the advantages and disadvantages of a mathematical approach in contrast with a

discrete and numerical ‘Austrian’ approach. I maintain that Wicksell, though high-

lighting much of Böhm-Bawerk’s obscure verbalization of his theories,

misinterpreted the Austrian’s intentions. I stress the merits of Böhm-Bawerk’s ap-

proach and conclude that one element of his reasoning that confused many later

commentators – Wicksell included – can be seen as the demand curve for capital in

a Marshall-like partial analysis. Some of Wicksell’s criticism is rejected.

Section 5 follows up on this and leads to the conclusion that, although a

mathematical analysis helped Wicksell to calculate market results by way of com-

parative statics, its very sophistication obscured the mechanical genesis of these

results. Böhm-Bawerk had not wanted to assume this ‘underlying’ mechanics

away. Against this background the Austrian method in general, often rejected as

clumsy, comes to stand in a different light.

Chapter II

2 The Subjective Value Theory as a unifying tool

I shall now look into the way Böhm-Bawerk defended the Subjective Value Theory

of his predecessor Menger – referred to below as SVT – against the classical theo-

ries summarized as Objective Value Theory – which will be referred to as OVT. A

core theorem in OVT was the Law of Costs. The debate over the Law of Costs was

Böhm-Bawerk’s major instrument for convincing the scientific community that

SVT was the only true and fully explanatory theory of prices. The first time that he

developed his argument was in his (1886) ‘Grundzüge’. A newer version became the

bulk of chapter II of Book three of Positive Theorie des Kapitales (1888, also of the

second volume of Kapital und Kapitalzins). He also gave a critique of the Law of

Costs in the excursion VIII: ‘betreffend den Wert von Produktivgütern und das Verhältnis von Wert und Kosten’ published in the third volume of the (1912) third

edition of Kapital und Kapitalzins.1 In this polemical essay he strengthened his

stance and responded to various criticisms.

2.1 The Law of Costs and the Law of the Marginal Agents

The Law of Costs (from now on: LoC) claims that end prices are equal to costs of

production. This appears to be confirmed by a tendency of prices to respond to

changes in productivity and output. The OVT theorists concluded not only that the

trigger of such a tendency lies at the input (cost) side but also that the cause of a

settlement of costs with end prices is to be found there. The trigger and the cause

are not one and the same thing. A succession of phenomena may be triggered by an

intervention (or by an ‘autonomous’ change in a variable) while the cause of a par-

ticular end state of this succession can best be explained by reference to a make-up

of the world at some place other than where the trigger strikes: Die größere Häufigkeit eines Produktivmittels ist (indirekt) Ursache des

geringeren Wertes des Produktes; aber der ebenfalls indirekt hieraus

entspringende geringere Wert der Produktivmittel ist trotzdem nicht Ursache,

sondern Folge des geringeren Wertes der Produkte.2

There is an endogeneity problem at stake here. Böhm-Bawerk’s point was

that the start of the causal chain with a change in the amount of inputs is not a at

one fell swoop also the start of another causal chain with a change in the value of

the inputs. How can this be? Ornate examples in PTK are furnished with such col-

ourful end products as tobacco, safety belts, ice-machines, Johannisberger wine,

distilled alcoholic drinks, and leaking ships. Here are three of them.

The Law of Costs, Böhm-Bawerk contended, would never satisfactorily ex-

plain the phenomenon of a rising value of land after the introduction of a fixed

1 The second and the third volumes of Kapital und Kapitalzins also appeared as the first and the second

volumes, respectively, of Positive Theorie des Kapitales, 4rth edition, 1921. As in chapter I, I shall refer

to the first volume as PTK and to the second as PTK, Vol. 2. 2 PTK, Vol. 2, p.188.

(high) export price for agricultural produce.3 The objective value theorist would

have to start his explanation at the higher price of land and next move to the legal

minimum price of exports, that was introduced by the administration. Böhm-

Bawerk called in David Ricardo as witness, who had written that ‘Corn is not high because a rent is paid, but a rent is paid because corn is high’.4 As it happens, this is an

exception to the very law defended by the objective value theorist Ricardo! More-

over, Ricardo leaves the exception unexplained. To be sure, he shared with

Malthus and Von Thünen the view that a high demand for agricultural produce

leads to the employment of less fertile portions of land. This process will thus

make those sections of land, which have previously been taken into possession,

scarce. This scarcity is in turn the source of income for the land owner. The Ricar-

dian explanation runs in the same direction as SVT, but is plainly inconsistent with

the objective approach! Böhm-Bawerk held that explanations of real market phe-

nomena, which situated causes at the subjective end – the end where individual

choices are made – are more consistent and unifying. Ricardo’s alleged mistake

had brought his very own theory in jeopardy.

A second example starts by asking what the causal origin is of the practical

worthlessness of a new but leaking ship. It can not be the costs incurred in making

it, for sure, which cause the low value of the ship.5 The OVT theorists had to reach

out for an explanation of the paradox and contended that costs are the cause of

value and price, but with the exception of goods without utility.

The third example is a simple question: whence the deviation between the

historic production costs of an irreplaceable good, like a rare jewel, and its high

price? Well, the OVT theorists contended, costs do still determine value and price,

but only of replaceable goods. OVT theorists, Böhm-Bawerk diagnosed, piled

clause upon clause to get it their own way and they were thus prone to get entan-

gled in ever more complex modelling. But “[j]ene Klauseln markierten die Bedingungen unter welchen die Kosten selbst in Übereinstimmung mit dem Grenznutzen bleiben“.6 The conceptual problems OVT theorists got entangled in already show

the way to SVT. Böhm-Bawerk was convinced that the SVT-inspired way of ex-

plaining was also capable of dealing with the Reibungswiderstände (friction): no

further complexities had to be added to the mechanistic explanations, even of cases

in which friction was allowed, because the relevant assumptions were dispensable

with regard to those explanations.

Menger had proposed the Law of Marginal Agents (from now on: LoMA) as

a theorem of SVT. The example of the horse market in the previous chapter (section

4.2) illustrates this explanatory law. LoMA says that the agents – buyers and sellers

3 PTK2, p. 191. 4 Ricardo (1996 [1817]), p.50, quoted by Böhm-Bawerk in PTK2, p.175. 5 Böhm-Bawerk (1886), p.72. 6 Böhm-Bawerk (1886), p.73. Böhm-Bawerk often stressed the impossibility to consistently follow the

objective reasoning to its end. See particularly pages 61-73, where he develops his views on the

value of inputs in complicated cases.

Chapter II

– who ‘operate in the margin’ determine price.7 But if the market is crowded

enough, it is ultimately the marginal utility of the demander that sets the objective

value, or price, because demanders are not prepared to pay more for a good than

the marginal utility of fulfilling an extra need for that good (see table II.2). So how

does LoMA deal with friction?

In everyday practice market equilibria continue to be disturbed. Indeed,

friction and disturbance is rule rather than exception:

Solche „Reibungswiderstände“ gibt es in der Praxis unzählige. […] dadurch

nimmt das Kostengesetz seinen bekannten Charakter eines bloß beiläufig

geltenden, über und über von Ausnahmen durchsetzten Gesetzes an.8

Whenever indispensable assumptions threatened OVT, classical economists dog-

matically tried to rescue this theory. As Böhm-Bawerk’s examples show, defending

that costs cause the value of end products requires all sorts of clauses, which mend

the explanations. So to the Law of Costs the disturbances are exception instead of

rule. But how can this be if costs rarely are equal to prices?

Mengerian SVT, in contrast, explains the precise succession of events during

disturbances in the following manner. The marginal utilities of final buyers deter-

mine the price of inputs. A greater abundance of inputs (a fall in scarcity) makes it

possible that a larger share of end users’ needs are satisfied, a process by which the

utility of an extra unit of satisfaction drops. Hence, the most competitive buyer is

not prepared to pay as much as he did before the fall in scarcity. (I have phrased

this as a lower intensity of demand.) The confrontation of the amount and intensity

of demand and of supply results in (objective) market prices. Thus, the increased

productivity of an ore mine decreases both the price of steel and of the ore itself;

but it does so with the utilities of the end users as prime causal instance, in spite of

the fact that the trigger of the process is the discovery of a new lode or the design

of a newly organised production process; that is, notwithstanding triggers lying at

the cost side. As a regularity, rather than as an explanatory law, the ‘Law’ of Costs

is a theorem of SVT (in conjunction with the assumptions 1 to 8; see section 2.4 of

the previous chapter) too because, with these assumptions holding, a convergence

of end prices and costs is to be expected.

But to Böhm-Bawerk it was clear that the classical theorists assumed the Law

of Costs to be confirmed if it was taken as a mere ‘tendency’ toward equilibrium. If

costs and prices deviate too much, disturbing circumstances allegedly obscured the

picture. Rid these and initial conditions are stable: no friction. This is the case if

assumptions hold.

In sum, the LoC is implied by OVT in conjunction with the assumptions 1-8

only, because it is a theorem of the ideal, limiting case. If the Law of Costs is con-

7 Huncke, in the 1959 translation of Böhm-Bawerk’s Positive Theorie des Kapitales calls it the law of mar-

ginal pairs, after the German ‘Gesetz der Grenzpaare’. Huncke thereby refers to the marginal goods

(the cost good, like money, and the good for sale) instead of to the agents. The Austrian value the-

ory focuses however on how demanders and suppliers take decisions in the margin. Therefore the

term Grenzpaare must be read, I believe, as referring to the marginal buyers and sellers involved. 8 PTK, p.316.

firmed, so is OVT. If the assumptions are false there is no longer such implication.

Hence, an actual unstable environment falsifies the Law of Costs and, hence, OVT.

Compare this to LoMA as a theorem of SVT. Even without the assumptions, LoMA

is true, and it is explanatory. This means that Austrian SVT encompasses both ideal

cases and real world cases. The theory Menger had proposed before Böhm-Bawerk

had the power to bring together descriptions of aspects of real world cases with

aspects of strongly idealised cases. One can say that SVT unified characteristics of

the actual and of hypothetical worlds under one explanatory principle. The scheme

above gives Böhm-Bawerk’s idea in relation to the key assumptions. It shows the

advantage of SVT over OVT: that the question whether assumptions 1 to 8 are ful-

filled is irrelevant. If the assumptions do not apply, the processes triggered by

changing initial conditions, or by market imperfections, can be described by SVT,

just as well as the end results of these processes, viz. the new equilibria.

[Arrows indicate deductive inference]

scheme II.1 The Law of Costs: Explanatory power of competing theories

The current subsection investigates how Böhm-Bawerk could use this unify-

ing explanatory power as a confirmation of the truth of SVT. Menger’s theory

explained the order of the phenomena if something changed at the input side of

production. If this worked, whence the appeal of the competing OVT for classical

economists? The answer lies in the observable tendency of prices and costs to often

(though not always) converge. As I try to explain in the following subsection, this

is a regularity that the OVT theorists mistook for an explanatory law.

…do not hold

Objective Value Theory (OVT)

“false”

Subjective Value Theory (SVT)

“true”

…hold …do not hold

assumptions 1 to 8:

OVT accepted

by the supposed

’experience of a

tendency’

(price equal to cost);

…hold

LoC is confirmed

assumptions 1 to 8:

LoC is falsified LoMA is confirmed

OVT defended

against apparent

deviations from

the tendency;

SVT best explanation

Chapter II

2.2 Layman’s observation and empirical justification

To recapitulate, objective value theorists believed the Law of Costs to hold at all

times, not only in very special cases. But it turns out to loose its explanatory power

whenever markets are out of equilibrium. This is the point where Böhm-Bawerk

took on the perhaps appealing but false theory.9 He welcomed any disturbance one

thought of and set it to work for mechanistic reasoning.

Objective value theorists could only sustain their theory, Böhm-Bawerk

pointed out, under the condition that all the idealising assumptions held. The Aus-

trian SVT had no such problem. These assumptions are dispensable with regard to

SVT and not dispensable with regard to OVT. Dropping an assumption does not bring the subjective value theory into jeopardy, but, on the contrary, provides extra support for it. The problem with the Law of Costs is that it makes it seem true that costs determine

price, if the price is considered as a mere result of a process: Ginge die Produktion in – praktisch undenkbarer – idealer Vollkommenheit

vor sich, ungehemmt durch die Schranken des Raumes, zeitlos, ohne jede

Reibung, […]: dann würden auch die originären Produktivkräfte mit idealer

mathematischer Genauigkeit in die lohnendsten Verwendungen investiert,

und das Kostengesetz würde […] in idealer Reinheit gelten. Es würden die

komplementären Gütergruppen, […] genau denselben Wert und Preis

behaupten, also das Genußgut genau seinen Kosten […] gleich gelten, bis zu

den letzten originären Produktivkräften herauf, […]. Diese ideale Symmetrie

wird aber durch zwei Störungsursachen durchkreuzt.10

The first ‘disturbing cause’ he alludes to in the last line of the quote refers to

friction. Allow friction and you will allow sluggishness in market adjustments and,

hence, time, the existence of which is implied by a mechanistic analysis.11 But

prices of goods – as end states of market adjustments – are more easily visible than

the mechanisms by which they come about. In consequence, the Classical Econo-

mists tended to impose a direct causal relationship on the phenomena of rising

supplies and lower prices, or on the equality of costs and end price. It is easy to be

seduced by the appeal of the Law of Costs as an alleged causal relationship. The

appeal can be strong, but that does not make the causal relationship in any sense

‘fundamental’. Hume saw our knowledge of causes to be ‘a determination of our

mind to form a more lively idea of an effect out of an impression of the cause’.12

With Hume, then, we could say that the Law of Costs appears to reflect a causal

9 For instance in PTK, p.427, note 1: Die Unrichtigkeit einer Theorie erweist sich daran, daß sie nicht für alle

vorkommenden Fälle eine befriedigende Lösung zu geben vermag. 10 PTK, p.315. 11 The second disturbance is the course of time needed before a means of production yields its fruit [der

Ablauf von Wochen, Monaten und Jahren, die verstreichen müssen zwischen dem Einsatz der originären Produktivkräfte und der Darbietung ihres genußreifen Schlussproduktes (PTK, p.316)]. Note that this is

not time as an existence condition of mere sluggishness of adjustment processes in the market, but

time due to which the genesis of interest is possible, i.e. as a productive factor. This concerns his in-

terest theory, not his value theory. 12 Hume (1890 [1739]), Book 1, Part 3, Section 14, and the appendix to the first volume.

ordering.

But clearly, ‘the mind’ may be wrong in seeing causes and this is what

Böhm-Bawerk maintained with respect to the Law of Costs. He offered an alterna-

tive explanation, which was empirically more successful, although the evidence is

rooted in layman’s observation. As an economist, he transcended lay knowledge

not by more scientific ways to observe, but by ‘redescribing’13 the phenomena in

terms of some underlying mechanism. As the mechanistic treatment of economic

phenomena enables the SVT theorist to account also for other than only limiting

cases it is an explanatorily better theory. It describes what happens if market parties

meet regardless of idealised equilibria, and how this happens, while it also tells us

what the result will be in the ideal case. OVT, and its quasi-law LoC, only tell us

about a particular result, not about what happens if the process leading to this re-

sult is disturbed. LoC cannot be cashed in as a true law. Or, in other words,

whereas LoMA is an explanatory law, LoC merely describes a weakly instantiated

regularity.

The claim of comparative success of SVT relative to OVT finds justification

in its being instantiated by the evidence twice instead of once. OVT is confirmed by

the occasional tendency of prices and costs to converge only, but it is falsified

when they diverge. SVT is confirmed in both cases; SVT, in combination with the

appropriate respective initial conditions, allows for both convergence and diver-

gence.

The following table summarises the positions.

OVT ╞ LoC SVT ╞ LoMA

convergence explanation explanation

divergence disturbance explanation

table II.1 Explanation of evidence by the two theories and their key laws

Böhm-Bawerk displays an interesting way to exploit the evidence in support

of the subjective camp. On the one hand the reader is often persuaded by the refer-

ence to claims, which are supposed to be ‘obviously’ true, about everyday

experience. Böhm-Bawerk regularly expressed himself in terms like ‘[h]ierüber hat, wie ich glaube, jedermann reichliche unmittelbare Erfahrungen’, and so on, to prove his

point14. On the other hand, appearances could be deceptive and layman’s observa-

tion was to be distrusted in favour of scientific vision. Es gibt eine Fülle von Beispielen, in denen die von mir behauptete

13 The term has been introduced by Uskali Mäki in the context of Austrian economics. See Mäki (1992a).

I shall deal with it more extensively in appendix 3. 14 PTK, p.123.

Chapter II

Abhängigkeit des Wertes der Produktivmittel vom Wert ihrer Produkte auf

das sinnfälligste zu Tage tritt, und es gibt – bei sorgfältiger Analyse – keinen

einzigen Fall, der sich als Probe für ein paritätisches Verhältnis zwischen

beiden bewähren würde: ein flüchtiger Anschein hiefür kann entstehen, der

aber einer sorgfältigen Analyse nicht lange stand hält. 15

Böhm-Bawerk’s frequent reference to friction accords with his frequent use

of the method of introducing and subsequently dropping assumptions that helped

demonstrate that SVT is true. Dispensability of assumptions was the key justifica-

tion criterion. He thereby defended his claim that SVT could explain better than

OVT by referring to the scope of the explanations in terms of a subjective approach

in comparison to those in terms of an objective approach. All possible disturbances

could be dealt with by the true theory. SVT was able to cover a manifold of phe-

nomena as instantiations rather than as disturbances, while the competing theory

could not do this.

3 Essentialist metaphysics

Above I contended that Böhm-Bawerk did not pretend to scientifically transcend

lay observational methods but that he did believe to offer really scientific explana-

tions. The description of the explanandum was not better from what anyone could

tell, but the description of the explanans. In the famous review of Schmoller’s Zur Litteraturgeschichte der Staats- und Sozialwissenschaften, a book celebrating the fiftieth

anniversary of Roscher’s doctorate16, Böhm-Bawerk acknowledged the importance

of phenomenal reality – accessible to non-scientists – as a guide for the scientist,

but he stressed its uselessness if not helped by what he called geistige Durchdringung.17 This argument gave him ammunition to defend the so-called ‘de-

ductive method’. There are superficial and deeper ways to observe and, noticeably,

his was deeper. His convictions are directly related to his very strong kind of real-

3.1 The ordering principle of the phenomena

Böhm-Bawerk accused not only objective value theorists of mistaking more acci-

dental phenomena for the expression of leading principles. He did so, throughout

15 PTK Vol. 2, p.187. 16 Böhm-Bawerk’s review was translated in the same year by Henrietta Leonard in the Annals of the

American Academy as ‘The historical versus the deductive method in political economy’ and came to

stand as the fiercest expression against the German historical method of political economy. Wilhelm

Roscher was the predecessor of Gustav Schmoller and he had expressed his views on the historical

method in a much more moderate fashion than his pupil Schmoller did. See Böhm-Bawerk (1890b). 17 Böhm-Bawerk (1890b), p.85.

PTK and in other work, with regard to all of his opponents. An instructive passage

can be found in his discussion of Ricardo’s interest theory: Die Signatur der Rententheorie Ricardos, die im wesentlichen bis heute

herrschend geblieben ist […] ist: eine Fülle von Wahrheit in falscher prinzi-

pieller Formulierung; eine glänzende Kasuistik, die nur die Anknüpfung an

das erleuchtende richtige Prinzip nicht finden kann, und die darum, nachdem

sie ein Stück der Erklärungsbahn hell beleuchtet hat, in Dunkelheit und

Irrung ausmündet.18

Böhm-Bawerk’s point is that facts without some ordering principle do not indicate

anything as to the direction of, for instance, causes. This ordering principle is un-

derlying in the sense of ‘not immediately visible’, it can be hypothesized, it can

even be seen by some ‘inward eye’, though it must be somehow tested by verifica-

tion - and for this one can turn to the casuistry again.

The metaphor of ‘seeing’ is strengthened by those of light and darkness.

However, the ordering principle that helps one to clearly see is to be found in real-

ity rather than just in the economist’s mind, the appeal to introspective efforts

notwithstanding. Take the following lengthy quote. …die Thatsachen sind nicht so gefällig, sich von selbst Glied für Glied in

einem übersichtlich geordneten Stufenbau, der von den speziellsten bis zu

den letzten allgemeinsten Thatsachen hinaufleiten würde, dem Blicke des

Forschers zu präsentieren. Sondern wenn auch freilich jener geordnete Aufbau in der Wirklichkeit stets vorhanden ist, so ist er selten oder nie in seiner

Vollständigkeit unmittelbar sichtbar. Einzelne Zwischenglieder, einzelne

Stufen sind uns fast immer vorerst verborgen, und ihre Existenz muß erst

durch deduktive Operationen erschlossen werden, deren Ergebnis dann

durch Verifikation an der Erfahrung sichergestellt werden mag. […] Wenn

man nur die Thatsachen, die äußerlich wahrnehmbar sind, für sich sprechen

lässt, so wird man niemals erfahren, ob der Wert der Kostengüter die Ursache

oder die Wirkung des Wertes ihrer Produkte ist. Man sieht leicht, daß ein

Zusammenhang zwischen ihnen besteht; aber durch die Komplikation der

Verhältnisse ist der ganze Aufbau für unsere Auge so verdrückt und

verschoben, daß man nicht ohne weiteres erkennen kann, welche der beide

Thatsachen die niedrigere und welche die höhere Stufe des kausalen Baues

einnimmt. Was hier die körperliche Auge nicht sieht, muß erst das geistige Auge durch eine Reihe verwickelter abstrakter Schlüsse rekonstruieren. Dies hat

bekanntlich die Theorie des „Grenznutzens“ und erst sie in einer Weise

gethan, daß die innere Ordnung der Gedanken hier wirklich vollständig

hergestellt, und zugleich die Probe der Erfahrung vollständig bestanden ist.19

The emphasised sentences teach us two things. First, the ordering principle

that enables us to see is ‘out there’ in the actual world (first emphasis), so it is not a

Kantian principle. Second, we can get there by abstraction. In this quote, abstraction

is possible when we set our intellectual capacities in motion. It is an epistemic

mode which subtracts from the phenomena a multiplicity of superficial properties

18 PTK, p.425. 19 Böhm-Bawerk (1890b), pp.87-88. My italics.

Chapter II

as these blur the picture. After subtraction, the explanatory clues remain. Natu-

rally, our fallibility often induces us to erroneously abstract the non-instructive

aspects from observed cases; but the scientific target is the ‘deeply hidden’ keys

that can be reconstructed. The causal relations that matter can be known by the

discovery of the right objective ordering principles underlying the socio-economic

world.

Böhm-Bawerk often repeats the view that these principles lie covered, wait-

ing to be dug up: [I]ch glaube in der Tat, daß es weder eine einfachere, noch eine natürlichere,

noch endlich eine fruchtbarere Vorstellungsweise von Tausch und Preis gibt,

als wenn man die Preisbildung im Lichte einer Resultantenbildung aus den in

der Gesellschaft vorhandenen subjektiven Wertschätzungen betrachtet. Es ist

dies kein Gleichnis, es ist lebendige Wirklichkeit.20

Again, his metaphysics seems to assume that causal relations exist in the economic

reality economists explain. Concerning the scientific need to penetrate onto the key

underlying factors he said: Und wieder nicht anders steht es bei einer ganzen reihe theoretischer

Probleme, z.B. bei der Frage nach dem wahren Wesen des Einflusses von

Angebot und Nachfrage auf den Preis; nach dem wahren Wesen der Funktion

des Kapitales in der Produktion; nach dem Ursprung des Kapitalzinses, nach

dem Zusammenhang zwischen Ersparung und Kapitalbildung u.s.f.21

Causal relations, I conclude, exist independently of scientific perspectives or

of scientific success or error. The term ‘wahren Wesen’ in this quote must be under-

stood literally, in a strong essentialist sense. It is at the level of causal relations that

we find metaphysical kinds.

3.2 A unified order

The belief that some phenomena are accidental (secondary) and other parts of real-

ity express leading (primary) principles applies not only to the issue of the Law of

Costs that has been discussed in section 2 above. Böhm-Bawerk believed this with

regard to the domain of the economic science as a whole. He assumed the eco-

nomic world to be essentially unified. A nice example is how he referred to the

unifying capabilities provided by his own interest theory.

As we have seen, he noted that the economists of his day had proposed the

Use Theory of loans and interest. According to this theory, money is handed over

to the one who borrows it, in order to temporally entitle the debtor the rights to

make whatever use of it, in exchange for a reward, this reward being the interest of

the loan. But if this were the correct description of the phenomenon of borrowing,

Böhm-Bawerk asked, how could anyone borrow firewood and use it? His alterna-

tive interpretation has substantial consequences. The importance of the exchange

20 PTK, p.285. My italics. 21 Böhm-Bawerk (1890b), p.85. My italics.

theory (Tauschtheorie), as opposed to the use theory (Gebrauchstheorie), is that it also

helped to unify the scientific descriptions of similar phenomena: for the same was

true in case of the market for credit and for the market for finished goods. The exis-

tence of interest is rooted in the difference in value between present and future

goods. This difference, as settled on the market for credit, is the interest. The mar-

ket of subsistence goods had to be understood as a market for exchange of present

(finished) goods against future goods (fruits of labour). The Tauschtheorie provided

for explanatory unification, because it showed how interest was a price concept

resulting from subjective welfare accounting.22 The Tauschtheorie subsumed the

horse market and the credit market under one and the same concept, the

Gebrauchstheorie presented the credit market as yet another institution.

Apparent deviations from what we may expect to be the case according to

the theory defended are explained by subsumption under the leading principle. A

negative (real) interest, for instance, can only occur when the proportion of needs

to satisfaction in the future is somehow perverse. Future goods will then be worth

more than present goods. In other words, interest will be negative whenever there

is reason to expect future levels of satisfaction to be grim. But negative interest is

rare. A zero interest rate can be explained along similar lines, consistent with the

theory defended. As a rule of experience, trade takes place under conditions of

positive interest. But Böhm-Bawerk again takes prima facie deviations from the

comprehensive rule, rare as they are, as confirming and not as falsifying instances.

Again, his theory is able to cope with what for the Use Theory are deviations. To

phrase it more evocatively, such facts of life, precisely due to their being eccentric,

confirm the Exchange Theory rather than falsify it. One notes the analogy with the

defence of the Law of the Marginal Agent. To Böhm-Bawerk, it is clear that the Use

Theory will never be able to explain deviant, special cases: Gerade das Vorkommen solcher Fälle scheint mir übrigens eine nicht zu

unterschätzende Probe mehr zu Gunsten meiner Darlehenstheorie zu bieten.

Denn wie wollen die Nutzungstheoretiker dieselben erklären?23

We can see that the theoretical unification brought about by the Exchange

Theory helps finding out what aspects of actuality must be understood as princi-

pal, displaying an essential causal structure with an abstract kind status.24

22 His ideal of explanatory unification can be interpreted in an analogy with Michael Friedman’s explica-

tion, who proposed to qualify an argumentation as more explanatory and unified if the number of

independent phenomena that we accept as ultimate in this argumentation is minimised (and, hence,

the number of conclusions maximised). See Michael Friedman (1974), p.15. He notes that the idea

that this is essential for scientific explanation has been observed by William Kneale already in 1949.

Philip Kitcher explicated explanation as unification by reference to an explanatory store, which forms

‘part of a systematic picture of the order of nature’ (Kitcher 1989). It seems to be appropriate to as-

cribe to Böhm-Bawerk the wish to give a systematic picture of the socio-economic realm. 23 PTK, p.373. 24 Chapter V discusses the status and meaning of abstract kinds.

Chapter II

3.3 Essentialism, reduction, abstraction

Böhm-Bawerk’s realism embraces a criterion of appraisal that compares theories on

the basis of unifying power and, by implication, explanatory success. His apparent

confidence that such success suffices to believe in the truth of one theory presumes

that the world itself is unified in the very same way as the conceptual apparatus

tells us it is. To underpin that this view on theorising is to be characterised as es-

sentialist, I now make more precise what essentialism means. Essentialism is the belief that there is one best way to conceptualize our theories about the world and that this ultimate conceptualization captures the joints of the world that awaits discovery.25 (Note

that this does not exclude fallibilism: one may be essentialist and prudent at the

same time.)

I hope to have shown that the epistemological and ontological underpin-

nings of the economics of Böhm-Bawerk are essentialist. Essentialism is a very

strong reading of the sort of realism held by him (and by Austrian economics more

generally). But, on the other hand, such an interpretation of Böhm-Bawerk’s meth-

odology is not out of the ordinary. Caldwell (2004) mentions Barry Smith26 as one

who favours to interpret Mengerian economics in an essentialist conceptual

framework of Aristotelian lining and coins it ‘the dominant reading’.27 He also

mentions Emil Kauder as an early interpreter who understood Menger’s ‘exact

types’ as Aristotelian essences.28 Uskali Mäki has written extensively about the

essentialist epistemologies of several Austrians – and of some other economists. I

treat his view in detail in appendix 3.

Nevertheless, in the same appendix, I also consider the contribution by Karl

Milford, against the idea that essentialist views underpin the Austrian method of

research: he instead says that there are better reasons to take the Austrians as em-

ploying a Kantian and the German historicists as taking an Aristotelian view. Had

the historicists indeed believed that there were essences (or kinds) to be found at

the social rather than at the individual level then it would have been rational for

them to exploit means of research that aim to find these kinds. Milford reasons that

the Historical School sociologists are holists and that, by implication, they must be

essentialists too. The Austrians, in turn, are said to represent a more Kantian view

of the preferred practice of social science. Thus, the Austrians and the Germans

allegedly represent the two broad positions of individualism and collectivism re-

25 This definition comprises essentialist epistemology to the extent that theories allegedly are meant to

dig up essences, or kinds. It comprises essentialist ontology to the extent that this world is indeed

supposed to be as it is, with all its joints in place. To conclude, it comprises a semantics that fits es-

sentialism too, as the key concepts must capture something from the world. I shall focus on

epistemological and the semantic aspects of it in chapter V. The epistemological aspects of essential-

ism are considered above all when the question of tenability is at stake (section V.2). The semantic

aspects of essentialism are given attention when discussing the Causal theory of Meaning. 26 Specifically Smith (1986) and Smith (1990). 27 Caldwell (2004), p.67. 28 Kauder 1957). Aristotle has also been ‘used’ to defend the opposite stance.

spectively. The former point to the atomistic constituents of social reality as the

keys to satisfactory social explanations, the latter point to social wholes as the start-

ing point and the end of social explanantia. Milford alternatively calls the latter

position Methodological Essentialism. This categorisation induces him to identify the

search for holistic explanations as essentialist (and microreductions as non-

essentialist, specifically Kantian). Milford’s conclusion is too hasty. To speak of

historicists as methodological essentialist theorists is to claim that they looked for

kinds at the social level. But, as I argue in the appendix, their methodological stance

was above all a result of their scepticism, not of any strong realist ontologies, and a

quest for kinds is hardly sceptical in orientation. Milford pretends to identify ho-

lism with essentialism, but he really confuses the two.

Much of the confusion over reductionism and holism lies, I believe, in a fail-

ure to specify one’s position in terms of strength. Is one defending a weak or a

strong form? Weak forms of reductionism, or Methodological Individualism (MI),

may well be compatible with weak forms of holism, or the belief that social wholes

can be focal points in explanantia too and not only in explananda. Another source

of confusion is over epistemological and ontological claims. Often, assumptions

about the structure of the world are thought to directly imply epistemologies.29

Consider the list of eight possible positions in the holism (H) and reduction-

ism (R) debate on the following page. Ontological Holism is referred to as HO,

Methodological Holism as HM. There are strong (s) and weak (w) versions. The list

enables me to make clear how best to characterise Böhm-Bawerk’s economics: I

believe he adheres to the strong positions ROs and RMs. An example of a kind at

the aggregation level of the individual is an intentional agent. I call it ‘social’ none-

theless because it figures in socio-economic explanations. A unified explanation of

socio-economic phenomena by reference to individual choice making need not refer

to essences or kinds if it is used as a tool in theory comparison. Even so, with his

strong allusions to truth, Böhm-Bawerk’s quest for theoretical unification presup-

poses that the world is essentially unified. This is an ontological commitment.

It seems to me that the broader Austrian praxeological view, on the nature of

the social and on social theory as it was put forward systematically by Menger for

the first time, does not differ from Böhm-Bawerk’s view. The capital theorist firmly

stood in his tradition.

There is, however, an indirect relation between ontology and methodology.

A prior essentialist ontology – the belief that the world has been made up of kinds

independently of efforts to conceptualize it – will be connected with the belief that

research must find these kinds and inspect their necessity. Essentialism, weak or

strong, commits the rational scientist to look for the clues where he guesses these

can be found. In other words, prior metaphysics delivers a perspective and, hence,

a heuristic. This perspective ‘indexes’ 30 any view of the world.

29 Appendix 3 discusses also this point and an interesting clarification by Harold Kincaid. 30 Chapter V treats the role of perspectives in essentialism, together with the term ‘index’.

Chapter II

So the essentialist economist is to identify the joints of the socio-economic

world in order to explain its workings. This process of identification is abstractive;

many of the phenomenal aspects are ignored or rated secondary so as to pick out

the characteristics that must not be abstracted from.

I wrap up. Böhm-Bawerk’s essentialist view on socio-economic science was

that there is an essential structure in economic reality, lying hidden beneath the

appearances. Abstraction is needed to produce a description that mimics this hid-

den structure. This is in the tradition of Menger. As to market prices, the latter

already had made a methodologically important distinction between real and eco-nomic prices. Real prices are those that actually come into being as the result of

concrete market forces, with their multiplicity of interactions and feedback loops.

They are an example of real types. Economic prices were those that come into be-

ing in the abstract. These are an example of exact types. The exact method is the

one that tries to find ‘exact types’, not unlike those commonly known as Max We-

ber’s ‘ideal types’. Exact types are abstractions of phenomena, stripped of most of

their rich empirical attributes.31 To make these hidden aspects of reality visible,

31 According to Bruce Caldwell, who follows Israel Kirzner in this view, the distinction explains why

‘Menger disregarded the problems of human error and ignorance that were so ubiquitous in the rest

of [the Grundsätze der Volkswirthschaftslehre]’.

HOs Some irreducible social entities are social kinds

HOw There are some irreducible social entities at the level of institutions

(without necessarily being kinds)

HMs Only social and no other entities should occur in the explanans of social

phenomena

HMw At least some social entities should occur in the explanans of social

phenomena

ROs Some individual entities are social kinds

ROw Some social entities are reducible to entities at the level of the

individual

RMs Only individual and no other entities should occur in the explanans of

social phenomena

RMw At least some individual entities should occur in the explanans of social

phenomena

Böhm-Bawerk needs descriptions of a modelled reality, viz. of aspects of economic

reality as they are expected to be under the force of isolating assumptions. As we

have seen above, the OVT theorists also described a world that is under the protec-

tion of isolating assumptions. But in Austrian eyes, they did so mistakenly. An

interesting example of an isolating assumption is that equilibrium in markets

comes into being instantaneously. It obviously does not, but under this isolating

condition, we can see interesting patterns and mechanisms. So contrary to the view

often coined ‘instrumentalist’ – that isolating assumptions take away the realistic-

ness of a model – the view expressed here is of a role of assumptions as high-

lighting aspects of (economic) reality that otherwise remain unseen. It is a

perspective, which helps interpreting Böhm-Bawerk’s particular heuristic.

4 Discrete narratives versus continuous mathematics

The whole of PTK, but also Böhm-Bawerk’s (1886) Grundzüge, made use of numeri-

cal examples in which much was exogenously given and in which figures were

altered just to inspect the consequences. It is easy to detect clumsy dynamics in it.

But whatever one thinks of the inefficiency of this method, logically a numerical

approach is in itself not invalid. At most, the use of these examples is risky in that

readers may take them for proofs instead of illustrations.32 They are of course not;

Böhm-Bawerk constructed the numbers in such a way as to fit his explanatory

needs.

In economics generally, the mathematical treatment of theoretical links be-

tween the elements of an economic system under research is supplemented by a

more concrete story about the referents of all the variables and their connections;

and so was Böhm-Bawerk’s description of interest in the market system. In the

combination of a unifying conceptualisation, theoretical links between the vari-

ables, and numerical examples, the latter played the role of what in Wicksell is pre-

empted by the mathematical equations. Discrete cooked-up examples did the work

that continuous mathematics did in Wicksell. These examples were adorned by lots

of story-telling.

In this section, I first present Wicksell’s mathematical re-interpretation of

Böhm-Bawerk’s system. In de second subsection, I shall show that the numerical

approach is necessarily associated with a partial analysis33. The last subsection is

devoted to a discussion of an advantage of giving a partial analysis. I shall draw

the conclusion one is led to once it is accepted that Böhm-Bawerk’s analysis is par-

tial. I shall claim that, implicitly, Böhm-Bawerk has been inventing a demand curve

for capital and that Wicksell failed to see this.

32 For instance, Bleicher (1890) accused Böhm-Bawerk of this fallacy. 33 In economics, ‘partial analysis’ is the term used for any model that disregards the feedback effects of

changes in the market under investigation as discharged by other markets.

Chapter II

4.1 Böhm-Bawerk’s interest theory in Wicksell’s Wert, Kapital und Rente

For the ease of presentation I redisplay table I.11 from the previous chapter. It

shows the case of the optimum roundaboutness of production with a market wage

level of 500 Austrian guilders.

Stock of capital K: 15 billion Share of capital needed: ½

Wage ℓ: 500 Stock of labour A: 10 million

period

pecuniary

marginal

interest

payable

surplus value

per worker

labour

demand

surplus

interest

‘z’

t p ∆p ∆p ⁄ ½ℓ p-ℓ K ⁄½ℓt (p-ℓ)K⁄½ℓt 2(p-ℓ)⁄ℓt

0 150 ** ** -350 - - -

1 350 200 0.800 -150 60.0 m. -9.0 b. loss

2 450 100 0.400 -50 30.0 m. -1.5 b. loss

3 530 80 0.320 30 20.0 m. 0.6 b. 0,0400

4 580 50 0.200 80 15.0 m. 1.2 b. 0,0800

5 620 40 0.160 120 12.0 m. 1.44 b. 0,0960

6 650 30 0.120 150 10.0 m. 1.5. b. 0,1000

7 670 20 0.080 170 8.6 m. 1.46 b. 0,0971

8 685 15 0.060 185 7.5 m. 1.39 b. 0,0925

9 695 10 0.040 195 6.7 m. 1.3 b. 0,0867

10 700 5 0.020 200 6.0 m. 1.2 b. 0.0800

table II.2 Market clearance at roundaboutness of t=6

Entrepreneurs demand subsistence means, or social capital, for the employ-

ment of workers. We assume now no other factors of production. Rising levels of

roundaboutness of production raises returns, but it also requires more labour-years

to be employed. Labour must be paid out of a wages fund, the social capital from

which entrepreneurs can take at market interest rates. This wages fund is exoge-

nous in the partial analysis and given by a perfectly inelastic capital supply

function as ‘the stock of capital’. The same is true for ‘the stock of labour’. As re-

turns are assumed to grow out of production and sales gradually, without any

cyclical investments in private stock, and as inputs of labour are also assumed to

be distributed evenly over time, the per new worker investment in increasing

roundaboutness is half the price of labour, or ½ℓ. If the individual entrepreneur

chooses t=0, labour intensity of his production process (labour relative to total in-

put) is 100%.34 Roundaboutness (in years, t) is the equivalent of what in modern

terms is called capital intensity. At t=1 and t=2, and given the production function,

he uses capital but the yield of this longer production route does not match the cost

of employing more workers. We can see this in the column (2), which gives pecu-

niary outputs, p, per worker. The surplus value per worker is given in column (5).

34 So capital for this entrepreneur is nil. Social capital, K, is still 15 billion.

In the third column marginal returns (or yield) is discretely given, whereby

each next row is supposed to represent the difference in values between two suc-

cessive rows in column (2). The same is true for column (4). Hence, the ‘interest

payable’ (to be referred to as zpay from now on) is what the entrepreneur can afford

to pay as interest if he increases capital intensity. If for instance he increased

roundaboutness from t=5 to t=6, zpay would be 12%. This is equal to the extra in-

come made possible by the raised – and, according to the third cause, more

productive – roundaboutness.

Given vertical factor supply functions, equilibrium exists in the labour and

capital markets at t=6. Columns (6) and (7) speak for themselves. Note that column

(6) shows that one effect of the numerical treatment by Böhm-Bawerk is that the

capital constraint is exercised over the labour market. Social stock of capital – or

the wages fund – determines the wage if labour stock is given. Column (8), finally,

represents surplus value in relation to the total capital market costs of the respec-

tive investments in roundaboutness. In what follows, I shall focus on columns (4)

and (8).

To illustrate how the economic process leads to individual optima in this

discrete production function, let us take a starting point in a situation when inter-

est is 10% and t=6. In the chapter of PTK about the level of interest in the isolated

case of a market situation, Böhm-Bawerk reasons:

Diejenigen Produktionslustigen […] welche Subsistenzmittel aus dem Markte

zu nehmen suchen, um die Produktionsperiode noch auf ein siebentes Jahr zu

verlängern, könnten aus dieser Verlängerung nur ein Mehrerträgnis von 20 fl.

(670-650) gewinnen.35

Apparently, there are those who might want to extend the production pe-

riod to t=7, for which they can afford to pay an interest of no more than 8%. That is,

an increase of t by one year would render an extra yield of 20 (indicated in column

4), earned over the extra investment of ½ℓ or 250. This outcome turns out to be

impossible in equilibrium, as interest to be paid in the market will be 10%. The

capital constraint comes to be realized by labour market forces. The wage level is

constant even if entrepreneurs change the roundaboutness of their production

process.

Wicksell’s Wert, Kapital und Rente (WKR) gave the mathematical interpreta-

tion of this theory of interest. The first possibility is that the worker is also

entrepreneur. So he has to borrow capital for paying his own wage ℓ, and the inter-

est z on the loan during the average production period.

The following legend of variables will be used for a representation of Wick-

sell’s reformulation of the Austrian interest theory and distribution theory. Note

that the choice of symbols is partly in accordance with the German of WKR.36

35 PTK, p.455. 36 Thus, interest is Zins, wage is Lohn, and land is Boden, and so on.

Chapter II

Interest: z Relative price of one commodity: π

Production period: t Real price of the product (numerary): n

Optimal production period: t* Individual yearly income: e

Increase of period t by one year: ∆t Countries: subscripts i and j

(Yearly) wage and land rent: ℓ and r Total amount of capital: K

Total revenue per worker: s Total amount of labour: A

Yearly revenue per worker: p Total amount of land: B

Optimal revenue: p* Units of land per worker, B ⁄A: g 37

Commodities: x and y Marginal utilities x and y: U’x, U’y

Briefly, Wicksell’s reformulation looks like this.

s = (1+z½t)ℓt (1)

Production per year – per unit of time t – is:

p = s ⁄ t = ℓ+½ℓzt (2)

which gives the production function. With t as instrument variable, the task is to

maximise ℓ by choosing the period of production well:

dp ⁄ dt = ½ℓz (3)

Wicksell turned Böhm-Bawerk’s discrete production function into a continuous

function. The figure Graph 1 below displays yearly production per worker, de-

creasingly rising as marginal returns diminish. This is the production function ƒ:

p=ℓ(1+½zt) in the way Wicksell depicted it.38

Note that the tangent line dp/dt (tangent ½ℓz) intersects with the horizontal

axis at −2 ⁄ z. This can be seen if it is noticed that ℓ is the vertical part of the tangent

running from this intersection to the vertical axis. Given ƒ, the relation between ℓ and 2/z+t is given. In the partial analysis, the interest z is given. Both a shorter and

a longer roundabout period than t* causes ℓ to fall. Equilibrium is settled on the credit market.

The other possible case is the entrepreneur-capitalist, the one Böhm-Bawerk

had described. In this case, wage ℓ is given (from the point of view of the individ-

37 Wicksell uses the letter ‘h’: hectares. I want to avoid the suggestion that per worker, the number of

hectares exceeds 1, and use ‘g’ instead: Größe . 38 See WKR, p.97. My graph has been based upon his.

ual entrepreneur – as the partial analysis suggests) and z is the variable to maxi-

mise, while t is still the instrument variable. (The tangent line dp ⁄dt is then drawn

from ℓ on the vertical axis, instead of -2/z.) Now equilibrium is accomplished in the labour market.

4.2 The limits of a discrete analysis

Böhm-Bawerk, working with a discrete production function, did not calculate

marginal production in optimum t*, like Wicksell, as ƒ’(t)dt but as ƒ(t)–ƒ(t–∆t). This

is much higher, especially with discrete steps as large as one year. Below, figure 1

shows the difference between the continuous and the discrete cases.

ƒl. ƒ(t)=p

p–ℓ ƒ(t)–ƒ(t–∆t) ƒ‘(t)∆t

t*–∆t t*

figure II.1 Total and marginal production per worker per year

In a continuous analysis, the increase in pecuniary output is (dp ⁄dt) ∆t,

while a discrete analysis leaves open two options: either ƒ(t)–ƒ(t–∆t) (as Böhm-

Bawerk did) or ƒ(t+∆t)–ƒ(t).

The first option gives a higher output differential than the continuous ap-

proach, the second option gives a lower output differential:

ƒ(t)–ƒ(t–∆t)>ƒ ’(t)∆t>ƒ(t+∆t)–ƒ(t)

where ƒ(t) is the production function denoted as p above, and ƒ‘(t) is dp ⁄dt. Filling

in ½ℓz for ƒ’(t), Wicksell shows that wages cannot be assumed constant, because

they will rise as a consequence of rising capital intensity.

Chapter II

Man könnte aber nach dem Wortlaute des Satzes verleitet werden zu glauben,

daß, wenn eine Vermehrung des Volkskapitals bei gleichbleibender

Arbeiterzahl zu einer Verlängerung der Produktionsperiode führt, das

Mehrerträgnis dieser Verlängerung, durch die bezügliche Kapitalvermehrung

dividiert, etwa die Zinshöhe liefere. Das wäre entschieden unrichtig. Letzteres

Verhältnis ist, wie wir sehen werden, immer kleiner als der Zins, und zwar

um eine endliche Größe kleiner, auch wenn von einer minimalen

Veränderung die Rede ist, was damit zusammen hängt, daß jene Vermehrung

des Volkskapitales von einer Lohnerhöhung begleitet wird, die sie teilweise

verschlingt, so daß die wirklich erreichte Produktionsverlängerung immer

hinter der bei unverändertem Lohnsatze möglichen zurückbleibt.39

His mathematical analysis allowed Wicksell to deal with more variables si-

multaneously and account for the causal feedback from the distribution theory (to

be dealt with below). Specifically, he could put into one system of equations both

the production period of the firm’s level of investments and the macro equilibrium

in income distribution. Any increase of the production period would thus raise

demand on the labour market, and increase the wage level. Consequently, due to a

higher wage to be paid by the entrepreneurs, any lengthening of the production

period would be inhibited. Indeed, under a regime of a flexible labour market the

production period will not rise as much as under the condition of fixed wages.

The tangent of the production function is lower at t* than at t*–∆t. But the

fact that Böhm-Bawerk’s ƒ(t)–ƒ(t–∆t) renders higher interest, as noted above, than

Wicksell’s ƒ’(t)∆t is a consequence first of all of the fact that the former used dis-

crete numerical examples, not continuous mathematics. In itself it is not in consequence of Böhm-Bawerk’s ignoring the causal feedback from adjustments in the labour market.

However, interest as he calculates it is overvalued. But it looks as if Wicksell

conflated the issues of the Austrian discrete analysis, on the one hand, and the

Austrian partiality of the approach on the other, as causes of the overvaluation.

Considering that Böhm-Bawerk gives a partial analysis, there is nothing flawed

about his method in principle, even though the continuous method renders more

precise outcomes than the discrete method. Böhm-Bawerk had hoped to reach such

precision by dropping the assumptions associated with the partiality of the ap-

proach in the closing part of PTK (viz. section III of chapter III, book IV: der Kapitalsmarkt in Entfaltung). Wicksell’s conflation of these two issues is remarkable

because he praised section II of this ‘Abschnitt über „die Zinshöhe im Marktverkehr“, an dessen genialer Gestaltung und überzeugender Kraft nur wenige Kritiker etwas auszusetzen gehabt haben’40, while section III contains the key to understanding what

Böhm-Bawerk had been out to do.

Why did Böhm-Bawerk, in PTK, not stress the consistency of his numerical

approach and try to justify it? He never responded to the criticism expressed by

Wicksell already in 1893, of the alleged imprecision that comes with the discrete

39 WKR, p.111-112, emphasis by Wicksell. 40 Wicksell (1928), p.199.

ƒ(t)–ƒ(t–∆t). But he could have easily done so. He assumed investment periods to

increase discretely in the partial analysis – the analysis of decisions taken by an

individual entrepreneur. So he could have easily defended his method. The inter-

vals between two successive roundaboutness periods – intervals of entire years –

were to narrow down in the macroanalysis he had proposed (though not devel-

oped formally) in the last chapter. Furthermore, Böhm-Bawerk could have made

remarks about Wicksell’s misunderstandings.

As he did respond to many other criticisms by Wicksell in his third edition

of PTK (1914), I think that Böhm-Bawerk simply never understood the point of this

one. Hennings says that ‘Böhm certainly learnt from him, but whether he under-

stood the full import of Wicksell’s reformulation of his theory may well be

questioned’.41 In a letter to the Swede, Böhm-Bawerk wrote: You are such a keen expert of the works of Pareto, which are less accessible to

me because of their mathematical form, that I would like to ask you to inform

me occasionally in a few lines to which group of theories in your opinion

Pareto’s doctrine regarding the cause of capital interest belongs.42

In recent months the problem was in particular a criticism that was put for-

ward simultaneously by Bortkiewicz and Irving Fisher [about the third cause

of the relative higher value of present goods as being dependent on the other

causes]. Both also use mathematics against me, and I have the suspicion that

the latter is somewhat abused by them.43

Gehrke also quotes a rather painful letter to Wicksell by Alfred Marshall: 44 A boy in a village school who made such a blunder in his arithmetic would be

punished: and [Böhm-Bawerk] knows I am a trained mathematician. If he

were really earnest in his desire to know what I mean, he would turn to my

mathematical notes.

The absence of mathematical treatment in Böhm-Bawerk’s models – due to his lack

of active mathematical skills – had caused his putting a large share of his intellec-

tual resources into explaining the mechanism behind economic processes. (See

section 5 below.) Meanwhile, his lack of mathematical understanding was not lim-

ited to active skills, but stretched to passive insight too.

4.3 The demand curve for capital

The strategy of Böhm-Bawerk to take the viewpoint of an individual entrepreneur

– taking wages exogenous and constant even if production processes with varying

roundaboutness were considered – was not just a second best option in the absence

of mathematical skills to deal with the problem of interest in a general distribution

theory. It did have an advantage. By his rhetorical strategy, Böhm-Bawerk had

41 Hennings, (1997), p.177. 42 Ibidem, appendix. Letter to Wicksell sent at 18 August 1899. 43 Ibidem. Letter to Wicksell sent at 5 July 1907. 44 Ibidem, Marshall about Böhm-Bawerk’s attack on the former’s abstinence theory of interest, p.231.

Chapter II

highlighted the underlying mechanism, which could explain how economic proc-

esses turned positions of disequilibrium into positions of equilibrium. Böhm-

Bawerk brought his own version of short run dynamics into the picture by telling a

story about maximizing entrepreneurs. And it was not just some story.

If we assume again that the data from table 2 above are given; what if an in-

dividual entrepreneur were in the position to marginally lengthen the period of

production beyond t=6? Then the maximum interest he could pay for a share of the

wages fund is 8%. Likewise, the entrepreneur who goes from a period of t=5 to a

production period of t=6 will even be able to pay a maximum interest for borrow-

ing capital of 12% (30 ⁄ 250) to his creditors, all under the assumption that wages

are constant and exogenous. The marginal demander, investing in a roundabout

production of t=6, is positioned exactly in between the former two and can afford

to pay a maximum of 10% (25 over 250), which, given the variables as Böhm-

Bawerk had drafted them, happens to be equal to equilibrium. (This may not be

obvious from table 2, as columns (3) and (4) refer to differentials. The interest level

of 10% lies in between 0.08 and 0.12 at t=7 and t=6 respectively.)

It seems that column (4) represents a demand curve for capital. Then again, it is a

demand curve established not by entrepreneurs’ marginal utilities, but by deter-

minants that lie in mere technical data.45 It is a function of roundaboutness of

production. Thus, it can be seen as a list of demanders’ (future) preferences, but

determined solely by the objective ‘third cause’. In the terminology of Marshall,

column (4) represents the ‘demand price of capital’. Below, figure 2 presents a

curve, progressively decreasing with rising capital intensity t: ‘interest payable’.

This curve is a reproduction in terms of a continuous function of what were dis-

crete figures in column (4) of Böhm-Bawerk’s table: zpay=(pt–pt–1)⁄ (½ℓ).

The ‘interest payable’ is the amount of output an entrepreneur can maxi-

mally afford to pay for borrowing subsistence means to maintain extra workers

needed to lengthen roundaboutness of production. It is a function that Wicksell –

and, as far as I can see, every other commentator – ignored. The parabolic line, in

its turn, represents column (8) of the table: the interest line z=(p–ℓ) ⁄ ½ℓt. It reflects

the market price of capital at a given wage. ‘Interest payable’ intersects at the top

of the interest curve z. This must be the case by implication, because every time an

optimum production period t* is achieved, the marginal demanders for capital are

able and willing to precisely pay the resulting interest.46 After all, this is what

makes them marginal demanders. Left of t* extra investments turn out profitable,

beyond t* production is too capital intensive. Roundaboutness t* is the entrepre-

neurial equilibrium.

45 This interpretation could have served as part of the defence, in PTK Vol. 2, Exkurs XII, of Böhm-

Bawerk’s much disputed claim that the third cause is independent from the first and the second. See

Bortkiewicz, L. (1906), Fisher, I. (1907), and Wicksell, K. (1928). 46 The intersection is z*=(pt*–pt*–1) ⁄½ℓ=pt*–ℓ ⁄½ℓt*. The optimum production period t*=pt*–ℓ ⁄pt*–pt*–1 .

column 4

column 8

figure II.2 Intersecting demand for capital and market rates of interest

There is an important similarity between the capital market and the horse

market, the latter of which had made its appearance in the chapters of PTK on the

value theory. In discussing the mechanism that translates individual decision mak-

ing into goods market outcomes Böhm-Bawerk had already given a step-by-step

alternative to the infinitesimal approach. We have seen that equilibrium pricing

takes place within upper and lower limits. The price (of horses) remains indeter-

minate if few market parties meet, but the boundaries converge as the number of

suppliers and demanders grows. In the limit, so to say, the degrees of freedom for

the price to come into being is zero. In similar fashion, as concerns the capital mar-

ket, Böhm-Bawerk proposed the reader to imagine ever more demanders-

entrepreneurs in the aggregated model. These demanders for capital are willing to

engage into ever more production processes, some with a roundabout period of

t=5, some of t=6, and some of t=7. We may assume the price of capital to grow to-

ward a determinate equilibrium. Below I propose to represent Böhm-Bawerk’s

method as revealing that a narrative approach – his use of narratio’s – about market

forces adds intelligibility to the process that leads to market outcomes.

Chapter II

5 The principle of sufficient complexity

Wicksell complained about the irrelevance of column (4): Merkwürdigerweise glaubt aber Böhm-Bawerk in den aufgestellten Zahl-

reihen „noch andere Beziehungen“ gefunden zu haben, die in positiver (?)

Weise auf den resultierenden Zinsfuß von 10 Proz. hindeuten. […]47

Wie könnte […] von der Verfügung über 250 fl. ein Mehrerträgnis von 30 fl.,

d. h. ein Reingewinn von 12 Proz. abhängen, da ja […] sich das Kapital

höchstens mit 10 Proz. verzinsen kann? […] Es ist somit allerdings wahr, daß

der Zins, auf die halbe Lohnhöhe berechnet, „zwischen dem Mehrerträgnis

der letzten gestatteten und dem der nicht mehr gestatteten Produktions-

verlängerung“ zu liegen kommt; zwischen diesen Grenzen aber wird seine

definitive Höhe nicht etwa durch Angebot und Nachfrage, sondern einfach

durch die Ergiebigkeit der lohnendsten Produktionsperiode festgestellt.48

Indeed, the demand curve for capital – as I now coin the fourth column – is not

needed to determine the interest rate. It can only intersect with the top of the hy-

perbolic curve that represents column 8; that is, logically so. What is the point of a

story on the decision process that is supposed to be going on in the entrepreneur’s

mind if the optimum production period can be calculated with a given wage and

production function?

I am not aware of any literature that explores this specific question, only of

writings that discuss the Austrian project of highlighting processes rather than end

states of these processes.49 Therefore I now discuss how economic narratio’s con-

vey intelligibility to process theory. An Austrian claim has since long been that a

study of processes by which equilibria come to be, rather than of equilibria per se,

is legitimate and even a pressing matter for an understanding of social reality. My

claim is that Böhm-Bawerk stood out in this tradition, because he so intensely ex-

ploited his narrative talents to this end.

The first subsection introduces As-If narrative reasoning (a notion that will

get extra momentum with a discussion of counterfactuals in chapter III). The sec-

ond subsection links this notion to an epistemological view on the constructs

historians make in their tales about the past. I shall pay special attention to the dif-

ference between the narratives of the German historicist economists and those of

Böhm-Bawerk. The last subsection draws the conclusion that Böhm-Bawerk ex-

ploited his narrative approach so as to suit research interests that required a certain

minimum of complexity, that is, an upper limit to the amount of disregard for the

very mechanics that shapes the market process.

47 WKR, p.108 Abbreviation, italics, and added question mark in the original. 48 WKR, p.110-111. Italics and abbreviation in the original. 49 See for instance Ullman-Margalit (1978), who discusses Hayek’s stance on unintentional consequences

of intentional action, Mäki (1990, 1991, 1992a), and Lachmann (1976). The latter has consistently

stressed the time dimension of subjectivism, viz. that individual knowledge (and, hence, individual

decision making) is subject to change. Although Lachmann critically stresses Böhm-Bawerk’s ‘Victo-rian’ view of a world with no error, ‘a world of steady progress through capital accumulation’

(p.61), he does admit that its Walrasian expression was not to Böhm-Bawerk’s taste at all.

5.1 Hypothetical worlds in the theory of the market

A story about what an entrepreneur would do if he were in the position to lengthen

roundaboutness of production fits into the subjectivist treatment of political econ-

omy. It pictures the choices individuals make in a changing environment. This

environment later turned much more dynamic and uncertain in von Mises’

praxeology, but already Menger and Böhm-Bawerk gave a subjectivist picture of

the economic process, i.e. a process as generated by individual choice making.50

The idea of choice splits up reality in actual and hypothetical situations, because

rational decisions are taken on the basis of hypotheses about what will (would) be

the case if a particular decision is (were) taken. Though not (yet) actual, these sub-

junctive or hypothetical worlds are real enough, at least in the sense that the

choices the entrepreneurs are facing are real.51 The story telling approach makes for

our imagination of a protagonist who really chooses. The narrative method helps

shape the subjectivist approach.

There is nothing mystical about this presentation in terms of possible

worlds. Column (4) of table 2 serves to explicitly report on possible configurations

of the capital market from the point of view of individual capital demanders. A

report on the consequences of other than the optimal choice depicts other worlds

than just the one that becomes actual under the force of the familiar conditions of

rationality and perfect markets. Such a report does not serve to contemplate the

possibility of error – as it would do later in the development of praxeology – but it

shows a world As-If demand and supply determine equilibrium demand for capi-

tal and As-If another than the optimal choice could be made. The model Böhm-

Bawerk built allows only for optima – there is no error – but the individual agents

deliberate, weighing alternatives against the optimum. They act As-If they were

given the option to make an error, even though there is no such thing under the

force of the assumptions.

There should be no confusion between the label ‘As-If’ in my understanding

of Böhm-Bawerk’s use of narratives and the same label in the familiar debates over

Milton Friedman’s (1953) essay ‘The Methodology of Positive Economics’. Fried-

man attached an As-If status to theoretical assumptions, which did not reflect true

descriptions of the world but nevertheless produced good predictions in economic

research; or at least so he conceived of these. In the present context As-If reasoning

is supposed to give true descriptions of real choice options.

Note that, in fact, any economics textbook treats markets – all markets – this

way. Consider figure 3 below, representing a familiar textbook demand-and-

supply scheme (of the now familiar horse market, if you like). P and Q represent

50 Praxeology is a general term referring to methodological individualist theories of human action,

which underlie typically Austrian explanations as to how institutions come about as unintended

consequences of intended action. The term is associated with Ludwig Von Mises and Friedrich

Hayek, rather than with Eugen von Böhm-Bawerk. 51 Hypothetical worlds are also real in a sense Tony Lawson has indicated. See his entire (1997).

Chapter II

demanders surplus P

actual point suppliers surplus hypothetical points intra-marginal area extra-marginal area

price and quantity. The figure represents the idea of economic rent in markets. all

the points on the demand line north-west of the intersection of the demand and

supply lines represent, one could say, the readiness of demanders to buy at prices

higher than equilibrium. As the resulting market circumstances allow them to pur-

chase at a price lower than the marginal utility they perceive for at least some of

their wants, they enjoy a demander’s surplus or rent. In a world where the equi-

librium is in fact generated the one point of intersection is actual. The points that

fence off demanders surplus (and, mutatis mutandis, those that fence off suppliers

surplus, southwest of the intersection) are then non-actual insofar as the commod-

ity in question is traded only at one price at one point in time: against the

equilibrium price. But the other pints on the lines refer to some state of affairs

nonetheless.

figure II.3 Supply and demand: hypothetical and actual

On the one hand these points on the demand and supply lines that fence off

the surpluses are hypothetical, for they refer to choices that would be made if it were

the case that the price turned out higher. This means that there is a hidden seman-

tics of counterfactuals at work in a simple demand-and-supply scheme.

Demanders would satisfy some of their needs at a higher price (and suppliers would

sell at a lower price). This way of putting it is like sketching a possible but non-

actual world.

On the other hand these intra-marginal market parties are, by implication,

demanders and suppliers who exist in the actual world. The same is true for the

demander’s surplus and for the supplier’s surplus. (The points, in turn, that refer

to the preferences by demanders and suppliers with regard to the excluded objects

of satisfaction (east of the intersection) do not fence off any actual surplus. As a

consequence, these seem to be merely hypothetical: they tell us how much suppliers

would offer if the market price were higher and how much would be demanded at

lower possible market prices.)

Textbooks treat intra-marginal demanders and suppliers as if they exist in

the actual world and place them, as one could say, in the nearest possible world

where equilibrium has not yet been attained. My interpretation of Wicksell’s com-

plaint quoted at the beginning of this section can now be expounded as missing a

point. Böhm-Bawerk’s narrative strategy must be read as the pretence to have epis-temic access to the hypothetical possible worlds where actors on the demand side and on the supply side act upon other (possible) prices than the equilibrium price settled in the actual world. I believe that it is the semantics of this As-If strategy – interpreted here as a

possible worlds semantics – that Wicksell failed to capture.

Böhm-Bawerk had read Wicksell’s WKR before the later editions of his PTK

and added eight full pages to further explain das Grundgesetz der Preisbildung, and

an extensive note in which he dismissed that there was a problem of matching a

narrative both with continuity on the factor market and with discontinuity on the

market for horses: Bei meinen späteren Darlegungen über die Preisbildung auf dem Kapital- und

Arbeitsmarkt hatte ich mit Märkten und Waren zu tun, die über jenen

einfachsten Typus hinauswuchsen, und deren Preisbildung daher alle

Eigentümlichkeiten der “reichsten Ausgestaltung des Sachverhaltes“

aufwies.52

He had made a point of this matching in the first and second edition, but now took

back the provisions in the third edition because he considered them unnecessary.

Indeed, in the last chapter, about the Hohe des Kapitalzinses, he explicitly referred to

Wicksell’s complaint, saying “Ein gegen […] Wicksell (Wert, Kapital und Rente S. 111) erhobenes Bedenken behebt sich wohl durch die inzwischen von mir oben S. 288ff. gegebene Erläuterung“.53

Böhm-Bawerk had thus put aside Wicksell’s interpretative mistake, without

much ado, as missing the point. The Austrian seems to have had a weakly devel-

oped sense of how much care a formal approach requires. But this seems to be

constitutive of his rhetoric: inserting and dropping assumptions at will, loosely

and conveniently.

5.2 Narrativism: a concept from historiography

Böhm-Bawerk had told narratives about decision making in an economic environ-

ment with the effect that it became clear to the reader what was the steering

52 PTK, pp.289-290. 53 PTK, p.454, note.

Chapter II

mechanism behind market outcomes. In the imagination of who reads these sto-

ries, the entrepreneur, however hypothetical his actions may be, turns into a living

person. But we have now reached a point in our interpretation of Böhm-Bawerk’s

rhetoric where it becomes delicate to talk of narratio’s, because the art of tale-

telling is, naturally, the prerogative of the historian while the Austrian method of

social study is opposed to, not in line with, the historical method advocated at the

time in Germany.

It is very important to observe the As-If character of the narrative, not the

truth of the stories told in terms of what did or did not happen in the actual past54.

Böhm-Bawerk did not carry out the sort of historical work the German historicists

did. His narrativist engagement differs completely from the motives behind his-

torical tales. Indeed, it is ironical that he is known to have once criticised the

German historical school for sticking to mere anecdotic work, as opposed to the, in

his eyes, much needed theory development. Every time I have spoken, above, of

Böhm-Bawerk’s ‘narrative approach’, then, I have been referring to what is the

opposite of the anecdotic and the superficial in nineteenth century German-style

socio-historical research. Böhm-Bawerk pretended to use the narrative method in

order, not to evade theory development, but to dig into a slice of reality that did not

come to the phenomenal fore so easily. For instance, with regard to the solution of

the scientific problem of the ultimate principle and measuring rod of our utility

valuations of economic goods, he wrote: 55

alle Elemente, aus denen die Lösung [to this problem] zu gewinnen ist, sind in

der ersten Million, vielleicht selbst im ersten Tausend der beobachteten Fälle

schon ganz ebenso enthalten wie in allen späteren. Die Kunst ist nur, sie

herauszulösen, und dazu kann nicht Verbreiterung, sondern nur Vertiefung

helfen […] eine tiefere geistige Durchdringung

This means that historical research may record actual preferences, but after

having accumulated many data on this issue the use of yet new data is very small.

But deductive research engages in setting free (herauszulösen) the underlying prin-

ciples according to which such data emerge: they are integrative for the data. These

principles lie hidden till the moment of discovery by the shrewd economist. It is

easy to see that in both areas of research – historical and deductive – the problem

of the truth of the story arises in some form. In both areas the question arises of

how the narrative integration of the facts is done. In other words, in both German

and Austrian story telling we are in need of an understanding of the truth condi-

tions of these different tales. Do these conditions differ?

In order to acquire some conceptual clarity on this question, I rely on Frank

Ankersmit (1983) concerning matters of the truth and cohesion of historical tales.

54 The concept of ‘actual past’ may sound paradoxical, but the use of the term ‘actual’ is not committed

to any reference to a moment in time. The concept entails that the actual world differs from hypo-

thetical worlds in that the latter are constituted by the course of things as these could have logically

been in the past, the present and the future, different form and in comparison to how they actually

have been, are, and will be. throughout this dissertation, I shall use the term ‘actual’ in this way. 55 Böhm-Bawerk (1890b), p.85. Italics are mine.

He has given some definitional ground for two possible epistemological positions

on the question of what it is that makes a story true or false, how the elements of a

story are put together, and where to find the rules for this narrative integration.

The positions are narrative idealism and narrative realism. We first have to well un-

derstand the use that is made of these two epistemological views by philosophers

of history. Let me go into this before investigating what these positions entail and

how they can serve us in explaining the nature of Böhm-Bawerk’s efforts to give an

insight into the choice making process of hypothetical entrepreneurs.

In tales told by historians, the past cannot simply be represented ‘as it really

was’. A complete description of (a particular aspect of) the past is never possible

and Ankersmit is out to find ‘the mechanism enabling the historian to give a narra-

tive representation of the past’.56 His conviction is that ‘a narrative structure cannot

be attributed to the past as such’. The past is not a tale but it is made into a tale.

Also, to produce our narratives, ‘we do not possess a set of translation rules either’.

Yet ‘there are certain rules governing the narratio which we cannot risk ignoring’57.

Now, essentialists would have it that ‘historians should indicate what is important

or essential for a correct understanding of the past’.58 This sounds plausible

enough, but turns out to be highly problematic in its application. So Ankersmit

defends an anti-essentialist position about history writing: Where is this “essence” [of the past] if there is such a thing? Surely not in the

past itself. The past itself has no “essence”: there are no episodes or aspects of

the past which for the historian’s convenience bear the label “this is the es-

sence”. [Otherwise] the writing of history would be a very simple affair. […]

Whoever says “this is the essence of (part of) the past” points to an interpreta-tion of the past and not to part of the actual past […]. [W]hen we speak of

‘essences’ of the past it is always narratios that we speak of and nothing else.59

Ankersmit concludes that the narratio has epistemic autonomy. The rules of

the logical structure of our narrative knowledge of the past can be found only in

the very narrative accounts of the past. The story told about the past obviously is

an interpretation by the historian. The facts about the past, drawn from historical

resources such as manuscripts, artefacts, oral history, and the like, have to be inte-

grated into one single story. The (supposed) facts are glued together by the

historical narratio. So we can see that the question does indeed arise as to which are

the rules according to which this ‘gluing’ takes place.

Ankersmit calls himself a narrative idealist. He warns that we should not be-

lieve narrative idealism to mean that ‘we should be able to discover the nature of

historical reality by means of an a priori inquiry into narrative philosophy’60. But a

narrative idealist looks for the rules of how to integrate the historical facts in the

story about the past, instead of in history itself. (Ankersmit compares this view

56 Ankersmit (1983) p.93. 57 Ibidem. 58 Ibidem, p.53. 59 Ibidem, p.54. All italics are Ankersmit’s. 60 Ibidem, p.93.

Chapter II

with the Popper’s position, who compared the empirical basis of science with a

swamp above which we erect our bold structures of our theories. There is also ‘no

past that could serve as bedrock for our narratios’61.) In contrast, narrative realism in

history writing is the view that the rules, which determine how the narrative

should be told, lie in historical reality itself and not in any epistemic order. Accord-

ing to this view, not only should the historian try to find particular facts in

historical sources, also the way in which these facts are to be arranged must be

sought in history itself.

The current insight into the nature of historical narratio’s seems to follow

Ankersmit in his narrative idealist orientation. With this weaponry of histo-

riographical epistemology brought to bear, I return to the original question: how

do German historicist and Austrian theoretical story-telling differ in terms of the

truth conditions of these stories? The German historicists appear to be committed

to narrative realism. (But, as shown in appendix 3, I reject Milford’s association of

holism with essentialism.) I endorse the interpretation of the Austrian outlook to

social research as being essentialist. Austrian narratio’s cohere and have structure.

There must be logical rules for these stories. It seems that, for Böhm-Bawerk, this

logical structure lies in economic reality itself.

The assumptions of the story and the three Böhm-Bawerkian causes of the

productivity advantage of present as compared to future goods determine a deci-

sion space from the entrepreneur’s perspective, i.e. in a partial model. The

deliberations of the hypothetical entrepreneur are bounded by this decision space,

so the story about these deliberations has a structure given by the way in which

Böhm-Bawerk models economic reality. But above I have claimed that he believed

a true theory to mimic the unified characteristics of the world. With the conceptual

explications of this subsection, we can now conclude that the narrative logic of

Böhm-Bawerk’s tale-telling is a realist one. This leads to the curious conclusion that

modern historians, though reporting about historical facts, see themselves mostly as em-ploying a narrative idealist logic, while an Austrian economist reporting about individual agents that never have really existed in any straightforward sense can be conceived of as applying a narrative realist logic.

However, the anecdotic character of a story, about an hypothetical entrepre-

neur who considers to augment capital intensity from t=5 to t=6 and then finds out

that the interest he could pay for this has an upper limit of 12%, is of course not

meant to suggest the actual existence, at any point in time, of an entrepreneur who

can be picked out as an historical individual. Nor has there been any real situation

that conformed to the assumptions listed, like perfect capital and labour markets,

perfect foresight, etcetera, and the initial condition of a wage level of 500. Let it be

clear that Böhm-Bawerk’s tales are no historical tales.

The narrative aspect of Böhm-Bawerk’s theory has often been taken to cover

up lacking active mathematical skills in developing an integrative model. Quite

61 Ibidem, p.92.

apart from the question whether this complaint is justified, the narrative practise

has the positive effect of making the decision space of an individual economic

agent perspicuous. This is a rhetorical and didactic advantage. Is there is a further

advantage of this strategy, for the justification of theory development? I have

hinted at an answer already. Let me consider this question more fully now.

5.3 Highlighting the mechanics from micro to macro: sufficient complexity

The reason why Böhm-Bawerk had inserted column 4 – the ‘interest payable’ de-

clared out of court by Wicksell – was not to explain the resulting interest level

itself, as his admirer-critic must have thought he was trying to do. The reason was

that he could use it as an exemplification of the subjective perspective. He searched

for the mechanisms that turned individual decision making into the social (equilib-

rium) outcomes found at the phenomenal level; outcomes that were, in his eyes,

essentially driven by subjective decision processes. The deliberations and account-

ing processes that steer decisions had to be (or to become) objects of study for

whom was interested in the essential character of social reality. Decision making

results are in turn inputs in the grand transition process of which we only can see

the superficial market phenomena. Although Böhm-Bawerk had a rather harmoni-

ous and relatively predictable reality in view – a ‘Victorian view’ as it was called

by Ludwig Lachmann – I believe he was strongly aware of the uncertainty an indi-

vidual agent had to cope with.62 The essentially subjective orientation Böhm-

Bawerk had inherited from Menger implies the view that a study of equilibria as

outcomes of the mere solution of equations would be prone to miss the essential

structure of complex social reality. It would fail to spot explanatory categories like

meaningfulness, individual decision taking and partial perspectives.

It was therefore imperative that economic theory would refer to the underly-

ing subjective micro-processes that come, both causally and analytically, before

macro outcomes. A theory that failed to do so was too simple. Economic theories

are too simple if they fail to identify precisely those aspects of social reality that we

should research in order to capture the essentials of its workings. True science can-

not disregard the underlying mechanics of the transition process that turns

individual decisions – themselves a product of deliberation in a meaningful deci-

sion space full of uncertainty and led by a partial view – into collective outcomes

such as ‘objective’ market prices. Böhm-Bawerk stressed the partiality, not so much

the uncertainty. The topic of uncertainty was for later Austrians to expand social

theory to. Their world view was less ‘Victorian’. However, as I hope to have made

clear by now, the interest for a partial representation of economic processes came

62 See note 49. Lachmann: ‘The kaleidic society is […] not the natural habitat of Austrian economics, but the

alien soil may prove nourishing. A model in which individual plans, each consistent in itself, never have time to become consistent with each other before new change supervenes has its uses for elucidating some striking features of our world.’ Lachmann (1976), p.61.

Chapter II

not out of modesty (as it perhaps did for Alfred Marshall). The partial view on

economic phenomena was to pay attention to social mechanisms, i.e. those mecha-

nisms that should always be the domain of social research. The ‘narratio intensity’

of his approach yields a positive (and presumably decreasing) marginal amount of mean-ingfulness to his capital and interest theory. Wicksell had invested more in

formalization, Böhm-Bawerk in story telling. Each of them found their own harvest

more palatable. But Wicksell did fail to see Böhm-Bawerk’s point. The identifica-

tion of underlying economic mechanisms was the ultimate project the latter was

engaged in. It is not possible to track down Wicksell’s misunderstanding of Böhm-

Bawerk without seeing the narrative realism which is involved in the latter’s study

of economics.

6 Conclusion

Böhm-Bawerk’s contribution to capital and interest theory uncovers essentialist

inclinations and a narrativist realist rhetoric.

As to ontology, or metaphysics, there are supposed to be structures, mecha-

nisms, and processes. Böhm-Bawerk assumed that the economic mechanism –

which transmit individual intentional behaviour into market results – is rooted in a

unified structure, which is there, waiting to be discovered. Hence, one of his self-

imposed tasks is to provide for conceptual and explanatory unification. In addi-

tion, economic theory has to describe this mechanism as fundamentally causal in

character. The belief that such a unified structure objectively exists and that proc-

esses are fundamentally causal in character entails a commitment to essentialism.

The economist, then, is to give a unified picture that reflects a unified world. The

drive to theoretically unify has been illustrated in section 2 by the discussion of

Law of Costs, but also by examples of the concept of interest in the Use Theory and

the Exchange Theory, in chapter I and section 3 of the current chapter, and by the

subsumption of many markets under one that trades present against future goods,

in chapter I.

As to methodology, Böhm-Bawerk walked a road largely rhetorical in kind.

Its persuasive force rests on the method of inserting and dropping assumptions at

will and on what I suggest to call ‘narrativist realism’. One of the assumptions that

mislead Wicksell in the interpretation of the demand for capital was that social

capital was exogenously given. This assumption enabled Böhm-Bawerk to adopt a

partial view on the level of interest and thus to undercut his lack of mathematical

skills. At the end of the very chapter of PTK that Wicksell said to admire the as-

sumption was dropped, though it must be admitted that Böhm-Bawerk did not

properly finish the job of nailing up a comprehensive macro model with an en-

dogenous wage level. Let alone that he would have liked to endogenise capital too.

The want for mathematical proficiency certainly prevented Böhm-Bawerk from

knitting the loose ends together.

Böhm-Bawerk sought to achieve, firstly, theoretical unification with, secondly,

sufficient complexity.

One tool for theoretical unification was that of unifying the microeconomic

conceptual apparatus, another was offering explanations that covered explananda

both in and out of equilibrium. As to conceptual unification, Böhm-Bawerk under-

took to subsume as many real world phenomena under one and the same concept.

Thus, in credit markets and capital markets, but also in labour markets, the traded

commodities are always future against present goods; wages are subsistence goods

and, hence, present goods, while labour is a future good. In accordance with what

later was coined praxeology much of the conceptual apparatus was entrenched in a

civil law framework, unifying praxis and theory too in a way that is, in view of the

divergence of disciplines today, uncommon.

Böhm-Bawerk’s hope to offer sufficient complexity lies in his use of an eco-

nomic narrative. The (comparative) static analysis helped Wicksell to calculate

market results in a mathematical model but it obscured the mechanical genesis of

these results. As Böhm-Bawerk chose the transmission of individual decision mak-

ing to market results as explanandum – which typically places him in the Austrian

tradition – he was concerned not to disregard the mechanics. This is true even

though Böhm-Bawerk’s system described stationary state economies: constant

capital.63 He saw that there is a dynamics at yet another level than that of an ex-

panding economy: the market process is dynamic even in a stationary state and

positive interest is possible even when capital does not accumulate.

In WKR Wicksell refined Böhm-Bawerk’s system. But, with regard to the

implicit dynamics of individual rational adjustment decisions and their collective

consequences, the narrative technique enabled the latter to formulate an explanans

of higher complexity. This level of complexity, springing from the inclusion of the

mechanistic underpinnings, was to satisfy his typically Austrian explanatory

needs. Wicksell misunderstood this aspect of Böhm-Bawerk’s work.

63 For Wicksell the assumption of a stationary state amounts to idealization. See chapter IV, section 2.

Chapter II

Intermezzo

We have seen that Böhm-Bawerk aimed to make scientific progress after Classical

Economics. He tried to do so by theoretical unification, that is by conceptual and

explanatory unification. He also tried to choose the ‘right’ level of complexity of

the explanans in order to leave nothing unexplained of what should be accounted

In what follows I want to show that Böhm-Bawerk’s endeavour to theoreti-

cally progress raises questions concerning (1) the empirical content of economic

theories, (2) the prior convictions economists can be assumed to have in this en-

deavour, (3) the feasibility of explanatory progress in contrast to predictive pro-

gress, (4) the role of abstraction in economic method and (5) the existence of

economists’ essentialist convictions.

These points have been discussed in the previous two chapters as well. But

they have significance beyond the case of Böhm-Bawerk and even beyond econom-

ics. They are not restricted to the idiosyncratic case of these chapters, which talk

just of a particular economist’s way of developing his research program in the

history of Austrian Economics. In this intermezzo I shall glance over points (1) to

(5) in a perspective that is broader in two ways: broader than that of an interest

merely in Böhm-Bawerk and Wicksell and broader than economics as a social

science. First, I believe that the five issues bear on the development of conceptual

apparatus of economics even today. But secondly, if this is true, it is evident that

the role of economics as a study of (an aspect of) the social is being contested;

certainly when its important role as producer of policy recommendations is con-

cerned.

This intermezzo serves as the path toward the chapters III to V, which dis-

cuss these matters more in depth. All three chapter to come are relevant for these

broader perspectives.

1 Empirical progress?

Many jokes about economics mock its alleged poor predictive record and its use of

strong assumptions such that it applies only to very special cases, which will never

occur in reality. Both allegations attack the irrelevance of economics as a source of

fruitful (economic) policy recommendations. At the same time, many research

institutions emit studies with predictions on the basis of economic indicators.

Generally, economists produce predictions about economic variables with much

more precision than sociologists do with regard to other social variables. This is

one of the causes of economics being classified as ‘the queen of the social sciences’

(be it that this appraisal is sometimes uttered with intended irony). If economics as

a discipline does so badly in prediction, it is a mystery why it does so well in the

market of advice and counselling.

Intermezzo

Many economic predictions go with a wide margin of error, which increases

with the length of the period the prediction applies to. This makes economics an

easy target for the joke that ‘God created economists in order to make weather

forecasters look fine’. Indeed, many similarities between economics and meteorol-

ogy strike the eye, including this curious mixture of public mockery and accep-

tance of its predictios. But economic prediction is not only in demand by the public

for practical reasons, it is also a scarce good for economists themselves. Predictions

are needed to test theoretical hypotheses.

However, Neil de Marchi has noted that economic theory alone yields few

unambiguous predictions. Economics has developed purely theoretically since the

eighteenth century as ‘political economy’ and until deep in the twentieth century

economists did not deduce testable hypotheses at all.1 De Marchi therefore ob-

serves that ‘Popper’s critical rationalism actually involves a change of style in

terms of the way economists engage in debate‘ [over the comparative appraisal of

two competing theories]. His history of the atmosphere at the London School of

Economics in the nineteen sixties depicts how after the publication of the transla-

tion of Popper’s (1934) Logik der Forschung, in 1959, the LSE economists wanted to

derive testable hypotheses from their theories, for instance in order to compare

their success relative to Chicago economic theories. It turned out very difficult,

DeMarchi concludes.2 Problems concerning the derivation of empirical tests had

moved into the focus of attention in London already over the fifties, after Milton

Friedman’s (1953) The Methodology of Positive Economics had stressed testability as

well. Furthermore, for qualitative predictions – predictions as to the direction rather

than the size of the change of a variable – one needed measurement of quantities.

As causal influences work in opposite directions, the specification of the direction

of the net outcome that theorists are interested in involves quantification.3

The want for testability and for quantification came with the insistence on

these practices by the LSE professors Archibald and Lipsey, respectively. DeMarchi

contends that Popperian philosophy ‘has no compelling answers’ to either need.

The Duhem-Quine problem of underdetermination in case of falsification and the

prior need to specify a model that fits the problem context stood in the way of

testing. The specification of error terms stood in the way of quantification.

The tools both for testing and for quantification come from econometrics.

According to Bert Hamminga, econometricians maintain a remarkable relationship

with theoretical economists. Econometric modelling, the argument goes, aims to

tract and identify variables in data sets. It requires very strict specifications while,

in contrast, economic modelling seems to engage in a loosening of the specifying

labour. Hamminga pretends to prove the autonomy of each discipline relative to

the other (called ‘the mutual independency thesis’; see below for a discussion). The

discipline that develops the fundamental theories and the discipline that enables

1 DeMarchi (1988), p.142. 2 Although Lipsey has a more positive memory of this episode (personal correspondence). 3 Ibidem, p.140. DeMarchi borrows this insight from Harry Johnson.

testing are said to stand isolated. What the one does is not recorded by the other.

He argues that, at least in the case of international trade theory:

neither some kind of realism [realisticness] of some kind of assumption nor

the conformity of predictions to reality, play a fundamental role in theory de-

velopment.4

If this is true, doubt is cast on the predictive record of at least the neoclassical

economics of international trade. Hammninga researched trade theory from the

factor price equalisation theorem5 to the Stolper-Samuelson theorem6.

2 Progress by appeal to prior plausibility?

’Every Ohlin-Samuelson programme participant seems to “have” a System of

Elementary Plausibility Convictions’ (SEPC), Hamminga says.7 These SEPC differ,

both among economists and among points in time of theory development. Clearly,

SEPC are a priori. They direct research in a context of discovery. Very little is said

about measurable economic processes although the SEPC always have some rela-

tion to the ‘real world’. Hamminga phrases a first, tentative approach to interpret

the notion of ‘the plausibility of a theorem’ as

the probability of the theorem being true in the “real world”, where the “real

world” is just one of the worlds that can be expressed by means of the econo-

mists’ language (production functions, utility functions, factor endowments).

If a theorem holds in many worlds that can be expressed in the economists’

language the probability of the theorem being true in our real world is high,

even without considering at all what our real world is exactly like.8

Purely mathematical methods can be employed so as to increase the plausi-

bility of an economic theorem, because a proposition of economic theory is a mere

metatheoretical provability theorem. Kuipers concludes:

[Hamminga’s] diagnosis of the mathematical nature of economics may be an

important underlying motive for the striking ambivalence of economists

about the question of whether economics is an empirical social science or not.

Moreover, the diagnosis illustrates that the cognitive aims of the social sci-

ences in general and of economics in particular appear to be less evident than

philosophers of science use to assume on the basis of an analogy to the natural

sciences.9

In his study, Hamminga describes four strategies of theory development in in-

ternational trade theory, none of which has anything to do with testing competing

hypotheses. Therefore his example shows that progress in theoretical economics

sticks to delivering fundamental explanatory hypotheses in a way which is quite

4 Hamminga (1983), p.160. 5 In consequence of free trade the same factor of production in different countries will be rewarded

equally. 6 International trade lowers the reward for the relatively scarce (more expensive) factor in a country.

Hence, a tariff benefits the scarce factor. 7 Ibidem, p.95. 8 Ibidem, p.71. 9 Kuipers (2001), p.34.

Intermezzo

disobedient to Popperian methodology. All four strategies build on alterations of

Propositions of Economic Theory (PET). A PET describes an implication with a field

and a number of conditions in the antecedent and a so-called Interesting Theorem (IT)

in the consequent. In Hamminga’s somewhat informal way, a PET has this form:

PET: Field, conditions, → IT

This formulation makes it clear that the PET is metatheoretical.10 I shall now

give a brief review of each ingredient of the conditional. Before that, let me just

give an example. A PET can be the meta-theorem that the Stolper-Samuelson IT

(that the scarcest factor will be remunerated less after trade opens) can be proved

on the domain of two countries, two goods and two factors, given the special con-

ditions that production functions are homogenous to the degree one and that the

factors are immobile between the two countries.

In the consequent of the PET we find the IT. This is the intended implication

of the conditions. For example, the factor-price equalisation theorem is an IT and

the Stolper-Samuelson theorem is another IT. Economists try to formulate such

theorems as surprising or at least new claims under conditions that are ‘plausible’.

A theorem is interesting if it influences discussions among economists or among

the public when economic policy measures are at stake.

Let me start with the antecedent of the PET. The field is the domain of the IT.

It is specified as c countries, f productive factors, and g goods: Fc-f-g. One strategy of

theory development by the alteration of the PET is field extension. If the IT can be

proved for a ‘F2-2-2’, and next also for ‘F2-2-3’ (or for Fc-f-g, 2 < c, f, g ≤ z; z being any

reasonable number in terms of the number of countries, goods and factors as we

generally encounter these in the actual world), it has become more interesting, be-

cause more plausible. Intuitively, this increased plausibility is related to the fact that

countries in fact do not trade bilaterally only, certainly not in only two goods, and

that productive factors are many (taking land, human capital, and risk taking into

account). The quest for ‘field plausibilism’, then, is one strategy to progress.

The conditions in the antecedent are (1) the set of fundamental (neoclassical)

convictions and concepts by which the programme is governed and (2) any num-

ber of special conditions. The convictions are called Fundamentals of Economic

Analysis (FEA), together with some Explanatory Ideal (EI). The special conditions

are for example strictures on slopes of functions, or on the available endowments.

The FEA are, it seems to me, much like symbolic generalisations and meta-

physical paradigms of the Kuhnian disciplinary matrix. For instance, we have

encountered the FEA of what I shall loosely call ‘the classical programme’ in the

previous two chapters. The classical programme is what Böhm-Bawerk attacked in

his critique of the Kostengesetz, that prices (and costs) reflect values and these in

turn reflect the (labour) costs of production. The ‘explanatory ideal’ (EI), in turn, is

10 The arrow ‘�’ can be read as a material implication here. Hamminga does not discuss the problem

that a false antecedent leads to a true implication trivially. See below.

a term from Toulmin’s (1972) Human Understanding. Hamminga characterizes it in

a way that makes it fit into what Kuhn called values (on how to solve puzzles) and

exemplars.11 Indeed, although Hamminga does not explicitly equate his circum-

scriptions to these distinct elements from the disciplinary matrix, he compares

entire paradigms (in the sense of ‘group commitments’ as a whole) with the com-

bination of FEA and EI.12 The classical EI was the canon answering the question

why countries engage in trade instead of being autarkic. Ricardo had said that this

was due to differing labour productivities (and therefore differing end prices) for

different end goods in the respective countries. The neoclassical EI is opposed to

this: assuming identical production functions, but differing factor endowments.

One country would produce more labour intensively than the other.

The three remaining strategies for theory development are all ways of look-

ing for ‘conditions plausibilism’. Hamminga distinguishes weakening of conditions,

replacing conditions by other conditions with deeper concepts, and constructing

alternative special conditions such that the former cannot be deduced from the

latter or vice versa. These strategies differ but I shall discuss weakening only.

The conditional PET is strengthened if the antecedent is weakened. The de-

creasing strictness on the number of countries, for example, comes with more

plausibility, as many more countries than two actually do engage in trade.13 An-

other strategy is weakening of the special conditions. These are substituted by a

new set of special conditions that can be derived from but are not equivalent to the

old set. One result of the Stolper-Samuelson investigations was the theorem that

the scarce factor in a country would benefit from an import tariff, under the rather

special conditions that one country is small and the other big (in terms of domestic

income) and hence that the terms of trade are constant.14 It could be shown that

these conditions rested on particular underlying conditions concerning factor

endowments, utility functions and production functions. Hamminga compares this

form of progress with the description of the earth as the centre of gravity for the

11 Hamminga equates the FEA with a ‘world view’. I do not see why he excludes the EI from the refer-

ence of this concept. Kuhnian exemplars are standard ways to solve puzzles. As the paradigmatic

‘world view’ includes exemplars, it should include the entire FEA-EI couple. 12 Hamminga (1983), pp.123-126. He notes that at a more specific level than that of an entire world view,

a description of economic research deviates from Kuhn’s very typical natural science examples. In a

normal science period, physicists for example do not look for so-called ‘Interesting Theorems’ or for

(the conditions that are true in) hypothetical worlds in which these theorems are true. In revolu-

tionary science, new paradigms give a blow to old ones due to the prediction of new phenomena,

but the trade theory programme did not look for (new or old) real phenomena at all. Hamminga

lists more differences. 13 Hamminga finds many allusions to the notion of ‘plausible’ in what he calls ‘a fountain of expres-

sions’ as economists employ these. Take for instance the likelihood that there are more goods than

factors of production; or the conditions having or lacking economic meaning. In the case of ‘lacking

meaning’ they use terms like pathological, artificial, and implausible. 14 A small country is one which cannot influence world market prices or the ‘terms of trade’. In fact, this

is one of the special conditions.

Intermezzo

moon orbit first under the special conditions that the moon does not also attract the

earth, and in the next step assuming that it does.15

I shall restate Hamminga’s account with the help of the following figure.

Note, meanwhile, that IT’s receive their plausibility from the conditions from

which they are derived. Weaker conditions are more plausible than stronger condi-

tions. The condition that countries do not specialize but diversify in production is

more plausible because it is weaker: its extension is wider, more possible worlds

satisfy it. At the same time, if conditions are weakened, the PET becomes stronger

(and the IT is more interesting16).

excluded possibilities of PET-C

excluded possibilities of PET-C*

Extension of conditions, IT’s and PET’s

PET-C is the conditional C→IT and PET-C* is the stronger C*→ IT. In the

figure, the box represents all logically possible worlds. The worlds where condi-

tions C and C* are the case form subsets of these logical possibilities. C = C1……Cj,

i.e. the FEA-, EI-, and special conditions that lead to the IT, are itself true of worlds

in the subset denoted C in the figure. The weaker the conditions are, the stronger

the conditional (the PET). C* (= C*1……C*k) is weaker than C, so C* has a greater

extension than C, it excludes less. The following holds:

15 See Hamminga (1983), p.52. New conditions with regard to demand and import elasticities were

introduced by Lloyd Metzler in 1940. They were weaker, because they allowed that an increase or

decrease in trade altered the rate of exchange in both of two countries instead of only in one of

them. The original Stolper-Samuelson result turned out to be a special case of the Metzler result. 16 Note that the IT need not be more interesting due to this. Some IT’s are not economically meaningful,

even though the PET may be very strong.

w’’ C

C ⊆ C* → [[PET-C]] ⊇ [[PET-C*]]

(The symbol [[ … ]] here means ‘the extension of …’, i.e. [[PET]] is the domain on

which the PET can be proved.) Increasing plausibility can be interpreted in this

representation as a growing number of worlds satisfying both the conditions and

the IT. Of course, each relevant PET is satisfied trivially anywhere else in the box.

But the economists’ feeling of likelihood of the interesting theorem is generated by

the proof that, besides C-worlds there are more hypothetical worlds such as w’ and

w’’, for which the interesting theorem is true. In other words: PET-C* can be

proved to hold in worlds w’ and w’’ too. The ‘trick’, then, is to increase plausibility

of IT’s by showing that there is an increased number of worlds in which these must

be true. In this sense, plausibility also has ontic implications.

So, if Hamminga is right, economists try to weaken conditions so as to raise

the plausibility of the theorems they find interesting. I have translated this into the

claim that economists aim for a greater strength of a Proposition of Economic

Theory. Although intuitive knowledge plays a role in this development strategy,

observable or measurable economic phenomena do not.

However, it is clear that the discipline also, and often, tries to measure real

phenomena. Without this no sensible policy advice is possible. Empirical work is

centred in econometric work. In order to measure economic phenomena, well

specified variables must be sought; and the more specific the claims of researchers

are, the stronger they are.

Weakening of conditions is precisely the opposite of strengthening the defi-

nitions needed to identify the data (already available or to be collected). As Ham-

minga says, in econometrics ‘minimum requirements for identifiability [of a vari-

able] are far stronger than the special conditions used in the Ohlin-Samuelson

programme’.17 For instance, production functions must be specified. But the trade

theory of Hamminga’s case says nothing about how to specify ‘identifiable mathe-

matical models’ for structural conditions under which production takes place.

There are also anchoring problems for econometricians. That is, it is not clear from

the theory how to select variables from data sets and interpret them correctly as

endogenous or exogenous. To quote Hamminga’s example, how should we know

whether the 1947 US economy approximates a two factor economy near equilib-

rium? So he asserts: One of the great misconceptions of theoretical economists on the nature of

their own enterprise is that the notion of “Heckscher-Ohlin theory” unambi-

guously denotes some element in the structure of economic expositions.18

Thus, Hamminga arrives at his blatantly dramatic ‘mutual independency

thesis’, which says that ‘results of econometric research cannot in the least affect

the dynamics of the Ohlin-Samuelson programme’.19 And although his conclusions

are derived from a case study of international trade, his pretension clearly is that it

17 Hamminga (1983), p.98. 18 Ibidem, p.99. 19 Ibidem, p.100.

Intermezzo

stretches beyond that and applies to all theoretical economics. Econometric practice

cannot help economics in theory evaluation. So while de Marchi concluded earlier

that deriving testable hypotheses from economic theories is difficult, Hamminga

believes that econometrics is not tailored to the practices of theoretical economists.

It seems to me that de Marchi and Hamminga refer to the very same problem; that

is, Lipsey and Archibald, who had wanted to do testing at the LSE, bumped into

the problems the origin of which is explained by Hamminga’s research. Thomas

Mayer apparently agrees. He contends in his (1993) that (what he calls) formalist

theory and empirical science theory ‘invoke widely different criteria in evaluating

theories’.20 Mayer proposed to ‘honour them both’, i.e. to allow each sub-discipline

to have a life independently from each other.

However, although evaluation problems related to the divide between the

need for theory and the requirement of tailoring an adequate basis for producing

evidence to it certainly exist, they are not only due to the different specification

orientations of economics and econometrics. Another problem is that the context of

justification and the context of discovery are severely entangled. Recent research

finds that, as regards economics, the usual assumptions that theories, models, and

data are clearly separated and that empirical assessment comes after model build-

ing are both false Boumans (1997). Criteria for the quality of mathematical models

are a priori. They stipulate requirements that models help find solutions to theo-

retical problems, explain phenomena, hint to policies, or simply provide a mathe-

matical conception of relevant phenomena without further derived uses.21 Eco-

nomic models transform ingredients such as empirical facts and theoretical ideas,

metaphors, stylized facts, mathematical concepts, and policy views into a coherent

mathematical form. Some properties of convenience are introduced into the model

as a special case of mathematical moulding. Furthermore, in order to develop the

best model, it has to be calibrated. The parameters must fit the data and all the

other ingredients, and make sure that the model is ‘true for all the ingredients’22.

But there is no unique manual for the construction of the model, and for each new

investigation there seems to be another recipe. Thus, Boumans compares economic

model building to baking a cake without recipe. The economist cooks and tastes at

the same time, adjusting the cooking process according to his liking: ‘a new recipe

is a manual for a successful integration of a new set of ingredients’23. The emphasis

is to stress that other procedures could have rendered perhaps not an identical but

at least an equally satisfactory result. It is a trial and error process.

It is safe to conclude that simple deductive-nomological schemata do little

justice to the sort of scientific development patterns we should be able to find in

economics. The strategy Böhm-Bawerk followed in order to do better than the

Classical Economists boils down – as I have shown – to trying to make explanatory

20 Mayer (1993), p.36. 21 Boumans (1997), p.28. 22 Ibidem, p.27. It seems to be more appropriate to speak of ‘valid’ instead of ‘true’ here. 23 Ibidem, p.2. See also Boumans (2003). He inserted the italic ‘a’ in order to stress that it concerns one out

of a set of several possible manuals.

progress. The proof for the comparative advantage of Austrian subjective value

theory pops up occasionally as having been provided by the observation of every-

day phenomena, and some other times in the form of higher conceptual and theo-

retical unification. The same is true for the interest- and distribution theory.

3 Explanatory empirical progress?

I have positioned Böhm-Bawerk’s interest and distribution theory as explanatory,

not predictive, but it does not follow that there is no empirics – that is, no basis for

producing evidence – to it. He subsumed many different phenomena under a

description by one set of hypotheses and with the use of one set of concepts. The

key hypotheses are LoMA and the law of diminishing marginal utility (for the

value theory), the productivity of roundabout production methods (for the capital

theory), and the three causes of surplus value of present over future goods (for the

interest theory).

The mechanistic approach offers a minimum level of complexity of the rede-

scription of economic phenomena that enables the economist to explain both how

markets tend toward equilibrium and how disequilibria recur. In other words, the

description of the mechanism explains stable patterns and instability. Classical

‘political economy’ (but also much of 20th century neoclassical economics) cannot

do this. As Hamminga phrases it, the Fundamentals of Economic Analysis of neo-

classical economics is ‘a functional framework which maps any data set on an

equilibrium set’.24 This means that markets are supposed to be in equilibrium,

rather than to be in perpetual flux. Neoclassical explanations perceive all market

phenomena as either ‘already’ or ‘not yet’ equilibrium phenomena. The resulting

focus on stability is apposite for explanations that come by lawlike regularities.

After all, social (quasi-)laws apply to relatively stable environments, free of exter-

nal shocks for some reasonable period of time. A body of phenomena must strike

the observer as constant to make lawlike parlance intelligible. (This is not to say

that the same cannot be claimed, to a certain extent, for the mechanistic ap-

proach.25) The point now is that the regularity account of economic phenomena

forces the theoretician to insert ceteris paribus clauses. Tony Lawson26 is vigor-

ously fighting the dominant constant-conjunction view of economics, as I under-

stand it, because it does no justice to the necessary ontology for an explanation of

socio-economic reality; it lacks metaphysical content. I am not sure whether this

24 Hamminga (1983), p.41. 25 Note that also explanations in terms of mechanisms must assume something to remain relatively

stable, otherwise there is no structural relationship between the composite parts of the mechanism.

Stathis Psillos has eloquently shown this as he noted that an analysis of causation in terms of coun-

terfactuals enters analyses of mechanisms. The mechanistic approach says that two events relate

causally if there is a mechanism that connects them. This makes sense only if there is some constant

conjunction at stake, and if we can think of what the mechanism would do if initial conditions were

different. See Psillos (2004). See also chapter V as I quote James Woodward on a related issue. 26 For instance in his (1997) Economics and reality.

Intermezzo

opposition is required. Stories quoting regular patterns in economic reality might

as well make explanatory sense even if that reality is in flux. Moreover, such pat-

terns may actually exist in a changing world, at least for a (policy) relevant period

of time. The point is not whether clauses theoretically exclude changes in initial

conditions, but whether the theorems in which these clauses figure are false or not.

The following chapter tries to prove that the use of such clauses need not involve

falsity in any fundamental sense.

This also sets idealization apart from abstraction as strategies of analysis.

Glanced over superficially, the product of either epistemic procedure sometimes

resembles the other, but they can be conceived of as each other’s inverse. In addi-

tion, this conclusion has consequences for the policy relevance of economics.

4 Does abstraction involve essentialism?

Explanatory progress, it seems, comes with the abstraction of concepts that help

identify particular patterns or constitutive elements from a set of economic phe-

nomena. The story which tells how these elements cohere and what the patterns

are like will be better as more phenomena can be subsumed under one single rede-

scription. Explanatory progress and explanatory unification come in a brace.

But abstraction involves priors. What to abstract and what to ignore is to be

decided intuitively, as with Hamminga’s SEPC, or on the basis of more explicit

methodological rules that are prior, at least analytically if not in time, to theorising.

Hans Radder has an interesting account of how concepts both abstract from ob-

served reality and structure the observations. I shall discuss his work in chapter V,

which investigates to what extent the essentialist inclinations, which apparently

buttress conceptual developments in economics, can be philosophically upheld.

Essentialism can for now loosely be characterised as a meta-judgement say-

ing that scientists refer to natural kinds in order to explain. In giving the notion

somewhat more precision I follow Kuipers’ hierarchy of epistemological positions.

He lists five questions, the answer to which helps identify the particular episte-

mology that some philosopher adheres to.27 The sequence of questions and an-

swers resembles a forensic inquiry: do you confess or deny? It is a hierarchy, for

any question presupposes that a previous question has been answered.

The first is a basic question in the sense that it ontologically grounds the

subsequent epistemological positions. To say no to this question is to be an onto-

logical idealist, or worse, a solipsist and it would leave very little room for a choice

between epistemologies. That is why it is counted (by Kuipers) as question zero.

0. Is there an independent natural (social) world W?

1. Are true claims about W possible?

2. Are true claims about W possible beyond the observable?

27 Kuipers (2001), p.60.

3. Are true claims about unobservables possible beyond mere reference of

terms?

4. Is there one best conceptualisation, or vocabulary, of W?

An epistemological relativist would say ‘no’ to question 1. An epistemologi-

cal realist says ‘yes’ and has the choice to be agnostic or not about the reference of

nonobservational terms. If not, he can still admit or deny. To say ‘no’ to question 2

is to be a particular type of realist (i.e. an observational realist; a constructive em-

piricist would be agnostic). Clearly, constructive empiricists and observational

realists have to form an opinion as to the status of the objects made visible by

telescopes, electromicroscopes and many other instruments whose working relies

on the truth of well developed observation theories. But stronger positions are

available if you answer question 3 positively too and thereby choose to be a theory

realist.28

Arrived at this position, if you believe that, in principle, an infinite set of vo-

cabularies might adequately help to describe the object of research – the social or

the natural – you answer question 4 negatively. It is the answer of a constructive

realist and this is the position Kuipers defends. But an essentialist prefers one

conceptualisation over the other and goes further than this by a quest for the best set

of concepts. The previous two chapters have claimed that Böhm-Bawerk believed

that this is a worthwhile operation and that he succeeded in effecting it. In addi-

tion, chapter II assessed Wicksell’s judgement as rather benign to this extremely

strong claim. In WKR, Wicksell praised Böhm-Bawerk for his concept of time as a

productive force and sharply contrasted it with competing conceptions, such as the

one developed by Jevons.

By the term ‘essentialism’ I refer to the belief that question 4 must be an-

swered positively: there is one best way to conceptualise the world in order to

research it. This fits the view that essentialist explanations refer to kinds, as these

are supposed to be fixed in reality. One may fight over the question whether a

conical form that appears in reality is best described as a triangle (seen sideways)

or as a circle (seen from below), but a better way to describe it would be as a cone

(depending on one’s description objectives, that is). Perhaps the best way is still

different, in terms of more than three dimensions.

If Wicksell is right in his judgement that ‘time-as-productive-factor’ is the

best explanatory concept, progress has been made possible by successfully ab-

stracting particular aspects of social phenomena from manifold appearances. These

aspects may comprise any combination of objects, properties, and relations. In any

case, they are aspects that identify social mechanisms shaping market outcomes:

market mechanisms. These are rooted in the stable social structure of modern

entrepreneurial capitalism. The social structures, in turn, are the ‘social kinds’

Böhm-Bawerk tried to dig up with his conceptual apparatus.

28 In more recent work Kuipers also proposes a refinement by which it is possible to distinguish entity

realism (terms for objects in reality refer) and structural realism (although such terms do not refer,

theoretical laws have truth value). Theory realism defends the combination of these two.

Intermezzo

The question is: are present day economists also inclined to conceive of theo-

retical economics as ‘finding the right vocabulary’? I think so. Let me devote some

reflection to realism – even essentialism – in economics, and to some more indica-

tions that the economic science’s tacit self-conception is often essentialist.

5 Is economics involved with essences?

In 1976, Robert Lucas wrote an influential paper, which is now referred to as the

‘Lucas Critique’.29 Already in 1971 he had suggested that, if we take the meaning of

the word ‘rational’ seriously, rational people use all the available knowledge,

including economic models. The 1976 critique holds that the outcome of policy

changes cannot be predicted without knowledge of the parameters that describe

individual behaviour. In the Keynesian tradition, macroeconomic models would

render policy advice assuming that markets (the agents who populate markets)

would be kept in a network of invariant relationships if authorities such as central

bankers changed policy. The exogenous variables of the model were supposed to

remain unaltered. Lucas understood that the ‘deep structure’ of economic relation-

ships between agents and authority would be affected by policy changes, for in-

stance because expectations would be affected. If this is correct, the conclusion

must be that macroeconomic policy requires microeconomic analysis even if the

relation between macro and micro is fuzzy. Macroeconomists must again focus

their curiosity on deep parameters concerning individual behaviour.

The Lucas Critique assumes that individuals entering (financial) markets

form expectations on the basis of the models that policy making authorities build.

In his Conversations with economists, Arjo Klamer reports an interview with Lucas.

He asks how we (or rather: his students) have to believe that agents can be ex-

pected to form expectations on the basis of economic models that are beyond their

comprehension. Lucas responds as follows.30

“I try to turn it around. People in business usually like to get into conversa-

tions about what they do all day and how they make their decisions. I’m al-

ways impressed with how sophisticated their thinking and information-

processing is. What puzzles me is the number of economists who seem to be-

lieve the reverse. It would be a miracle if I could write down a model for the

demand for shoes and the supply of shoes, cook up a little difference equa-

tion, solve it, and the solution would reveal profits available to me from the

shoe business that weren’t obvious to people working in the shoe business for

20, 30, 40 years. It seems ludicrous that we could discover sizable rents with

our simple equations without knowing anything about shoes. But some

economists think we can get an insight into someone else’s business without

knowing anything of the substance of his business.”

One page further down, Lucas says:

29 Lucas (1976). 30 Klamer (1983), p.47.

“There is nothing descriptive in demand theory in terms of the process by which

human people, families, and whole business firms make decisions. Econo-

mists have lived with that for years.”

It is laid out here that economists may be making models, but real people who

make the decisions know their thing. Economic actors are instrumentally and

cognitively rational. The theory makers have to model this situation for their own

purposes, but this modelling is not a reproduction of reality in any descriptive

sense.

This is reminiscent of PTK. Böhm-Bawerk also notes how ordinary people

know what to do in everyday decision making without the sort of human delibera-

tions that Austrian economic theory is modelling. Normal decision making is the

result of experience and routine. The models of Menger, Böhm-Bawerk and Mises

aim to reveal a formal structure of the decision making process without pretending

to actually literally describe what it feels like, as it were, to decide as a consumer,

as a demander or supplier on business-to-business markets or on the labour mar-

ket. It is a ‘re-description’31 of the story a real decision taker would tell you, even if

this agent would not redescribe it in the same way. The quote shows that Lucas has

similar reservations as these Austrians: he sees economic theory as a redescription

of the layman’s description, which deviates from a description that is truthful

about the concrete details of the world of economic agents. Moreover, he general-

ises this view so as to ascribe it to the broad ‘economists’ in the last quote.

It is possible to explain this view on the discipline with the help of the fa-

mous ‘instrumentalist’ account of economics known from Milton Friedman. His

essay ‘The Methodology of Positive Economics’ has had much impact on econo-

mists around the word, not least at the LSE. It puts forward the view that neither

economists nor their lay public need bother about the ‘realism’ (as he calls it) of the

assumptions of economic theories. The way in which theories withstand tests is

supposed to be the only criterion on the basis of which to judge their usefulness.

This critique has become known as the ‘F-twist’. Social scientists see the conformity

of the assumptions of a theory to reality as an additional requirement to the test of

the predictions. Dead wrong, says Friedman. He stresses that:

[t]ruly important and significant hypotheses will be found to have “assump-

tions” that are wildly inaccurate descriptive representations of reality, and, in

general, the more significant the theory, the more unrealistic the assumptions.

[…] The reason is simple: A hypothesis is important if it explains much by lit-

tle, that is, if it abstracts the common and the crucial elements from the mass of com-

plex and detailed circumstances[.]32

Friedman does not only claim that the realisticness of the assumptions of a theory

is not needed, but that it is not even desired. Scientifically interesting truth does

not come about by maximizing the accuracy of all the descriptions. The ‘crucial

elements’ (from the mass of complex circumstances) are doing the job. The postula-

31 Uskali Mäki shows how Austrian explanation can be seen as ‘redescription’. See appendix 3. 32 Friedman, Milton (1953), p.14. Emphasis is mine.

Intermezzo

tion of such a scientifically interesting truth has metaphysical bearing.33 Indeed, the

essay sometimes even suggests a strong type of realism:

A fundamental hypothesis of science is that appearances are deceptive and

that there is a way of looking at or interpreting or organizing the evidence

that will reveal superficially disconnected and diverse phenomena to be mani-

festations of a more fundamental and relatively simple structure.34

I draw the conclusion that Lucas has the same normative stance towards

economics as Milton Friedman, although he does not mention him. Like Böhm-

Bawerk, Lucas wants sufficient complexity, but he rejects descriptions and calls for a

redescription in relatively simple terms, at least compared to the actual complexity

of economic life. The question is if his critique is an attempt to get to the best the-

Klamer asks whether Lucas is after truth. He responds:

Yeah. But I don’t know what we mean by truth in our business. I don’t see

economics as pushing that deeply in some respects. We’re programming ro-

bot imitations of people, and there are real limits on what you can get out of

that.35

The answer is mixed. ‘Yes, I want true theories, but no, I don’t know whether I can

get that deep’. One can’t help feeling that yes, he looks for deep structures, but no,

for he is not as optimistic as Böhm-Bawerk with respect to the idea that economics

can ultimately succeed in doing that. He does not pretend to be infallible.

John O’Neill blemishes an inclination to confuse essentialism as a metapro-

position with false claims done by essentialist authors. Especially postmodernists,

he says, who ‘celebrate difference and diversity’ sometimes say that a belief in the

explanatory value of the ‘essential nature’ of things is to deny difference.36 True, it

is easy to fall into the trap of pinning down objects or worse, people, in the search

of their essence. This is how women have been held fixed in their roles by reference

to gender, or inhabitants of colonies by reference to race; it is how essentialist

expositions can help exploitation. But one need not be insensitive to possible mis-

takes in explanations by reference to kinds. Essentialism and fallibilism are com-

patible.

It is relevant to this thesis that O’Neill also observes the following about es-

sentialism and social science:37

Contractual and personal relations are essentially different in virtue of the

meanings constitutive of them. In this respect at least, the essential natures of

social objects are different from those of natural objects. However, the essen-

tialist need have no difficulty in recognizing that the objects of the human sci-

ences do have such distinct properties.

33 Dan Hammond observed the realist, rather than instrumentalist, orientation of Friedman’s essay as he

considered the other essays from the book. See Hammond (1990). 34 Friedman, Milton (1953), p.33. Emphasis mine. Mäki has noted the apparent essentialist, rather than

instrumentalist, import of this paragraph, especially due to the word fundamental. Eric Schliesser

reads Friedman’s essay as Kantian rather than essentialist. See Schliesser (2006). 35 Klamer (1983), p.49. 36 O’Neill (2001), p.170. 37 Ibidem, p.171.

Those opposed against essentialist views of science often deny that true

theories are an attainable goal in science. So many a theory seemed to give the

answer to lots of what scientists wanted to know, only to be replaced by another,

incompatible theory later. Newtonian mechanics was a paradigm after the once

robust Aristotelian world view, but it was replaced by Einstein’s General Theory of

Relativity. It is important to remark that essentialism merely holds that scientists

have to try and find the best vocabulary in order to formulate the strongest true

theory, not that they have already found it. So I say again: essentialism and fallibil-

ity are not contradictory.

The a priori plausibility convictions that Hamminga tracked in the work of

international trade economists fit into a view of economics as essentialist. The

world is believed to behave in a certain way and no other. Economists try to carve

up that world in such a way that this particular behaviour is made intelligible.

There is a truth to be discovered, if the carving is just right. One can encounter this

preconceived notion of science among economists all over the place. Insofar as they

propose new concepts and other explanatory building blocks that help erect alter-

native models, they insist that the new conception picks out something we lacked

so far in order to construct satisfactory explanations. This is a natural attitude not

reserved to Austrians. Take current experimental research aimed at feeding the

theory of incomplete contracts. There is a paper by Fehr, Gächter, and Kirchsteiner,

reporting an experiment by which subjects play the roles of agents in the labour

market. After deciding on a price for the service and a quality level supplied, the

subjects enter a second phase in which the labourer decides what level of quality to

actually deliver. A high quality means a negative payoff, a low quality means fines

and exclusion from trade. ‘Reciprocity’ is the crucial (essential) concept at stake in

this research. The neglect of the presence of motives determined by the existence of

reciprocity in human interactions is taken to seriously miss the point of the work-

ing of markets. Thus, Fehr et al. argue that:

the neglect of reciprocity motives may lead to wrong predictions […] recipro-

cal behavior may cause an increase in the set of enforceable contracts and may

thus allow the achievement of nonnegligible efficiency gains.38

[M]odern principal-agent theory has so far not been concerned with the im-

pact of reciprocity on contract terms and their enforcement. Our results indi-

cate, however, that the neglect of reciprocity may render principal agent mod-

els seriously incomplete.39

We can see that some of the building blocks of an explanatory theory are

seen as essential. Perhaps the use of the word ‘seriously’ expresses this most

clearly. All models, by their very nature, are incomplete. But Fehr et al. refer to

something more severe than just incompleteness of the description of an aspect of

the world. They find the missing links ‘serious’, they accuse competing economists

of having done a bad job. Fehr et al. want to explain by a better conceptual appara-

tus. Is it the best conception? It is not likely that they would be so daring as to claim

38 Fehr, Gächter, and Kirchsteiner (1997), p.833. 39 Ibidem, p.856. Emphasis mine.

Intermezzo

this. But I guess that they would probably hold that economic research, in the end,

is walking en route for the use of ever better conceptions.

I think it is only natural that any scientific work, including social scientific

work, is upheld by a tacit ‘feeling of essentialism’, for the following reason. Daily

scientific scrutiny into the nature of the material and socio-economic world does

not invite a scientist to linger much over the fact that one’s prior conceptual

schemes are contingent and that other people – the lay public, scientists from other

cultures, or of the other sex – may conceptualise the world differently. The natural

attitude of scientists will be to assume that their conceptual scheme is the most

obvious and the most convenient one, if not the only one that leads to true theories.

Only when we choose to consider, at the metalevel, as it were, the relativity of

schemes the appreciation for a pluralist world view is born. This is not to say that

scientists, by default, can never have such an appreciation. But it comes in very

handy to fail to engage ourselves with pluralism so long as one conceptual scheme

helps understand the meaning of our categorisations. I have little reason to engage

in pluralist metaconsiderations while observing the world in ways that are suited

to my interests, even if I can appreciate the possibility to do so; yes, even if I can

appreciate the fruitfulness of doing so in other contexts.

As philosophers we can look down on this attitude. But I think it is weird to

dismiss forms of essentialism purely on analytical grounds and just ignore the

ubiquity of it in and out of science. It seems more plausible to accept its descriptive

adequacy and see to what extent it can be made sound analytically. This is what I

to do in the last chapter.

III Abstraction and idealization

1 Introduction

A multitude of authors have contributed to debates over the use and alleged abuse

of isolational practices, that is, idealization or abstractive reasoning. Isolation is

here used as a general and rather inexact term. It is often referred to as the use of

‘closures’ in debates over methodological issues in economics. A quick glance over

one of my bookshelves helps me to mention those making up just over the first half

of the alphabet: Marcel Boumans, Nancy Cartwright, Victoria Chick and Sheila

Dow, Bert Hamminga, Wade Hands, Daniel Hausman, Maarten Janssen, Theo

Kuipers, Tony Lawson, Uskali Mäki, Thomas Mayer, Leszek Nowak. These are all

scholars who have discussed economics at a meta-level and it is difficult to men-

tion two authors using one such concept like ‘idealization’ in evidently the same

In the following section, I shall discuss the difference between horizontal

and vertical isolations as introduced by Uskali Mäki, because his discussion of

closures in economics is the most subtle I know of. Next, section 3 is to give a very

specific meaning of the notion of idealization as horizontal isolation. I shall try to

make this notion more precise than I did thus far. To this end, I shall make use of a

well documented and often used case from physics. Section 4 takes another look at

closures from the point of view of abstraction, or vertical isolation. It is often a

prerequisite for policy relevant research to allow for some form – but not for all

forms – of ‘unrealistic’ theorizing. The distinction between idealization and ab-

straction helps understanding how policy relates to theory. In addition, one com-

mon way of isolating fields of study from disturbing influences is the insertion of a

vague ceteris paribus clause. In section 5, I aim to show that it is problematic to

interpret the use of such vague clauses as a case of idealization. I shall distinguish

between abstraction from explananda for which there is, and from those for which

there is not an explanans available and claim that in using vague instead of well-

specified clauses, economists tend to abstract rather than to idealise. The policy

relevance of theories with such closures depends on the availability of additional

theories.

Chapter III

δ∆+

δ∆=∆

0OC =∆

2 Isolations: a simple taxonomy

Isolation is a common term for rendering unspecified sorts of closures. I shall in-

troduce a basic taxonomy of vertical and horizontal isolations, a distinction pro-

posed by Uskali Mäki.

2.1 Clauses concerning the ceteri

In the special sciences generally, and even more so in economics, there seems to be

little agreement on the meaning of such terms as idealization, abstraction, ceteris

paribus, concretization, and on what counts as unrealistic modelling. There are a

bewilderingly many meanings in which the so called ‘lack of realisticness’ is dis-

cussed in the literature. As can only be expected, confusion is ubiquitous.1

Ceteris paribus clauses tell us that, by assumption, certain possibly causally

relevant variables take a limit value. Marcel Boumans uses a total differential to

show how economists treat their cetera.2 According to him, a causal and invariant

correlation between two variables, y and x, can be written as:

in which OC is ‘other circumstances’. So then the clause

means ceteris paribus. However, the assumption that

may better be labelled ceteris absentibus. Thirdly, social sciences use a ceteris neglec-

tis clause meaning that a relation between two variables is relatively invariant as a

consequence of the negligibility of the causal influence of the ‘other circumstances’:

In social science, the scope for experiment – for control – is very limited. Boumans

notes that social scientists are therefore above all passive observers. Their models

serve as virtual labs, which comply with the ceteris neglectis clause.

1 Uskali Mäki has developed a taxonomy to weed out the ambiguities in the debate on the precise role

of isolations and closures in economics: “‘is unrealistic’ has been taken to apply to representations

that do not refer to anything real; do not represent any features had by their existing referents; are

false; are non-observational; are non-comprehensive; are simple; are abstract; fail in empirical tests;

are implausible; are practically useless.” Mäki (1992b), p.320. 2 Boumans (2003), p.11.

I take Boumans’ total differential to illustrate the diversity of respective

clauses that may be introduced in order to engage in an idealization. This means

that, under certain conditions, a ceteris paribus clause, a ceteris absentibus clause,

and a ceteris neglectis clause can all be instruments of ‘idealization’. But the notion

of idealization has as yet not been made precise enough.

2.2 Horizontal and vertical reasoning

Note that, in the total differential, the use of these clauses, which I believe can all

usefully be explicated as special cases of idealization, always involves a finite list-

ing of the ‘other circumstances’. This is because without clarity as to precisely

which variables are subject to the closure, it is impossible to know how ‘ideal’ the

object of idealization is treated. Take the example of market demand. If market

outcomes are different than expected on the basis of a market theory and a number

of initial conditions, one can try to look for circumstances explicitly referred to in

the clause so as to find the source of this deviation. The clarity of reference poten-

tially may help to solve part of the Duhem-Quine problem if it ever comes to test-

ing.3 The reference is clear so long as the clauses of economic theorems have a well

specified content. With the reference of the clause well indicated, the idealization

resulting from the use of such a clause renders a closure, which does not increase the

abstractive level of the idealised theorem relative to the non-idealised claim we

started with. That is why the epistemic operation under consideration here is

coined ‘idealization’, and not ‘abstraction’.

I want to borrow Mäki’s own example to explain the difference. A demand

function, which states a functional relationship between the quantity of a good or

service q1 and its own price p1 and the prices of other but related goods (p2, …, pn),

can be made more simple by rephrasing it in terms of a function of price only, un-

der the force of the assumption that the prices of other goods do not change. In

order to do so, the economist needs to introduce a clause concerning the causal

influence of the prices of the other goods and services and the prices of substitutes.

So the reasoning runs from:

q1=ƒ1(p1, p2, …, pn)

q1=ƒ2(p1).

In sum, well specified clauses help to make idealizations take form in eco-

nomics and the level of abstraction does not increase by the use of such clauses. For

this reason, Mäki calls this type of closure, or isolation, ‘horizontal’.

3 Briefly, the Duhem-Quine problem is that falsification of theoretical hypotheses is logically impossible

since the truth of an indefinite number of auxiliary conditions remains unsettled.

Chapter III

Vertical isolations, then, are those that do increase the level of abstraction of

propositions. Mäki gives an example from demand theory. The formula:

q=ƒ(p)

is more abstract than:

q=a + bp,

but this latter one is again more abstract than:

q=8.5 − 0.85p.

The move from the first to the third formula ‘is one of increasing concreteness’4

and the reverse way a case of abstraction. An economist – or indeed any cognisant

agent – who abstracts from particular propositions is leaving out some spatio-

temporal detail of what is referred to by the respective concreter or more abstract

claims. The object referred to acquires more generality, as Mäki’s example clearly

shows. Abstraction, then, is a sub-species of isolation and it epistemically moves in

a vertical direction. The importance of the distinction between horizontal and ver-

tical isolation becomes evident when one considers that de-isolation in one direc-

tion can be accompanied by isolation in the other direction.5

2.3 Abstraction and its use in explanatory theories

To say ‘the rusting of an iron nail is an instance of oxidation’ has aspects of what

above was identified as an abstractive move. This is because the properties of the

process of rusting, which include the discoloration of the iron, the loss of integrity

of the material and the swelling, are not referred to in a scientific description of

oxidation. These concrete properties, accessible for direct observation, are ab-

stracted from in the explication in terms of ‘oxidation’.

Of course, claims about oxidation are not directly implied by claims about

rusting without a chemical theory that explains why rusting is in fact oxidising.

Some more claims are needed to make it an abstraction in the sense I explained in

this section. These very claims in the theory make possible a conditional of the

form ‘if this nail rusts and if [the relevant chemical theorems], then this nail oxi-

dises’. The reason why abstraction is scientifically so interesting in this case is that

the similitude with quite another natural process now becomes apparent. Take the

burning of a piece of wood. This is also an instance of oxidation. If stuck to the con-

crete descriptions of the rusting of nails and the burning of pieces of wood, viz.

without the relevant theorems from chemistry, no living soul would see any simi-

larities between the two processes. Fires strike us with their fierce properties of

4 Mäki (1992b), p.323. 5 Mäki (1994a), p.152. Mäki takes an example from Haavelmo (1944), in which the author concretises (or

de-isolates) and idealises at the same time. Mäki borrows the term de-isolation from Adolfo García

de la Sienra.

heat, light, and even danger, properties that are so very different to our senses than

those of the slow process of rusting. In these examples of abstraction one can see

the use of what I shall explain below as existential generalization: there are particular

properties such that these help identifying oxidation, but in every instantiation of

oxidation these properties can be different. Due to this, an uninformed description

of the spatio-temporal detail of rusting and burning emphasises their differences

and obscures their similarities. The more abstract descriptions of both as a process

of oxidation makes us see their fundamental likeness.

It seems to me that this is what Böhm-Bawerk meant to convey as he

stressed so often that the point of scientific (and by implication, economic) research

leading to truth was to find deeper lying processes, the explanation of which

would unify our vision of the more apparent phenomena. To give a unified de-

scription6 of social phenomena such as borrowing money and trading goods, ab-

straction was required. So forms of abstraction can be a tool in explanatory unifica-

tion. However, I shall use abstraction in a very specific sense in section 4 below.

3 Idealization

In the course of the history of chemistry and physics three related laws have been

discovered that constitute physics textbook knowledge today. They are obvious

candidates to illustrate problems of idealization and abstraction. The laws concern

the behaviour of gases in a closed container. The discoveries took place in an order

from simple to more complex, and, as I shall maintain, from idealised to de-

idealised. My aim will be to get an understanding of the difference between ideali-

zation and abstraction.

3.1 From Boyle to van der Waals – and back again

Robert Boyle is the British chemist (but born in Ireland) today considered the father

of chemistry. For instance, he gave the first modern definition of a chemical ele-

ment. Boyle’s Law appeared in a 1662 appendix to his New Experiments Physio-

Mechanical, Touching the Spring of the Air and its Effects, (1660). (In fact, Mariotte was

the original discoverer.) The 1660 text was the result of three years of experiment-

ing with an air pump. The law states that the product of the volume of a container,

such as a pump, and the pressure of the gas in the container is constant. In an

equation:

6 I do not wish to distinguish between description and explanation here. An abstract description of a

real process will, as a rule, strike the scientist and the layman as explanatory because of the scope

for unification that the abstractive move opens up. Strictly spoken, successful explanations are de-

scriptions; they are very good descriptions, at least from an explanatory point of view.

Chapter III

( ) TcnbVV

PV=c (1)

In other words, P1V1=P2V2. But Boyle had some luck in his experiments. The con-

tainer he used was by no means adiabatic. In consequence, temperature changes

were offset nicely; so a spontaneous material isolation from these temperature

variations and not any theoretical closure enabled him to discover his law.

The three gas laws: progressed de-idealization

Results of experimentation by Jacques Charles in 1787 (and, independently, by the

French chemist Joseph Gay-Lussac in 1802) showed that there was also a fixed

relation between the temperature of a gas and its volume, if the pressure was kept

constant. Every degree Kelvin increase of temperature gave nearly a 0.4 per cent

expansion of the volume. (The more precise present day textbook figure is 0.3663

per cent.) The combination of Boyles law and Charles’ Law leads to the Ideal Gas

Law, which says that the product of the volume of a container with gas and its

pressure divided by temperature is constant; or:

PV=c*T (2)

The constant c* is generally written as nR, i.e. the number of moles of gas mole-

cules n, multiplied by the ideal gas constant R7. It is most accurate at conditions

that correspond to low pressure and moderate temperature, like the atmospheric

conditions.

Nobel prize winner (1921) Johannes Diderik van der Waals modified the

Ideal Gas Law in 1873. He did so by theoretical reasoning, not by experiment. The

van der Waals equation is more complex but also (roughly) correct for a liquid: it

is, in other words, not phase specific. This is because in a liquid phase, the size of

(and the forces between) the molecules relative to the total volume of a mole of

substance cannot be reasonably neglected any more and the van der Waals equa-

tion takes the properties of the individual molecules into account. The equation is:

in which c* is a constant again generally written as nR, a is a measure of the attrac-

tion between the particles due to pairwise attractive inter-particle force (an exam-

ple of the van der Waals force), b is the volume enclosed within the molecules (per

mole of substance, n).

However, the Ideal Gas Law is often used as a rough approximation in

science and engineering calculations. It is said to be an idealised form of the law of

van der Waals. If the non-zero size of the particles (causing repulsion) and their

attraction are both assumed to have the limit value of zero, equation (3) reduces8 to

7 R=8.314472 Joule per degree Kelvin per mole. 8 Reduction is used here in a somewhat vague sense, not necessarily in the more precise sense that laws

can be reduced to explanatory theories.

( ) TcnbVV

naPTcPVcPV

20b0a0T *,*

+ ←= ←=

===∆

equation (2). Moreover, if temperature is assumed constant in the Ideal Gas Law,

equation (2) reduces to equation (1).

Idealization and falsity

Some considerations are in place. First, Boyle’s law can be seen as expressing a

range of instantiations of the Ideal Gas Law. At one chosen temperature (in the

case of the actual experiment by Boyle, around room temperature9) the Ideal Gas

Law is instantiated in the form of Boyle's temperature specific Law. At another tem-

perature it is instantiated again. Secondly, both the Ideal Gas Law and Boyle's Law

are phase specific: they are not accurate enough for liquids. But note that they are

not molecule specific. In contrast, the parameters a and b in the van der Waals equa-

tion have a value related to the sort of matter subjected to changes in temperature,

pressure, and volume. So, as a and b are taken into consideration, the van der

Waals equation is molecule specific. However, and this is a third point to note, the

phase non-specificity of the van der Waals equation suggests that it represents a

lawlike statement (in a Böhm-Bawerkian sense) more fundamental than the other

two equations, or at least more unifying.10

What connects the three equations is that they form a derivational sequence

if some variables are assumed to have a limit value. This, then, must amount to

idealization in the sense of the taxonomy I discussed in section 2. With the size of

the molecules (or the repulsion force) and attraction forces assumed zero, we get

the Ideal Gas Law. In turn, by experimentally holding temperature constant we get

Boyle’s Law. This is a case of limiting its change to a value of zero, or the use of a

ceteris paribus clause – and, moreover, a case of material or experimental isolation.

In sum, we get the following picture.

An arrow indicates what is derived under the assumption specified on top of the

arrow. They point into a direction counter the historical development of physicist’s

understanding of the gas laws.

We can also think of this in terms of formulas at a higher level of abstraction,

as follows.

9 An interesting aspect of the spontaneous closure that occurred in the experiments Boyle did over three

years is the following. Suppose a scientist was not aware – as Boyle himself wasn’t – of the precise

closure needed to get to the simple law that he discovered: the constancy of temperature. Would he

not try to isolate as many possible causal influences as possible? In order to do so, the scientist

would probably prefer to use insulating materials for the gas container, like some synthetic fabric

and not a metal that conducts heat. But in that case the rise in temperature, triggered as pressure

goes up, would not be offset in the way that it was in Boyle’s experiments. It was the very lack of

modern insulating materials – that is, his inability to carry through an isolation – that caused his

luck and permitted him to discover the law. 10 The Böhm-Bawerkian sense in which it is more fundamental is that successful conceptual unification

must be the product of successful research into what is a ‘fundamental’ aspect of reality.

Chapter III

( ) ( ) ( )baTVPTVPVP 30b0a

1 ,,,,,

ƒ= ←ƒ= ←ƒ====∆

The more abstract equations are entailed by the concreter equations, the latter en-

tail the former. Hence, the step from PV=c to P=ƒ1(V) is deductive. It is truth preserv-

ing, for it only leaves out some hard detail from the description; it does not intro-

duce the claim that such detail has in fact somehow vanished from our world.

There has been, one could say, some ‘spatiotemporal’ detail left out; this is the pre-

cise form of the function ƒ1 or, in physical terms, the constancy of the product of P

and V is disregarded. The same is true for functions ƒ2 and ƒ3. In section 0 I shall

engage in further considerations about abstraction.

Reasoning along the arrows – from right to left – develops conditionally de-

ductive. To deduce the Ideal Gas Law from the van der Waals law requires the truth

of the (false) condition that a and b have a value of zero and to deduce Boyle’s law

requires acceptance of the condition that temperature remains perfectly constant.

Two issues now arise. Firstly, the introduction of a false condition touches upon a

key issue in contemporary economic methodology. It is well known that many

methodologists ask how blatantly false claims can help the science to produce true

theories. The approach to sufficiently truthlike theories seems to require an in-

crease rather than a decrease in the number of true propositions. As stated in the

introduction to this chapter, this is the problem of how to deal with ‘closures’.

There also is the logical problem, that when conditionals – conceived as material

implications – have a false antecedent, logically, the conditional claims themselves

are trivially true. My aim in this chapter is to solve both the more general and the

logical problem. Secondly, there is an ambiguity between the process and the

product of idealization. The sort of reasoning which goes under the name of ideali-

zation must be distinguished from the claim we have after carrying out an idealiza-

tional procedure. I shall deal with these issues in the respective order.

As to the first issue, of falsity, suppose that we believe that the van der

Waals equation expresses a relationship true of the actual world (we do not; we

know that the van der Waals equation is an approximation too). We can formulate

the idealization of the van der Waals equation as a counterfactual:

“if the value of a and the value of b both were equal to zero, then the Ideal Gas

Law would be true”.

Note that this is a subjunctive conditional with a false antecedent, which is pre-

cisely what turns it into a counterfactual. In which sense can we speak of the intro-

duction of a false claim? It is important that the falsity is introduced before the

entailment ‘then’. In using a counterfactual, it is as if we have stepped from the

actual world to a hypothetical world, which is close enough to the actual world in

the following respect. All physical laws are more or less the same as they are in the

actual world (in ‘our world’) except that molecules are not subject to attraction and

repulsion. Clearly, this hypothetical world is only logically and not physically pos-

sible. Therefore the description of such a world can be judged unrealistic. Strictly

spoken, the Ideal Gas Law would be true if the van der Waals equation plus the

specified conditions were true. Likewise, Boyle’s Law is entailed by the Ideal Gas

Law under the condition that temperature remains constant. So now I ask: given

the physical (near) possibility of keeping T constant, can one simply say that in the

idealization of the van der Waals Law toward the Ideal Gas Law there is plain fal-

sity involved, whereas in the idealization of the Ideal Gas Law towards Boyle's

Law there is not?

Boyle did keep temperature roughly constant, so perhaps equally roughly

the assumption of the constancy of temperature is physically possible to mimic.

What really happened is that in compressing the container, temperature was in fact

rising. But it went unnoticed as the variations in temperature were offset easily due

to the lack of thermal insulation. So the gas temperature stayed on one isothermal

curve during the experiments, once again roughly. The point is that the more trou-

ble one takes to perform the experiment accurately, the better one approximates an

isotherm. Ultimately, it is physically impossible to keep temperature constant other

than by approximation. This, now, is true of the Ideal Gas Law as well.11 It is how-

ever a better approximation than the ideal gas law. It is physically impossible to

exclude repulsion and attraction between molecules, but with a bit of effort, one

can approximate such a situation, by working with gas under low pressure up to

the point that the factors assumed to have the limit value cannot be materially be

manipulated further as with helium gas. In this case there seems to be a ceteris ne-

glectis clause involved, in the former case of keeping temperature constant a ceteris

paribus clause.

Note that it is the result of reconstruction that Boyle’s law can be understood

as an idealization of the Ideal Gas Law. Boyle himself did not know the need for

the specific ceteris paribus clause, for he did of course not start reasoning from the

Ideal Gas Law upwards to more idealised cases. He had just been experimenting

within his focus of research: the relation between pressure and volume.

Let me now turn to the second issue, the process-product ambiguity. Depending

on the intentions of the inquirer, the answer to the question ‘what is an idealiza-

tion?’ can now plausibly both be ‘the reasoning process by which a scientist introduces a

false clause into the antecedent of a counterfactual in order to arrive at a true claim about

(an) hypothetical world(s)’ and ‘the form a lawlike proposition takes under the pressure of a

false clause’ 12. The reasoning is that given the van der Waals law, if the clause a=0

and b=0 holds, the Ideal Gas Law follows deductively. The product of the idealiza-

tion is the form the van der Waals law takes if a=0 and b=0, that is, the Ideal Gas

Law. Indeed. the counterfactual associated with the reasoning process follows de-

ductively from the van der Waals equation.13

11 Even the van der Waals equation is only approximately true, as can, for example, experimentally be

shown with substances in a process of phase transition. 12 It is perhaps plausible to speak not of ‘the pressure of’ but rather of ‘the degree of freedom created by’

the false clause. 13 I thank Erik Krabbe for the formulation of the proof, in terms of the same the same possible worlds

Chapter III

We now have, I believe, a background example against which we can study

questions of truth and falsity in the use of closures and discuss my alternative ver-

sion of idealization and abstraction.

3.2 Idealizational conditionals as conditional implications

The obvious occurrence of ‘falsity’ in scientific propositions is not just a reflection

of the difficulty to create ‘nomological machines’ – as Nancy Cartwright calls these

– in the natural or the social world. There is more to the common use of false

clauses, that is, of idealizational theorising. This ‘more’ is that the make-up of non-

actual but (physically, economically) possible worlds can be scientifically interest-

ing, even if these hypothetical worlds are dissimilar in some relevant respects to

the actual (or ‘real’) world. Why is this so?

I think the reason is this. I have just proposed to phrase idealizations as

counterfactual propositions, the antecedent of which contains a false clause, that is,

false with respect to the actual world. But, although the antecedent is false, the

counterfactual itself may well be true! To show what this means I shall subject my

general idea of idealizational propositions to a scrutiny somewhat more rigorously

than I did so far.14

In this subsection I shall propose an understanding of idealization in terms

of hypothetical worlds. Next, I shall discuss the issue of the external validity of

theories that idealise and provide a definition of ‘external validity’. Thirdly, the

difference between idealization and abstraction will be taken into consideration

once more, but now with some more precision.

In general. idealizations can be written in the form of a counterfactual ‘if D

then if it were the case that … then it would be the case that DD’. Proposition DD is

what results after the idealization of another proposition D. Above, for example,

the van der Waals law is D and the Ideal Gas Law is DD. Furthermore, deliberate

falsity is introduced in the open space of this counterfactual; without falsity the

conditional would not be a counterfactual, but a subjunctive conditional.

semantics I apply below in subsection 3.3, that the counterfactual, which leads to the Ideal Gas Law

(IGL), follows from the van der Waals equation (W). “Assume that W is true in the actual world (@).

Although the clause (CL) is false in @, the content of W and CL is such that CL-worlds where W

holds are closer to @ than CL worlds where W does not hold. It is then a materially deductive con-

sequence that CL □ W is true in @. Since CL, W ⇒ IGL, it formally follows that CL � IGL".

(The symbol ‘⇒’ is used here as an implication operator. The block-plus-arrow operator indicates

that this is a counterfactual conditional.) 14 Quite some time after engaging in this idea I noted, very much to my surprise, that Ilkka Niiniluoto

had proposed the same: to see idealizations in general as true counterfactuals with false antece-

dents. See Niiniluoto (1990). However, his formalization is strongly rooted in Nowak’s notion of

Idealization and Concretization and, secondly, I here try to develop the idea further with respect to

the special case of economics and its relevance for policy. Pietrosky and Rey proposed to link ceteris

paribus-laws with subjunctive (rather than counterfactual) claims. See Pietrosky and Rey (1995),

p.104.

Very successful experimental set-ups can be seen as hypothetical worlds

“made” (nearly) actual, so that the antecedent, where the clauses are located, is

(approximately) true. Such set-ups are precisely the nomological machines I un-

derstand Cartwright is referring to as showing (true) laws, which are not universal

– the universality lacking precisely due to the spatio-temporal uniqueness of the

experiments, or the instability of the material conditions that keep the details of the

experiment together.15

In idealizations, the falsity is introduced in the antecedent as a consequence

of the use of a well specified 16 clause, CL. The clause may for example be a ceteris

paribus clause, or a ceteris absentibus clause, or any other clause which deliber-

ately describes a non-actual state of affairs. In my discussion of the gas laws, the

clause was that the attractive forces between molecules and their repulsion both

have a value of zero (in deriving the Ideal Gas Law), or that temperature remains

constant while pressure and volume change (in deriving Boyle’s Law).

Thus, an idealizational counterfactual can be written like this:

D&CL □ DD (4)

The box-plus-arrow sign again distinguishes this conditional from the much

weaker material implication. The conditional reads ‘if D is (were) true17 and if CL

were true, then D would be true’. It interests us what is so special about CL. To make

this clear, I shall now introduce some model theoretical notions in the spirit of

Lewis and of Wolfgang Balzer et.al.18.

A theory T (here taken to be a set of propositions) is phrased in a particular

language, L. In other words, the language L generates a (large) set of models, M(L).

Any consistent theory in L has models M(T) for which T is true19. The set M(T),

then, is a subset of M(L). The models of a theory T, elements of M(T), can be inter-

preted as hypothetical worlds; worlds that have properties described by T of which

these hypothetical worlds are models.

A true theory – a theory any scientist with realist inclinations may be sup-

posed to aim for – has (a set-theoretical representation of) the actual world among

its models. That is what truth simpliciter means. I shall refer to the actual world by

the symbol @, and to hypothetical worlds as w. Worlds wi may be relatively similar

to @ in that the same laws of nature hold in them, or that the same true proposi-

tions of economics are true of them. Hypothetical worlds wi may also be very dis-

similar to @, in that they are, for instance, physically impossible.

15 See for instance Cartwright (2001) and (1989). 16 In section 5 we shall see why a clause must be well specified in order to render idealizations. 17 In the treatment of the gas laws, even the most de-idealised van der Waals equation was only ap-

proximately true. But it is conceptually possible that D is true of the actual world. For our purposes

the conditional can loosely be read as ‘if D is true, …’ so long as we read ‘if CL were true…’. 18 See Lewis (1973) and see Balzer (1982) and Balzer, Moulines, and Sneed (1987). 19 An inconsistent theory contains contradictions; hence, such a theory has no models.

Chapter III

The question, then, of what makes clause CL so special can now be an-

swered. Like a theory, the propositions D, CL and DD have models on which these

propositions are true. The point of CL is that among its models, the set M(CL), we

do not find @. The actual world is not an element of the set of models of CL. The falsity of

the clause CL – in the sense that it does not accurately describe the actual world –

is not a nasty coincidence, or a necessary evil. On the contrary, we have seen that

scientists – and surely economists – deliberately insert falsity into their theories, so

as to acquire an idealizational description of reality. If an economist were able to

specify all the causally relevant variables such that they would in conjunction be

true of @, his theory would turn so complex as to become intractable.

The frequent use of ceteris paribus (and other) clauses in economics is an

expression of the idealizational character of economic theories. Allow me to use a

very simple example, which perhaps is instructive for its very simplicity. Econo-

mists may hypothesise that people will cut on many sorts of satisfaction of needs

before they stop renting a house. In consequence, they will perhaps say that ‘ceteris

paribus, the demand for housing in Amsterdam in relation to its price is inelastic’.

If however demand for housing happens to be price elastic, it is not implied that

the assumptions about the preference ordering is mistaken, or worse, that the the-

ory of the market that is used in the background is wrong. It may be that other

circumstances have changed. Suppose that around Amsterdam small towns and

villages have developed cheap housing while public transport facilities to Amster-

dam have improved. Under the pressure of rising house rents in the densely popu-

lated capital, people decide to move out. The ceteris paribus clause turned out

false, because these circumstances typically are the ones that are supposed to be

referred to in the clause. The formulation of the clause in this example may not

look well specified at all – counter the condition I stressed above – but if an

economist can roughly sum up which circumstances are supposed to fall under the

clause, this will perhaps do well enough.

The point I want to make here is that scientists will not generally phrase the

idealization as a counterfactual proposition of the form given here. They will rather

mention the clause CL, insert a comma, and then claim DD: ceteris paribus, all F’s

are G’s. We do not find much of the starting proposition D. This may belong to

tacit knowledge or to background information, but more often, scientists do not

know D, as was the case with the gas laws. (I shall return to this point in the next

subsection.) My reconstruction of an idealizational reasoning step is to highlight

the underlying structure of it.

I have so far stated that idealizations are a particular kind of counterfactuals.

The clause CL helps formulating an idealizational proposition in the sense that CL

is false (of @). However, I also stressed above that an idealizational move is a con-

ditionally deductive operation. This can be written as follows.

D & CL ⇒ DD (5)

The double arrow stands again for ‘implies’. The relation between on the one hand

the not-yet-idealised proposition D and the false clause CL and on the other the

idealised proposition DD is that of entailment. The falsity of CL explains why an

idealization can be interpreted as the counterfactual above. Nevertheless, we have

noted that idealizations can be scientifically interesting because, as a counterfactual

proposition, they can be true. So now we can ask a further question: In what sense

can the idealizational counterfactual conditional – despite being counterfactual, that

is, despite having a false clause in its antecedent – be true? For the answer to this

question I need to develop some intuitions on external validity. I shall propose a

strict definition of external validity which shares some morphological characteris-

tics with Lewis’ definition of the truth of a counterfactual.

3.3 Economics and policy

In case anyone wants to use propositions of economics for policy, the clause CL is

of course a nuisance, because its falsity creates a certain distance with the current

make-up of the actual economic world. Nevertheless, the belief that a particular

idealizational proposition is (approximately) true can be of utmost importance,

both theoretically and in terms of policy relevance.

It is merely theoretically interesting if it helps explain observed phenomena.

An example of idealizational theorems that are explanatorily interesting could

perhaps be this one: ‘this group of high income earners cuts down in working

hours as wages rise, because ceteris paribus, the labour supply curve is backward

bending.’ If the ceteris paribus clause mentioned here amounts to saying that a

specified number of variables remain constant, although it is clear that these vari-

ables will not remain constant at all, then this theorem is idealizational. The clause

is true of hypothetical worlds – its models – that are dissimilar to the actual world

@. What is more, we are quite aware of this dissimilarity. Nevertheless it is theo-

retically interesting insofar as it gives even a partial explanation of observed phe-

nomena.

The belief that the labour supply curve is backward bending can also be in-

teresting in terms of policy relevance because it may help policy makers, say, to im-

pose or (in case the observed behaviour of high income earners is considered un-

desirable) remove restrictions in a market that mimic the conditions expressed in

the clause as much as possible or as much as socially desirable. But of course, not

all interesting explanations have such a direct bearing on policy. The extent to

which theorems of economics are relevant for policy depends, among other things,

on their external validity. Therefore, I shall now expand on the concept of external

validity.

Idealizations: truth and external validity

In methodological discourse about social science, the concepts of internal and ex-

ternal validity have a common but somewhat loose meaning. In order to develop

Chapter III

an understanding of the difference between idealization and abstraction I need to

make the concept of external validity more precise.

The ‘internal validity’ of a study often refers to the correct use of data, or the

absence of confounding variables. For instance, if there are non-random patterns in

the groups of people that partook in the study, the study is internally invalid. The

‘external validity’, in contrast, refers to the generalisability of a study. When cause

and effect relationships between the independent and dependent variables are

demonstrated to be present in a certain experimental situation, i.e. a ‘nomological

machine’ for which clauses CL are (approximately) true, the scientists who de-

signed the experiment would like these relationships to be present outside the

experiment too. Social scientists, for instance, want to say something useful about

groups of people at large. An experimental study that allows its findings to gener-

alise to people at large is said to be externally valid. Clearly, the concepts both of

external and of internal ‘validity’ are gradual, not absolute notions. I want to give

the concept of ‘external validity’ a somewhat different use. I shall give it a meaning

in the model theoretical terms presented above, but the notion of generalisability

remains part of it.

In a Lewis semantics for counterfactuals20 hypothetical worlds are hierarchi-

cally ordered. The hypothetical worlds that are relatively closer to @ are also the

worlds that share more properties with @ than those that are more alienated from

@: they display relatively many similarities to the actual world.21 According to the

concept of ‘external validity’ I am putting forward here, an idealised proposition

DD is more externally valid if the distance between @ and the models of the ideal-

ised proposition that occurs in the consequent, M(DD), is smaller.22 (This notion of

external validity shares properties with the more traditional notion. Among other

things, we can see that it is gradual too.)

External validity, then, is a likeness function, which maps ‘a degree of simi-

larity to the actual world’ to the models of certain (sets of) propositions that we call

externally (in)valid. This function can be seen as describing the distance of the

worlds, in which these (sets of) propositions are true, to the actual world. Some

sets of propositions are labelled theories and some theories are externally more

valid than others, as these are true of possible worlds situated more closely to the

actual world. As idealizations involve falsity in the way explicated above, the sci-

20 See Lewis (1973). 21 Of course, there is no conceptual possibility to see a hierarchical order of worlds of which properties

have been altered one-by-one. As Lewis noted, ‘if kangaroos had no tails, they would topple over’

and this is true in possible worlds that resemble ‘our actual state of affairs as much as kangaroos

having no tails permits [them] to’ (Lewis 1973, p.1.). These possible worlds must somehow differ

from ours in more respects than just the one under consideration, for in this counterfactual situa-

tion, the kangaroo tracks in the sand must have been produced by something else than their tails.

(cf. Lewis 1973, p.9). For my present purposes, this is of no consequence. What matters is that close-

by worlds have things ‘pretty much as they [actually] are’ (Ibidem). 22 The distance d between single worlds is a function d(wi, wj). The distance between the actual world @

and a set of worlds V is the function d(@, V) defined by min{d(@,v)v∈∈∈∈V}.

entist using an idealised hypothesis (DD) is aware of the relative lack of external

validity of this hypothesis. (Below I shall insist that knowledge of the precise con-

tent of the clause is essential.23) But he may judge it valid enough for the purposes

at hand.24

The figure below makes the notion intelligible, showing that the nearest

models of D, for which CL is true, are models of DD. In the proposed Lewis se-

mantics, if the area in the figure indicated to be empty is in fact not, the counterfac-

tual is false. This is because in such a case, given that the DD area is depicted as it

is in the figure, the nearest D worlds for which CL is true are worlds where DD is

not true. The elements of this empty set are conceptually impossible.

figure III.1 Sets of models of D and DD

So here we can clearly see what it means for a counterfactual to be true, it

only requires a restatement of the counterfactual. If D, and if it were the case that

CL, then it would be the case that DD. This is the situation depicted in figure 1. Let

us rephrase this in terms of the example used above. Given demand theory, if peo-

ple put housing very high on their preference list (if D), and if it were the case that

no development of satellite towns and of public transport took place (if it were the

case that CL), then demand for housing would be and remain price inelastic (it

would be that DD). The figure makes clear that if the counterfactual (4) is true then,

and only then, seen from @, in the closest D-worlds where CL is true, DD is true. This

is what the truth of a counterfactual amounts to in the Lewis semantics.

23 Thus, the clause must enable the scientist to judge whether ‘any counterinstance of the law [is] inde-

pendently explicable’ (Pietrosky and Rey, (1995), pp.81 and 90). 24 In this way of putting it the reader may hear the resonance of some familiar statements done by Mil-

ton Friedman in his famous essay on the methodology of positive economics. But the idea of the fal-

sity of assumptions or of hypotheses is fundamentally different here. In my treatment, it merely is

the antecedent of a true idealizational conditional in which the falsity is involved. Therefore we

speak of an idealizational counterfactual.

Chapter III

It should also be noted, secondly, that the models of the relatively non-

idealised (or the not-yet-idealised) proposition D are closer to actuality than the

models of the idealised proposition DD. This can be formalised with the distance

relation just formulated. As @ is the model of true statements about actuality, it is

to say that d(@,M(D))<d(@,M(DD)).

Close inspection of the figure induces one to note, finally, that the set of

models of proposition D – the proposition we had before the idealization – does

not include @. However, it is not excluded a priori that the actual world is also an

element of set M(D), viz. that D is true. In spite of this, I think that such a situation

is unlikely. As we have seen in the treatment of the gas laws, even the van der

Waals equation describes reality approximately only. If we assume, plausibly, that

proposition D is an approximation of a true proposition or only partially true, then

actuality is not to be depicted as an element of the set of models in which the (rela-

tively) non-idealised proposition D is true.

One might wonder why, in case researchers have access to an un-idealised

true proposition, they would want to idealise at all. As a first answer, note that the

historical order in which science produces knowledge of the truth of these proposi-

tions is often reversed. Of the gas laws Boyle’s is the most idealised, but also the

first known. Boyle did not have this access. But, in addition to this, idealised

propositions – true in the Lewisian sense outlined – may be interesting even if the

de-idealised alternative is available. The consequents DD of idealizations are nec-

essarily true of their models, but they are not true simpliciter (or true of @), else they

would not be the consequents of counterfactuals. Their falsity with regard to the

actual world gives an instructive insight into what it is that makes, in the relevant

respects, our actual world. Some detachment of complex knowledge of actuality,

even if readily available, may help us see more simple patterns. Likewise, Margaret

Morrison and Mary Morgan have noted that a ‘model’, in order to make us learn

about the world and about a theory, must have a partial independence from both

the world and this theory.25 For example, knowledge of an un-idealised true

proposition may not be useful if we cannot connect it to further knowledge. In the

same bundle of essays edited by Morgan and Morrison, Nancy Cartwright puts it

as follows: In exact science we aim for theories where the consequences for a system’s

behaviour can be deduced […] But so far the kinds of concepts we have de-

vised that allow this kind of deductibility are not ones that easily cover the

kinds of causes we find naturally at work bringing about the behaviours we

are interested in managing. That is why the laws of our exact sciences must all

be understood with implicit ceteris paribus clauses in front.

25 Morrison (1999), p.17. Note that the concept of ‘model’ as used by Morrison and Morgan is prone to

cause confusion in the present context. I talk of idealizational propositions, like theorems, that are

true of their models. For Morrison and Morgan a model can be a set of equations, a replica, a story,

anything that mediates between theory and the world, is partially independent from both, and that

has some autonomous elements too. Still, also in my more strict use of the concept it is plausible to

speak of an ‘idealizational model’ in case it differs from @.

[…] our best and most powerful deductive sciences seem to support only a

very limited kind of closure.26

Cartwright claims to disagree with Ronald Giere who, to her understanding,

believes that ‘it is instead the truth of the hypotheses of the theory that should concern us

since these indicate the degree of similarity or approximation of models to real systems’27. It

follows from my treatment above that I agree that it is the clause CL, not the theory

itself, which determines the degree of similarity – or difference – to ‘real systems’

(or, in my terms, to the actual world). I baptised this degree ‘the degree of external

validity’.

There is one more clarification to be given. It has to be made clear how the

degree of external validity of the consequent relates to the truth of the entire ideali-

zational counterfactual.

External validity of true counterfactuals and economically possible worlds

What does it mean to say that an idealised theorem of economics DD is externally

valid? And, secondly, what does it mean to say that the (idealizational) counterfac-

tual is true? We have just answered these two questions. The third question is how

the answers to these two questions are linked.

To answer the third question, regarding the relation between external valid-

ity and the truth of counterfactuals, I define the set of ‘economically possible

worlds’.28 These are such that they bear sufficient similarity to @ in the sense that

the ‘laws of economics’ – as these are vigilant in our actuality and whenever such

laws exist – are true in these worlds. One may plausibly maintain that there are no

economic laws, neither phenomenal, nor theoretical, but that there are only physi-

cal laws.29 But this does not formally speak against the idea of a set of economically

possible worlds. From a realist point of view, economics is an enterprise aiming to

capture the boundaries between the economically possible and the economically

impossible. The set of models that exactly exhausts all the economic possibilities is

formed by those models, of which the strongest true economic theory is true. Of

course, we do not know the strongest true theory; it is possible to be mistaken

about it. No strand of realism is incompatible with fallibilism.

I can now say that an idealised theorem in economics, which is true of eco-

nomically impossible worlds only, is externally invalid in the sense and to the ex-

tent that these hypothetical worlds differ from the actual world in their law-like

structure. But economists model worlds in which agents have an infinite life span

or who are endowed with infinite information processing capabilities. Such condi-

26 Cartwright (1999), p.254. 27 Ibidem, p.249. 28 According to what Kuipers calls ‘the nomic postulate’, confrontation of the set of conceptual possibili-

ties with the domain of economic phenomena leads to the subset E of economic possibilities or the

intended applications. See Kuipers (2000), p.147 for a similar set-up of the subset Ph of physical pos-

sibilities. 29 This means that the difference between the set E of economically and the set Ph of physically possible

worlds is the empty set, so E and Ph are identical.

Chapter III

tions are incompatible even with the physical laws as these happen to take form in

the actual world. The worlds economists model often lie somewhere in the large

subset of physical (and, hence, economic) impossibilities. The idealising but true

counterfactuals involved in this modelling practice specify distant hypothetical

worlds for which the clause CL is true: the set M(CL) is not even a subset of the

physical possibilities. Do such modelling practices amount to zero external valid-

ity, i.e. are they necessarily fruitless?

The answer must be subtle. Theories true only of economically impossible

worlds are externally invalid in the sense that these worlds differ from @ in their

laws. This is due to the sense of ‘external validity’ that I use. It does not follow that

nothing can be learned from this modelling practice in economics. Some laws in

such economically impossible worlds will be different, but others will not be dif-

ferent from the lawlike structure of the actual world @. In terms of the semantics

here proposed, the following is the case. The closer to @ the models with, say, im-

mortal agents are, the more their laws are like those in @. The point is which laws

are different from those in @. To illustrate my point, let me use a brief exemplifica-

tion of the so-called ‘irrelevance theorem’ of Nobel Prize winners Franco Modi-

gliani and Merton Miller.30

Economically impossible worlds. The case of capital structure.

The ‘irrelevance theorem’ says that given a definite number of assumptions, the

market value of a firm is equal to its discounted expected rate of return, regardless

of its capital structure. This is due to arbitrage by investors who can choose to ei-

ther invest their private capital in a mixed debt-and-equity firm or invest a mix of

private capital and a privately borrowed amount in a 100% equity financed firm.

The theorem means that it does not matter in what proportion the firm is financed

by equity and debt; even an unlevered firm has the same value as a firm financed

by debt alone. Two of the crucial assumptions are that the interest part of the cor-

porate debt servicing are not tax deductible and that there are no bankruptcy costs.

In later publications, Modigliani and Miller successively dropped these two as-

sumptions in two steps. They so derived the theorems, respectively, that 100% debt

financing maximizes the firm’s market value (step 1) and that there is an optimal

capital structure at a particular debt-to-total-value ratio between 0% and 100%

(step 2). What interests me in this case is not the two assumptions Modigliani and

Miller dropped but the other assumptions, the ones they did not drop in the 1958

paper. Cools, Kuipers, and Hamminga, who treat the three capital structure theo-

rems as a case of ‘Idealization and Concretisation’, list nine other assumptions31,

one of which is that capital markets are frictionless. Without this assumption the arbi-

trage argument used in deriving the ‘irrelevance theorem’ does not work.

It may appear that the proposition about frictionless capital markets violates

economic laws, so that the theory models worlds beyond the boundary between

30 Modigliani and Miller (1958). A correction appeared five years later in the same journal. 31 Cools, Kuipers, Hamminga (1994), p.211.

the economically possible and economically impossible. However, modern elec-

tronic information processing and the possibility of instant information flows are

aspects of an actually changing physical and institutional environment, dissolving

friction in financial markets. It follows that the assumption of the absence of fric-

tion may be interpreted as a proposition about initial conditions, rather than about

economic laws. At least this seems to be the case with regard to capital markets. It

is hard to know what is an economic possibility tomorrow. (I repeat that realism is

compatible with fallibilism.) The same problem, of assessing whether the assump-

tions of a theory are non-actual so as to violate economic possibilities or just be-

cause the initial conditions are different, turns up with the other assumptions that

Cools, Kuipers, and Hamminga list:32

• Firms negotiate a risk-free rate;

• Individual people negotiate a risk-free rate;

• Risk-free debt and risky equity exhaust the financing alternatives;

• Firms operate in the same risk-class;

• Discounted cash flow behaves as a perpetuity;

• Equal access to information for insiders and outsiders;

• There are no agency costs, i.e. stakeholders receive their legitimate share;

• Complete contracts are enforced;

Without some conjectures about what it is that makes a generalization a law,

it is pointless to decide which of these assumptions concern the lawlike make-up of

the world and which only render clauses about initial conditions. What matters

much more is the question whether the interesting aspects of the actual world

come into focus by theorising about models with otherwise non-actual properties.

And there is no a priori rule which tells us what aspect of an hypothetical world is

interesting. This is to be decided by the theoretical economist who is driven by his

curiosity for the comparison of formal properties of the actual world and the world

he models.

The bottom line is that an idealizational counterfactual must be true and that

an idealised theorem (the consequent part of the counterfactual) must have models

that are sufficiently like the actual world, even if these models are, strictly spoken,

economically or even physically impossible. The closer the models of this theorem

are to the actual world, the more externally valid it is. But closeness to @ in itself

does not entail the theorem being interesting. Whether a theorem is interesting

remains a function of the research aims and the background information of the

economists in question.

32 Ibidem, p.211-212. I reformulated the assumptions.

Chapter III

4 Abstraction

Whereas with idealization there always is some form of falsity involved, this is not

the case with abstraction. We have seen that abstraction merely involves the disre-

gard for some (but not all) spatio-temporal detail. In consequence, the result of

abstraction is always a weaker proposition. The relation between a concrete and an

abstract proposition is that of entailment simpliciter.

4.1 Abstraction as existential generalization

Consider the example of the gas laws again. A restatement of Boyle’s law could be

that the pressure of a mole of gas is a function of its volume. Analogously, a re-

statement of the Ideal Gas Law is that pressure is a function of volume and tem-

perature, and the van der Waals equation can be restated saying that pressure is a

function of volume, temperature, repulsive and attractive forces between mole-

cules. These restatements are more abstract than the formulation under which the

respective laws are generally known. They keep the precise form of the functions

in a black box. It is possible, of course, to abstract from a proposition that is already

false at the start. But in abstraction, there is no additional false claim involved.

I now come to what I have already hinted at: the facts that, first, abstraction

is a merely deductive operation and, secondly, that the sort of epistemic move that

materialises the entailment is that of an existential generalization. In case of the Ideal

Gas Law, the restated form – that pressure is a function of temperature and volume

– only implies that there is some function for which it is true that the law can be

expressed. This is of course rather trivial. But many consequences of deductive

reasoning schemes are trivial due to their being deductive. Abstraction is however

interesting when compared to idealization. To show why, I shall confine myself to

Mäki’s example quoted above. The relation of entailment that comes with abstrac-

tion is as follows:

Xd = –0.85P+8.5 ⇒ ∃∃∃∃a,b(Xd = a⋅P+b) ⇒ ∃∃∃∃ƒ (Xd = ƒ(P))

In more general (one may say: abstract) terms the relation of entailment between

the concreter starting point and the more abstract result can be written as:

A ⇒ AA (6)

As we can see, there is no introduction of a false clause. Any falsity of – any dis-

tance between @ and the models of – the abstract proposition must have been car-

ried over from the antecedent of the entailment. In this sense, we can see again,

abstraction is at least truth preserving.

We can immediately draw three conclusions about the external validity of

abstract propositions. Recall that I defined external validity as a gradual notion,

viz. as the distance between the models of an hypothesis or theorem and the actual

world @. The first conclusion is that, as abstraction is at least truth preserving, ab-

stracted but externally invalid propositions derive their invalidity from the those

concreter propositions the former have been abstracted from.

Secondly, and more importantly, abstraction as an epistemic operation is si-

lent about the degree of external validity of the resulting proposition. It is clear that

the abstract but true formulation of an actual state of affairs is perfectly externally

valid, because @ is an element of the set of models of this proposition. But without

a further specification of the extent to which a state of affairs is indeed a property

of the actual world, and not, counterfactually, of an hypothetical world, we do not

know anything about the external validity of the abstracted proposition. In com-

parison, idealization involves the explicit introduction of a false but well-specified

clause, which allows us to direct our attention to a finite number of causal vari-

ables that could be responsible for our idealised explanations failing to refer to the

actual world, or our predictions failing to be confirmed. A specified claim to falsity

helps us solve (part of) the Duhem-Quine problem, for it is logically possible to

check the truth of a finite list of propositions.33 In laboratory settings, for example,

each causal variable can be considered for subjecting it to manipulations such that

the false proposition about it can be made (approximately) true. This is precisely

what has been done with the Ideal Gas Law.

It is the third conclusion, however, which is most of all interesting for issues

of policy relevance. In the act of abstraction, we have seen, we produce weaker

claims about the world. This means that the set of models of which these weaker

claims are true grows by the very level of abstraction. It is conceivable that, ulti-

mately, the actual world finds itself among the members of this set, even if the

abstractive operation took off at a proposition true only of some hypothetical

world. Or, to dwell on the example of the demand function, suppose, however

unlikely, that in @, demand behaviour with regard to a certain specified market is

such, that its proper function has a constant tangent, just as we have been doing all

along; and that the actual function satisfying the demand line is Xd=–1.61P+8.2. In

this case, the line Xd=–0.85P+8.5 does not accurately describe the actual demand

behaviour for the good in question. But Xd=a⋅P+b does.

4.2 A concise antithesis of idealization and abstraction

Wrapping up, I see idealization as an epistemic move away from one logically

possible world to others and abstraction as moving from a smaller set toward a

superset of logically possible worlds. Possible worlds that differ in any minor re-

spect from our single actual world are hypothetical.

I propose to categorize idealization as an epistemic move, materialised by a

clause, which specifies the state of affairs in an hypothetical world in a transparent

– that is, well-specified – way. The idealised description of that state of affairs - the

33 See footnote 3.

Chapter III

clause – lists the differences with the state of affairs in the actual world. In conse-

quence, idealizations imply a claim about the relative external validity of the theo-

rems derived. Furthermore, the level of abstraction of an idealizational conditional

is left untouched, so idealization is a case of horizontal isolation.

Abstraction, in turn, increases the level of abstraction by definition. It is a ver-

tical type of isolation and it cannot be taken to imply anything about the external

validity of the more abstract proposition derived. If the starting point of abstrac-

tion – the more concrete proposition – has models at some distance to the actual

world @, then the weaker entailments may be true of models that are closer to @.

Consequently, more abstract theorems may or may not be true representations of

the relevant state of affairs in the actual world. Furthermore, if the (relatively) ab-

stract proposition AA has its roots in a (relatively) concrete and true proposition A

about the actual world, the truth value is preserved. Abstraction is a deductive

operation and the consequent AA is weaker than the antecedent A. Therefore, if

the antecedent A is false, the consequent of the conditional may or may not be true.

Let us now return to economics. Uskali Mäki used the example of a demand

function to illustrate the difference between horizontal and vertical isolation, and

he coined the latter ‘abstraction’. I repeated the example in order to illustrate the

same, but I called it the difference between idealization and abstraction. Further, I

proposed what the clause is supposed to do and I put it into a semantics appropri-

ate for the aims of this chapter. Now, according to the example, the proposition

that the quantity demanded of a good is a function of the own price of this good is

relatively idealizational compared to the proposition maintaining that it is a func-

tion of the own price and the prices of complements and substitutes (and, we may

add, of the income of the demanders). Formally:

The step represented by the arrow is conditionally deductive, the condition is

specified on top of the arrow. It is clear that idealization leaves the level of abstrac-

tion of the propositions in the antecedent and in the consequent the same.

In contradistinction, the proposition that the quantity demanded of a good is

a function of the own price of this good is relatively abstract compared to the

proposition that it is a specific function with a negative tangent (the function de-

scribes demand for a ‘normal’ – non-perverse – good), or with a function that

specifies that the tangent is minus 0.85 and that quantity demanded is 8.5 units at a

price of nil.

The idealization of income and other prices can also be carried out at a lower

level of abstraction as was done above. Expanding on Mäki’s example, we could

formalise it this way:

ƒ= ←ƒ=

ypppqpqn211

0yppn2

,......,,

,,...,

Thus, one can have an idealised and yet relatively concretised demand function.34

4.3 Abstraction and policy relevance

Jokes that mock the supposed irrelevance and failing predictive power of economic

research abound in the critical literature and, inevitably, on the internet. Mostly,

the jokes express the rather common idea that, for an economist, real life is a spe-

cial case for which theory cannot account. This is a sorry idea for whom hopes that

economic theory can help economic policy advice. Allegedly, economic models are

unrealistic. Economists are said to produce policy recommendations that they are

unable to make operational. Economics is also said to be too ‘abstract’. So what is

wrong with abstraction in economics? To bear relevance for policy recommenda-

tions, an economic theorem must be weak enough to be true, and strong enough to

have a bite. I shall first consider one side of this issue, the weakness part, and discuss

what abstraction has to do with it.

Demand equations express behaviour. Assume that the income effect of a

price change for a particular group of low-income consumers in the actual world is

so strong as to cause a perversely related demand-price couple, as with so-called

Giffen Goods. The related demand behaviour is then falsely described by a concrete

and normal demand function (viz. with a negative tangent). But the more abstract

formula, which merely says that ‘demand is a function of price and income’, may

be true of this behaviour. In this sense abstraction can be said to sometimes also

introduce truth, rather than that it only preserves truth.35 An abstractive move ren-

ders weaker propositions (in the consequent of the strict conditional) that are, by

definition, true of more possible worlds than the more concrete propositions (in the

antecedent of the strict conditional). But this notion becomes interesting if it is also

noted that some weak propositions are strong enough to be interesting for policy.

This consideration brings us to the second, the strength part of the issue. If we

allow for abstract but interesting propositions, we reach a remarkable conclusion.

The more abstract a proposition, the greater the chances that it is true of actuality, so the

more probable that it is a (true) proposition suitable for policy considerations. This is re-

markable given the profusion of jokes about the irrelevance of economics for our

34 This conclusion confirms an observation that Mäki does in his (1994), p.152. He says that ‘vertical de-

isolation is often accompanied by horizontal isolation’. Note however that it runs counter the stan-

dard view of ‘Idealization and Concretisation’, which tells us that idealization and concretisation

are opposite movements and that one first idealises, only to concretise later. See the appendices 7-8. 35 A possible confusion is over generalization and abstraction. Let me just note that (1) abstracted proposi-

tions are more general, but not all generalizations involve abstraction and that (2) abstraction is a

deductive move while some forms of generalization involve induction.

ry p n p c p b a q p b a q m 2 1 1

0 y p p n 2 + + + + + = ← + =

, ,...,

Chapter III

lives. It is after all the level of abstraction of economic theory about which critics

often jump to conclusions: that the use of abstract theory in policy matters neces-

sarily presents an obstacle for a debate on policy intervention.

Under what conditions is a relatively abstract theorem interesting enough –

that is, strong enough – to have a policy bite? It seems that this is a contextual mat-

ter. I do not believe we have the luxury of general guidelines for policy limits to

abstraction. One could object that theoretical hypotheses, even if true, do not di-

rectly apply to the domains of our desire to intervene in the world. This objection

has in fact been put forward by Cartwright in the context of natural science. But

she uses the concept of ‘abstraction’ differently:

Many of our most important descriptive terms in successful physics theories

do not apply to the world ‘directly’; rather, they function like abstract terms. I

mean ‘abstract’ in a very particular sense here: these terms are applied to the

world only when certain specific kinds of descriptions using some particular

set of more concrete concepts also apply.36

The quote shows that there are two differences. Cartwright speaks of the abstrac-

tion of terms, not of propositions as I do and she uses the notion of applicability,

rather than of truth as I do. The two items are related.

Let me first say something about the difference between abstraction of con-

cepts and of propositions. The extension of a concept (Cartwright says ‘term’) con-

sists of those concrete objects in the world that the term refers to. In turn, the inten-

sion of a concept is what we may call ‘the conceptual content’, i.e. what the term

‘means’. We see here Frege’s distinction between Sinn and Bedeutung. Clearly, ab-

stract concepts have a wider extension and a narrower intension than more con-

crete concepts. But the matter looks differently in case of abstract propositions. The

intension of a proposition is a function that assigns truth values to this proposition

in particular contexts;37 the extension is the number of world on which the proposi-

tion is true. In other words, the abstraction of a proposition does not extend the

reference to a greater set of objects in the actual world, but to more possible

worlds. In my approach, more abstract propositions can be derived from more

concrete propositions by the relation of existential generalization.38 Abstract propo-

sitions will more easily be true than concrete descriptions.

Concerning, secondly, the difference between applicability and truth, note the

following. When Cartwright speaks of ‘applicability’ of terms she seems to mean

the ‘reference’ of them. The so-called ‘applicability’ in this sense has to do with the

extension of concepts (‘terms’). The quote from Cartwright above says that abstract

terms are applied to the world only if there are propositions (‘descriptions’) with

more concrete terms are available such that these concreter terms do apply. Com-

pare this to my version. I say that the terms that figure in a proposition may well

36 Cartwright (1999), p.255. 37 See for instance Gamut (1991), p.14. 38 Hamminga and DeMarchi stress the peculiar relationship between abstraction of concepts and of

propositions. Their examples make use of derivability in terms of instantiation, not, as I do, in terms

of existential generalization, which runs the opposite way. See Hamminga and DeMarchi (1994).

refer (‘apply’) even if the proposition with concreter concepts is not available (‘not

true’).

Can we conceive any abstractions that, despite the decreasing strength that

they carry, are interesting in the sense outlined? I do think so. It is a matter of po-

litical debate which ones are interesting and of scientific debate which ones are true.

It is, in addition, a matter of philosophical debate whether any example of an ab-

stract but interesting proposition is abstract enough to impress.

I would like to rhetorically stress that the point seems to be, not whether any

such propositions are conceivable, but rather that it is difficult to believe that there

are none such. Let me finish with some candidates, without further economic ar-

guments. I offer them as suggestions for defence by economists. My aims are phi-

losophical, not economic. The first two are mine. The following four are not mine.

They all belong to the domain of macroeconomics.

The first is: ‘ the peoples who engage in trade instead of in autarky will en-

joy a greater welfare than those who do not.’ The second is: ‘there is a natural rate

of unemployment, which is insensitive to budget policy’. I believe that these claims

are true, generally subscribed to by the core of economists and relevant. So are the

following four, expressed by Eichenbaum who listed ‘four widely agreed-upon

propositions’ in macroeconomics39. I think that his list offers candidates for the

predicate ‘weak enough to be true, strong enough to be relevant’.

(1) monetary policy is neutral in the long run;

(2) monetary policy is not neutral in the short run;

(3) persistent inflation is a monetary and not a real phenomenon;

(4) most aggregate economic fluctuations are not due to monetary policy

shocks;

Then again, one may disagree over my list, or over Eichenbaum’s, or over both.

5 Vague qualifications

Many isolative clauses seem to tell us that some possibly causal factors are as-

sumed to show no changes or to be plainly absent, whatever these factors are. If a

ceteris paribus clause (or a ceteris absentibus clause) quantifies in an indeterminate

way, it has an unclear reference: the factors in question are not explicitly listed at

all. In contrast, a precise (or well-specified) ceteris paribus clause sums up, one by

one, all phenomena that must be assumed, for instance, constant. A precise clause

has a determinate reference, in the sense outlined above: the factors that are object

of a false claim are listed. But what if in economic thought clauses are often in use

that do not sum up the finite set of all possible causally relevant disruptions, how

can we say that these disruptions are idealised? How must we understand that the

39 Eichenbaum (1997), p.238.

Chapter III

factors in question are referred to? The answer I shall give is that they are not re-

ferred to. This has an important consequence.

5.1 Hausman: Vague laws as inexact generalizations

Daniel Hausman has tagged the use of such clauses, i.e. without a clear reference,

as vague qualification.40 He notes that, in economics, idealising conditions like the

ceteris paribus clause are sometimes qualified precisely but more often vaguely.

Vague clauses have an open domain: they are in use whenever it is uncertain

which disturbances are to be idealised. In social science, this is typically the case.

He demonstrates it as follows. Take the law:

Ceteris paribus (C) everything that is an F is a G.

Vague qualifications do not unambiguously separate between the F-worlds for

which the law is true and for which also the C-clause is true and the F-worlds that

have to be treated as an exception by the clause (because not-C is true in these

worlds). In other words, F’s are G’s in all worlds, except in those worlds where,

regrettably, G is not true after all. These turn out to be not-C worlds and the clause

pushes them by the wayside. In addition, we do not know which worlds are C-

worlds and which are not. This is the very reason why Hausman calls it an inexact

law. If an inexact generalization in economics serves as a law, it has simply to be

assumed that the (ceteris paribus) clause refers to a predicate, which, if added to

the antecedent of this unqualified generalization, turns this generalization into an

exact law.41 It is unclear which predicate does this. The economist does not know in

this case which causally relevant variables will set in motion the disturbances of

the phenomena described by the generalization. In brief, a vague ceteris paribus

clause gives limit values to some variables, but it is not clear to which variables. This

makes theories vulnerable to the empirical problems referred to by the Quine-

Duhem thesis (see note 3).

I find this approach highly unsatisfactory, for it does not allow one to decide

anything about the external validity of the alleged ‘law’. A vaguely formulated

clause (henceforth labelled ‘VCL’) does not have terms that clearly refer to any of

the specific spatio-temporal properties subjected to this clause. The use of a VCL

does not meet the requirement I have mentioned above, that it amounts to a finite

list of variables – of ceteri – that are to be assumed absent or to take a limit value,

or to play any other role different from what they actually do.

More generally, it seems to me that, if a theory does not refer at all to some

(large) set of causally relevant factors of an economic process, for example, because

they are unknown, then, in a way, the theory seems to be abstracting from these de-

tails. But this appears to be at odds with the semantic analysis I have given so far.

40 Hausman (1992), pp.132 ff. 41 Ibidem, p.137.

After all, I have stated that the use of a clause gives rise to idealization, not to ab-

straction. Still, I do think it makes sense to somewhat stretch the concept of abstrac-

tion as developed so far and reserve a special place for VCL’s.

5.2 Abstraction as focussing in scientific practice

In order to illustrate my point, let me return to the story about the gas laws. Above

I have already noted that the precise closure needed for Boyle’s experiments with

the air pump was unknown to him in 1660.42 It was partly by luck that he had dis-

covered an empirical relationship, which actually turns out not to be hard-wearing

at all. In retrospective, knowledge of the Ideal Gas Law allows one to spot a well

specified ceteris paribus clause. The fact that Boyle’s law can be derived from the

Ideal Gas Law by idealization is discovered by way of a reconstruction of the facts.

Indeed, thanks to the spontaneous material isolation, discussed earlier,

Boyle could concentrate on the relation of the pressure of a gas with its volume.

His luck had partly been triggered by his focus. As he could never know what

clauses are involved in the resilience of gas behaviour along one isotherm, he can

be understood to have in fact used a vague clause. And note that with such a VCL

there is no well-specified claim to falsity involved. Hence, there was no clarity

about the external validity of the lawlike relationship that Boyle thought to have

discovered. It was not clear what closures were in fact needed for the resilience of

the relationship. But such clarity is there now, so it appears that issues of external

validity can be decided in retrospective.

Do the models of a theoretical hypothesis, hedged with a vague clause, have

a specified distance to @? Apparently not, because the clause is vague precisely in

the sense that this distance cannot be specified. But then, do these models form a

superset of the set of worlds for which the theoretical hypothesis would be true if it

were hedged by a specified clause? This is indeed the case because a specified

clause fixes the distance to @. The vague clause invites the scientist to disregard

some causally relevant variables and concentrate on a particular relationship hy-

pothesised to be of interest – it includes anything, also the variables that could be

subject to a specified clause.

Using a VCL shares many characteristics with abstraction, as follows from

what I have said so far. Above I have characterised abstraction, in accordance with

Mäki, as (1) vertical isolation and as (2) a case of existential generalization. Let me

consider each in its turn.

(1) With regard to the aspect of vertical isolation, let me defend my view as fol-

lows. The use of vague clauses in the sense of Hausman enables the scientist to

pick out whatever he believes is a crucial relationship for the explanandum that is

subject to theoretical fascination. The theoretical hypotheses about these interesting

42 See note 9 above.

Chapter III

relationships have models for which – in the semantics presently discussed: worlds

in which – they are necessarily true. My claim, then, that the use of vague clauses

helps to abstract rather than to idealise is a claim about the relationship to the ac-

tual world, @, that the models of these hypotheses have. A vague clause attached to

a theorem calls to disregard spatio-temporal detail and, thus, to consider ab-

stracted aspects of the models of which the theorem is true. Whether the set of

these models includes the actual world or not is a matter for consideration of dis-

crete theoretical deliberations. Some of these deliberations may have resulted in a

theory that now makes part of the scientist’s accepted background knowledge.

Some are to result in future accepted theories.

(2) In connection with the aspect of abstraction being a case of existential generali-

zation, let me suggest the following. When an abstracted theorem is true in a par-

ticular set of hypothetical worlds (but not necessarily of @) due to the fact that its

more concrete counterpart is true of a proper subset of these hypothetical worlds,

the theorem generalises over this subset. This is clear given that there are no clues

provided of the distance relation of @ with the subset. To say that there is some

domain for which it is true that an x having the property F must also have property

G may seem trivial. But I see no reason to assume that this is more trivial than the

use of a vague clause. The use of a vague clause as represented by Hausman is like

saying that there is some domain of which x is a member and for which it is true

that if x is an F, then it is a G.

VCL, ∀x(Fx Gx) ⇒ ∃D, ∀x∈D(Fx Gx) (7)

Does the use of a vague clause not have too many characteristics of a closure

typical of the idealizational description of non-actual states of affairs instead of a

closure typical of abstraction? I do not think so.

5.3 The ‘Social Model’

Policy relevant theoretical hypotheses say something about how to deal with real

life situations we might want to intervene in. Economic policy may for instance

intervene in areas like market failures, inflation rates43, and trade imbalances. If

these theoretical hypotheses are hedged with an isolating clause, they are true of

models other than the actual world if the clause does not actually hold. It is a pre-

requisite for the policy relevance of these hypotheses that we know precisely what

they are hedged against. This is because only then the clause explains why ob-

served social phenomena deviate from the economic quasi-reality of the theoretical

models. We tend to call such deviations ‘disturbances’ and they need to be ex-

plained; they form an explanandum. In other words, even if our theoretical hy-

43 It is of course dubitable whether inflation rates are phenomena. For an interesting treatment of this

question, see Hoover (2001).

potheses aim to explain our actual world, the clauses that hedge them are explan-

antes with ‘disturbances’ as explanandum. But if the clause is indeterminate, it

cannot serve as an explanans.

If the world we want to intervene in is full of ‘disturbances’ and if we have

no determinate or precise clauses to deal with these, we feel the lack of sufficient

information to explain what makes the real world so different from our models.

We can only hope there is some theory capable of dealing with the multiplicity of

phenomena that disturb our simpler theoretically established regularities. This

additional theory may already exist as part of another discipline, so that it does not

make part of the existing background knowledge of the economist. It may also be

completely absent. It seems that interfield theorising is needed to deal with policy

relevant phenomena that escape our explanatory hypotheses and their clauses that

serve to hedge them.

The need to do interdisciplinary research has been stressed by many econo-

mists of heterodox breed. Those who feel that theoretical economics leaves too

many aspects of the social world unexplained tend to be the same scientists who

also look out for alternative intellectual sources.44 As it does not serve my present

aims to go into this issue very deeply, let me just use one example set by a Dutch

economist who contrasts purely economic models with what he calls ‘the Social

Model’. Folmer calls for ‘a stronger integration of the social sub-disciplines than

currently is the case.’ The proper approach is a simultaneous analysis and estimation of the inte-

grated economic-sociological-psychological model, or, in other words, the So-

cial Model. […] In case a study conducts surveys, information should be col-

lected on economic as well as sociological variables as well as on the percep-

tion of these variables.45

What Folmer calls the ‘Social Model’ could highlight a lot of the interrelationships

between an observed and ‘disturbing’ phenomenon and abstract theory.

Let us summarise the results so far by way of the table below.

WAYS TO DEAL WITH A CLOSURE REGARDING POLICY RELVANT PHENOMENON PH

Explanans Explanandum Epistemic operation

1 Hypothesis X Ph is an instantiation explanation

2 Isolating clause to hypothesis DD Ph is a disturbance idealization

3 Hypothesis AA does not explain Ph Ph is out of sight abstraction

44 A call for interdisciplinarity often comes with an even stronger rejection of mathematical formalism.

See for example the online journal of the ‘Post-Autistic Economics Network’ (see

http://www.paecon.net/). However, a possible limited or orthodox focus is logically independent

from the ubiquitous use of mathematics in economic theory. 45 Folmer (2000), p.881.

Chapter III

A phenomenon Ph may be explained by hypothesis X (row 1). But ever so of-

ten the explanation is not available. In such a case, the phenomenon is pushed

aside, as it were, either by an idealising clause CL, as represented in row 2. The

clause explains why Ph is at odds with hypothesis DD: that is, DD is hedged by

this clause. But Ph can also simply be neglected. In this latter case, the phenomenon

constitutes no explicit part of the explanandum of any theory, even if, evidently,

the phenomenon cannot but play some or another causal role in the world mod-

elled (and quite probably in the actual world too). This possibility is mentioned in

row 3. The hypothesis AA abstracts from the disturbing peculiarities of the phe-

nomenon. Once the disregarded variables turn out to be relevant for the scientist’s

aims after all, the economist hopes to draw from other theories that do take up the

phenomenon in the explanandum. In order to do the job, these (possibly sociologi-

cal or psychological) theories must give rise to theorems of type X. Folmer’s Social

Model would in this case comprise the combined hypotheses X and AA.

It remains to be seen whether an appropriate background theory is always

available. If not, one alternative option to deal with a phenomenon seems to be that

the scientist explicates a (vague) clause after all. If this option is interpreted as a

case of idealization, the clause has to be made precise. We have seen that, without

a clause that is at the same time precise and within approximate reach of being

made true, there is little hope for policy relevance. Without precision, viz. without

appropriate specification of a finite list of variables, there is no way of finding out

which ‘other circumstances’ must be checked to save the theory. In the following

subsection I shall propose another alternative and call this option abstraction ‘ante

explicationem’.

In the following subsection I shall call into consideration a case from the ear-

lier treatment of Böhm-Bawerk’s Interest Theory.

5.4 Böhm-Bawerk: what closure is it that explains the explanandum?

Böhm-Bawerk developed the Austrian Interest Theory presupposing the truth of

Menger’s Subjective Value Theory. He maintained that equilibrium market interest

is equal to the ratio of the marginal yearly yield of lengthening the production

period and the price of labour, according to a particular production function of the

time needed for goods to mature while still in process.

However, changes in initial conditions, for instance in patterns of productiv-

ity and marginal costs, would, with a time lag, work their way to the results pre-

dicted by the Interest Theory. Especially the theory of perfect markets explains

how production costs and market prices tend to converge, a phenomenon that has

given rise to the allegedly mistaken conviction, among most of the Classical

Economists, that the ‘law of costs’ be true. The amount of time needed for adjust-

ment to new equilibria prevents these very equilibria to materialise phenomenally.

And, as changes in initial conditions are profuse in reality, disequilibria shape our

view of the economy: a superficial scrutiny appears to deny the truth of the theo-

rem of equilibrium interest. In order to be able to neglect these disturbances,

Böhm-Bawerk assumed that the time of adjustment was nil, or its speed infinite, or

that the course of time was altogether absent, or that the aforementioned initial

conditions did not change, or perhaps something else. The assumption was im-

plicit and, by any means, rather unclear. He did nowhere really mention that he

wanted to leave the course of time out of consideration or explicitly assume the

speed of adjustment to be infinite. He possibly was not even well aware that he

isolated a system from influences from within that system. Whatever clause he

assumed or did not assume true, he did presuppos the truth of the Subjective

Value Theory, which explained the mechanism by which markets adjust to

changes. And this Subjective Value Theory was not only in the back of his own but

also in the reader’s mind, for the 160 pages in PTK preceding the treatment of the

Interest Theory aimed to convince his audience of the Austrian idea that this is the

ubiquitous economic mechanism establishing any market process.

Apparently, there are two explanations on offer as to which particular im-

plicit closure it was – idealization or abstraction – that Böhm-Bawerk exploited.

The first option is that it was an idealization serving to explain the effect of disturb-

ing influences. In this case, then, there is a ceteris absentibus clause in use (absence

of time) or perhaps a ceteris paribus clause that imposes a limit value to the vari-

able in question (the speed of adjustment is infinite). Deviations from some pre-

dictable equilibrium were the object explained, or explanandum, and the idealiza-

tional use of a clause – an indication of the conditions that are admittedly not ful-

filled in actuality – was part of the explanans.

The relevant theorem says that, given some initial conditions, a particular

equilibrium interest level will come to be. But as the actual world might show

some other outcome, the theorist is in need of an explanation for this deviation. He

does not challenge the Interest Theory, but alludes to the disturbances, expressed

by a clause, to explain the distance between prediction and actual outcome. In sum,

an idealising clause excludes possible disturbances from a theory.

But as we have seen, Böhm-Bawerk, squarely in the tradition of Karl

Menger, took these disturbances instead as explanandum of the mechanistic Sub-

jective Value Theory, rather than of clearly explicated clauses that could otherwise

hedge this theory. Assuming the truth of the Subjective Value Theory, an idealiza-

tional clause was not really needed in the Interest Theory. In other words, slug-

gishness of adjustment simply did not play any role, neither as an instantiation of a

presumed lawlike generalization derived from the Interest Theory, nor as a dis-

turbing candidate for idealization. For the explanation of interest, markets were

interesting for their results, not for their adjustment processes.

My view is that the disregard for the actual deviations from what the Interest The-

ory puts forward as the truth serves to abstract the result of the adjustment mechanism on

goods markets and on the labour market from the process of adjustment. This,

then, is the second option and the one I prefer to explain how Böhm-Bawerk

Chapter III

brought about the particular closure. The advantage of abstraction by failing to

explicitly refer by way of a specified clause is that the scientist is not impeded in

theorising by his lack of comprehensive knowledge about the complex processes

he is making an explanatory theory about.

I conclude that Böhm-Bawerk permitted himself the thesis that, given par-

ticular initial conditions, his theory predicted certain equilibria, regardless of the

effect the course of time has on market results. He could do so because the equilib-

rium producing mechanics had already been described and explained by Menger’s

theory. The so-called disturbances simply were not relevant to him any more: he

abstracted from them. So also here – where unspecified and implicit clauses exercise

their closure in the background – we can see the ignoring of spatio-temporal detail;

we can see abstraction at work. The particular closure that Hausman discusses, the

use of vague clauses, is not simply a case of horizontal isolation. It is more like

abstraction.

5.5 Abstraction ante explicationem and abstraction post explicationem

I want to characterize Böhm-Bawerk’s reasoning strategy as abstraction post explica-

tionem. This means that he abstracted from the course of time in his interest theory

after the developments of Mengerian Subjective Value Theory, due to which he

could be confident that the explanation of mechanisms, that take time to evolve,

had been taken care of.

But note that an additional theory, which takes the detail of a deviant case as

part of its explanandum, is not always available. Suppose that the economist needs

such an additional theory and that actual disturbances imply the risk of falsifica-

tion (without which there will of course be no problem). In this case he can only

hope that additional explanations will be developed at some later day. The reason-

ing strategy now is one of abstraction ante explicationem.

This means that scientists may focus on the causal relationships that, so he

hypothesises, will tell us much about phenomena that arose his curiosity, without

paying too much attention to the many other variables that influence the phe-

nomenon under study. If the ‘other variables’ are limited in number he could insert

a clause to account for deviations from what he describes the world is like. But as

they are, by default, too numerous to formulate a proper clause – a well specified

clause – he can ignore them. Of course, the disturbing content of the actual world

will often call for scientific attention. Suppose that the economist needs such an

additional theory and that actual disturbances imply the risk of falsification (with-

out which there will of course be no problem). In this case he can only hope that

additional explanations will be developed at some later day. It may take long be-

fore an additional theory that takes this disturbing content as explanandum.

If the theorist wants to save both the theory and the phenomena, he will

have to explain any deviations by reference to the idealising practice of economic

research. If, on top of this, the theory pretends to be relevant for policy, de-

idealization is a prerequisite. Cartwright calls this ‘customizing’.46 The policy

maker cannot abstain from considerations of how to deal with the circumstances

listed in the clause. He will above all be interested in the question whether these

circumstances can actually be given the (approximate) value they assume in the

idealization. As we have seen in the case of the gas laws, some approximation is

not excluded a priori, although it is clear that in economic policy – and in social

engineering in general – this encounters practical, ethical, or legal difficulties. But

note that if we believed that the social world was unconditionally inaccessible for

making any inroads at all, there would be no point for any policy whatsoever.

Alternatively, I characterised Böhm-Bawerk’s reasoning strategy in the in-

terest theory ‘abstraction post explicationem’ because he could be confident that

some explanatory hypotheses had been developed before and that these hypothe-

ses satisfactorily explain how deviations come about, such that the theory at issue

can be saved. Every economist builds on former work and thereby assumes the

accepted theoretical hypotheses from the past. In general, Austrian mechanistic

market theories provide us with the tools to scrutinize the precise effect on market

outcomes of sluggishness and the time needed for deliberations by entrepreneurs,

workers, and consumers. An economist who dislikes Austrian economics, quite

regardless of his reasons, may not disregard time lags themselves but may never-

theless ignore the underlying mechanism that explains the time lags. He abstracts

from this mechanism. If he explicitly dismisses the truth of the relevant Austrian

hypotheses, his epistemic move can be classified as abstraction ante explicationem,

because we are not told what the lags’ provenance is, hence this explanation is to

be awaited. If, in contrast, he explicitly accepts such Austrian explanations, it can

be classified as abstraction post explicationem.

The distinction between ante explicationem and post explicationem reason-

ing brings in considerations about background theories and research aims. These

help solving the problem of inexact laws. They do so in the following way.

An inexact economic theorem is a product of abstraction ante explicationem.

The unknown and fuzzy set of details, captured by the vague clause, remain

unmentioned. It follows from my critique of Hausman’s discussion of vague quali-

fication that, in this case, this set of details is being abstracted from. More generally,

it follows that the judgement as to which epistemic operation is in use is a matter

of a background of research interests. The aims Böhm-Bawerk must have had in

mind can be found out on the basis of our knowledge of the developments of the

theory of the day and the tradition to which he belongs. Hausman believes that the

growth of economic understanding over time will decrease the vagueness of inex-

act laws.47 This is in entire agreement with a thesis I can now defend: that post ex-

plicationem abstraction fixes part of the meaning of a clause by (possibly implicit) allusion

46 Cartwright (1999), p.250. 47 Hausman (1992), p.136.

Chapter III

to some existing theory. The more scientific knowledge expands, the more post expli-

cationem allusions we will be able to allow; and the more precise our existing

clauses can be made.

6 Conclusion

The efforts to make a gas law complex enough to deal with an idealizational clause

involved in a simpler law provide an example of theoretical de-idealization in the

course of the history of science. The clause attached to Boyle’s law is very specific

although the relative constancy of temperature in the experiments was more or less

a fortunate coincidence. The clause attached to the Ideal Gas Law is also well speci-

fied. The variables subjected to the clauses are undisputed referents of the terms in

the idealizational law-like hypotheses; i.e. we know what these clauses are about.

All three gas laws are approximately true, the extent to which the approximation is

acceptable depending on the respective aims the engineers of the physical world

happen to have in their (design) research.

The engineers of social science are the policy makers (or their advisers.) And

social reality generally presents difficult cases of intervention. The mere fact that

economic policy exists reflects that we believe to have at least some economic

knowledge of what is economically possible and what is impossible. Even if our

economic theorems are idealizational, we think that these are true enough. We also

think that their external validity is such that, on these theorems, we can reasonably

base decisions to intervene in economic processes. But we may be dead wrong in

this belief. The idealizational clauses tend to be a nuisance. They describe non-

actual states of affairs, and the scope for moulding the actual world such that it

counts these states of affairs among its properties is very limited. We can try to

reorganize a particular market such that it mimics a perfect market, but few mar-

kets can be made like a perfect market in all respects. We can also try to control a

market and erect prohibitive tariff walls but contraband will make us face reality.

For social design research, it would be helpful if we could de-idealise en-

tirely. De-idealised theorems would guide us in interventions. At least, one advan-

tage of a well specified idealizational clause is that, if predictions concerning eco-

nomic policy outcomes turn out wrong, the cause of deviant phenomena can be

attributed to variables referred to in the clause. Specified clauses can therefore be

taken to indicate something about the external validity of the idealizational theo-

rem. In this chapter, idealizations are interpreted as counterfactuals. In their se-

mantic status as conditionals, these can be true. Why? Because in the actual world

we are able to imagine possible worlds; we can imagine the set of economically

possible worlds (even if we do not know how to demarcate this set).

In contrast, abstraction, as outlined, involves no propositions about the ex-

ternal validity of the model in which the abstraction occurs. The object abstracted

from is an indeterminate set of details that presumably play a role in the possible

world the concreter description is true of. So what is the relevance for our actual

world of abstract theorems about a model world? The answer lies in the extension

of abstract theorems. Abstract theorems count among their extension a set of pos-

sible worlds larger than those referred to by more concrete theorems. But this im-

plies that abstract theorems tend to have a higher external validity than their more

concrete counterparts. More strongly put, abstract theorems sometimes make a

good chance of being true of the actual world.

The use of vague clauses in what apparently are idealizational theorems is to

be interpreted as a case of abstraction, rather than of idealization. The reason is that

vague clauses tell us nothing of the exact domain of models for which the theorems

are true. This is a similarity with the case of abstractive scientific claims: one can

abstract over propositions that are true of the actual world but also over those that

are false, without being aware of it. This what we see with theorems that are sus-

tained by vague clauses too. The point of tacit or explicit ‘open clauses’ is not that a

false clause is involved as such and that, hence, the theorems be idealizational. I

want to show that the point is that there must be some explanans that takes the

discrepancy between the state of affairs in actuality and the state of affairs pictured

by the theory as explanandum. In the abstraction ante explicationem case this ex-

planans has to be awaited. In the abstraction post explicationem case (part of) the

deviant differences are explained by another, already existing, theory.

IV Four examples

1 Introduction

One important conclusion of the previous chapter was that abstraction is helpful

rather than an obstacle in policy relevant scientific research. Without abstractive

hypotheses there is no interesting theorizing and, hence, no policy relevant re-

search beyond pre-scientific intuitions. Idealization, in contrast, is likely to cause

trouble if it comes to intended interventions in the social world.

I shall consider four economic policy relevant economics cases, trying to de-

tect whether they aim at vertical or horizontal isolation. In case of abstraction they

are in need of the bite that makes abstract hypotheses interesting. In case of ideali-

zation they have to be de-idealised in order to apply to concrete situations.

First, I discuss Wicksell’s attempts to develop a theory of capital building to-

gether with the interest theory he inherited from Böhm-Bawerk. Heinz Kurz has

noted that Wicksell’s ultimate aim was not to study the stationary state stricto sensu

but to make use of a static view of the economy in order to ultimately highlight the

dynamics of capital growth. However, the distinction between these two types of

theoretical stance remains cloudy in his paper. The concepts of abstraction and

idealization come in handy to explicate the distinction more clearly. The second

case is a rather simple textbook example about the Keynesian macroeconomic ex-

planation of involuntary unemployment. I shall use my explication of idealization

and abstraction as a conceptual engine and note that an often repeated claim about

Keynes’ idealizational assumptions is false. The third case is a more complex

treatment of theory and policy interventions in the inefficient Dutch market for

health care before the latest restructuring of this market.1 Finally, I shall look into

the tacit idealizations in an argument against buying organic and fairtrade food.

Their being tacit makes it look, superficially, like the use of a vague clause, about

which I have commented that it might better be treated as abstraction. But a closer

look reveals that some very well known facts are ignored such that a finite list of

excluded variables figure in the background. The assumptions in the argument

help idealise, not abstract a case.

1 As of 2006 the Dutch have reinsured themselves against the costs of health care in a new system with

more elements of a market economy.

Chapter IV

2 Wicksell’s stationary state

In a stationary state, capital is constant. This implies that the entrepreneurs exhaust

all earned interest for consumption. They do not add saved interest to the principal

sum: they do not accumulate. Wicksell follows Böhm-Bawerk in considering the

way in which non-compound interest has to be calculated.

2.1 Building on Böhm-Bawerk: rent

In his Positive Theorie des Kapitales, Böhm-Bawerk assumed land ‘away’ for the ease

of representation. He thought this would do little harm to the external validity of

the model, because of the durability of this factor. It is a sort of perpetual.2 The

fruits of its use stretch out from the present to the far future, so the discounted

value of these fruits are now negligible.

In section 6.2 of chapter I it has been explained that perpetuals only earn net

yield as these are not depreciated: Der Grund ist immer der gleiche wie früher3. Sup-

pose the yearly yield is indexed at 100. With a discount of 4%, the fruit of last year

is worth 100/(1.04) = 96.15, of the year before it is now 100/(1.04)2 = 92.46, etcetera.

The total value is the discounted value of all the yield calculated as the sum of an

infinite geometric series. In the example it is of course 100 / 1.04 + 100 / 1.042 + … +

100 / 1.04∞ = 2500 or the yearly yield is multiplied by 100 / 4.

Wicksell allowed the amount of labour allocated to one hectare of land to be

endogenous. He could do so thanks to his mathematical treatment, which does not

encounter the problem of a multitude of simultaneous equations – as long as the

number of variables does not exceed that of equations. Böhm-Bawerk had been

limited by his narrative approach. But he was grateful in the newer editions of

Ich kann nur bedauern, daß mir, als Nicht-Mathematiker, die mathematischen

Darlegungen Wicksells nicht völlig zugänglich sind; ich kann sie deshalb we-

der in ihrem Detail genauer verfolgen, noch mir für meine eigene Darstellung

zu Nutze machen, und muß mich begnügen auszusprechen, daß sie nach ih-

rer mir erkennbaren Hauptrichtung eine jedenfalls in hohem Grade dankens-

werte Ergänzung meiner Theorie zu bieten scheinen.4

With land as a separate factor, the subsistence fund is now needed to pay for

wages and rent during the production process. (I continue the counting of the

equations as started in chapter II, with the same choice of symbols.)

2 Perpetuals are bonds without closing date. This means that they will bear a fixed interest forever but

are not paid back. 3 PTK, p.423. He goes on (p.424): Daß [der Grundstück] für den Eigentümer ein reiner Ertrag, ein reines

Einkommen wird, das hat gar nichts mit Fruchtbarkeit, Lage, Bodenklasse usw. zu tun, sondern lediglich mit der geringeren Wertschatzung künftiger Güter… Die theoretische Erklärung der Grundrente fällt also in ih-rem Schlussstücke mit der Erklärung ausdauernder Kapitalstücke zusammen.

4 PTK, p.426, note 1. Italics are in the original.

K = ½tA(ℓ+gr) = ½t(Aℓ+Br) (4)

Like in figure 1 of chapter I, production per worker per year is

p = (ℓ+gr)(1+½zt) (5)

The optimum is given by:

dp ⁄dt = (ℓ+gr)½tz (6)

But, as B ⁄A is endogenous, revenue per worker can also grow due to fewer work-

ers per unit of land (or more land per worker)5:

dp ⁄dg = r(1+½tz) (7)

2.2 Land as capital: the insertion of heterogeneity

Böhm-Bawerk had allowed to drop the assumption in the isolated model of homo-

geneous production functions in the last chapter of PTK. This not only involves

that all marginal yields are the same (which has been listed as assumption (4) in

chapter II), but it also opens up the possibility that end products (assumption (2)),

labour (assumption (5)), but also land and capital goods are heterogeneous. It trig-

gers the idea that there is a loop structure of capital (assumption (8)). However,

Böhm-Bawerk had not been able to incorporate the consequences of all de-

isolations into a coherent mathematical system, or General Theory. In Wicksell’s

model the allowance of heterogeneity of production caused to increase the number

of equations and variables. Wicksell’s trick was to work with the assumption of

two countries with mutually different (due to different outputs) but internally

homogenous production functions. Country 1 produces good x; country 2 pro-

duces good y.6

Endowments and production period must receive indices referring to coun-

tries i, j, marginal utilities will carry indices x and y. Under appropriate assump-

tions (related to market perfection), after opening trade these countries would ap-

pear as one single economic system with two types of production functions. Factor

5 Wicksell remarks that in the theory of the marginal productivity of land, Von Thünen did thought

experiments with several doses of capital, so as to calculate rent. Wicksell criticises him for not yet

endogenising capital. This looks very much like Böhm-Bawerk’s stepwise approach with several

‘doses’ of capital – or years of roundabout production. However, I believe the difference is that

Böhm-Bawerk did want to endogenise t (it was his key to the mechanism) but lacked the mathemat-

ics. 6 As we shall see below, Wicksell came to see how much more difficult this was by the time he pub-

lished the original (Swedish) edition of the Vorlesungen, in 1901. Note the difference with the Ricar-

dian trade model, in which each country produces the two commodities. Wicksell had a very differ-

ent aim than Ricardo: the latter wanted to analyse the welfare effects of trade as compared to au-

tarky, and for that the realistic assumption is required that domestic economies bring forth more

than one product. The point of Wicksell is to trace how one economy with different production

functions behaves. Hence his strategy is to merge two models of economies with homogenous pro-

duction functions into one.

Chapter IV

incomes thus loose their indices. In other words, one crucial new assumption is

that of factor price equalisation, without which for each factor no single markets

would exist in the two countries together7. Further assumptions remained:

• Output consists of non-durable consumer goods8;

• The relation of capital to land is fixed;

• There are no futures and forward markets;

The set of equations, then, is:.

A = A1+ A2 (8)

B = B1 + B2 (9)

K = K1 + K2 (10)

g1 = B1 ⁄A1 (11)

g2 = B2 ⁄A2 (12)

Equations (4) to (7) above are doubled in number in the two-country two-

commodities model:

K1 = ½tA1(ℓ+gr) = ½t(A1ℓ+B1r) (13)

K2 = ½tA2(ℓ+gr) = ½t(A2ℓ+B2r) (14)

p1 = (ℓ+g1r)(1+½zt1) (15)

p2 = (ℓ+g2r)(1+½zt2) (16)

So are the derivations:

dp1 ⁄dt1 = (ℓ+g1r)½t1z (17)

dp2 ⁄dt2 = (ℓ+g2r)½t2z (18)

dp1 ⁄dg1 = r(1+½t1z) (19)

dp2 ⁄dg2 = r(1+½t2z) (20)

We have until now thirteen equations, (8) to (20). The thirteen variables are

A1, A2, B1, B2, K1, K2, t1, t2, g1, g2, ℓ, r, and z. The marginal utilities of x and y relate in

7 See WKR, p.130. Note that, in the later Heckscher-Ohlin international trade models, this is not an

axiom, but a theorem. Showing how axioms of models can be theorems of other models seems to be

one important form of progress in economics. 8 Durable capital goods are, so to say, home made. In fact, as the non-durable consumer goods are

defined as the wages fund and, hence, as capital, Wicksell did at this stage not yet cope with the

problem of the heterogeneity and the loop structure of capital. Once he tried to take this step, he

ended up in the trouble of measuring the value of capital and the average period of production.

the same way as the prices of x and y, respectively. So Wicksell operated largely in

a Walrasian way, assuming equilibria on all markets, without reference to the

mechanism by which such equilibria come into being. This is the key difference

between his and Austrian economic theory.

Nevertheless he said that the price relation of the commodities ‘kann aber hier nicht als bekannt vorausgesetzt werden, es ist vielmehr eben unsere Aufgabe, zu zeigen, wie es durch das Spiel der sämtlichen volkswirtschaftlichen Kräfte bestimmt wird.’9 He

goes on to say: ‘[Wir müssen] uns auf den Markt des Tausches der beiden Waren versetzt denken, und die Bedingung dafür aufstellen, daß auch auf diesem Markte Gleichgewicht herrscht’10. So there is a need for another equation, but the price relation of the

goods is lacking. If π is the price of y in terms of a quantity of x, or x ⁄ y, then in-

come e of an individual worker equals x + π∙y or

e = x + (U’y ⁄ U’x) ∙ y (21)11

So x and y can be expressed in terms of the factor rewards. Aggregate consumption

of these commodities will be

Σx = A1∙p1 (22a), or

Σy = A2∙p2 ⁄ π (22b)

so that

Σe = Aℓ + Br + Kz (23)

The system is now closed. If free trade takes place in a model under the as-

sumption of two-countries and two-commodities, the two countries are similar to

one country with markets for two final goods: there is one market for each factor of

production. Hence, ℓ, r and z are not indexed for the type of industry. The ex-

change relation between the two goods is endogenously fixed when12: any of the

factor incomes is maximised given the other two (partial analysis); factor incomes

are equal in the two lines of production; available capital suffices to finance wages

and rents (suggested by equations (13) and (14)); the goods are distributed such

that their relative marginal utilities are equal to the exchange relations.

As this model represents a stationary state economy, the equilibrium situa-

tion is thought to persist. No new fixed capital13 is added to the stock of durable

production goods. Capital, labour, and land are taken as given and savings, other

9 This phrasing does suggest a certain interest in mechanisms, but this is not what Wicksell is after when

he merely paraphrases Böhm-Bawerk’s system mathematically. 10 WKR, p.133. This is like the Clarkian approach in the short run. See subsection 2.4 below. 11 In equilibrium the goods are traded at a price relation that equals the inverse of the relation of their

marginal utilities. 12 WKR, p.135. 13 Recall that the end goods that form the subsistence stock to finance labour and land are Böhm-

Bawerk’s floating –- or working –- capital. Tools, machinery, buildings and the like are called fixed

capital, which nowadays stands in contrast with non-durable capital, such as combustibles.

Chapter IV

than gross savings used for maintaining capital stock, are absent. There is as yet no

short run or long run dynamics in the model. Another way of saying this seems to

be by calling the method static.

Böhm-Bawerk’s assumption had been that land can be treated as a special

case of durable capital goods. However, the loop structure of investment and pro-

duction14 enters already at the level of land itself and not just at the level of capital.

Böhm-Bawerk defined capital as piled up (or frozen) land and labour15, but land, in

turn, can be defined as a sort of capital good. This, then, implies that land partly

consists of land, contradicting Böhm-Bawerk’s other claim that land is an original

factor. In other words, land cannot simply be calculated as the sum of an infinite

geometric series in the way Böhm-Bawerk had done16. Wicksell conversely pro-

posed to treat durable capital as a rent-earning good. He entered such goods as

productive factors into the equations as ‘B’ and as ‘g’. Due to Wicksell land got its

autonomous status in the general theory.

The assumption of a stationary economy had allowed Böhm-Bawerk to cal-

culate the average investment period as a measure of the quantity of capital: in

PTK more capital is synonymous with longer roundaboutness. But with two origi-

nal factors some capital may consist of more frozen land, other of more frozen la-

bour. The heterogeneity of capital then props up more pressingly. If the project is

to also make the model dynamic, tracking down the quantity of capital becomes a

real problem. Kurz discusses the attempts by economists to repair Wicksell’s

model, which have turned into the debate on Wicksell’s ‘missing equation’. Kurz’s

analysis is interesting for two reasons. The first reason is that he misinterprets

Wicksell on the period of investment and the failure to come to terms with the

simple interest – as implicitly used by Böhm-Bawerk and explicitly in his own

equations. The second reason why I am interested in Kurz’s contribution is his

treatment of the difference between the idealization of economies as being in a sta-

tionary state and the abstractive method of a static analysis.

One must distinguish between short run dynamics (or business cycle analy-

sis) and long run dynamics (or growth trends). Capital changes in the longer term

only. Hence, the most dramatic assumption dropped was a sort of domain assump-

tion, that the models described the life of capital stock as being in stationary state.

Many other assumptions subsequently had to be dropped with this one, the most

important of which is the assumption that growing capital stock is heterogeneous.

This means that the variable for the average period of production cannot be equal

to ½t, so Wicksell substituted the symbol ε for ½ in WKR. But that would not solve

the problem, for long term growing economies present us with the loop structure

of capital. The investment period becomes intractable. This, then, is the nastiest

problem Wicksell faced (and never really solved).

14 See note 76 of chapter I. 15 Note that this definition applies both to durable capital goods with their key role in the capital theory

and for the subsistence fund that was assumed in the interest theory. 16 See chapter I, section 6.2, page 42.

2.3 Simple interest as a dispensable assumption

Some passages from Wicksell’s work seem to imply strictly stationary conditions,

numerous others cause ‘this impression [to be] quickly dispelled’17. Heinz Kurz

concludes that Wicksell did not want to describe stationary economies. There is additional evidence that Wicksell did not intend to study the problem

of distribution in terms of a strictly stationary state of the economy. To see

this we have to turn to his criticism of Walras and his successors.18

But does the quote really prove that Wicksell was not interested in the stationary

state at all? Let us see what else Kurz and Wicksell claim about the conceptual

struggle with the role of capital. Kurz indicates two ways for Wicksell to choose

from: either capital is measured by some value unit, or capital – heterogeneous as it

is – is represented by the average period of production in the economy (which is in

fact Böhm-Bawerk’s solution). This is in accordance with Blaug’s analysis who

quotes a paper by Wicksell claiming that either one can represent capital in the

model as a cross section of existing capital (fixed and floating) goods or longitudi-

nally, as ‘having a time structure’.19

Kurz says that Wicksell started off using the average period of production as

an indicator for the capital intensity of the economy. Allegedly, Wicksell wanted to

explain the distribution of factor incomes in a static framework, even assuming

fixed factor supply.20 But to deal with growing capital stock, Kurz goes on, requires

calculating compound, not simple interest. So Wicksell did not see the importance

of the use of compound interest for the analysis he ultimately was after.

But this, I believe, is not true! Another quote can show this. Wicksell con-

tended that assuming a stationary state makes the use of simple interest harmless,

even ‘essentially’ so:

Das Produkt εt, d.h. die Investierungszeit des Kapitals, kann aber hier als eine

einzige Variable aufgefaßt werden, so daß wenigstens bei Berechnung einfacher Zinsen die Ausdrücke nicht wesentlich verändert werden.21

The words I put in italics clearly show that, given simple interest, and only under

that condition, the quoted factor εt can be taken as exogenous. Wicksell does not

17 Kurz (2000), p.774. 18 Ibidem, p.776. 19 Blaug (1997), p.537. This paper is reprinted in the appendix of the Vorlesungen I. Blaug subsequently

judges it unclear whether Wicksell’s intentions embrace Walras’ idea – expressed in his Éléments d’économie politique pure – that interest merely originates out of growing capital stock. This would

however imply that capital in a stationary state has no time structure. Wicksell has been very clear

about this from the start. Time and again, he endorsed Böhm-Bawerk’s insight as groundbreaking

that capital has a time structure even in the absence of growth, i.e. in a stationary state. 20 Kurz (2000), p.775. Note, however, that the assumption of fixed factor supply is not required for either

an analysis of a stationary state or for a static analysis of a dynamic model. 21 WKR, p.101. Italics are mine. Wicksell referred here to the fact that the average production period

(which boils down to the ‘investment period’) need not be ½t, as the distribution of labour and land

over the process of production can be subject to choice by the entrepreneur, or to technical condi-

tions, giving rise to a period εt instead. The coefficient ε supposedly had to be empirically deter-

mined. It seems to me that the term wesentlich must be taken literally.

Chapter IV

claim at all that the use of simple interest were of no matter for the essence of the

analysis. Otherwise stated, he merely says that the conditional, with the assump-

tion of simple interest in the antecedent, is true.

The whole idea was not to stick to the analysis of a stationary state, Kurz

claims, but to a comparative static analysis. In that case the use of simple interest is

not harmless at all, let alone essentially harmless. He says:22 While he saw that compound interest was necessitated by the assumption of

free competition, he seemed to think that using simple interest involved an

admissible simplification and no 'essential alteration'. As we know, this pre-

sumption cannot be sustained.

But I believe that Wicksell knew that the use of simple interest was not

harmless in general, i.e. in case one wants to study growing economies. In WKR,

Wicksell shows that, taking into account compound interest, the end sum of capital

(principal plus interest) will be equal to

This is a revision of equation (1) in chapter II23. This equation allows us to

calculate the same endogenous variables in the stationary state analysis. Both in

WKR and in the major part of the Vorlesungen I, Wicksell explicitly assumes the

stationary state. (It must be admitted that, if t and z both are large, the deviation

between calculating simple compared to compound interest becomes very large

too. This is the reason why Mark Blaug judges this a ‘serious failing in Böhm-

Bawerk’s calculations’.24 But the simplification is not so serious in principle and this

is what Wicksell contended as he judged it as admissible.)

Kurz’s claim, that Wicksell was not really interested in the stationary state, is

untenable if it is taken to describe all phases of Wicksell’s his career. He was inter-

ested in the properties of Böhm-Bawerk’s idealised model with time as the essen-

tial factor, even if it described a stationary state. In the last chapter of the Vorlesun-gen I, on Die Kapitalbildung, the stationary state is finally abandoned. But that is

precisely the passage which marks the next step in Wicksell’s research.

Böhm-Bawerk, in his capital theory, had engaged in many considerations of

positive and negative growth of the capital stock. This required a dynamic view. A

static analysis of cross sections of a growing economy is of course conceivable but

it does not allow for all questions of growth that one might want to take into con-

sideration. Indeed, a static analysis highlights some aspects but covers up others.

For example, a static description exposes no transition mechanisms that connect

one state of the economy with a next state. It also curbs the discovery of the differ-

ence between a change in initial conditions and a change caused by agents’ re-

22 Kurz (2000), p.781, referring to the (English translation of the) quote from the preceding note. 23 WKR, p.98, note 1. See chapter II, p.73. If z is small and t is large, s will of course approximate

(1+z½t)ℓt, as equation (1) indicates. 24 See Blaug (1997), p.496. Blaug rhetorically suggests (though does not literally say) that Wicksell also

judged it a serious failing.

( ) dtz1s t0

t∫ += l

sponses to unstable initial conditions. Perhaps even worse, a static analysis fails to

show the difference between variations in initial conditions and in the very way in

which agents tend to organize their responses to changing initial conditions, for

example given the institutional environment. Many subtleties are suppressed be-

cause these remain, so to say, undiscovered.

2.4 Statics versus dynamics, stationary state versus long run growth

The stationary state: condition for static analysis or limiting case? How do Böhm-Bawerk and Wicksell deal with statics and dynamics and with the

stationary state and growth? We have seen that Kurz stresses both Wicksell’s and

Böhm-Bawerk’s interest in growth rather than in the stationary state. I qualified

Kurz’s claim by noting that, concerning Böhm-Bawerk, the interest theory in PTK

treats the stationary state up to the last chapter, only to turn to considerations of a

growing economy in that chapter, and that the capital theory – largely discon-

nected from the distribution theory – tries to analyse long term growth of capital as

well; and, concerning Wicksell, that his early WKR essentially analyses distribution

in a stationary state, so as to merely clear the path for a study of capital growth in

his later works. In addition, Wicksell praised Böhm-Bawerk (and criticised Walras)

for his crucial insight that positive interest is possible even in a stationary state.

There is a potential confusion over the terms ‘stationary’ and ‘static’. It is

possible in principle to treat a stationary state dynamically, viz. to highlight short

run market events in an economy at long run standstill. Also, it is viable to discuss

growth by a static view if one is interested in the properties of a growing economy

synchronically and not diachronically. Kurz’s project (in his (2000)) is meant to

show that much of the research on Wicksell’s failure to develop a complete distri-

bution theory for the case of modern economies is misguided due to its confusing

the two concepts. This has induced the researchers to look for some missing equa-

tion in Wicksell’s work, which is, he says, not missing at all. He draws on the clari-

fication of (and the warning to observe) the distinction of the concepts already

given by Lionel Robbins (1930): we must recognise not one general class of “static states” and “static laws”,

but two: the classical conception in which the condition of stationariness is the

resultant of the balancing forces tending to change, and the Clarkian in which

the factors of production are stationary by hypothesis, and equilibrium is at-

tained within these conditions. Both rule out inventions and fundamental

changes in nature and human beings. But the one admits the possibility of

variations of labour and capital, the other excludes these by definition. […]

The modern economist […] will recognise in the two constructions we have

been examining, not competing abstractions, but successive stages of exposition.25

The differences are subtle. The Classical economists held a view of the economy as

a system that produces long run equilibrium and, hence, of a stationary state until

25 Robbins (1930), pp.206-7. Italics in the original.

Chapter IV

interventions or spontaneous changes in initial conditions reset it; and they won-

dered about the conditions for such a stationary end state. John Bates Clark looked

upon it as a system that produces long run equilibrium only under conditions of

fixed endowments: he studied the economy under the given axiomatic condition of

a stationary state. Thus, the picture seen by the Classical economists is a limiting

case of Clark’s approach. Kurz, then, notes that Wicksell did not assume a strictly

stationary economy in the sense of the Classicals, but maintained a static point of

view ‘designed to throw some light on the actual, growing economy in terms of a compara-tive static analysis of consecutive states of the economy characterized, inter alia, by differ-ent “quantities of capital” in existence’26. Wicksell did ultimately not aim for describ-

ing the limiting case.

It is important to note that it is not useful, perhaps not even intelligible, to

speak of ‘a static economy’. This is not just because economies are essentially dy-

namic systems, even if in a stationary state, but because it is only the view of such an

economy – captured in a model – which can be static in kind. ‘Static’ refers to the

property of a method, not to a property of a really existing or even hypothetical

economic system.

Idealization of the stationary state In the Lectures on Political Economy, Kurz says, Wicksell gave rise to later misunder-

standings due to his ambiguous treatment of equilibrium. At one time Wicksell

said that (1)

‘augenblicklich begnügen wir uns jedoch mit dem, was man die statische Seite

des wirtschaftlichen Gleichgewichtsproblems genannt hat, d.h. mit den Be-

dingungen der Erhaltung oder periodischen Erneuerung stationärer wirtschaft-licher Verhältnisse.’27

But at another time he referred to (2)

nimmt man als einfachste, grundlegende Hypothese die stationäre Volkswirt-

schaft an, wo das Kapital und die übrigen volkswirtschaftlichen Faktoren an-

nähernd je als eine unveränderliche Summe aufgefaßt werden können.28

The first quotation talks of the conditions for the possibility of a stationary state. The

second quotation involves a stationary economy as the simplest hypothesis and as a

mode of thought. In fact, however, Wicksell looked for a way to treat growing

economies by a comparative static method, although he alluded to the ‘stationary

state’ to refer to that method.

The stationary state analysis assumes stable endowments, reflected by fixed

factor supply. But no such long run standstill economy was like what Böhm-

Bawerk and Wicksell observed. In his WKR, Wicksell introduced the assumption

of a stationary state as an isolative strategy. He studied a hypothetical economy.

26 Kurz (2000), p.777. I endorse this interpretation, but I repeat that I differ in opinion as concerns Wick-

sell’s first phase, the early period in which he wrote WKR. 27 Wicksell (1984 [1913]), p.163. Italics in the original. Kurz quotes the English version; see Wicksell

(1951 [1901]), p.105. 28 WKR, p.ix. Italics in the original. Kurz uses the English (1954) Value, Capital, and Rent (p.22).

This model economy in perpetual stationary state equilibrium, gave him an inter-

esting research object. It allowed him to learn about the sources and level of inter-

est and about the determinants of income distribution, without any form of capital

growth. But it did not take him to his final goal. Kurz’s main point shows that it is

wrong to take Wicksell as trying to present a stationary state analysis ‘stricto sensu’. Indeed, for a theory of capital and distribution in the long run Wicksell dropped

the assumption of a stationary state, changed his strategy, so to say, and started of

with a static economic model of what actually is a long run dynamic situation. Let

me now to express this in terms of the conceptual apparatus we have developed.

From de-idealization to abstraction Wicksell dropped the clause of vertical supply curves so as to further approach

what he thought was the kernel of actual economies. Part of the actual state of af-

fairs is that capital and labour inputs rise with time. The false clause was relaxed;

the external validity of the theory as developed after WKR had to increase. But the

more sophisticated model was not strictly dynamic: it was an instance of compara-

tive statics. Each step of the analysis was studied with a ‘bereits angesammelten Kapi-talvorrates’.29 Apparently, the result of every such step was an abstraction in the

sense that Wicksell now disregarded the mechanics of the process.

I here present the de-idealizational reasoning step of dropping the clause

about vertical supply curves and the abstractive step in the comparative statics as

two distinct phases but in practice they really form one single step: their being one

and the same thing de facto does not render the analytical distinction wrong.

Comparative static analysis is a repeated static analysis. The use of the in-

strument of successive synchronic views however does not imply the assumption

of a world in long run arrest. Each synchronic slice of economic reality from its

diachronic flux is an abstraction in the sense that the dynamics of the system is

ignored. Nevertheless, the aim of it all is not to ignore but to study it. That is why

the mode of analysis is repeated.

Recall that the proposed way to characterise ‘abstraction’ is by its ignoring

details judged irrelevant for the issues at hand, just as when inferring existential

generalizations. Wicksell ignored the dynamics but was nevertheless interested in

(long run) economic development: comparative statics came in the place of real

dynamics. Still, he did not introduce falsity by this mode of analysis. The process

moving the world from one moment in the course of its development to another

only remain unexplained. I can now qualify this mode of analysis as abstraction

ante explicationem: a theory explaining the mechanics was not readily available but

he awaited one. On the other hand, Böhm-Bawerk’s mechanisms explained the

stationary state, not a developing economy. Had he been able, at the time, to ex-

tend his mechanistic narratives, explaining an expanding economy, then perhaps

Wicksell’s efforts would have presented a case of abstraction post explicationem.

29 Wicksell (1984, [1913]), p.219

Chapter IV

3 Keynes and involuntary unemployment.

The case of pure theory.

A neoclassical microeconomic orientation has it that without market rigidities,

labour market adjustments come quickly enough to solve the problem of unem-

ployment, provided joblessness is not voluntary. Involuntary unemployment is the

phenomenon that jobless people would want but do lack a job at a competitive

wage rate. In unison, neoclassic cannot see voluntary unemployment as an unem-

ployment problem at all. Involuntary unemployment clearly cannot exist in a neo-

classical world with perfect markets and complete contracts. This means that the

existing wage rate is too low for an incentive to work more hours in a job. So the

marginal utility of leisure exceeds that of earning a higher income. Unemployed

are not willing to take a job or start a business of his own.

This, then, is the typical neoclassical explanation of voluntary unemployment and it explains why it is often assumed that Keynes, not the neoclassical theorists

themselves, assumed labour market rigidities to develop a disequilibrium theory.

Neoclassical economics is equilibrium economics. However we shall see shortly

that, in the view of Keynes, permanent unemployment can only be explained in the

neoclassical theory with reference to the existence of labour market rigidities.

Many textbooks give a loose presentation of Keynesian economics, habitually

claiming that the assumption of rigidities in the labour market is crucial to explain

Keynesian disequilibria. I believe that this assumption is wrong-headed. According

to the Keynesian concept, if unemployment is involuntary, it is not the conse-

quence of a rigid suboptimal wage level.

Moreover, this is precisely the reason why neoclassical economists did not

perceive the possibility of involuntary unemployment. Later theoretical develop-

ments explained the dynamics of incomplete and complete contracts.30 A contract

is incomplete if at least one of the stipulators aren’t certain about the level of utility

of his part of the deal. Thus, employers run the risk of a lower than expected qual-

ity of work done. A long lasting labour market relationship is efficient in the sense

that the demander can trust the seller to deliver a desired level of quality in return

for which the demander offers a rent on top of the payment. It is productive in the

sense that under these circumstances more deals will be struck. The resulting wage

level would be super-optimal for the worker in the model with incomplete con-

tracts. It is this rent, which the unemployment foregoes. Hence, there is an inter-

pretation of the concept of ‘involuntary unemployment’ as being unemployed and

lacking the opportunity of displaying one’s trustworthiness, and as missing the

rent. Involuntary unemployment is then suboptimal for the supplier.

But unemployment can be involuntary in other, macro-economic contexts.

Apart from frictional unemployment and voluntary unemployment, Keynes wrote:

‘[t]he classical postulates do not admit of the possibility of the third category, which I shall

30 See for instance Brown, M., Falk, A., and Fehr, E. (2004).

define below as ‘involuntary’ unemployment’ 31. The postulates are paraphrased as: [T]he wage of an employed person is equal to the value which would be lost if

employment were to be reduced by one unit […],; subject, however, to the quali-fication that the equality may be disturbed, in accordance with certain principles,

if competition and markets are imperfect.

[T]he real wage of an employed person is that which is just sufficient […] to

induce the volume of labour actually employed to be forthcoming; subject to the qualification that the equality for each individual unit of labour may be disturbed

by combination between employable units analogous to the imperfections of

competition which qualify the first postulate.

3.1 The neoclassical model

Let us first briefly expand on the neoclassical economic model32 to explain

voluntary unemployment. Unemployment is defined as a labour supply surplus.

See figures 1 and 2 for a schematic representation of the model.

figure IV.1 The labour market with voluntary unemployment

In the figures, ℓrig is rigid wage level, the asterisk (*) indicates full employment

levels, X is real output, Ad and As are demand and supply of labour, respectively.

The starting point of neoclassical analysis is that the employers determine em-

ployment given a rigid wage level that is beyond their control. The idea is that

31 Keynes, J. M. (1973), pp.5-6. Note that Keynes’ report on these postulates already strongly suggests

that the classical economists looked in market failures for disturbances of the laws they proposed. 32 Unlike in the previous chapter, in this one I conform again to the economists’ use of the term ‘model’

in the more loose way as something like ‘formal representation’.

ℓrig

A* Arig

U’L/U’A

labour

Chapter IV

unemployment must be due to the problem that the employer’s marginal labour

cost (i.e. the extra wage paid) when an extra worker is set to be equal to the mar-

ginal pecuniary revenue. Keynes called this ‘the first postulate of the classical the-

ory’ (respectively so listed above). Insofar as employees do not want to be unem-

ployed (no involuntary unemployment), and in the face of being jobless, the rela-

tion of the marginal utility of leisure time, L, and of extra income – U’L/U’A – will

be set equal to wage ℓ again. In other words, workers will decide to give up leisure

time, even at lower wage levels. This is what Keynes called ‘the second postulate of

the neoclassical theory’.

figure IV.2 (Lack of) labour market adjustment, classical analysis

If workers do not wish to give up more leisure, a labour supply surplus per-

sists due to voluntary unemployment. (In the case of monopolistic market power

of ‘closed shop’ trade unions, individual workers may be unemployed involuntar-

ily, but as members of the union they are seen as voluntarily unemployed.) The

wage level is instrument variable.

Wage level ℓ is to fall to the point where marginal revenue of labour equals

marginal disutility or pain of loosing an extra hour of leisure. So in this model,

persistent unemployment indeed is due to market imperfections. For example,

wage rigidity can manifest itself in the form of trade union power that creates

ℓrig > ℓ*

As* > Ad(ℓ)

X(Ad(ℓrig)) < X*

ℓrig = c (wage rigidity)

U’L/U’A = ℓ*

∆ℓ < 0 (wage flexibility)

instrument variable

As* = Ad(ℓ)*

U’L/U’A < ℓrig

(voluntary unemployment)

As* > Ad(ℓ)

(persistent unemployment)

some form of oligopoly in the labour market. The solution to this problem, from

the point of view of policy makers, is to be found where the point of application of

policy tools lies: in making the market more flexible.

3.2 The Keynesian model

In Keynes’ view, the problem of unemployment is as in figure 3.

figure IV.3 Involuntary unemployment, Keynesian analysis

This view is, in a sense, diametrically opposed to this analysis and so is the

policy recommendation that one could derive from Keynesian economics. It is im-

portant to repeat that Keynes ascribed explanations in terms of market imperfec-

tions and wage rigidities to the neoclassical economists. Hence, it is unlikely that

Keynesianism can be associated, ever so often, with reference to wage rigidity. The

key mistake of neoclassical economic thought was, as Keynes believed, that it gave

domestic income a role exogenous to the microeconomic model and determined by

given factor endowments, like As. It ignored the macroeconomic feedback of wage

changes onto aggregate effective demand, Xed, and, hence, on domestic income.

The starting point of the Keynesian analysis lies at the highest level of ag-

gregation. Aggregate effective demand for commodities, Xed, lags behind aggre-

gated supply, X*. The typical economists’ meaning attached to this phenomenon is

‘sub-optimal usage of domestic productive capacity’; in Wicksell’s terms it would

Xed < X*

As* > A(Xed)

ℓ(A(Xed) > ℓ*

∆ℓ < 0

(wage flexibility)

∆Xed > 0 instrument variable

As* > A(Xed)

(persistent involuntary

unemployment)

Chapter IV

be failing ‘product exhaustion’. The most flexible factor is labour as capital and

land are fixed for some time. Labour functions as a buffer. Also in this model lei-

sure L has low utility, which is consistent with involuntary unemployment (there

is a low disutility of working). It must be added that this is false, strictly spoken,

for Keynes objected to the second postulate of the classical theory on several

grounds, one of which was expressed by the argument of a reductio ad absurdum33.

More fundamentally, Keynes did not believe that real wages are directly deter-

mined by the wage bargain. But in order to ease the exposition of the contrast be-

tween the neoclassical analysis and the Keynesian analysis of unemployment I

shall assume – as Keynes sometimes did – that the second postulate holds good,

only to investigate what this would amount to. The argument is that it amounts to

even more unemployment.

With a cheaper wage bargain, entrepreneurs are not (or not sufficiently) in-

duced to raise their demand for (now cheaper) labour, as sales tend to drop even

further: the downward pressure on money wages causes a further erosion of ag-

gregate effective demand. The output gap depresses nominal prices, which in turn

determine real wages. Demand is an instrument variable. Keynes opened his The General Theory thus:

Most treatises on the theory of value and production are primarily concerned

with the distribution of a given volume of employed resources34.

[…] whilst Ricardo expressly disclaimed any attempt to deal with the amount

of the national dividend as a whole, Prof. Pigou, in a book which is specifi-

cally directed to the problem of the national dividend, maintains that the

same theory holds good when there is some involuntary unemployment as in

the case of full employment.35

The very point Keynes made is that it doesn’t. It must be read as a proposal to en-

dogenise aggregate income.

3.3 Idealization and abstraction in the analyses

I repeat that the classical economists had to explain disturbances by reference to

market failures. The very assumption that markets, ideally, are without failures

buttresses the various general equilibrium theories. Such theories have a normative

bearing because they point to the relevant instrument variable: ‘if you want to

solve the economic problem of unemployment, take away market failures’.

We can now say that the General Theory explained the possibility of a labour

supply surplus. In contrast, neoclassical explanation is about the same explanan-

33 Ibidem, p.13. The argument is that it would be absurd to conclude that labour supply decreases in case

real wages fall as caused by rising prices. If the formula were true, a rise in the cost of living would

cause workers to withdraw their labour offer. In fact, the argument disputes the realisticness of the

assumptions. 34 Keynes (1973), p.1. 35 Ibidem, p.5, note 1.

dum, but by reference to strict assumptions. A condition in the clause (i.e. flexible

prices) was not fulfilled. This is also how they could stick to their concept of volun-

tary unemployment. An unemployed person is someone who refuses to work for a

competitive real wage or who organises his bargaining power in the form of an

monopoly, i.e. by trade union representation. Such behaviour results in surpluses.

Assuming that this is not the case renders an idealization. The list of possible market

failures is provided by the theory and it is a finite list. For instance, the ceteris ab-

sentibus clause takes care of it. Keynes’s theory helps deduce the theorem of invol-

untary unemployment, which serves as explanans of labour market disequilibria.

This theory in turn abstracts from labour market failures.

A misunderstanding looms that I should prevent from occurring here. I did

certainly not mean to maintain that abstraction be somehow related to endogenising

the factors, which a theory abstracts from and that idealization be associated with

keeping variables exogenous to the economic model. Not at all. Although Keynes

endogenised aggregate income in order to have a theory which – he supposed –

could explain persistent and demand-induced unemployment, endogenising itself

is not the issue in abstraction. He developed a theory about a type of (non-

frictional) unemployment, which does not evolve out of any labour market rigidi-

ties. This enabled him to ignore the question of whether such rigidities were actu-

ally present in various sections of the labour market. The General Theory does not

deny that there also may exist unemployment induced by inflexible markets. Mar-

ket failures had already been studied by contemporary economists in Cambridge,

such as Joan Robinson. But for the explanation of the type of unemployment that

Keynes studied it is simply irrelevant whether other types than demand induced

involuntary unemployment actually occurred as well.

Let me summarize the roles of explanans and explanandum, seen from the

viewpoint of Keynesian explanations of unemployment, in the scheme below.

Economic

phenomenon In Keynes In neoclassical theory

Aggregate income Explained (endogenised) Abstraction ante explicationem

Labour market

imperfections Abstraction post explicationem Idealization

To explain unemployment Keynesian theory endogenised aggregate income.

It was a key concept in his macroeconomic theory. The neoclassical economists, in

turn, abstracted from aggregate income: this is abstraction ante explicationem, be-

cause there was not yet a theory which could take care of the endogeneity of ag-

gregate income. However, the opening pages of the General Theory acclaim that

income is not to be ignored. Keynes, in turn abstracted, from something else: mar-

Chapter IV

ket imperfections. He could do so without selling their actuality short. He could

also admit that imperfections do cause forms of unemployment (that we now call

‘structural’). But these particular forms of unemployment had already been stud-

ied in other research contexts. This is why I classify these as subject to abstraction

post explicationem. But, quite apart from this, labour market imperfections did not

play any role in Keynesian explananda or explanantia. It was irrelevant for Keynes.

4 The Dutch market for health, care, and cure.

The third case is about the ubiquitous problem of inefficiency in health care. In the

Netherlands, several political and administrative committees have chosen an eco-

nomic perspective in the formulation of policy proposals for solutions. The com-

mon denominator of the successive recommendations was the plea for more mar-

ket incentives in health care.36 The hopes were that, in this way, surpluses and

shortages together with the rise in costs would disappear. The dominant idea be-

hind this is that the more a market looks like the ideal of a free, complete and perfect market, the more efficient this market will be.37 Particularly important is the desire

to decrease the grip of the administration on the market. The ideal of a free market

is centre stage in a liberal policy orientation.

A market is supposed to be efficient if it generates Pareto-optimality. The

operative assumption in advising to allow for market incentives is not that ideal markets are efficient – they are by definition – but that any real market is efficient

to the extent that it looks like the ideal. Let me put this with some more precision.

Often, in some markets at least, disequilibria vary directly as non-market

conform government intervention intensifies. One can think of the markets for

housing or agricultural produce. Furthermore, incomplete markets tend to create

monopolies. Fewer bakers in an area goes with higher prices for bread, and not just

in theory. I think we may conclude that the ideal type of freedom, completeness,

and perfection is a fair compass reading for the orientation of microeconomic pol-

icy. Still, more market conform incentives have produced nothing of the expected

results in the Netherlands so far.38 This has been caused by the use of some quite

different assumptions than merely those of freedom, completeness and perfection

36 Mostly, the idea was to combine market elements with solidarity. Advisers recommended a role for

competition between medical staff and between health insurance companies and for freedom of

choice for the insured. See Ministerie van WVC (1988) and (1990). 37 Not all textbooks use these terms in the same way. I take ‘free’ to mean freedom from government

interference, ‘complete’ is any market on which no one can affect prices individually, and ‘perfec-

tion’ is absence of friction. 38 At the time of writing this, the organisation of Dutch health care is in a major transition towards a

system with a greater role for market incentives, the effects of which cannot yet be clearly seen (at

least not by me). Some of the descriptions in this section may have lost their relevance to the situa-

tion of today.

in the sense I have used these in discussing Austrian economic thought. The point

with the market for health care is not that it suffers from imperfections (although it

does) but that there are also some other properties that do not necessarily amount

to imperfections. They are to be qualified as ‘idiosyncrasies’, rather than as imper-

fections. Therefore, they remain rather implicit in general economics textbooks.

Economists explain the problem with regard to this specific market explicitly with

reference to these particularities.

4.1 Idiosyncrasies of the health care market

I shall list three idiosyncrasies of health care systems that help explain their fate.

For my aim, I shall abstract from some properties, like the institutional fact that

governments use health care as a means to redistribute disposable income. The

heavy government intervention in this sector is a consequence of the obligations

article 22 of the Dutch constitution imposes on the administrative authorities. The

institutional environment causes the market of health to deviate strongly from a

‘normal’ market, let alone from some ideal of a market. So I do not go into every

aspect of the complex institutional environment.

Let me discuss the following three special features of the market. The first distinctiveness has to do with one aspect of the institutional arrangement. The

second concerns the specific perspective demanders engage in when they experi-

ence utility. The third involves suppliers behaviour.

As to the first, a necessary condition for optimization is that the patient, the

decider, and the payer are joined in one and the same individual. But in health care

these roles are distributed over three different economic subjects exactly so. This is

easy to see. The patient uses medicine, the physician prescribes it, and the insur-

ance company pays for it. This is of course caused by the inevitable role of insur-

ance companies. Compare this with a normal market. If you buy your pint of lager,

if it is you who has decided to do so, you pay for it and you are drinking it as well.

This property, of what I should like to call ‘the split in roles’, can be judged a real

imperfection. Health care is not the only market with such a property. Dutch par-

ents pay for the schoolbooks their children will use, whereas the choice of these

books are decided upon by the teachers.39 Then again, we can see the publishing

houses making supernormal profits. Of course, some other institutional elements

add to this sub-optimal outcome. In the Netherlands, for instance, all books have

been subject to legal resale price maintenance for a long time after WW2.

The second particularity is the strong tendency towards demand-inducing

supply: once it is there, people want it. This is a form of preference drift seen in any

market, but it takes a prominent shape in health care. Viagra presents us with a

telling example. This is often not understood as a means to relieve a medical prob-

lem, but as an aphrodisiac. Other examples of medical treatment abound. The

39 See also Rol (1998).

Chapter IV

point here is not that marketing techniques in general try to raise a certain prefer-

ence drift. But demanders of care have a very special view of the meaning of

therapies and health. A related aspect is that satisfaction given by health care

products is a derived utility. Patients do not feel to purchase a cure of a disease,

but indeed health. There seems to be no presumption that they themselves must

take care of their health. If the user were also the one who pays, this awareness

would arise quickly. But now it does not and demand behaviour goes off the rails.

The final distinctiveness concerns satisficing supply behaviour. Such behav-

iour has been observed and measured in the behaviour of New York taxi drivers.40

Medical doctors in the Netherlands have always been paid by the number of

treatments they do. Though they are entrepreneurs within the hospital, they do not

seem to maximize their incomes. In debates on the causes of the emergence of wait-

ing lists, some economists have mentioned satisficing as one of them. Also this is

not exclusively a property of the health care market.

4.2 Health care and ‘the social model’

So what does all this mean for this market? Perhaps health care is simply no exam-

ple of anything like a market. But this objection is inconsistent with the proposal to

introduce ‘more market’ in the industry. To exaggerate it perhaps, ‘more market’ is

to serve a policy maker’s desire to make a market distribute health care as effi-

ciently as wide screen television sets. Thus, a policy orientation towards economic

liberty praises withdrawal of government intervention (other than that of antitrust

legislation) in the market. However, as explained, at least three properties of the

trade in care and health have been identified by economists as impeding a quick

settlement of problems of government intervention. First, it presupposes that de-

manders and users decide on the basis of the same considerations as the financier

does. Second, utility of health care would derive from what the product actually

does, not from what it is unreasonably trusted it does. Third, medical staff would

be eager to maximize both income and effort to treat as many patients as possible.

The question I now want to answer is this. Where in the scheme of epistemic op-

erations – of abstraction and idealization as explicated above – do we have to situ-

ate the complex of assumptions under review?

I have claimed in chapter III that (1) this phenomenon could be explicitly de-

fused by a clause. Further, I have said that (2) some phenomenon, which is seen as

potentially disturbing, could also simply be ignored. The first method is called

‘idealization’. The second strategy has been marked ‘abstraction’.

40 See Camarer et al (1997). One may of course wonder whether this is a case of satisficing in Simon’s

sense (see Simon (1957)). It can as well be conceived as a particular form of their (the taxi drivers’

and medical doctors’) utility curves. Whatever the ways in which we can classify this behaviour, the

effect is the treatment of fewer patients than what seems to be optimal from the point of view of the

demand side of the market. This is a rather old concern – see for instance Anderson (1979) – but still

in debate – see for instance Marelich et. al. (1998).

In what follows, I shall consider the phenomenon that effective demand for

care appears to have no limit and assume, for the sake of argument, that this is

exclusively caused by the split in roles of patient, physician, and financier. Most

prominently, the one who pays for the treatment lacks the means to curb demand.

Ignoring it is simple. But an idealizational clause would say that, given an integra-

tion of the three roles in one person, the market would benefit from decreased gov-

ernment interventions. Now, if theory has to generate proposals for policy, this

clause has to be coped with somehow.

If the phenomenon in question in fact makes part of a domain of desired

policies, abstract theory (let me denote it again AA) will only serve as a useful

source of policy recommendations if there is (or will soon be) a complementary

theory (X), which teaches us something about the roles of patient, physician, and

insurance company. So assume that theory AA tells us nothing about this aspect of

the real market. As government policies are embedded in a complex environment,

such a complementary theory will not rarely originate in other disciplines, like

sociology, business administration, the study of law, and psychology.41 The table

below summarises the options. I shall consider them in the respective order.

WAYS TO DEAL WITH A CLOSURE REGARDING A POLICY RELEVANT PHENOMENON

Explanans Explanandum Epistemic operation

1 an idealising clause to a theory DD The phenomenon is

a disturbance idealization

2 absent: theory AA does not refer to

the phenomenon

The phenomenon is

out of sight abstraction

3 Hypothesis X The phenomenon is

an instantiation explanation

A theory AA, which is essentially about ideal type markets, does not explain

the cited phenomenon.42 One can think of a theory idealising from these effects, that

is, to introduce an idealizational clause (CL) into the idealised perfect markets the-

ory (denoted DD in chapter III). The clause explains how the disturbing phenome-

non, call it a demand surplus, occurs despite the theory’s predictions. In other

41 The morale of an article by the Dutch economist Folmer (2000), in a journal on economics and policy,

is that social sub-disciplines have to be integrated much more than actually is the case. (The reveal-

ing title of it is ‘Why economists blunder so often’.) Folmer calls the desired integrated complex of

theories ‘the social model’. I must say that I am not sure whether such an integration of sub-

disciplines is too much to ask. There is a certain disinterest for, or even derision of, other social sci-

ences in mainstream economics. It seems to me that mere scientific cooperation would do much of

the job. 42 The term ‘ideal type’ bears reminiscence to Max Weber, and it is to be loosely understood in this way.

Hence, it should not be associated with idealization.

Chapter IV

words, CL is explanans of the phenomenon and the perfect markets theory merely

refers to – is true of – (a particular subset of) the set of models of CL. This need not

be problematic for the idealised theory as such. But for policy relevance, this prac-

tice boils down to treating these characteristic effects as disturbances. One can en-

tertain the belief that CL – counterfactually saying that the roles of patient, physi-

cian, and financier are integrated in one and the same person – does not hold back

policy relevance. The belief that it does not entails the assumption that the ideal-

ised state of affairs under description sufficiently approximates aspects of the real

world that form the domain of policy. If this confidence is however not justified,

the idealization must be understood as delivering externally invalid policy rec-

ommendations.

In case of abstractive operations, on the other hand, the economist assumes

that there is or can be an other explanans than the specified clause. This other ex-

planans deals with the problem of the split in roles. The theories underlying this

other explanatory hypothesis may have been developed before, or we can expect

these theories to be available at some future date. In the fist case we have abstrac-

tion post explicationem, in the latter we have abstraction ante explicationem. A theory

(labelled AA in the table) that abstracts from the market peculiarity in question is

obviously oriented at other aspects of the market than this very peculiarity. Once

the problem of the split in roles turns out to be relevant for policy, one has to draw

from an other (complementary) theory that does take this idiosyncrasy of the

health care market as its explanandum. The table refers to such an alternative

complementary theory in terms of ‘hypothesis X’.

5 The economics of environmental governance

In 2006, under the cynical slogan ‘Buy organic, destroy the rainforest’, the weekly

magazine The Economist contended that the production and consumption of or-

ganic food is bad for the environment.43. Generally, the argument to buy organic

refers to conventional intensive farming, which uses chemical inputs. ‘But’, the

editors say, ‘it all depends what you define as “environmentally friendly” ‘. Ever

since humans farm, deforestation has been the result. Although chemicals are bad,

an extensive way of farming which satisfies global nutrition needs would require

several times the currently cultivated land. More rainforest would have to be given

up. Chemical fertilizer has contributed to multiplying agricultural yields with very

little increase of land under cultivation. Hence ‘If you think you can make the

planet better by clever shopping, think again. You might make it worse’.

Something similar as with organic food is argued as regards fairtrade food.

Farmers in Third World countries suffer from the exposure to volatile world mar-

43 ‘Good food?’ (2006)

ket prices. Their income marginalises as prices drop. Fairtrade food is sold at a

higher than market price, subsidising the farmers. ‘But’, it is claimed, ‘prices of

agricultural commodities are low because of overproduction.’ Indeed, low prices

are supposed to be signals of supply surpluses and the appropriate reaction to this

should be a cut in production. The fairtrade subsidy deprives the price of its func-

tion as a messenger of scarcity relations. What poor farmers should do is not pro-

ducing more of the same but diversify. To bolster the price is to depress it further,

as output rises in the face of surplus. Fairtrade food creates its own strain. (In addi-

tion, but not of importance of the argument in this section, ‘the system gives rich

consumers an inflated impression of their largesse’.)44

5.1 The adverse effects of ethical food in your trolley

Clearly, the issues, so provocatively presented in this article, are relevant for public

or private policy. In the light of the categorisation I presented in chapter III, the

question arises whether the inevitable assumptions underlying the economic ar-

gumentation against ‘Good Food’ generate isolations of the horizontal or of the

vertical tailoring, or of both. This is an important question. As I argued, abstractive

isolations need not render policy irrelevance. Abstraction often helps focus on cru-

cial relationships of factors that may otherwise remain out of sight due to the

wealth of detail of the issue under study. But then again, abstraction may also re-

sult in propositions too weak to have a bite. The trick, so to say, is to abstract inter-

esting economic relations between variables without losing grip on the handles for

intervention. On the other hand, idealizational isolations result in propositions true

only of hypothetical worlds. For any kind of involvement of policy, the idealiza-

tional clause must be dropped if reality does not approximate the nomological

machinery of the economic theory containing the clause.

The Economist recommends to abstain from ethical shopping. This will ad-

mittedly not be sufficient to create a finer world; ending protective agricultural

policy comes into the bargain too. In 2007, out of the European Union budget €55

billion will be spent on the Common Agricultural Policy (CAP). In real terms the

CAP budget has increased from € 330 billion over the last seven-year budget to

€371 billion over 2007-13. The costs of CAP will also account for a bigger share of

the new EU budget than it currently does (from 42.6% in 2005 to 43% on average

over the 2007-13 total).45 After a gradual decline of this share over the eighties of

the previous century, it is on the rise again. The French government has already

made it clear that they will still veto any cuts in the CAP. All this tax payers money

(approximately 0.4% of the total Gross National Product of 27 European countries)

results in artificial pricing, making it impossible for Third World farmers to com-

pete. Meanwhile, agricultural prices are set too high, so that Europeans, after con-

44 Another category the article treats is local food, produced closely to the consumer. 45 Open Europe bulletin (2005).

Chapter IV

tributing to an unfair system as taxpayers, pay too much for their food in their role

as consumers.

I shall not go into the obvious need for abolishing unfair trade. The article

concerns private action (‘voting with your trolley’). Given that the ends are worth

to be pursued, are the recommendations to consumers – to be liberal rather than

ethical – a good idea? There are three tacit assumptions at stake. Two deal with

organic food, one with fairtrade food.

The first assumption is that chemical fertilizer has less of a detrimental effect

on the ecosystem than deforestation. But there is no argument sustaining this

proposition. The absence of such an argument is in line with the assumptions being

tacit. Only if deforestation is worse does intensive farming have the biological ad-

vantage over extensive farming. Whether this is the case depends on the ways in

which extensive farming is put through. Chemical fertilizers have long run effects

that we only are beginning to understand, inclusive of soil exhaustion. A compari-

son with the harm done by deforestation is much needed.

The second assumption is connected to the first. There is not just the choice be-

tween using chemical fertilizers and manure. One must also select between mono-

culture and crop biodiversity on a pasture. Recent experiments in Wageningen, the

Netherlands, suggest that growing several types of crops on one piece of land

keeps out diseases and increases yields. The assumption, then, is that agriculture is

one-dimensional. You either farm and destroy the forest, or you don’t and keep the

forest intact. In this reasoning, there is no room for speculations about simultane-

ous farming and foresting.

The third assumption plays a role in the argument against fairtrade. Low

prices reflect decreasing scarcity, so subsidies finance overproduction. It is implic-

itly assumed that these prices are competitive-market prices. However, the criti-

cism against protective agricultural policy by the big trading blocks is that they

cause artificial pricing, so it is hard to see how the price level, which harms the

farmers, must be interpreted as a signal. But I am interested in a further assump-

tion. The article talks of overproduction while famine is ubiquitous. Apparently,

effective demand is taken to be the same as ‘need’. It is a remarkable plea that pro-

duction should be cut while people are hungry. The point is not overproduction

but access to markets. Moreover, the argument does not take into account the ex-

ternal effects of farming, such as jobs, infrastructure, and an economic environment

that drums up a little hope to the poor. Some of these external effects can be stud-

ied with economic theories, but especially the latter, hope, is a psychological cate-

gory economists have typically little to say about.

5.2 Idealizational assumptions and their policy implication

The assumptions seem to abstract from the details that I mentioned in the above

discussion. They focus on particular properties of a market mechanism, which

seemingly are taken to be essential for understanding what ethical shopping brings

about. No assertions seem to be expressed about the positive externalities of farm-

ing, or about known long run effects of chemical fertilizers, so also no false asser-

tions. A theoretical hypothesis about the alleged mechanism by which ethical

shopping causes harm rather than fortune is an explanans, the harm itself the ex-

planandum (the description of it thought to be true). A defender of organic and

fairtrade food would then have to propose a competing theoretical analysis.

I believe, however, that the editors idealise over the subject matter that I dis-

cuss here. Idealization, I proposed, is an isolation by which one uses a clause with

a finite set of false propositions. Much – though not as much as we should like – is

known about the effects of the prolonged use of chemicals. We can also be sure

that biodiversity on the pasture helps protect crops in better ways than pesticides

do. The list of phenomena about which one does better not to speak untruthfully is

already there. Assuming that the ecological harm of chemicals is less of a problem

than deforestation is risky, because we lack a decent yardstick. Assuming that any

farming necessarily damages the forest is plainly false. De-idealization may well

induce the theorist to favour ethical shopping.

But then again it may not. It is possible, at least in principle, that the advice

of the article is right. After de-idealization, for example, one may conclude that,

given everything we know, the amount of harm done with extensive farming is

still more severe than using chemicals. But under the force of the idealizational

clause we shall never know this. We must assume true claims about the other cir-

cumstances, and only then theorise about them. Idealization summons irrelevance.

V Abstraction, vagueness, and social kinds

1 Introduction

Abstraction, like idealization, does not preclude truth. But it allows for truth in a

way quite different from idealization. Full concretisation, meanwhile, is not nec-

essary for policy relevant science. The point of policy irrelevance is not abstract-

ing too much, but abstracting without explaining.

Saying that there are good and bad abstractions in science would for

Böhm-Bawerk be like saying that some explanations are true and some others are

not. In this respect he clearly is a representative of Austrian theory. As we have

seen, and as stressed again in appendix 3, Uskali Mäki believes that other econo-

mists besides Austrians were looking for the essence of the market. Transaction

costs, or market dynamics were identified as being essences. I expressed the sup-

position, in the intermezzo, that Böhm-Bawerk’s hunch was not some personal

aberration, but that it reflects a rather ubiquitous feeling in science. I have the

intuition that, no matter the opinions sometimes explicitly ventured, scientists are

realists of a strong variety. Although the world is often startling and inaccessible

due to its notorious flux or its unsuspected complexity, for a single mind there

seems to be just one best way to conceptualise it. From the perspective of one

scientist, instrumentalism seems to be a luxury product, avowed at best after sat-

isfactory theories have come to be.

So how to abstract the right concept? There is no rule to guide scientific

creativity and I shall propose none such. Instead, I want to investigate to what

extent the realist pretension indicated above can be made tenable. This is not an

easy task. A host of authors have made a strong case against essentialist episte-

mologies. I shall discuss one at some length in section 2. But it seems a sterile sort

of philosophising to dismiss essentialism on purely logical grounds when even

economists can be said to have essentialist inclinations. Section 3 addresses an

interesting endeavour to resuscitate essentialism of a Putnamian flavour, which

however sees such scope only for natural kinds. Section 4 deals with an even

more interesting attempt to weaken Kripke’s philosophy to the point that it ac-

commodates even social kinds. It will be claimed that it is not the essences them-

selves that are discovered but their vagueness. Section 5 stipulates the sort of

social kinds we may think are the sought-after things in economics.

Chapter V

2 Is essentialism tenable?

Many philosophers have rejected essentialism. To propose a form of essentialism

in social science, i.e. to propose that social scientists are really trying to find the

holy grail of a uniquely true conceptualisation of social reality may look like

overshooting the aim, expressed in the introduction, of finding the logical limit to

which an essentialist philosophy of science in general can be stretched. Hence,

the announcement that I shall investigate the tenability of social kinds seems to be

startling. Nevertheless, this is what I shall do.

I start with the anti-essentialist case presented by Hans Radder as he pre-

sents a sophisticated account of abstraction; and this is the key term that brings

me to pose questions concerning essentialism in science. In consequence, quoting

Hans Radder’s work neatly suits my efforts. His case helps me to explain mine.

2.1 The theory of extensible concepts: essentialism is not tenable

Hans Radder draws his conclusions from the research he did into the epistemol-

ogy of experimentation. The central claim he defends is that concepts have two

faces. They ‘not only structure the world but also abstract from it’1. In both orien-

tations, as structuring elements creating a phenomenal world and as abstracting

from a given world, concepts are ‘nonlocals’. The nonlocal aspect of an abstract

concept, as Radder sees it, seems to come close to what an essentialist would

point to speaking of universals. But Radder rejects essentialism. So let us take a

closer look at the arguments he develops for this two sided orientation of con-

cepts and investigate what the nonlocal aspect of abstraction is.

The typically Kantian understanding that it is impossible to know the

world in the absence of structuring concepts is well illustrated by an experiment

that Herman Koningsveld2 did with his students. Koningsveld presented his

students with a number of written signs with no apparent meaning and the task

was to arrange them into three groups according to criteria unknown to his sub-

jects, but known to himself. He knew the relevant characteristics for the categori-

sation: that some signs were only curved, others only straight, and that a third

group was composed of both curved and straight elements. The students took a

long time before they managed to separate the set of symbols into these three

categories. They had not been told the meaning of the structuring concepts in a

clear language (‘curved’, ‘straight’, ‘both’), but in a strange con-language (‘urve,

‘raig’, ‘thob’). The subjects of the experiment could only find out about the mean-

ing of these terms by the teacher who would affirm or reject the categorisation

chosen in the course of the experiment. For them, it was not possible to carve up

the world in the right way, so to say, without understanding the meaning of

1 Radder (2006), p.99. 2 Koningsveld (1973), pp.7-10.

these unfamiliar concepts. The difference between the curved and straight parts

of the symbols did not strike them as relevant if they tried to look for other classi-

fications, like ‘mathematical symbols’, ‘Arabic letters’, and ‘Latin letters’. But it

was possible to acquire the understanding by experimenting with the symbols

and drawing conclusions from the comments of the teacher. More generally, it is

impossible to observe and interpret the world without a prior conceptual struc-

turing of the groups of things to be observed, but one can gradually acquire such

an understanding by having to cope with the pressing circumstances of nature

and culture that one is in. And not everyone will end up having carved the world

by the same conceptualisation.

Starting from Koningsveld’s experiment, Radder conducts a thought ex-

periment. Suppose the experiment would be repeated with blind students. They

would have to be given these urve, raig and thob symbols not on a piece of paper

but in tangible form, made from soft and hard materials. They could learn the

relevant concept in the same way as Koningsveld’s students. If this new experi-

ment would actually take place, the concrete details of the experimental set-up

would be quite different, but the result of the observation process would be the

same: the intended categorisation of the signs. Radder now draws two important

conclusions. The first is that although a new observation process, leading to the

same result, will differ as regards many concrete details of the conditions for the

materialisation of it, the resulting conceptualisation is meaningful in abstraction

from these conditions. The second is that it is a contingent matter whether the

observation result can and will be materialised in the future again.

These two conclusions are crucial to his epistemology. Radder formulates

his first conclusion as follows:

The sensible attempt to enlarge the scope of a concept by applying it to a

materially different domain, including its novel relevance and irrelevance

conditions, involves abstracting from the original realization context.3

He goes on to define the notion ‘extensibility of concepts’, on the same page:

I call a concept extensible if it has been successfully applied to a certain do-

main and if it might be used in one or more new domains.

And he moves straight to his second conclusion:

The notion of an extensible concept is a modal notion […]. Both practically

and philosophically speaking, the realizability of observational processes

cannot be taken for granted. […] the notion of possible extension refers to de

facto extensibility in the material and social world.

The first conclusion has to do with the fact that concepts, forming the re-

sult of observation processes, are abstract. Hence its importance for my distinc-

tion between abstraction and idealization. The second conclusion is of conse-

quence for its import pertaining to essentialism. The extensibility of concepts

toward the results of new observation processes is not something fixed. Essential-

ism presumes that the extension of a concept is fixed by natural kinds, so that it

3 Radder (2006), p.103. My emphasis.

Chapter V

can be established, in principle (not necessarily also in practice), whether the

concept will be applied to another observation process. The belief that kinds exist

in nature seems to be the natural ally of the Causal Theory of Meaning.4 Radder

also rejects the Causal Theory of Meaning.

Radder then investigates the precise meaning of the term ‘abstraction’.

From everyday (English) language he retrieves three possible meanings. One is

‘leaving out’, one is ‘setting apart’, and one is ‘summarising’. First, ‘abstraction

means that we leave out from the conceptual interpretation of the original proc-

ess everything but its result […] with a view to a potential extension of the con-

cept through realising a novel observational process’5. Secondly, and similarly,

‘the notion of abstraction entails that the (conceptual) result of an observational

process is set apart, or separated, from the original process’. These two meanings

fit well into Radder’s comprehension of how concepts are abstracted and ex-

tended toward new observational situations. But the third notion, that of summa-

rising, is problematic for its implication that two particular situations be observed

nonconceptually, so as to abstract what is common and general about it. This is

impossible. There is no such thing as an nonconceptual observation, because

concepts are not only abstracted from the world, they also structure the world.

Hence the title of Radder’s (2006) book The world observed / The world conceived.

Radder brackets an understanding of the concept of abstraction as summa-

rising together with the claim that the notion of an nonconceptual interpretation

of the world is ‘often taken to give us the essence of the kind’6. This is of course

not a necessary association, for one can at the same time believe, without incon-

sistency, that (1) the world is readily given, to be interpreted in a nonconceptual

manner and that (2) whatever is given cannot be a kind. Such a view would

probably stress the informational content of the wealth of concrete detail. But

true, in the inverse case, if one holds for example an Aristotelian epistemological

view, then one must also believe that abstraction as summarizing the relevant

and common aspect of different observation processes does not suffer from the

problem of the prior conceptual structuring labour that our concepts do for us.

By contraposition, this makes a rejection of both the idea of nonconceptual obser-

vation and of the explanatory value of kinds plausible.

Radder disapprovingly quotes Nancy Cartwright’s (1989) Nature’s Capaci-

ties and Their Measurement to mention one author who employs the Aristotelian

notion of abstraction. This is his quote from her book:

‘I should like to reserve the word ‘abstraction’ to pick out a more Aristote-

lian notion, where ‘abstraction’ means ‘taking away’ or ‘subtraction’. […] we

begin with a concrete particular complete with all its properties. Then we

strip away – in our imagination – all that is irrelevant to the concerns of the

4 See below. I shall discuss Joseph LaPorte’s claim, that this view mistakes rather than rejects the

Causal Theory of Meaning, in subsection 4.2. 5 Radder (2006), p.109. 6 Ibidem, p.110.

moment to focus on some single property… ‘.7

Cartwright takes the abstract concept of a Coulomb force in relation to the con-

crete instance of such a force as the paradigm example of Aristotelian abstraction.

She ‘argues that the relation between abstract and concrete concepts […] in sci-

ence is (like) that between a general truth and a specific or local model for that

truth.’8 This type of ‘setting apart’ is imagined, like an ideal circle can be imag-

ined, while the concrete particular (near) circles, such as the outer boundary of a

compact disc, are material. The account of abstraction proposed by Cartwright

serves her focus on methodological questions. In this light, she wants abstraction

to do the same work as Radder, who now approvingly quotes her again:

‘A … major argument for capacities concerned the exportability of informa-

tion: we gather information in one set of circumstances but expect to use it

in circumstances that are quite different.’9

The correlate of this is my contention, in chapter III, that abstracted propositions

are true of more worlds than the original more concrete propositions. The set

M(AA)10 contains models for which much of the content of the original proposi-

tion is not true. Cartwright’s term ‘exportability’ seems to be synonymous to

Radder’s term ‘extensibility’. Her use of the ‘capacities’ concept explains how it

can be that our knowledge of a state of affairs in world w can be used to under-

stand what happens in the actual world @. If statements are abstract enough, they

tend to render true hypotheses – true, that is, of the actual world. Cartwright

believes that certain claims about capacities are abstract enough.

We can see how Radder’s idea of the extensibility and nonlocality of con-

cepts partly (but only partly) coincides with that of universals. He endorses Cart-

wright’s account of the use of abstract concepts. Nonetheless, he does not endorse

her essentialist account of how they are formed. So we can now summarise Rad-

der’s view on the issue as follows. The use of kind terms presupposes that the

world can be interpreted in the prior absence of structuring concepts. This

presupposition is unsound, because we can only observe the world and focus on

the relevant aspects of it if we dispose of these structuring concepts. Without

these, we make mistakes and we need a teacher who tells us what we are doing

wrong. The book of nature is one such teacher, but He is less benign than the

teacher Koningsveld. The reality we research tends to behave in a more complex

way. But this is not to deny that we can acquire the necessary concepts that are

entrenched in our culture in order to adapt our behaviour. So although we need

the concepts to observe, we can form them – abstract them – from the world. This

process is conceivable (!) even without the false assumption that the right

7 Cartwright (1989), p.197 quoted by Radder, ibidem, p.140. 8 Ibidem. 9 Cartwright (1989), p.227 quoted by Radder, ibidem, p.143. Italics are mine. 10 Recall from chapter III that M(AA) is the set of hypothetical possibilities allowed by a proposition

that has been abstracted from another – concreter – proposition, the latter of which is true of

M(A): M(A) ⊂ M(AA).

Chapter V

right conceptualisation is eternally given and fixed. There is no unique right con-

ceptualisation; at most, there is a limited set of conceptualisations suitable, even

available, for our contingent purposes as cognisant humans.

2.2 The theory of redescription: essentialism is there

Economists may believe that their explanatory concepts are abstracted from the

world and deny that these structure it. Menger’s lifelong research to understand

what markets are ‘made of’, how they function, how the Invisible Hand outcome

of markets is related to individual choice making, would seem to be a trifling

endeavour if he had assumed that there is no such thing as a given reality which

commands just one conceptualisation. That is to say, regardless how philosophi-

cally untenable, Menger did best by adhering to an essentialist working assump-

tion if only because of its convenience. In appendix 3, I have tried to indicate the

ways in which one can interpret essentialism in social research, by comparing

Mäki’s and Milford’s account. In this appendix, I discuss Mäki making the point

that the scientific redescriptions of the market mechanism, as proposed by the

Austrians, are ways to reach for the alleged essence of the market. I inserted the

appendix in order to illustrate that economists may well be essentialists by de-

fault. Therefore, I shall very briefly present the core of the concept of redescrip-

tion in Austrian economics that Mäki has introduced.

Science is worth its salt if it does more than what any layman can do by a

superficial common sense scrutiny of the world. Apparently, scientists tell an-

other story about the world than laymen. What story is this? Mäki says that it is

that they theoretically redescribe it. This presupposes a prior description of the

world, in lay terms, or ‘phenomenologically’.11 A theoretical redescription – rede-

scribing, that is, the initial phenomenological description – pretends to point to

the essence of the object of description. Why should anyone do this and why

should a scientist have this pretension? Because the theorist believes that the

object of research appears to be different to the common sense eye than what it

really is like. Otherwise he would have no reason to engage in theory making at

all. In his (1992a) Mäki refers to Austrian economists specifically. These scientists

conceptualise markets as causal processes. Mäki warns us that the type of cause

we have to bear in mind when Austrians talk of causal processes is what Wesley

Salmon called causal propagation.12 This conceptualisation of markets stands in

stark contrast with Walrasian General Equilibrium Theory (GET). There are no

causal claims in GET. Mäki alludes to Frank Hahn to explain that GET and, more

particularly the Arrow-Debreu construction, does not inform us what the work-

11 Mäki (1992a), p.44. 12 Salmon distinguished causal propagation from causal production to indicate that the influence of

our past learning is somehow transmitted into the future as we abductively make decisions in

new learning and decision processes. See Salmon (1984).

ing of the world is like but what it would need to be like if market equilibrium

would (instantaneously) come to be.13

This is not to say that only GET is isolative and Austrian theory is not. (I

hope it is clear by now that any theory is isolative in some sense.) Austrian theory

redescribes the market as a causal process in isolation of all sorts of disturbing

influences. ‘Walrasian theory describes an ideal state, Austrian theory describes

an ideal process’, Mäki says.14 The pretension of, for instance, Menger is that the

tendencies described in the redescription are real and that this redescription is

true of the essence of the market. Moreover (and this is Mäki’s most important

comment for my aims) the ‘strongly isolative character of the theory serves pre-

cisely this idea. Instead of detaching the theory from reality, the revised isolation

allegedly brings it closer to the essential aspects of reality’15

Mäki’s claim about economic theorising is that it is theoretically realist in

the sense that ‘good theories are purportedly true descriptions of the way the

world works’.16 This he calls the ontological, or ‘WWW’, constraint. The assump-

tion of the availability of such a WWW constraint is essentialist. Note that this

assumption is not Mäki’s, but one that is simply implied by the apparent self

image economists have of their methodology given the way they speak of truth

and falsity. I do not ascribe essentialist convictions to Mäki.

As will have become clear in all of the previous chapters, I think that

Mäki’s interpretation of Austrian economics as entailing an essentialist episte-

mology is correct. Moreover, I have already gone beyond that claim. As I put

forward in the intermezzo, it is quite natural for a scientist to be a tacit essentialist

in scientific practice.

2.3 The theory of exact types: essentialism in economics

It has been shown in chapter II (and alluded to in chapter III, sections 5.4 and 5.5)

that the Subjective Value Theory (SVT), which Böhm-Bawerk inherited from the

Austrian economist Menger, has the advantage of explaining two sorts of evi-

dence, both rooted in layman’s observation. It can show how, in cases that ap-

proach the ideal of a stable environment and perfect markets, a tendency may

come into existence for prices to come close to input costs. The stricter formula-

tion of this phenomenon is the Law of Costs (LoC); this law says that end prices

are equal to inputs. As a tendency, it can be seen as a quasi-law, or perhaps a rule

of experience.

Recall from chapter II that it is a merit of the classical Objective Value The-

ory (OVT) too that it predicts this tendency. But on top of this, SVT explains wide

13 Mäki (1992a), p.47. The reference is Hahn (1973). 14 Ibidem, p.53. 15 Ibidem, p.57. 16 Mäki (2001), 285. Italics mine.

Chapter V

deviations of these end prices compared to the costs of inputs, for instance if the

environment is highly unstable. OVT, meanwhile, does not. The objective ap-

proach holds LoC to be an explanatory law, not a mere observational (weak)

regularity. But at most, it is a tendency difficult to experience. Viewing it as such

a tendency, one sets aside frequent actual deviations. SVT, in turn, can explain the

deviations as a consequence of the postulation of the role played by what Böhm-

Bawerk called the ‘Law of the Marginal Agents’ (LoMA). To use Mäki’s terms,

this law helps redescribe the mechanism by which prices of finished goods re-

spond to changes at the input of a production process. And it is this explanatory

law that constitutes SVT’s empirical success relative to OVT.

Although Menger (and Böhm-Bawerk) redescribed this process, in Mäki’s

formulation again, ‘as really being something else’, a lay observer may describe

such a process of change ‘phenomenologically’. So, apparently, we have to as-

sume that what ‘really’ is the case need not always be what we think we perceive

is the case. The phenomenal description presents the process of price adjustments

due to changes in costs as follows. A cost alteration directly feeds into the prices

of the goods by successively altering prices in input markets at intermediary

stages of production and finally altering prices in the consumer markets. These

latter changes take place at the level of the purchase by end users, when the

goods have ‘matured’. This phenomenal representation was the one proposed by

the OVT theorists, who – I just contended – took the Law of Costs as more than a

shaky regularity.

In Menger’s redescription, in contrast, LoMA is the explanatory theoretical

law pointing to an essential underlying structure in social reality that gives rise to

a causal (market) mechanism. (Here and below, a structure is underlying in case it

is not immediately visible, or even intelligible, without the theoretical redescrip-

tion that proposes the existence of the structure. The term referring to this struc-

ture is a theoretical term.) This may be interpreted such that the object of rede-

scription is supposed to have an essence, or to exhibit a social kind and that this

essence is abstracted from all information that is available about market phenom-

ena. This is the interpretation of Austrian methodology by Mäki and Caldwell.

The underlying objects seem to be those that Menger identified as ‘exact

types’. ‘Exact types’ are found by what is called the ‘exact method’.17 They are

abstractions of phenomena, stripped of most of their rich empirical attributes.

Real types, in turn, can be identified in the real world. Real types form the con-

tent of empirical laws.18 A very important aspect of exact types, for the point of

17 See chapter II for a treatment of Menger’s exact types and the similarity with Weber’s ‘ideal types’. 18 According to Bruce Caldwell, who follows Israel Kirzner in this view, the distinction between real

and exact types explains why ‘Menger assumed away the problems of error and ignorance that

were so ubiquitous in […] [the Grundsätze der Volkswirthschaftslehre]’. See Caldwell (2004), p.67. It

also fits in Mäki’s view, expressed in the previous subsection, that causal process theories isolate

just as well as General Equilibrium Theory, be it that the former isolate ideal processes instead of

states.

the present section, is that these shape exact laws that hold with causal necessity,

much in the way the laws of geometry hold with mathematical necessity. Thus,

Menger says:

Die reine Theorie der Volkswirthschaft an der Erfahrung in ihren vollen

Wirklichkeit erproben zu wollen, ist ein Vorgang, analog jenem eines Ma-

thematikers, welcher die Grundsätze der Geometrie durch Messungen rea-

ler Objekte berichtigen wollte, ohne zu bedenken, dass diese letzteren ja mit

den Grössen, welche die reine Geometrie supponirt, nicht identisch sind,

auch jede Messung nothwendig Elemente der Ungenauigkeit in sich

schliesst. Der Realismus in der theoretischen Forschung ist gegenüber der

exacten Richtung der letztern nicht etwas höheres, sondern etwas verschie-

denes.19

On the other hand, though the exact types may not be something ‘higher’,

the search for such types certainly amounts to the investigation of some deeper

lying reality. This is because what cannot phenomenologically be established (but

is nonetheless real) must be hidden. Witness the scientific effort by the Austrians,

its being hidden does not prevent the underlying reality from being known. A

claim of this chapter is that if we have to believe that there are hidden social kinds,

there is no reason to assume that these cannot be uncovered.

3 Essentialism is tenable, but not for social science

Taking Austrian economics as an essentialist programme may be historically

adequate, this does little to support the normative weight of essentialism as an

epistemology. But as I think that an essentialist epistemology underlies most

methodologies in science, I am interested in an inquiry into the extent to which

such an implicit essentialist assumption could be made tenable. I do so in two

steps.

The first step builds on the arguments put forward by Jessica Brown

(1998). She proposes to interpret essentialism in a way that I see as an amend-

ment of Putnam’s position rather than as an attack on it. The second step builds

on the arguments of Joseph LaPorte (2004), which imply that the point with the

specification of (natural, social) kinds lies with rigid designation, while both rigid

designators and the Causal Theory of Meaning should be understood in quite

another way than habitually proposed. This section discusses the first step; sec-

tion 4 treats the second.

19 Carl Menger (1883), p.54.

Chapter V

3.1 The theory of recognitional capacities: essentialism is tenable

Let me start with the well-known argument about natural kinds put forward by

Putnam. If on twin earth twater looks like water on earth, in all possible respects

but its chemical structure, which is XYZ instead of H2O, we would be inclined to

say that twater is not water. Seemingly, we intuitively conceive of water rela-

tively independently of its immediate appearances. If this is true, then we believe

that the fundamental properties of a substance are the discriminating ones to be

able to speak of a natural kind. This raises the question whether the people who

lived before us, who could not know of molecular compositions, would have

been able to recognise the natural kind ‘water’. Fetched somewhat further, it

raises the question whether we are able to do so now. It might be the case that the

most fundamental properties of a substance lie deeper than the molecular level

and are to be sought in the subatomic world or still deeper down, at levels cur-

rent science has as yet no knowledge of.

Brown develops an interesting answer to these questions. She points out

that, to be satisfactory, any answer should be able to solve in fact two problems.

The first is that items, like water, typically instantiate more than one natural kind,

like liquid, water, and perhaps heavy water (D2O, with deuterium, i.e. hydrogen

atoms containing a neutron). The second problem Brown wants to see solved is

that natural kinds occur in impure form, like water in tea or aluminium in rubies.

Brown presents her solution as an alternative not only to Putnam’s but also to

Kripke’s account, both of whom are unable to solve the two problems20. The way

in which she proposes to solve these two problems is as follows.

It is possible that a community has natural kind terms before having scien-

tific knowledge about the chemical properties of substances. How can this be?

Putnam’s answer is well known. Even before the chemist’s knowledge of the

molecular composition of water, the term ‘water’ referred to a substance with the

composition H2O. The assumption that most of what is called water before 1750

is in fact water is perfectly plausible. The reference of ‘water’ is fixed by the defi-

nition that x is water iff x is of the same kind as most of the substance English

speakers have called ‘water’. But the definition fails to fix the reference of ‘water’

to be water, because it applies to both water and to liquid. Brown’s own example,

of gold, shows the point more clearly. Gold naturally occurs as only one isotope,

so ‘gold’ refers to the naturally occurring isotope gold, as well as to gold, and to

metal. ‘Thus, on Putnam’s account […], the reference of gold is indeterminate’,

she says. She calls this first problem the higher-level natural kinds problem; it also

comes under the name of the ‘qua-problem’.

As to the second problem, water has long been referred to as the naturally

occurring water, like that in rivers, which contain a host of other substances, for

example minerals. Putnam seems to be more successful in solving this one, be-

20 For my aims it suffices to concentrate on her treatment only of Putnam, for in her opinion, Putnam’s

approach withstands the problems better than Kripke’s. See Putnam (1975).

cause his definition simply allows that a community has a term referring to a

natural kind although the people in that community only have impure samples

of that kind. But Brown points out that in case of the term ‘ruby’, and not of wa-

ter, Putnam’s approach fails. According to him, the term ‘ruby’ is defined by

someone’s pointing to a ruby and saying, ‘This solid is called ruby’, where the

force of this definition is to say that x is a ruby iff x is of the same solid-kind as

most of the stuff English speakers have called ‘ruby’. However, ruby is a mixture

mostly composed of the compound alumina with a few impurities.21

So ‘ruby’ would refer to alumina. Applying Putnam’s theory to this second

problem – called the problem of compositions – amounts to the counterintuitive

claim that ruby has the same reference as ‘sapphire’ (which is also mainly com-

posed of alumina), namely a substance that lacks the characteristic properties of

both ruby and sapphire. This is unacceptable, especially if one considers that

‘scientifically ignorant communities’ — Brown’s term for peoples that do not

know the scientific account of a particular substance’s fundamental properties —

cannot abstain from reference to the superficial properties of substances. Other-

wise they have nothing left to refer to.22 Brown proposes to solve both problems

by what she says is a recognitional capacities account.

Without recognising it explicitly, Jessica Brown in fact uses three building

blocks that I identify as separate constitutive elements of her argument.

A first building block of the recognitional capacities account is that al-

though a scientifically ignorant community recognises things on the basis of their

appearance, members of such a community may have a recognitional capacity for

natural kinds. A second building block is that the claim that someone has a recog-

nitional capacity for a natural kind is not denied by the postulation of the possi-

bility that this person may sometimes fail to distinguish that natural kind from

some other possible natural kind. The third building block consists of the insight

that, apart from a discriminatory capability, a member of a scientifically ignorant

community must have an appreciation of the metaphysical nature of natural

kinds. Let us take a closer look at each of these elements of Brown’s account of

natural kind terms.

The first building block is both simple and important. We are able to dis-

tinguish more than appearances. We have direct access to appearances, but that is

not all; we also have an access to the kinds that give rise to these appearances.

Brown gives the example of persons. We ‘can recognise people despite the fact

that we recognise them on the basis of their appearance’23, she points out. A large

share of what something is, is given by what it looks or feels like, what sensory

input it arouses. Thus, we are able to separate water and rubies.

The second building block is also sustained by the example of persons. My

21 Brown (1998), p283. An important impurity is chromium. 22 The scientifically ignorant community can have science, and even chemistry; their ignorance is

relative with respect to the chemical composition of the substance under consideration. 23 Ibidem, p.286.

Chapter V

recognitional capacity for my neighbour is not rejected by my incapability to

distinguish her from a copy that walks around in a possible world (or from a

look-alike of my neighbour).

Finally, the third building block, which says that we must have some un-

derstanding for the metaphysical nature of natural kinds, can be explained best

by the use of Brown’s own example, that we shall not identify a snail five miles

from our garden with the snail we saw half an hour ago in our garden. This is

true even though the two animals that biologists call Orthalicus Reses look very

similar. This last element of Brown’s argument, a claim about man’s metaphysi-

cal appreciation for the properties of the world, is important for what follows. I

shall therefore dive a bit deeper into it here.

3.2 Our metaphysical appreciation

The point of the metaphysics is that we appreciate that instantiations of natural

kinds are of such kinds because of their fundamental properties even though a

natural kind may be common to two fundamentally different substances (like the

kind water to both D2O and H2O) and even though the problem of compositions

looms everywhere. We shall see that Brown’s account withstands both problems

that Putnam’s does not.

The reverse case of the higher-level kinds problem is that fundamentally

equal substances may have very different appearances. That is, in turn, well in

line with the intuition that if we have a recognitional capacity for a natural kind,

fundamental properties matter. In contrast, as Brown remarks:

[w]hen a subject has a recognitional capacity [merely] for an appearance

kind, information about the internal constitution of an item is irrelevant to

whether that item is of the appearance kind in question.24

An appearance kind (which remains undefined by Brown) must be taken

to be determined by what triggers our capacity to tell objects apart. The point of

the metaphysical appreciation is that we are prepared to revise our belief that

two objects be of the same kind whenever new information about their funda-

mental properties tells us to do so. To know that fundamental properties matter

is sufficient. We need not know these properties.

So how does the recognitional capacities account survive the two prob-

lems? Can humans have recognitional capacities for natural kinds despite the fact

that one object often instantiates more than one natural kind and despite the fact

that natural kinds mostly occur in impure form? The higher-level natural kind

problem is solved if the demands for what may be counted as the capacity to

recognise a natural kind are slightly attenuated. Take the example of an element

24 Ibidem, p.287. Brown does not define ‘appearance kind’, but I take it that she refers to any integra-

tive concept that helps us distinguish the phenomenological aspects, but not the fundamental

properties of an entity. Thus an appearance kind is what triggers my recognitional capacity for

the cow I would like – but lack the gift – to draw a picture of.

and its isotopes. Isotopes have the same superficial properties as their element.

The distinctive fundamental properties of the isotope are not required to explain

its superficial properties. Those of the element will do. The point Brown makes is

as simple as efficacious. Which of the kinds is recognised, that of the isotope or of

the element, depends on (1) which set of superficial properties it is in virtue of

which the items trigger the recognitional capacity; and (2) which of the natural

kinds […] has fundamental properties which explain the items’ possession of the

superficial properties25. In other words, we recognize the carbon-14 isotope since

we are capable of doing so, and we recognized the atom or the substance before

So to claim that humans have always had the capability to distinguish

natural kinds of substances, a scientifically ignorant community need not have

access at all to the knowledge that there are isotopes of elements. That would be

too much to demand. The possibility to cut down on the demands in this way

rests on the first and the third building block26, because these tell us that recogni-

tional capacities for appearance kinds, in addition to an appreciation for the idea

that the appearances of an item are associated with some fundamental properties

of that item, are sufficient to recognise a natural kind.

As to the composition problem, it is solved in very much the same way as

the higher-level kinds problem, although the second building block also plays a

role here. The second building block says that if I may sometimes fail to distin-

guish a natural kind from some other natural kind, it does not follow that I have

no recognitional capacity for that kind. Note that the fact that superficial proper-

ties of a natural kind are often not exhibited by impure samples of it is inconse-

quential to Brown.

Moreover, she asks whether mixtures are natural kinds. She believes not,

but, interestingly, neither a positive nor a negative answer to the question bears

on the way the recognitional capacities approach treats natural kinds. All we

need to say is that we do not expect someone to recognise alumina if that sub-

stance occurs in ruby and sapphire and in other mixtures that hide the typical

appearance kind of alumina. The limits to our ability to develop a recognitional

capacity for kinds are set by the world. These will be especially pressing if the

only available samples are impure. So whether or not mixtures are natural kinds,

i.e. whether ‘ruby’ refers to ruby or not, Brown rightly notes, ‘the recognitional

capacities account does not have the counterintuitive consequence that “ruby”

refers to alumina’27.

In sum, it is fair not to expect too much of our recognitional capacities of

natural kinds and still believe in such capacities. There are limitations to our abil-

ity to recognise natural kinds by their appearance. The superficial properties of a

25 Ibidem, p.295. 26 Brown does not note this explicitly. 27 Ibidem, p.297.

Chapter V

sample may not help us to discriminate the kinds in the sample. To say that natu-

ral kind terms refer, I do not need to expect people to distinguish nitrogen from

carbon dioxide or from noble gases in air. Without scientific methods, we can’t

develop a recognitional capacity for these natural kinds, but we can for others.

The gist of Brown’s argument is that, counter many anti-essentialist pleas,

there can be a coherent account of natural kinds. It seems to me that her account

is a step towards a defence of essentialist convictions in science. But she sees no

scope for kinds fixed by subjective interests and thus, as I believe is implied, nei-

ther for social kinds. This does not sit easily with the fact that Mäki sees much

evidence for such essentialist convictions in economics. In the following subsec-

tion I shall look into this matter.

Mäki’s point was that true economic theories isolate the relevant aspects

from the irrelevant aspects of concrete social reality. To be able to do this, econo-

mists must have a metaphysical appreciation for whatever makes up the realm of

the economy. This cannot be anything but an appreciation for abstract socio-

economic kinds, as a focus merely on the wealth of concrete detail would uncover

nothing about this realm. The appropriate reading of Austrian theory, then, is

such that it ‘has the character of a causal process theory’, supporting ‘a realist

reading of the theory’ which is ‘not undermined by the theory also having the

character of an isolative theory’.28 It has the self-image of approaching the truth

by the isolation of abstract kinds.

3.3 Exact types and social science

As announced above, I want to tease out how far an essentialist interpretation of

social research can be stretched to the limit where it is still tenable. Let me con-

sider this passage of Brown’s paper:

[T]he kind whose extension includes both pure samples of water and some,

but not all, impure samples of water, seems to be a gerrymandered kind

whose extension is fixed by our subjective interests, not by the world.29

The kind water whose extension includes pure and impure samples is indeed

unhelpful. It is not implied that the specification of kinds being subjectively mo-

tivated is gerrymandered. But I shall show why research interests and other sub-

jective motivations will actually guide the specification of kinds in science, in-

cluding in social science. New (empirical, but more often conceptual) discoveries

induce theoretical economists to abstract interesting aspects from economic real-

ity and at the same time to stipulate new social kinds. (The way this sentence has

been formulated is meant to bear reminiscence to Radder’s claims. But note the

difference. Radder would not underwrite the talk of kinds.)

The exact types Menger said we should try to find are the candidates I

28 Mäki (1992a), p.57. 29 Brown (1998), p.281.

propose for social kinds. I still have to discuss what sort of objects in the world

may turn out to instantiate these kinds, but here I can say that they appear in the

redescriptions economists propose. These redescriptions tell an alleged true story

about the object of research and they help redescribe how the world works. Con-

ceptualised thus, Mäki used the notion of the WWW constraint for pointing to

the condition for truth in the scientific specification of social kinds. That is why I

propose to phrase the interpretation, by Mäki, of much of economic theorising as

in fact essentialist, in the terms Jessica Brown formulates for her epistemology of

natural kinds. To do so I repeat Brown’s three building blocks:

B1 Recognitional capacities for some sort of fundamental kinds are possible

even if we do not have much more than appearances to give access to these

kinds;

B2 Our failure to distinguish fundamental kinds in some situations is no coun-

terexample to the possibility of such capacities;

B3 We additionally need an appreciation for the metaphysical character of

such kinds, because that is what makes them refer to some fundamental

reality.

Compare Mäki’s views, reformulated as follows:

M1 Apparently, although a common sense description of the social world is

close to the phenomena, to appearances, it does not prevent us to know

more about that world because a redescription tells us about social kinds

on which the appearances supervene;

M2 A scientist may fail to abstract the right social kind, i.e. he may still be cut

off from the targeted essence, but this does not imply that he has no capaci-

ties to increase the essesimilitude30 of his hypotheses;

M3 Reference to economic Invisible Hand mechanisms, as giving rise to social

processes, expresses an appreciation of the metaphysics involved in such

mechanisms, because it acknowledges the way the world works as a con-

straint on the choice of social kinds in theorising.

I conclude that my objective to find the edge to which an essentialist inter-

pretation of science can be stretched need not be confined to natural science and

natural kinds. We have to think in terms of social kinds.

4 Essentialism is tenable, also for social science

Often, essentialism is argued to be untenable on the basis of assumptions that

associate essences with (1) relatively enduring (non-historical) properties of ob-

jects, such as molecular make-ups, and (2) with a determinate reference of kind

30 This is Mäki’s variation on Popper to indicate the content of abstracted truth.

Chapter V

terms such that vagueness is excluded. However, this is not the most obvious

depiction of essentialism.

As to (1), natural kinds are indeed often supposed to be spatio-temporally

unrestricted. The institutional environment is spatio-temporally very restricted.

Institutions change, people make their deliberations in a changing environment,

new ways of coping with the environment are invented. Thus, it seems that es-

sences cannot have any explanatory power in economics, or in any other non-

naturalistic social science. However, I shall use the work on natural kinds in biol-

ogy and chemistry by Joseph LaPorte in order to draw the conclusion that, in fact,

any projectable properties are candidates for a treatment in terms of kinds. In the

discipline he uses as core example – biology – even historically delimited species,

as these came to exist in some limited period of evolution, can be biological kinds.

The only postulate he needs for this conclusion is that naturalness, like flatness or

vagueness, comes in degrees.

As to (2), LaPorte notes that the reference of kind terms need not exclude

vagueness. He does believe that the meaning of a kind term comes into existence

by a speaker pointing to an object (a mammal) and says, “this is … (an instantia-

tion of a mammal)”. This is in accordance with Kripke and Putnam’s views. But

he does not believe that this is all. There is plenty of room for indeterminacy.

Why should a ‘causal baptism’ not allow for additional descriptions that make

clear what a term precisely refers to. LaPorte manages to show, I believe, that if

we allow for this we can also see that the Causal Theory of Meaning (CTM) is not

so problematic as it is taken to be. It does not face the problem that kind terms are

often used vaguely and that they are so often subject to revision. Kind terms do,

but CTM should not be taken to do the job of eliminating indeterminacy and

vagueness.

4.1 Kind terms specified de jure

So, according to LaPorte, there is no contradiction between a causal account of

how words get their meanings, properly understood, and an account of how we

deal with uncertainty and vagueness and how meanings change. This proper

understanding alluded to, then, is as depicted in the following two examples.

In chemistry, to start with, the specification, that H2O is the substance

‘H2O’ designates, is done intentionally by the scientific community. In biology,

the specification of ‘mammal’ or ‘whale’ is intentional too. And this can be un-

derstood as having been done by a sort of causal baptism. Hence,

‘a term for an individual is typically coined in the presence of the individual

[…] the speaker says something like “this person is to be called ‘Cicero’”

(the speaker points). A term for a natural kind is typically baptized in the

presence of samples of the kind’.31

31 LaPorte (2004), p.5. Italics by the author.

There seems to be, however, a difference between ‘Cicero’ and a kind term,

in that the latter appears to be discovered, while Cicero got his name by deliber-

ate specification. This is how the use of kind terms is indeed mostly interpreted.

And here LaPorte’s argument already deviates from that of Putnam.

The central point LaPorte makes is that, even if the molecular structure of

water as such is a discovery, and even if ‘water’ is a rigid designator, it has not

been discovered that ‘water’ designates rigidly. If a community of speakers

makes the term ‘water’ refer to water, and the term ‘H2O’ to H2O, then in all pos-

sible worlds where these terms refer at all, they refer to their referents necessar-

ily. In other words, if these terms are no rigid designators de facto, they are so de

jure. Terms like ‘water’ and H2O, which are coined by reference to a sample of the

kind, are rigid in the same way as names are rigid de jure. Let me explain this

point further.

Before people were aware that water has the microstructure we refer to if

we use the chemical formula, they already knew, more or less, what water was.

Quite apart from the possibility to assume that we engage, as human beings with

a complex organisation of our life, in a linguistic division of labour, we can say

the following. Man has always had (in Brown’s words) a recognitional capacity

for the appearance kind of water at least.

The reference of the term ‘Hesperus’ is rigid too, but its extension is one

concrete object. As the same goes for ‘Phosphorus’, after the discovery that both

celestial bodies are one and the same planet, in all worlds where there is Phos-

phorus as we understand it, it refers to Venus. So ‘Phosphorus’ and ‘Hesperus’

are rigid designators. “Hesperus=Phosphorus” expresses the discovery of a nec-

essary truth. LaPorte agrees with Kripke’s explanation, first, that the proposition

“Hesperus=the brightest nonlunar object in the evening sky” is contingently true.

This is because ‘Hesperus might have been overshadowed by some other still

brighter object, had our solar system had more planets’;32 and, second, that “Hes-

perus=Phosphorus” is necessarily true.

But things are different with ‘water’ or ‘H2O’. The reference of these two

kind terms is not the concrete objects that instantiate the kinds, i.e. their exten-

sion, because the actual samples of water may be different in the future or in

other possible worlds. There is nothing rigid in the term ‘water’ if its reference is

the concrete instantiations. But ‘Water’ or ‘H2O’ are terms that designate abstract

kinds. Meanwhile, it has been discovered that water consists of molecules H2O,

and so “water=H2O” was the discovery of a necessary truth.

A question now emerges: is it not contingent that the substance that we re-

fer to with the term ‘water’ happens to be such that we can also refer to it as

‘H2O’? The answer argued for below is again that, given our use of these terms,

“water=H2O” is necessary if true at all.

In biology, although the kind ‘mammal’ has deliberately been specified as

32 Ibidem, p.35.

Chapter V

well, in each and every world where the term has the meaning it actually pos-

sesses, a mammal is ‘lactating animal’. Nevertheless, the extension of the term

differs across possible worlds. To what does a kind term refer if it is not to its

extension? The answer again is: to an abstract kind. Moreover, the kind term re-

fers to the same abstract kind across possible worlds.

All this may look suspicious. The treatment seems to allow for a specifica-

tion of just any term and counting it as rigid: there being no difference any more

between rigidity and non-rigidity. It also looks as if Jessica Brown is right in say-

ing that our subjective interests in fixing kinds are going to deliver gerryman-

dered kinds. But this is not true. An example of a non-rigid designator LaPorte

gives is: ‘the largest dinosaur on a 1989 US post stamp’. This designates Bronto-

saurus, but not rigidly, because if there had been, in some world, two instead of

three US post stamps with pictures of dinosaurs on them, the largest might have

been Tyrannosaurus Rex. The individual is picked out by a particular accidental

description. In contrast, in different counterfactual situations we keep on using a

kind term, like ‘Brontosaurus’, for the same object. We do not so as regards a

term designating non-rigidly, like ‘the largest dinosaur on a 1989 US post stamp’.

So, clearly, some terms designate their referent rigidly, and others do not. There

remains a difference between rigid and non-rigid designators.

If all that is needed for a term to be a kind term is that it designates the

same kind in all possible worlds, then artificial kinds, like toothpaste, exist too.

This is because artificial kind terms do designate the same kind in all possible

worlds. And so do social kinds. This result is precisely what I have noted above

(by means of the quote at note 29) that Brown objects to. The relevant distinction,

then, is not between natural kind terms and artificial kind terms over rigidity, but

between on the one hand rigid and on the other hand non-rigid natural kind terms

and, likewise, between rigid and non-rigid artificial kind terms. That is why we

are asking too much if we want rigidity to make a difference between, say,

‘whale’ and ‘bachelor’, or between ‘gold’ and ‘public good’. Rigidity lies not in

the make-up of the world but in the way language functions.

Given this account of rigidity and kind terms, it turns out that speakers

make rigid designators refer to their referents. ‘Gold’ and ‘public good’ are rigid

de jure, like names. But if this is so, how can we expect a kind term to explain?

LaPorte holds that ‘a natural kind is a kind with explanatory value’.

A lot is explained by an object’s being a polar bear. That it is a polar bear

explains why it raises ice cubs as it does, or why it has extremely dense fur

[…] The polar bear kind is a useful one for providing significant explana-

tions. It is a natural kind.

Compare a highly unnatural kind, such as the named-on-a-Tuesday kind[.]33

And I want to add: similarly, a social kind is a kind with explanatory value

too. Of course, one can make the context very strict, so as to deny abstract social

kinds their existence. As we would have had a set of completely different species

33 Ibidem, p.19. Italics by LaPorte.

of living things on earth if evolution had followed another course, LaPorte notes,

Paul Churchland denies biological species their kind status. If you want kind

terms to be associated only to very basic laws underlying the universe, the spe-

cies kind can be rendered non-natural even if H2O (or something deeper still) is

allowed its natural kind status. But there is no objection against setting lower

standards, and then one can also propose that ‘’[d]epression’, ‘inflation’, ‘monop-

oly’, and ‘money’ are good candidates from economics’.34

But are these candidates for the social kind status explanatory? I should

maintain so. The monopoly kind, to use one of the examples, is explanatory:

some producers have market power and earn supernormal profits. How come? It

is because they are of a monopoly kind. This is true even if a relatively non-

enduring environment – in which man-made institutions dwell – characterizes

market forms; or even if the boundaries between competition and monopoly are

vague. I conclude that the light-weight approach of LaPorte helps see that it

makes sense to talk of social kinds.

Section 5 aims to motivate my choice of terms’ acting as social kind terms.

This choice is based on the mechanistic view on the economy held by the Aus-

trian theorists.

Nonetheless, contrary to what is often assumed, vagueness of a kind term

– natural, social, or artificial – remains possible – and shall be a frequent (or even

ubiquitous) phenomenon. An interesting example from the history of science is

the confusion between what LaPorte calls ‘Hopkins-species’ and ‘Darwin-

species’. Darwin, LaPorte says, discovered that species evolve from ancestors.

William Hopkins was a critic of Darwin, making the point that Darwin’s theory

denied the existence of species, which seemed absurd to Hopkins. As the defini-

tion of ‘natural species’ stipulated that these have a separate and independent

origin, species simply could not have derived from the same origin. So did Darwin

discover the correct designation of the term? No. Darwin did not discover what

the kind term ‘species’ means, he discovered that there was vagueness in the use

of the term. He specified ‘species’ in a way tailored to his needs, and it was a

specification de jure. But this does in no way deny ‘species’ the status of a natural

kind term. Given the way in which Darwin newly specified the term, in all possi-

ble worlds where there are species in the sense specified, they derive from a com-

mon ancestor. ‘Species’ is one rigid designator under the rule of Hopkins. It is

another rigid designator under the rule of Darwin.

I discuss vagueness of kind terms in subsection 4.3. But allow me to first

treat the mistaken role of CTM in this idea of vagueness.

34 Ibidem, p.25

Chapter V

4.2 The exaggerated demands on the Causal Theory of Meaning

We have seen that LaPorte maintains the familiar point that kind terms derive

their meaning by a sort of causal baptism. This is interesting because in section 2

it was explained why Hans Radder rejects CTM. The extensibility of the result of

an observation process is not fixed, Radder says, while CTM fixes the meaning

and reference of concepts rigidly. Radder associates CTM with rigidity.

But LaPorte does not. CTM says that names are given by baptising cere-

monies and that kind terms are given their meaning by speakers pointing to sam-

ples. This creates plenty of room for indeterminacy, because later speakers may

be uncertain whether a new observation results in the same kind. When the

properties of the organisms known as Platypus were discovered – lactating but

no womb, with cloacae and laying eggs – it became tricky to use the term ‘mam-

mal’. That is one reason why the practice of causal baptism is often enhanced by

that of describing. (The second reason is that it solves the so called ‘qua-problem’,

that it is not clear of which kind a concrete object, like a tiger, is an instantiation,

of the kind tiger, mammal, organism, or living thing.35) As LaPorte says, the

proper job of CTM is to recognise, not to prevent indeterminacy. If a new species,

or a newly discovered species, causes trouble for our use of the term ‘mammal’,

we can specify it anew, with a new baptism ceremony pointing to mammals like

whales; and a ceremony in which we point at a platypus and say ‘this is a platy-

pus’; and one in which we decide what ‘fish’ means (whales not instantiating this

kind). Scientists can find out about confused suppositions underlying the terms

that are nevertheless rigid designators, be they rigid de jure.

It is often assumed that the coining of kind terms by pointing to samples

precludes vagueness. According to this understanding of CTM, the sample fixes

the kind. But, evidently, it does not. It is evident because causal baptisms are

performed by speakers typically with unsophisticated knowledge, so later speci-

fication in the conceptual development of a term changes its meaning, a happen-

ing which results in a new kind term, rigid de jure again.

That an ostentation can be supplemented with descriptions of the referent

(thereby not implying a descriptive theory of meaning) does not give sufficient

fire power to reject an understanding of science as using kind terms; and not

even to reject an understanding of social science as using social kind terms.

4.3 Vagueness of kind terms

Vagueness, like necessary truth, can be discovered and if such a discovery is made,

there is scientific progress. Böhm-Bawerk discovered that the concepts of ‘value’,

of ‘market’, and of ‘borrowing’, for instance, were vague. A ‘market’ (even a so

called ‘abstract market’) is the institutional setting that enables exchange. How-

35 LaPorte (2004), p.6.

ever, before Böhm-Bawerk’s new conceptualisation, the market for credit had

been understood as the institution that enabled the use of capital, rather than the

exchange of future against present goods. Hence, there was a considerable amount

of vagueness both concerning ‘market’ and ‘borrowing’ and Böhm-Bawerk’s con-

ceptual unification added precision to the concepts as part of what was at the

same time a meaning change and theory change. Take the statement ‘the credit

market enables the entrepreneur to acquire present goods’. This proposition

could be seen by other economists as a conceptual falsehood. According to

Böhm-Bawerk, however, it was an unrecognised truth. Likewise, in natural sci-

ence, ‘species’ acquired a new meaning by the theory change initiated by Darwin.

‘Evolving from ancestors’ was a meaning of ‘species’ that fitted Darwin’s theory

but was nearly impossible to understand by creationists. Thus, (with Kuhn) we

can say that theory change comes with conceptual upheaval.

It is due to vagueness that competing theories may induce different under-

standings of the world while using the same term. The term then marks different

concepts. But vagueness does not only cause a rift between competing theories.

The very same vagueness that parts paradigms also provides the common

ground for communication between opposing theorists. Let me explain.

Essentialism and pluralism as concordant

Pluralism says that the world does not dictate one unique correct conceptualisa-

tion of itself. Another way of putting this is that no truths exist outside concep-

tual schemes: there is a diversity of conceptualisations to categorise the world,

and though they are incompatible, they can all be in a sense ‘true’. Conceptual

schemes can be seen as indices of propositions.36

The obvious objection against pluralism, if it is interpreted in this way, is

that the pluralist ends up with a completely relativist theory of truth, or worse,

with idealism. Michael Lynch (1998) stresses that this objection need not be

raised. He defends a theory of truth, inspired by Wittgenstein, which defuses the

pluralism-realism contradiction. So after LaPorte, who defuses the divide be-

tween the possibilities of meaning change and of essentialism, we have here an-

other author who dismounts the opposition between pluralism and realism.

On the one hand the alleged knowledge of kinds presumes a robust onto-

logical commitment and, on the other hand, there is communication possible

between two or more incompatible such ontological commitments. What is

needed for this is the understanding that we can step outside the scheme of our

immediate community while not outside of all schemes. Lynch gives us this un-

derstanding. (It must me admitted that it is not clear whether Lynch would agree

36 One may object that indices cannot have truth value. But ‘conceptualisation’ is in itself a relatively

vague concept. If it means ‘vocabulary’ (Kuipers, 2001), then it does not have truth value and the

previous sentence must instead be: ‘they can all be appropriate’. If it means something similar to

‘synthetic a priori’, then a conceptualisation can be true or false. For the present argument this is

of no consequence.

Chapter V

with LaPorte’s defence of essentialism. He rejects that talk of ‘the essence of exis-

tence and of object’ could be coherent.37 But he does not reject that talk of the

essence of some referents – other than those of ‘existence’ and ‘object’, in what he

says are robust extensions of concepts, see below, – could be useful.)

The core of Lynch’ idea is that robust conceptualisations are always rooted

in the use of ontologically more minimal concepts. It is at the level of these mini-

mal concepts that vagueness appears downstage. We extend the use of minimal

concepts from one paradigmatic example into metaphysically more loaded direc-

tions as we commit ourselves to a particular ontology. My conviction is that it is

in the loaded versions that we specify kinds. So is Lynch’ idea, but he does not

count with the possibility of conceptual upheavals in science, such as when Dar-

win specified the kind ‘species’ anew. I have to propose an amendment to Lynch’

story. But let me first tell this story.

In some contexts, Lynch says, concepts are such that we do not know pre-

cisely what these refer to. The familiar example by Wittgenstein is that one can

disagree over the meaning and the reference of the terms ‘game’ and ‘sports’.

‘Game’ allows us to appeal to several paradigmatic examples. This is because

there are family resemblances between the referents of this term. But natural

kinds like gold are ‘nailed down’38 by the division of conceptual labour. It is clear

that Lynch follows Putnam in this respect. Apparently, Lynch sees natural kind

terms as precise due to the expertise of specialists. They know what a whale is,

and a mammal, so they tell us which paradigmatic example to use. This view

accords with LaPorte’s in the role of CTM, but it contrasts with what LaPorte

reports about the disagreements between Darwin and Hopkins: that vagueness is

not precluded even in specialist science. Vagueness, I note again, is discovered

frequently, and precisely by these scientists.

I continue, following Lynch. With the conceptual schemes by which we

talk about the world, we extend the use of concepts into a particular direction,

and pluralism acknowledges that different traditions, or paradigms, do this in

radically different directions. But all these extensions share a minimal concept of

the object in a dispute over how to best conceptualise something, and so we share

a minimal concept if we engage in such a dispute. This minimal concept implies

relatively little concerning ontology. As Lynch puts it, the minimal concept

‘”floats free” of metaphysical questions’39. Hopkins disagreed with Darwin, but

he did not have something completely different in mind when he used the term

‘species’. In using the term, he did not see an English playwright or a Scottish

loch before his eyes, but a group of phenotypically related animals or plants:

something in many respects similar to what Darwin had in mind. So Hopkins

and Darwin could talk without talking completely past one another (if Hopkins

had only wanted).

37 Lynch (1998), pp.88-89. 38 This is Lynch’ phrasing. 39 Ibidem, p.68.

But once a metaphysical commitment comes in (recall the similar phrasing

by Jessica Brown, above), we make a concept more robust. In such a case, a rela-

tively vague and minimal ‘species’ becomes a Darwin-species, for example, or a

Hopkins-species, depending on the chosen ontology. Thus, metaphysical com-

mitments trouble the communication between two scientists who extend their

concepts into different directions.

There need not even be one single shared property for all robust uses – for

all extensions – of a concept. This is the Wittgensteinian element of Lynch’ ap-

proach. There may be no more than family resemblances. But there can be shared

properties. In other words, the distinction between minimal and robust exten-

sions is a matter of degree. The minimal concept, however, provides the common

ground for communication.

Minimal concepts, rigid designation de jure, and kinds

In sum, in the case of the term ‘species’, Hopkins and Darwin did not take them-

selves to discuss completely different subjects. Hopkins could step outside his

creationist scheme and try to understand how Darwin extended his notion of

‘species’ into that of evolutionism. It is true that, as humans, our degrees of free-

dom in choosing other world views are restricted. But the concept of ‘species’

was discovered by Darwin to be malleable enough to extend the use into the di-

rection of evolution. For the extension he had to start from a minimal concept of

species, more minimal than the concept Hopkins and other creationists used, so

he had to push it away from the extension toward the creationist interpretation.

Hermeneutics seems to be the study of this process of stepping outside of

one’s own scheme. A necessary condition for hermeneutics is the assumption of a

shared reality, and shared minimal concepts. These provide the common ground.

Competing specialists in scientific theorising can enter common ground guided

by curiosity for the meaning of competing concepts. Thus, they need not accuse

each other of conceptual falsehood. This is why I believe that Hopkins could have

been convinced by Darwin, at least in principle.

What about Böhm-Bawerk and his conceptual unification? What about so-

cial kinds? He believed that the credit market is essentially a market where pre-

sent against future goods are traded; and that the same is true for the labour

market and for any other business-to-business market. He provided conceptual

unification by subsuming all these markets under one and the same super type,

the ‘market for trade in present-against-future goods’. He coined this concept

with sufficient robustness so as to be able to unify the knowledge of what in his

eyes was a given ontology. He coined a kind term. This means that, given the

Böhm-Bawerkian meaning of this kind term, in every possible world where there is a

market of future and present goods, the referent of this concept has the properties

specified by Böhm-Bawerk. If the referent has the properties of a labour market, it

is an instantiation of the market of future and present goods. If it has the proper-

ties of a credit market, it is another instantiation of the kind ‘market of future and

Chapter V

present goods’. Just like humans and whales instantiate the kind ‘mammal’.

Böhm-Bawerk believed that his conceptualisation was the only correct one.

He also thought that with this conceptualisation he was picking out necessities de

re concerning social and institutional reality. This is a deep ontological commit-

ment and, hence, an engagement in a very robust extension of the concepts famil-

iar from Austrian economic theory, as he borrowed these from Menger or as he

specified these himself. Nevertheless, the specifications, either by Menger or by

himself, were nothing but the de jure specification of kinds. This is the maximal

extent to which Austrian essentialism can be stretched.

In the intermezzo I had John O’Neill entering the stage with his defence of

essentialism in social science. ‘Wittgenstein’s argument does highlight the open-

ness of essentialist claims to empirical criticism’, he says.40 In his essay it is not

clear – for not argued with sufficient precision – how O’Neill squares his meth-

odological position with his endorsement of Wittgensteinian philosophy. An

account along the lines of Lynch and LaPorte is capable of providing this clarity.

5 Social kinds as structures of the social

So far, we have used the term ‘social kind’ loosely, meaning a kind the instantia-

tions of which cannot be found in the natural world but in the world of institu-

tions and social phenomena. The market for trade in present-against-future

goods has just been presented as an example. But a market is not a simple object.

It is a complex set of reciprocal relations, which in the Mengerian eye can best be

described referring to intentional agents, endowed with the capabilities to order

preferences and to compute the best alternative in the light of scarcity and the

objective of utility maximisation. In this Austrian view and in the appropriate

institutional context LoMA (the Law of Marginal Agents) is the principle govern-

ing market processes.

So social kinds refer to complex objects. In this section, I endeavour to

sketch which objects in social reality can be plausible candidates for a kind status.

My proposal is that social kinds are structured entities that give rise to mecha-

nisms. Of course, this merely lies the burden of the explanation elsewhere;

mechanisms are to be typified in their turn. James Woodward has an interesting

notion of what the crucial characteristics of a mechanism could be because he

compares mechanisms in natural science with mechanisms in behavioural sci-

ence. I shall tell my own story about mechanisms, and merely mention Wood-

ward’s specifications in passing.

In brief, this section defends the view that social kind terms are specified to

refer to the structural characteristics of complex objects in socio-economic reality,

40 O’Neill (2001), p.172.

such that they give rise to causal mechanisms. I exemplify these causal mecha-

nisms as hypothetical processes. Given certain initial conditions, one of the pos-

sible hypothetical processes turns actual: the actual causal process. A three lay-

ered ontology thus emerges: of process, mechanism, and structure.

5.1 The idea of an actual causal process

In explaining my view in this subsection, I shall use a forceful example meant to

trigger the imagination: that of a river. In doing so, I discuss some considerations

on the alleged causality that seems to be involved in many actual processes.

Standing alongside a river bank in the Cantabrian mountains, one cannot

but understand the way in which the water washes over the rocks and coils

through the next bend as somehow causal. The observer is not habitually im-

peded by Humean contemplations: his observations make him think that the

water is being caused to run as it does. He is inclined to contend that the move-

ment of river water is a causal process, or the effect of a causal process. So what

kind of thing may a causal process be?

One thing we can say about processes, causal or otherwise, is that they are

actual. They are observed, if at all, in the actual world. Now, suppose that a par-

ticular process is observed and that, in addition to this, it is understood by the

observer somehow as essentially causal. To interpret the actual process as causal

is to add some aspect essentially belonging to this process. There must be some-

thing on top of its being a mere diachronic succession of states so as to make it

causal. So what is on top?

The literature on causation has tried to answer this question, among other

ways, by the use of counterfactuals and a possible world semantics.41 Clearly, the

use of counterfactuals to track down causes is prone to many difficulties. One is

that in trying to pick out causes as an ontological project, philosophers have

ended with the mere epistemological result of how to recognise causes, not a char-

acterisation of what causes are. To point to the same difficulty, Psillos distin-

guishes the inference problem from the identification problem.42 More impor-

tantly, it is easy to construct examples that show that the truth of a relevant coun-

terfactual is neither a necessary nor a sufficient condition for what we would like

to portray as a cause. To my knowledge, Jaegwon Kim has rejected the fruitful-

ness of counterfactual reasoning in causal talk most efficiently.43 But Psillos notes

41 For an extensive overview see for instance Psillos (2002). 42 Ibidem, p.165. 43 Kim (1973). A sophisticated and strongly essentialist approach by Stephen Yablo (1992), however,

counters the evidence against the use of counterfactuals in these contexts. He makes a particularly

forceful use of the Stalnaker-Lewis toolkit to demarcate between the categorical likeness of similar

causal events and the hypothetical difference these otherwise very similar events may constitute

nevertheless. In his view, ‘causal properties must be hypothetical to tell categorical duplicates

Chapter V

that counterfactuals are going to be needed in one way or another: sustaining

either explanations in terms of lawlike behaviour, or explanations in terms of

mechanisms. It seems to me that this claim is justified. Psillos also remarks that,

though counterfactual reasoning needs no story about mechanisms, the latter do

need counterfactuals.44 Below I shall make use of the same possible worlds se-

mantics as I did before and distinguish hypothetical from actual situations in

order to be able to talk about processes as ‘causal’.

So, to return to the example introduced above, the observer must have

epistemic access to other possible worlds where the river water washes over the

rocks in a different way. Why? Because if some state of the world is the cause of

the actual flow of the river, then, in the absence of that cause, it would have been

otherwise. The concept of counterfactual reasoning operates, so to say, in the

background. It is the form of the riverbed and the pull of gravity, which makes

the current go as it does. Whether the riverbed has its particular form is a contin-

gent matter. In other words, there are hypothetical but possible worlds where the

riverbed is different and where the current runs differently, or not at all. Due to

this, for a process to be interpreted as causal, the observer must base this inter-

pretation on ideas about the non-actual world. To interpret a process that comes

about to our eyes as causal means to have a notion of alternative possibilities. But

if we can imagine possible (non-actual) states of the world, it does not follow that

we accept anything as possible. We imagine that the world has structure.

The particular form of the riverbed delimits the possible currents of the

river. Given this structure and given gravitational pull as we have this on earth, a

restricted set of currents is possible depending on the amount of water that

washes downhill. Other circumstances, such as the force and direction of the

wind, do matter. In the example exploited here, this means that the riverbed is

the structure that makes possible several possible flows of the water, depending

on initial conditions such as the amount of water that is to be channelled down

the slope; much in spring, when ice melts, and little during fall, when a summery

drought has caused the greener landscape to pine away. Every season produces

its own current through the riverbed and every year identical seasons will not

even produce completely the same actual flows. This is because more initial con-

ditions matter, for instance the strength of winds. But still not anything is possi-

ble. The riverbed does not cease to be a structure enforcing the direction of the

current (until of course the initial conditions cause the water to burst its banks,

when other structures take over, new mechanisms and processes emerge).

The water supply and the other circumstances make the initial conditions

of the process. Given particular initial conditions, a process emerges. If these

initial conditions are actual, so is the causal process that is given rise to. The fol-

lowing scheme represents the core idea. The causal process of the river current,

apart’ (p.413) Yablo makes essence causally relevant. Although promising, I shall not dwell on his

contribution to the discussion, for it would take us too far from my objectives. 44 Psillos (2004), p.315.

then, is what emerges with the presence of the riverbed, gravity, water supply,

and other circumstances.

The causal mechanism can be described by its constitutive elements: the riv-

erbed and gravity. A mechanism generally is described as a structured make-up

of components that somehow interact in a predictable way. For example, look

what Woodward says about how to represent mechanisms:

a necessary condition for a representation to be an acceptable model of a

mechanism is that the representation (i) describe an organized or structured

set of parts or components, where (ii) the behavior of each component is de-

scribed by a generalization that is invariant under interventions, and where

(iii) the generalizations governing each component are also independently

changeable, and where (iv) the representation allows us to see how, in vir-

tue of (i), (ii) and (iii), the overall output of the mechanism will vary under

manipulation of the input to each component and changes in the compo-

nents themselves.45

The parts of a mechanism that are described as required by necessary con-

ditions (i), (ii), and (iii) lie in what I call the structure (or the riverbed). Condition

(iv) reflects what I just said about a hypothetical aspect of a mechanism. It is at

(iv) that the truth of counterfactuals matters. A description of the causal mecha-

nism that ‘see’ if we look at the river is true of a set of hypothetical worlds, which

only differ in the initial conditions, viz. the water supply and the other circum-

stances. This set contains the possible particular currents given the riverbed and

gravitational pull. It is delimited by the physical structure of the riverbed. If the

description is true simpliciter, the actual world is an element of this set.

PROCESS

Actual flow of the river

given the riverbed, gravitational pull,

water supply, other circumstances

MECHANISM Possible flows of the river

given the riverbed and gravitational pull

STRUCTURE Structural characteristics of the riverbed

table V.1 Structure, mechanism, and process in geology

The structure, finally, is given by the characteristics of the riverbed itself.

The scheme above represents the core idea. Another structural make-up of the

riverbed will give rise to another set of possible worlds, i.e. of other possible

flows of water. In every world in such a set this particular riverbed exists; this

structural make-up is necessary insofar as we only consider this set. LaPorte

would say that, if we give this riverbed a name – let us call it ‘Ebro-1’ – then

‘Ebro-1’ rigidly designates this riverbed. Small wonder: we termed it ‘Ebro-1’, so

this name designated rigidly de jure.

45 Woodward (2002), p.S375.

Chapter V

5.2 The idea of an hypothetical causal mechanism in Austrian economics

In Austrian economics the structure, in which a causal mechanism is rooted, is a

social kind. It gives rise to a restricted set of possible causal processes. Therefore,

this set comprises the possible worlds where representations of the mechanism

are true. This means that there are (allegedly) true counterfactuals about the be-

haviours that the mechanism could give rise to given the structure. The set of

worlds alluded to, the possible or hypothetical behaviours, or outputs, is delim-

ited by the structure.

Note that Woodward’s account evades the term ‘causal’. He chooses to

keep his account metaphysically thin (although, as Lynch would say, his account

is not metaphysically invisible). Furthermore, he makes use of a particular type of

counterfactuals, those related to interventions only. This is the consequence of his

invariance-under-intervention account of causes. But although social mechanisms

are ultimately brought about by intentional human action, many such mechanisms

are in itself the unintended result of the (intentionally motivated) interventions by

human agents. Therefore the mechanisms I have in mind – those that Menger and

Böhm-Bawerk tried to explain by reference to social kinds – are to be discovered

rather than to be intervened in. No matter the world of individual choice making

underlying social aggregates, social mechanisms are much like mechanisms in

nature in the respect that they appear to exist independently from the individual

observer’s intervening powers. Thus, I reformulate Woodward’s clause (iv) as:

‘the representation allows us to see how, in virtue of (i), (ii) and (iii), the overall

output of the mechanism will vary under changes in the input to each component

and changes in the components themselves’.

The hypothetical aspect of a description of a mechanism, according to

Woodward, is invariance (under interventions) of a generalization. This generali-

zation describes the behaviour of a component of this mechanism. It is the very

same hypothetical aspect that makes a description of a causal mechanism refer to

a set of worlds. A mechanism in Woodward’s sense is instantiated also by the

structural elements of the riverbed and gravity. Mechanism and structure con-

flate in his necessary conditions to accept a descriptive model to be one of a

mechanism. But I want to differentiate between the riverbed itself (that I wish to

call the structure) and the thing a riverbed can bring about if there is a particular

supply of water and other circumstances. What it can bring about is the extension

of the description of the mechanism.

So what makes the mechanism causal? Among other things, the fact that

the output of each component forms the input of the next and the independent

changeability of the interacting components. This independent changeability of

Woodward’s mechanism seems to be an attractive notion for the problems dis-

cussed in this thesis. Austrian economists have been trying to abstract mecha-

nisms from social reality. These were Invisible Hand mechanisms and they are

independently changeable, at least in principle. Or are they?

How come these are independently changeable in principle? Think of the

modular organisation of the mechanisms. First there is the decision making, for

instance by a potential buyer, on the basis of preferences. Next there is the infor-

mation gathering on the market, together with the deliberations about budget

constraints. Another component is the confrontation with potential sellers. Under

some well specified market conditions, marginal agents will be singled out such

that the quantities demanded and supplied equalise. Both changes in the ordered

set of preferences (changes in the input), and cognitive irrationality on the side of

the buyers (changes of the component itself) affect the number and identity of the

marginal agents and the market price (the overall output). Any change in input of

the structure or in a component itself of that structure will give rise to another

possible outcome. The mechanism is given by the set of possible outcomes, while

this set is constrained by the structure.

It is not always clear whether the change of a component leaves the struc-

ture intact (whether it will still be the same Invisible Hand structure) or will

change the given structure into another one. The riverbed metaphor comes in to

explain this too. Over time, the current of the river changes the banks. After a

long enough period of time, one has reason to wonder whether it can be identi-

fied as the same river. But the fact that structures change does not imply that

these do not deserve to be termed ‘structure’. Just like the fact that changes in the

referents of kinds do not deny their kind status.

So the main point I want to make remains unaffected: we can see how,

within one structured institutional environment, human agents have a restricted

(though often large) set of options to choose from for action. So their appreciation

for the metaphysical character of social kinds allowed Austrian economists to

understand how the organisation of a world of acting individuals constrains the

available options for the individuals and the number of social results of their

choice. Reference to economic Invisible Hand mechanisms – as giving rise to one

particular social process if the initial conditions are there – expresses an apprecia-

tion of the metaphysics involved in such mechanisms, because it acknowledges the

way the world works as a constraint on the choice of social kinds in theorising.

5.3 Kind terms for social structures

Essentialism, not infallibilism

Woodward doubts that it is possible to develop models of psychological mecha-

nisms that satisfy the condition that the behaviour of the components of the

mechanism can be interfered with ‘without necessarily interfering with the be-

haviour of others’46. Only if a system meets this condition can it be modular. If no

46 Woodward (2002), p.S374.

Chapter V

such modular systems can be modelled in psychology, i.e. when there is a lack of

decomposability, then there are no genuine psychological mechanisms. Wood-

ward’s complaint is that psychologists do rarely deliver convincing evidence that

their ‘boxes’, describing the operation of psychological mechanisms, satisfy the

conditions (i) to (iv). This gives these conditions ‘a real normative bite’.47 I can see

that Woodward’s conditions for fruitfully talking of mechanisms make sense.

If they do make sense, are there any genuine economic mechanisms? Does

Woodward’s proposal, if justified, harm the idea of social kinds referring to struc-

tures, like the Invisible Hand mechanism, which in turn steer social mechanisms

and actual causal processes as their actual output? I think not. I have suggested

that there is decomposability in the different modules of what Menger and

Böhm-Bawerk called the ‘market mechanism’. It is possible, for instance, to single

out the decision making process, the information gathering, and the meeting of

market parties. Changes in initial conditions trigger a step-by-step change in the

input and output of each module, eventually leading to a new equilibrium.

Again, the economist may be wrong in his identification of a mechanism,

but it does not follow that no social mechanisms exist. Science is continuously in

construction and social structures may be discovered and rendered kinds to re-

place kinds previously pinned down. Böhm-Bawerk was strongly convinced that

he and Menger had excavated truths about the market the OVT theorists had

failed to see. His rhetoric makes one think he thought of himself as perhaps falli-

ble in principle, but beyond failure in fact. In terms of what I have said above, the

Austrians had chosen to load the minimal and vague concept of the Invisible

Hand with a metaphysically more robust meaning. Thus, the use of a social kind

term implies a lot about the chosen ontology. Other extensions of the minimal

and vague concept must be proposed, giving rise to new ontologies. All propos-

als are expressions of the effort to find out how the world works. Ultimately, the

constraint on theorising lies in the social world itself. Whatever its fate in psy-

chology, the notion of a mechanism seems to be appropriate in economics, as the

example of the Invisible Hand illustrates.

In sum, we have a structure, referred to by a kind term; we have a causal

mechanism, the description of which fixes all the possible courses of the world

insofar as these are governed by the structure; and, finally, we have the phe-

nomenon that is picked out of this set of possibilities as being the actual one, a

‘causal’ process. Its being causal lies in the fact that this process is one actual re-

alization of the mechanism.

Market structures

The present subsection aims to explore how Böhm-Bawerk’s example of ore min-

ing fits with the concepts thus far developed. I shall tell the story entirely from

his viewpoint. Recall that the claim by SVT theorists is that, contrary to many an

47 Ibidem, p.S377.

intuition, if a rise in the productivity of ore mining is followed by a price decrease

of iron consumer goods, the cause of the price change must be sought at the side

of the demanders on the consumer market. The satisfaction given by such goods

will cover more needs, also those with a lower utility rating. The lower subjective

utility determines the price buyers are prepared to pay.

Under strict stability assumptions (almost never met in the actual world)

input and output prices will be equal for a reasonable period of time. In this rare

case, LoC holds, at least as a rule of experience. To the layman’s eye, it seems to

be a causal law, identifying the ultimate cause of this adjustment process at the

side of the productivity of the mine. However, the Law of the Marginal Agents

(LoMA) really does the mechanical job. The market structure is composed of

intentions, ordered preferences, and market forms. It gives rise to LoMA, which

describes a mechanism, referred to by Adam Smith as ‘the Invisible Hand’. The

mechanism at work can give rise to a set of prices at which input and output

prices are equal, one element of which is the actual price that emerges. The actual

process is the coming into being of one input=output price level. Table 2 summa-

rises this.

PROCESS Actual output and input prices equalise according to LoC;

MECHANISM LoMA in the particular market structure;

KIND STRUCTURE Free, perfect, and competitive markets, intentional agents

with decision making capabilities, well ordered preferences;

table V.2 Kind, mechanism, and process in Mengerian economics;

the ideal case turned actual due to stability of initial conditions

Under the assumption of instability, output prices seem to sluggishly ad-

just to input prices at most. Let us suppose this adjustment does take place. Care-

ful scrutiny is needed to see a tendency emerge in the course of time, but the OVT

theorists like Ricardo noticed it. They saw LoC at work. Again, LoMA really is at

work. See table 3.

PROCESS Actual output and input prices converge with sluggishness;

KIND STRUCTURE Free, perfect, and competitive markets, intentional agents

with decision making capabilities, well ordered preferences;

the case of instability of initial conditions

Chapter V

Under the assumption of imperfection things are different. This concerns

another kind. Take the example of monopoly. Price setting prevents from show-

ing any tendency toward input-output price equalisation. One or more monopo-

lists abuse their market power to calculate a mark-up and boost profits. In any

possible world with this structure, input-output price differentials will endure.

Table 4 below summarises the idea.

PROCESS Output-input price differential equals a particular mark-up;

KIND STRUCTURE Monopolistic markets, intentional agents with decision making

capabilities, well ordered preferences;

the case of monopoly and stability

6 Conclusion

Radder terms the property of abstract concepts, that these can be used in other

observation processes than the one that gave rise to the concepts, the ‘nonlocality’

of concepts. The very contingency of whether a new observation process with the

same result is going to take place grounds Radder’s anti-essentialism. Another

ground is the Kantian part of his philosophy: the structuring function that con-

cepts have.

Nevertheless, economists may believe that their explanatory concepts are

abstracted from the world, rather than that these structure it. Austrian theorists

had a rather strong essentialist view of their science. Mäki has noted that they

redescribed a more phenomenal description of the social world in terms of

mechanisms and ‘as really being something else’. This may be interpreted as im-

plying that the object of redescription is supposed to have an essence, or a social

kind and that this essence is abstracted from all information which is available

about market phenomena. Thus, Menger identified hidden or underlying objects

as ‘exact types’: abstractions of phenomena, stripped of most of their rich empiri-

cal attributes.

Jessica Brown claims that 1) a direct access to more than these appearances

is not a necessary condition for our recognitional capacities for fundamental

kinds; 2) neither is infallibility of these recognitional capacities; 3) the only neces-

sary condition is that we have an appreciation for the metaphysical character of

the world around us. In line with this epistemology, economists can also be un-

derstood to have an appreciation for the metaphysical character of the social. We

understand that a supplier in a very competitive market cannot make profit, ce-

teris paribus, without entering a market niche. This illustrates our recognitional

capacity that induces us to specify social kinds.

Often, essentialism is taken to be untenable, for essences tend to be associ-

ated with (1) relatively enduring (non-historical) properties of objects, such as

molecular make-ups, and (2) with a determinate reference of kind terms such that

vagueness is excluded. However, this need not be the case.

From Joseph LaPorte we learned that even if the molecular structure of wa-

ter as such is a discovery, it has not been discovered that water designates rigidly.

If a community of speakers make the term ‘water’ refer to water, and the term

‘H2O’ refer to H2O, then in all possible worlds where these terms refer, they refer

necessarily. In other words, if these terms are no rigid designators de facto, they

are de jure, in the same way as names are rigid de jure. ‘Water=H2O’ is necessary

if true at all. The fact that water=H2O was discovered to be a necessary truth. But

it is contingent that the substance that we refer to with the term ‘water’ happens

to have the particular molecular make-up it has’.

Furthermore, LaPorte notes that a kind term refers to an abstract kind. But

all this does not allow for a specification of just any kind term and counting it as

rigid: there is a difference between rigidity and non-rigidity. ‘The largest dino-

saur on a 1989 US post stamp’ refers to Brontosaurus, but not rigidly, because the

individual is now picked out by a particular accidental description instead of by

causal baptism. As indicated in previous sections, the relevant distinction is not

between natural kind terms and artificial kind terms with respect to their rigidity,

but between on the one hand rigid and on the other hand non-rigid natural kind

terms and, likewise, between rigid and non-rigid artificial kind terms.

Vagueness, like necessary truths, can be discovered and if it is, there is scien-

tific progress. Thus, Böhm-Bawerk discovered that the concepts of ‘value’, of

‘market’, and of ‘borrowing’, for instance, were vague. In trying to explain the

workings of the world, scientists need metaphysically loaded and robust concep-

tualisations of this world. They do so by specifying kind terms. And as stated, in

all possible worlds where the meaning of the term is as it actually is, the referent

of the term possesses all the properties associated with the term. But, contrary to

what is often assumed, vagueness of a kind term shall be a ubiquitous phenome-

non. Yet, well chosen kind terms are explanatory interesting.

This light-weight approach to essentialism makes it palatable and helps see

that it makes sense to talk of social kinds. It makes essentialism compatible with

pluralism. For instance, if vagueness creates a rift between paradigms, at the

same time it provides the common ground for communication between opposing

theorists. Pluralism can be compatible with the use of kind terms in trying to

understand the world.

Pluralism, according to Lynch’, says that the world does not dictate one

unique correct conceptualisation of itself. No truths exist outside conceptual

schemes, or outside the scheme-indices that come with propositions. We nor-

mally extend the use of minimal concepts from one paradigmatic example, in

Chapter V

accordance with the Causal Theory of Meaning, into metaphysically more loaded

directions as we commit ourselves to a particular ontology. It is in the loaded

versions that we specify essences. We can step outside the scheme of our imme-

diate community while not outside of all schemes. As discussed by Joseph La-

Porte, the creationist Hopkins could have done so with regard to Darwin’s new

theory.

So how does the market mechanism relate to social kinds? A social kind

term refers to an underlying structure that exists in economic reality. Thus, the

use of a social kind term implies a rather robust ontology.

The key kind is an assemblage of individuals’ capability to order their

preferences and engage in marginal calculation, and the communication between

input and output markets in which prices serve as ‘information carriers’. In a

particular institutional environment such a structure is a relatively stable object

(much like a riverbed in geology). The social structure delimits the set of hypo-

thetical processes. (The riverbed sets a limit to all the possible flows of river water

from times of relative drought to times of winter inundation.) SVT allows for

many (but not just any) prices for the same input goods and the related end

goods, depending on initial conditions. The actual causal process, finally, is just

one of the possible processes given the mechanism and given the market struc-

ture. (It is the way in which the river actually flows, given the initial conditions as

these apply, gravity, and the riverbed.) In sum, we have a structure, referred to

by a kind term; we have the modular make-up of a causal mechanism, the de-

scription of which has a set of possible states of the world as models; and, finally,

we have the succession of phenomena that is picked out of the set of possibilities

as being the actual one that we call the causal process.

Conclusion

The first question I asked in the introduction was what it is that makes the Austrians

so different from economists for whom equilibrium is a key notion as well. The

answer is that Austrian analysis, so vigorously defended by Böhm-Bawerk, opens

the black box hidden under the Invisible Hand. I suppose that Adam Smith had no

explanatory pretensions when he used this mystical Hand but merely used it as a

manner of speech. But an Austrian elucidation of mechanisms talks of intentional

individual behaviour, which translates into unforeseen market outcomes. The mar-

ket structure gives rise to a mechanism and this mechanism, given initial condi-

tions, to market processes. Both this mechanism and the socio-economic market

structure it depends on lie hidden and Austrian analysis digs deep to uncover

these: Smith’s Hand made visible.

The second question was how the ubiquitous use of idealizations in economics

can be squared with the quest for truth. I did not seek the answer in a description

of the discipline as a non-truth seeking enterprise. I defended my conviction that

many economists, regardless their occasional messages to the contrary, do seek

truth in theorising. In addition, and stronger, I suggested that they often tend to be

tacit essentialists. The answer to the second question is that falsity is the disguise in

which truth enters. The falsity an idealizational clause is involved with is inserted

in the antecedent of a possibly true counterfactual conditional and it is this falsity,

which makes the entire subjunctive proposition a counterfactual. Often economists

do however not idealise but abstract; they do not hedge their theories by well

formed clauses but they just ignore some causally relevant aspects of the world. In

this case there is no clear indication as to the precise circumstances or causes that

are counterfactually being made provisos about. Only in retrospect – after more of

the subject is understood – a past abstractive theoretical endeavour can sometimes

be reconstructed as the use of a particular clause.

The accusation of theorising being ‘too abstract’, habitually uttered by policy

makers, is often premature. Judgements of this sort must be placed in the context

of the question whether it is ante explicationem or post explicationem abstraction

that one wants to attack. Economists, in their turn, must take care not to disregard,

by abstraction, the crucial aspects from the explanandum without paying attention

to what other social disciplines have to offer. I return to this plea in the last para-

graph of this conclusion.

The third question asked what part of the social world it is that economists fo-

cus on when they abstract it from all this spatio-temporal detail in which it is em-

bedded. The answer seems trivial: whatever they believe is important for explana-

tory purposes. It is however not trivial that economists describe an explanandum

such that they can gather the ingredients for abstract explanatory social kinds. In

the act of abstraction and concept formation, they use pre-theoretic notions to

structure the explanandum but, once they have strengthened their ontological

commitments to more robust world views, they tend to behave like essentialists.

Conclusion

The fourth question was how the idea of abstraction as ‘summarising’ copes

with the analytical problems essentialism tends to run into. Other philosophers

have shown that there can be a degree of essentialism sufficiently weak to buttress

a social scientific epistemology in accordance with natural kinds as explanatory

tools. That is why I speak of ‘social kinds’ and I suggest that Austrian analysis

finds these in those structural aspects of economic reality that give rise to mecha-

nisms. Böhm-Bawerk explicated Menger’s market mechanism in terms of the Law

of the Marginal Agents. The modularity that one expects to find in a genuine

mechanism is a property of the mechanism-cum-structure. The mechanistic part is

pronounced by its hypothetical character as expressed by the idea of Jim Wood-

ward’s ‘invariance under interventions’.

I also say that a ‘perspectivist essentialist’ epistemology is thus conceivable.

It is now expedient to refer to Kuipers’ scheme of the four epistemological ques-

tions, as summed up in the intermezzo. Recall from my intermezzo that one can

ask, firstly, whether scientific propositions have truth value and, secondly,

whether, if so, truth can also be a property of propositions with respect to unob-

servables. The third question one can ask is whether, apart from mere entities, en-

tire structures as posited in true theories really exist. Is it fruitful to engage in

claims to knowledge about the (natural, social) world even beyond the mere refer-

ence of theoretical terms? If so, there is reason to adhere to realism of an impor-

tantly stronger sort than referential (or entity) realism.1 In Kuipers’ scheme, two

more degrees of freedom exist if one endorses this, given the fact that realists (and

both structural and referential realists) are faced with a fourth question: is there

one best conceptualization of theories? Kuipers’ own preferred option is to answer

this question negatively. This is what he calls constructive realism. This is the view

that although one type of vocabulary can be more useful than another, there need

not be one unique vocabulary that carves the world at its joints better than any

other. But essentialists precisely claim that there is one best conceptualization.

I propose a further refinement. Perspectivist essentialism can be incorporated

in Kuipers’ scheme as providing yet an extra option. Perspectivist essentialism is

different from constructive realism in the belief that scientists must look for kinds

because of their explanatory value and in that these kinds dictate the conceptuali-

sation. It is however also different from full blown essentialism in that it accepts

that there are minimal and robust versions of essentialism, such that any essential-

ist inclination in science is a matter of rigid designation merely de jure. This view

allows that one particular perspective on the world loads the meaning of concepts

with one set of ontological presuppositions, while another perspective loads it with

another.

One may object that perspectivist essentialism has no teeth. Its being differ-

ent both from constructive realism and old fashioned essentialism shows that it

has. There are reasons of descriptive adequacy for inserting this refinement. Phi-

1 I pass over the complicating possibility of structural realism here, i.e. the view that terms do not refer

but that laws nevertheless have truth value.

losophy of science is more than ornithology (‘scientists lay their eggs anyway’) but

its normative bite is toothless if it bears little relation to what scientists (econo-

mists) in fact are doing. My emphasis in philosophy of science is therefore that a

quest for normative adequacy must start with descriptive adequacy. Since I believe

scientists to act as if they are essentialists of some sort, the question in chapter five

was how to normatively defend this. The concept of perspectivist essentialism does

the job.

Rests the scope for further research.

The need for a scrutiny of what further kinds scientists propose in their ex-

planations stands out. I have here provided little more than a hunch of what a so-

cial kind is in Böhm-Bawerk’s theorising. Other examples have to be looked for in

the work of any economists who complain that some essential explanatory aspect

of the world has not been taken into account by existing analyses. In appendix 3, I

discuss Uskali Mäki’s favourite cases of George Richardson, Ronald Coase, and

James Buchanan. All three of these stress the importance of ‘a missing link’ in eco-

nomic theorising, all three engage in the intuition that particular explananda must

be endogenised. (I refer to this appendix for the respective links that these econo-

mists tell us are missing.) Quite another proposal (albeit without the notion of per-

spectivism) I know of is to give emotions a natural kind status in psychology.2 We

have seen that the possibilities are infinite but that we have to guard the explana-

tory value of the whole enterprise. I have not discussed what counts as explanatory

and what not, so there is work to be done here too (not least about the question

what it is that makes an explanans explanatory).

The notion of abstraction as defended in this thesis can benefit from a fur-

ther refinement. In my treatment abstraction comes with generalization, as it leads

to logical weakness and an increased chance of truth; so concretization, conversely,

comes with specification. But some very abstract and explanatory propositions of

science – like Newton’s laws of mechanics – are in fact strong. Or at least, the meta-

statement that a very precise pronouncement is true in many cases is stronger than

that it is instead true in only a few cases. So why are the best theories the most

abstract and the strongest at the same time? I am convinced that this has to do with

their unifying power. The role of abstraction, then, in theoretical unification spe-

cifically has to be investigated. To allow myself a hint to reflexivity, the concept of

‘abstraction’ can be concretised further.

This brings me to an issue that I most dearly wish to examine. Contrary to

Henk Folmer’s recommendation3, there is still little communication between the

social sciences. But worse, there is little communication even between the respec-

tive research programmes both within psychology and within economics. As to

psychology, this seems to be due to the lack of conceptual unification that one notices

when talking to, for example, neuroscientists, cognitive psychologists, and psycho-

analysts. In the case of economics, in contrast, there is little communication for

2 See Charland (2002). 3 See Folmer (2000).

Conclusion

example between institutional economics, the neoclassical research programme,

game theory, and experimental economics. This may be the consequence of a

strong orthodoxy in the science, one making use of a well unified conceptual appara-

tus. If this is true, then both a low level and a high level of conceptual unification

causes separate paradigms between which there is little interaction. If it is indeed

true that both conceptual unity and disunity render excommunication within the

two disciplines, the thought of an approach directed between these extremes is

exciting. For instance, I would like to find out whether it makes sense to inspect the

conditions for an optimum level of unification in each of the two sciences. This is

especially interesting given the recent developments in experimental economics,

for the experiments with real people are psychological in kind but they bear on our

ideas about individual economic behaviour.

Appendices

1 Saving and dissaving in Böhm-Bawerk’s capital theory 2 Capital gains of productive goods - a refinement 3 On essentialism. Two competing conceptions

Appendix 1

Dissaving in Böhm-Bawerk’s capital theory

Consider some capital structure and a possible change in that structure. Böhm-Bawerk considers economies of which the population wishes to consume less, and also those of which the population wants to consume more than the 10 million years of available labour we assumed in chapter I.

For instance, he takes the case when consumers demand 12 million labour years worth of goods under the given circumstances of a current labour supply of 10 million years. This obviously means that the economy will face capital contrac-tion, because overconsumption implies dissaving. The table shows the initial or-ganisation of capital under step 0.

million years of labour STEP 0 STEP 1 STEP 2 Replace Amassed Replace Amassed Replace Amassed

Maturity class t=−1 t=0 t=1 t=2 t=3 t=4

1 1 6 1 7 0 4 2 1 5 1 4 1 5 3 0.5 4 0.5 4 0.5 4 4 0.5 3.5 0.5 3.5 0.5 3.5 5 0.5 3 0.5 3 0.5 3 6 0.5 2.5 0.5 2.5 0.5 2.5 7 0.3 2 0.3 2 0.3 2 8 0.4 1.7 0.4 1.7 0.4 1.7 9 0.3 1.3 0.3 1.3 0.3 1.3 10 1 1 1 1 1 1

Cap. goods 6 30 6 30 5 28 Cons. goods 4 10 4* 11 5* 12 Total goods 10 40 10 41 10 40

Table app.1 Capital structure and dissavings

Each step has been split up in a t=i and a t=i+1, i.e. a moment of shifting accumu-lated labour years into a new maturity class and into replacement of these labour years, and a moment representing the result of that process, respectively. The ma-turity class number 10 is the capital that has just been brought into existence by investment. Maturity class number 9 is the capital that has been invested the year before and for which it takes another nine years before the accumulation of labour enters consumer goods. Assume the economy at t=0, has 10 million labour years to supply, 4 million of which are used for the consumer goods industry. The second column shows how the replacement of each segment of capital, at t=−1, has been

Appendices

distributed. It takes one million labour years to invest in the crudest and most re-cent segment, 0.3 labour years to add value to this segment and move it one period closer to maturity, etcetera. Total replacement adds up to 6 million years. This is also the amount of labour years that has been accumulated in the oldest segment (t=0, maturity class 1). The total capital stock is the aggregate of the ten maturity classes, each with a higher level of accumulation: 30 million labour years. Note that the population consumes a production that embodies a value of 10 million labour years (t=0, consumer goods). This means that social production equals social in-come because the labour force also comprises 10 million years. If this were to go on without fluctuations, size and structure of capital would be kept up just as it is for all steps t>0

Steps 1 and 2 show the alternative possibility for t>0, of overconsumption. This type of dissaving is conceived in two steps as indicated by the shaded area in the table.1 In step 1, at t=1, capital goods will be subtracted from maturity class 2 and moved toward maturity class 1. To vary on one of the examples discussed above, sowing-seed is made brandy of instead of sown to grow new cereal.2 The accumulated total of capital goods at t=2 remains the same, because the equivalent of what is added to class 1 has been subtracted from class 2. The consumer goods industry now has at its disposal inputs worth 7 million of labour years. The 4 mil-lion labour years of value is added to these inputs, thus raising their finished product value to 11 million labour years. Without any extra replacement invest-ment to keep up the oldest segment’s value of 7 million labour years, i.e. without changing the value of 1 million into 2 million in maturity class 1 at t=3, the accumu-lated value of capital in that class will fall back to 5 million. That is 1 million less than the stable state of step 0. The value of the total of segments of capital will fall to 29 million. But, according to the thought experiment, the economy needed 2 million extra for consumption. Therefore, in step 2, an additional 1 million of la-bour years will be subtracted from its allocation to the replacement of the oldest segment of capital. In other words, no adding of value takes place, at t=3, to the segment that belongs to maturity class 2. The labour set free by this process can be used to raise social income. In the table, this is shown by the figure 5* (as opposed to 4* under t=1) in the consumer goods rows. The end of the story is that these 5 million labour years of value will be added to the 7 million labour years worth of mature capital goods, so as to convert these into consumer goods. This will enable the population to consume an income of 12 million labour years, while only 10 million years of work is done.

1 Böhm-Bawerk proposes to consider my step 2 first and step 1 afterwards. The above treated order

turned out to ease the representation. No aspects of the theory are violated by my representation. 2 …mit solchen Kapitalgütern, welche eine mehrfache Verwendung zulassen, vielleicht noch den Ertrag einer […]

Million von Arbeitsjahren aus höheren in die erste Reifeklasse dirigieren… PTK p.150. Böhm-Bawerk uses the term Ertrag and Erträgnis in this context so as to refer to the stock of capital itself, and not for what it means in modern business economics, viz. ‘revenue’.

Appendix 2

Capital gains of productive goods - a refinement

The value of durable goods is equal to the sum of the services it offers to the owner. However, these services wear out over time, so that a discount is needed to set the present value. In case the durable good in question is fixed capital, like a machine, the renditions of service are twice dissociated from their role in the satis-faction of final needs. After machines have paid their service in production, the finished good or semimanufacture goes on to mature further.

Suppose, as in chapters I and II, the rate of discount is 5%, the capital good of which we want to know the value bears fruit (viz. another good) over six years, and that good takes another two years to mature. With this setting we follow Böhm-Bawerk’s own numerical example, but develop it in a spreadsheet. Dis-counting by the formula 100 ⁄(1+0.05)2, which covers the assumed two years of ex-tra maturity the owner of the original capital has to wait for, gives 90.70. The inte-grated interval in which the final produce of the capital goods (the renditions of service born by the product brought forth) reach their mature status is thus 8 years. Therefore the depreciation in the first year is equal to 100 ⁄(1+0.05)8 = 67.68, and the total discounted value of the durable capital goods at the start of the first year equals 460.38. As semimanufactured goods, the value of the produce matures fur-ther outside the reach of the entrepreneur’s capital, to 100 in steps via 90.70 and 95.24. The increase in value, then, develops in steps of 4.54 and 4.76, respectively.

Net yield from durable goods; evenly distributed gross yields; 5% discount; (1) (2) (3)

Year 1st Jan 31st Dec (4)

Gross yield (5) = (2) – (3) Depreciation

(6) = (4) – (5) Net yield

1 460.38 392.70 90.70 67.68 23.02 2 392.70 321.63 90.70 71.07 19.63 3 321.63 247.01 90.70 74.62 16.08 4 247.01 168.65 90.70 78.35 12.35 5 168.65 86.38 90.70 82.27 8.43 6 86.38 0 90.70 86.38 4.32

7 90.70 95.24 90.70 8 95.24 100

(value of

produce) 95.24

TOTAL YIELD: 544.20 83.84 Table app.2 Yield of capital goods

It follows that durable capital goods, as distinguished from durable goods in gen-eral, give rise to interest due to two causes. Produce to be rendered in the future bears interest originating in the mere current of time involved in the exhaustion of

Appendices

the capital good. In chapter I we call this ‘the growth of value of future goods into presentness’.

For instance, produce already delivered bears interest due to the further de-velopment of semimanufactured into finished products, after it has left the owner of the capital goods under consideration. The former only must formally be im-puted to durable capital, for the entrepreneur closes the account at the moment when the produce is separated from the machines.

Appendix 3

On essentialism. Two competing conceptions

Essentialism entails the belief that good science, in order to explain, picks out the joints of the world. So, in the realm of the economic, at which level of aggregation should one look for them? Two possibilities present themselves. One is the level of the individual with his intentions, the other is the level of social wholes. Austrian economics typically is methodologically individualist (MI). One of the theses of chapter II is that Böhm-Bawerk’s MI was essentialist. Uskali Mäki believes that this can be said of Menger, Mises, and Hayek as well. But this notion, viz. that of Aus-trian microreductionist essentialism, has been challenged. Below I shall defend Mäki and discuss different possible ways to express both reductionism and holism, ordered along scales from weak to strong. I first treat Mäki’s analysis of the Aus-trian theory of entrepreneurship as a causal process theory. Next, I shall briefly outline what his account of essentialism in general means for isolative reasoning in economics. Thirdly, I shall treat the concept of explanation by redescription as Mäki has developed this with regard to Austrian economics. All this is helpful to understand the way in which wholes and individuals relate in Austrian economics. The fourth subsection considers the question whether social wholes are in any sense real. The fifth subsection pays attention to Karl Milford’s alternative view of Austrian economics. He says that methodological individualism must be inter-preted as anti-essentialist, and German historicist analysis of social wholes in turn as essentialist. Next, in the sixth subsection, I shall look into some possible degrees of reductionism, which help qualify Milford’s notions. I shall endorse Harold Kin-caid’s warning, in subsection seven, that ontology and epistemology should not but are often conflated in issues about holism and reductionism. It is claimed that Milford falls victim to this conflation and that, hence, he is mistaken.

1 Mäki on causal process theory

Austrian economic theories stand out as a causal process theories. In order to com-pare such theories with equilibrium ‘model theories’, Uskali Mäki uses two con-cepts of causation developed by Wesley Salmon to make the notion of ‘causal process’ precise.3 Salmon distinguished production from propagation. Causal pro-duction concerns ‘bringing about effects’. Causal propagation is ‘transmitting in-fluences’ from one point (in time, in place) to another. These two concepts of causa-tion are also covered by the terms ‘agency’ and ‘transmission’, respectively.

Agency in the Austrian theory of entrepreneurship, for example, involves (1) the capability of acting, (2) the property of being alert, (3) the presence of incen-

3 Mäki (1992a), p.40.

Appendices

tives, (4) ambition or the intrinsic nature to act in a certain way.4 Perhaps alertness is the most typical for entrepreneurs. True, if agents engaged in the market process had been machine-like perfect decision makers, they would not need to be alert. They also would not make any error (by implication), and hence, not learn.

Transmission, in turn, is an important notion in the Austrian conceptualiza-tion of the market, because the praxeological approach to the domain of the social, as was advocated by von Mises and Hayek, involves the transmission of informa-tion. ‘Disequilibrium market prices are vehicles of conveying information across time and

space’ says Mäki.5 Indeed, if the only existing prices would always present equilib-ria, no incentive for action could exist on either side of the market. By any interpre-tation of the notion of ‘perfect information’, if agents can dispose of it there is no point in conveying information to them. We have to understand that error is a necessary condition for learning. And a market concept that allows error together with the four conditions for agency just mentioned provide sufficient conditions for learning. For sure, a ‘process theory’ – as opposed to a static or ‘model theory’ – can help understand such learning processes. In Mäki (2001), it is discussed how less than perfect circumstances are essentially those that make possible human action; and I add that it is human action that a praxeological, typically Austrian conception of social research takes to be the starting point for explanations of social phenomena.

2 Mäki on isolations and essentialism

In his (2001) Mäki studies the criticism of current mainstream economics by George Richardson, Ronald Coase, and James Buchanan. He interprets claims by these economists concerning a supposed missing link in mainstream economic theory as missing the essence of good explanations. Richardson, for instance, complains that by taking perfection of markets as an ideal type there can be no analysis of the transmission of information. Coase, in his turn, wants to see an endogenous treat-ment of the costs of making transactions. Buchanan, finally, objects to the exogene-ity of behavioural patterns in equilibrium theory. The Austrian conception of the market process precisely provides for the demanded endogeneity. So Mäki asks:

Such grounds [for complaining] are far from intuitively obvious given that there are highly respected scientific theories that study planets without exten-sion, planes without friction, and molecules without color. Why are some ex-clusions suspicious while some others are not?6

The answer Mäki gives is that there is an ontological constraint on theorizing and this constraint is given by the way the (economic) world actually is:

4 Ibidem, p.42. 5 Ibidem, p.43. 6 Mäki (2001), pp.380-381.

The idea is not […] that these “imperfections” are real and causally influen-tial, and therefore should be included […]. The idea is rather the stronger one that the imperfections play a necessary or essential role in the working of the world7

There are two philosophical positions involved in this way of looking at the mar-ket. One is a version of ontological realism and the other is a version of epistemo-logical realism.8

The first position, ontic realism, says that the world has a characteristic way of functioning. Thus, ‘there are ontic necessities and impossibilities in the world’, and this is true ‘independently of what we believe of it or how we represent it’. The second position, theoretical realism, says that ‘good theories are purportedly true descriptions of the way the world works’.

Richardson, Coase, and Buchanan in fact distinguish well between isolative assumptions that do and isolative assumptions that do not bar the essentials of the system the character of which economists claim to explain. This is quite apart from the question whether they are justified in abstracting these alleged essentials. The interpretation of their views as to what counts as a fruitful explanation in econom-ics as essentialist does not rule out infallibilism. It is possible to believe that picking out essentials, or kinds, or necessities de re9 from the economic realm is feasible and worth while and at the same time that one may be dead wrong in the actual picks.

3 Mäki on scientific theorising as redescribing social phenomena with microphenomena

The self-image of Austrian theory – and the self-image of Böhm-Bawerk specifi-cally – is that it is a scientific theory. Science is different from common sense knowl-edge. Mäki writes:

Let us refine [my] interpretation of Austrian theory by employing the idea of theoretical redescription. Theoretical redescription is a matter of redescribing, in theoretical terms, what is already empirically or “phenomenologically” de-scribed, as really being something else – this something else constituting the “essence” of the object of (re)description.10 It is only by means of the conceptual resources of a theory – not being reduci-ble to the observational language of empirical facts and generalizations – that empirical facts can be redescribed in a way which reveals what those facts really are.11

Austrian theory is a theory of action, which (re)describes how social phenomena come about as a result of individual causal agency (causal production), and the

7 Ibidem, p.383. My italics. 8 Ibidem, p.385. 9 This is how Mäki labels it. In chapter V a tenable form of essentialism is analysed as the proposition of

rigidities de jure. 10 Mäki (1992a), p.44. The emphasis is Mäki’s. 11 Mäki (1990), p.321. It is not clear why we have to assume that nothing of the new conceptual appara-

tus, of the redescription, can be reduced observational language.

Appendices

market is the typical institution where we can see instantiations of this agency. Austrian theory also (re)describes the market as the institution where we can find forms of the propagation of information (causal propagation). These (re)descriptions make use of a conceptual apparatus not reducible to observational language.12

Let me explain Mäki’s point. The (re)descriptions are explanations. The ex-planandum of the (re)descriptions are the phenomena known by using our com-mon sense ways of observing. Mäki coins the initial description of the explanan-dum the ‘phenomenological description’.13 The so called ‘phenomenological‘ de-scriptions (may) involve wholes, like the entire market, objective market prices14, social capital, etcetera. The explanans, in turn, is the ‘deeper’ story. It demarcates Austrian economics as ‘scientific’: it tells what the objects of scientific (re)description really are. As I see it, such explanations of the market make Smith’s ‘invisible hand’ visible. Moreover, the explanans is methodologically individualist: the scientific explanations start at the level of individual agency and explicate how, by some causal mechanism, phenomena at this level generate phenomena at a higher, non-individualist level. The redescriptions are microreductions.

4 Do social wholes exist too?

The question is, next, how Austrian realism can be understood: does it involve the claim that the referents of the explanans really exist or also that the referents of the explanandum really exist? The first is certainly correct. Mäki says:

[M]any fundamental objects of economic theory are claimed to be subjective in nature; they are “made of subjective stuff” […] those subjective entities are maintained to have an objective existence, they exist independently of econo-mists’ theories of them.15 So this is ontological realism with respect to the micro-entities. As to the sec-

ond, the claim that the object of common sense description – the social explanan-dum – is real, the twentieth century Austrian praxeologist Ludwig von Mises is agnostic.

There is no need to argue whether a collective is the sum resulting from the addition of its elements or more, whether it is a being sui generis, and whether it is reasonable or not to speak of its will, plans, aims, and actions

12 Note that the term ‘reducible’ here does not rest on the concept of reduction as specifically microre-

duction. The common sense terms (or observation terms) referred to are terms having a meaning only at the ontologically relevant level of the social: that is, a level higher than the level to which methodological individualism tries to reduce social phenomena.

13 Mäki (1990), p.320. Phenomenological descriptions form what Böhm-Bawerk called Kasuistik. 14 The term ‘objective’ used here has nothing to do with how it is used in the phrasing, which serves to

denote aspects of essentialism, like ‘objective existence’. We have seen that in PTK objective value is the market price coming into existence by trade. It is equal to the marginal utility of the marginal buying agent.

15 Mäki (1990), p.336.

and to attribute to it a distinct “soul”. Such pedantic talk is idle. A collective whole is a particular aspect of the actions of various individuals and as such a real thing determining the course of events.16

While the microreduction carried out in praxeological social theory postu-lates the referent of the explanans as objectively existing, the existence of the refer-ent of the explanandum is an irrelevant matter for his purposes. Anyway, the enti-ties and properties at the individual level are supposed to really exist. These are referred to in methodologically individualist theories, hence in micro-reductive explanations of social phenomena.

Mäki sees sufficient cause to view Austrian economic methodology as essen-tialist. I have concluded the same concerning at least Böhm-Bawerk, in chapter II. If we now place the discussion about the objects of redescription of the phenomena against the background of this possible essentialist reading, we complete the pic-ture of what Austrian explanation amounts to.

Individual agents ‘invested with meaning’ must act in explanations that are to be the best descriptions of what socio-economic really is like. This, then, is meth-

odological individualism cum essentialism. I stress that this also nicely fits Böhm-Bawerk’s implicit convictions underlying the rhetorical strategy he employed in theory appraisal.

5 Milford’s anti-reductionism as essentialism

There is, however, an alternative characterisation of what essentialism in social research amounts to. This other interpretation – by Karl Milford – gives an explica-tion of the Austrian project which is diametrically opposed to the view defended by Uskali Mäki. Karl Milford (1989, 1997) says that essentialism in social science implies anti-reductionism or some form of holism. This is remarkable. It runs counter any of the views expressed so far.

The question I shall try to answer is whether the emergent social properties to which initial common sense descriptions and historicist studies refer, holistically described, are indeed required in essentialist explanations. Are holist entities, such as institutions, a necessary condition for essentialism in social theory? I shall argue that Karl Milford is wrong due to a confusion of holism and essentialism.

Milford starts off with a classification of epistemological positions: Überblickt man die methodologischen Positionen zur Frage befriedigender Erklärungen in den theoretischen Sozialwissenschaften, so kann man zwei große Gruppen unterscheiden: die Position des methodologischen Individua-lismus und die Positionen des methodologischen Essentialismus, die manches Mal auch als methodologischer Kollektivismus bezeichnet werden (Popper 1969).17

16 Von Mises (1949) Human Action, p.43. 17 Milford (1997), p.101. The reference to Popper is to Poverty of Historicism.

Appendices

Not only Austrian economic theory is Methodological Individualist (MI), also modern game theory is committed to MI.18 Milford counts the German Historicists as adhering to Methodological Collectivism or Essentialism (ME).

Bruce Caldwell, in his biography of Friedrich Hayek, notes that the single most important protagonist of the younger Historical School, Schmoller, gratefully had taken over the heritage of Wilhelm Roscher in the idea that theorizing about the social could only be premature.19 The German Historicists were above all scep-tical towards the optimistic theoretical endeavour of Menger. Redescription by MI methods was seen as ‘using theory’ and therefore dangerous.20 The so called His-torical Method was holist in the sense that it prescribed the search for aggregate social patterns at the level of observable regularities in the historical development of societies. Is the holism of ME then the same as the holism of the Historical School?

6 Reductionism and holism: strong and weak

Here it is important to distinguish Methodological Holism from Ontological Ho-lism. The first is the prescription, at its weakest, that social science must also look for explanations at the level of institutions or social wholes. The strong version has it that social science must only explain at the level of the social, not at the level of individual agents. In neither of these two versions of holism has anything onto-logical been a priori claimed, that is, about what social reality is like. This is impor-tant. It allows that even if one believed that social entities – or properties – are con-stituted by aggregates of ‘atoms’, one could nevertheless prefer to set as a scientific aim to highlight regularities at the level of the social. One could do this for all sorts of explanatory reasons. Even when I understand that my car keys are an aggregate of atoms, I may be wise to explain its working in a lock in macro terms.

Ontological Holism, in contrast, embraces the a-prioristic position that as-pects of social reality simply have no individual constituents. But even in this case the direct inference that scientific method should confine itself to research at the institutional level only is unjustified. Also if one believed that the social is made up of irreducible entities, there need not be anything irrational in certain contexts about looking for individual agents and their role in a social world nevertheless.

One can think of a continuum of stronger and weaker versions of ontological holism. A middle position would be that some aspects of social reality have indi-vidualistic constituents, but not all. Whichever the position one adheres to, it is a mistake to assume that ontological positions as regards holism or individualism imply anything definite about required scientific methods of research. Ontology

18 See for instance van Hees (1997) for an interpretation of methodological individualism in this context. 19 Caldwell (2004), chapter 2. 20 The German Historicists adhered to the idea that political preferences should be prevented from

entering theory choice. Their opposition to the ‘theoretical’ Austrians was partly out of Ideologiekritik.

has some bearing on proposed methods of research, but claims about epistemology do not directly follow from claims about ontology. This is the reason why I have held, above, that claims about the existence of the emergent properties of a social reality are not crucial to the essentialism implied by Austrian social theory.

The a-prioristic character of the mistake is notorious. Indistinctness about the relative strength of any such position is often part of the mistake. In a more sensitive attempt to draw proper distinctions between holist and reductionist posi-tions Harold Kincaid has listed seven charitable interpretations of MI and anyone who contemplates on these for a moment will be struck by the obviousness that the question whether MI methodologies are correct is ultimately an empirical matter:

1. Social theories are reducible to individual theories. 2. Any explanations of social phenomena must refer solely to individuals,

their relations, their dispositions, and so on. 3. Any fully adequate explanation of social phenomena must refer solely to

individuals, their relations, their dispositions, and so on. 4. Individualist theory suffices to fully explain social phenomena. 5. Individualist theory suffices to partially explain social phenomena. 6. Some reference to individuals is a necessary condition for any explanation

of social phenomena. 7. Some reference to individuals is a necessary condition for any full expla-

nation of social phenomena.21 In this list, the strongest methodological positions of MI are perhaps number 2 and 4, while the weakest seems to be number 7. (It is impossible to exhaustively order the seven interpretations in terms of logical strength.) The methodological recom-mendations of this list are ultimately empirical; that is, aprioristic assumptions cannot produce any advice on method.

7 Conflation of positions

Naturally however, there is always some bearing of the truth of ontological claims on the truth of methodological claims. But the most one can hold is that, in certain contexts, some aspects of the ontic organisation of the social will have some impli-cations for the question which method is wisest to employ. Stated in this way, clearly, this is a very weak claim. Thus, Kincaid concludes:

[b]road conceptual facts about the primacy of the physical, about human agency, or even about the ontology of society will not tell us how the sciences must relate. Like other failed attempts to show a priori how human inquiry must go, arguments of this sort give philosophical or conceptual considera-tions a power far beyond their means.22

I couldn’t agree more. As so many approaches can be hid in the simple guise of the label ‘Methodological Individualism’, it can only lead one astray if these issues

21 Kincaid (1997), p.31. 22 Ibidem, p.143.

Appendices

must be decided a priori. They are to be decided as a logical consequence of exten-sive empirical social research (of whatever holist or atomist kind) and of the rela-tive success or failure of conceptual frameworks to satisfactorily explain. What counts as a satisfactory explanation is another matter (but see de Regt (2004)).

Milford is aware of the danger of conflating believes about the character of the

social with prescriptions about methods of research. As to holism, he says that selbst in einer Welt, in der es keine Ganzheiten und keine historischen Ent-wicklungsgesetze gibt, könnte es vernünftig sein, nach Regelmäßigkeiten die-ser Art zur Lösung praktischer Probleme zu suchen, sofern nicht andere Ar-gumente, z.B. logische, gegen ein solches Verfahren sprechen

and on the same page he proceeds in similar terms about reductionism.23 Then he connects this point with the good old problem of induction:

Beide metaphysischen Positionen behaupten die Existenz von Regelmäßigkei-ten, doch für die Suche nach wahren Gesetzmäßigkeiten oder Regelmäßig-keiten bedarf es keiner solchen Voraussetzung, zumal es ohnehin keine Veri-fikation gibt, und man nicht wissen kann, ob man ein wahres Gesetz entdeckt hat.

The fallacy to try and directly deduce a methodology from an ontology is easy to commit, as Milford explains, because it is clear ‘daß methodologische Forderungen

vielfach eine Umdeutung von metaphysischen Behauptungen sind.’24 It is curious that, while he is so well aware of the fallacy, he seems to confuse methodological and ontological issues nevertheless when he connects essentialism in social science with (epistemological) holism. Austrians, Mäki has convincingly shown, are in-clined to essentialism and reductionism at the same time.

23 Milford (1997), p.105. 24 Ibidem, p.106.

References

Anderson, C. E. (1979) ‘Hospital Production--Can Costs be Contained?’ The American Economic Review, 69, No. 2, (Papers and Proceedings of the Ninety-First Annual Meeting of the American Economic Association), pp.293-297.

Ankersmit (1983) Narrative Logic, Den Haag: Martinus Nijhoff Balzer, W. (1982) Empirische Theorien: Modelle, Strukturen, Beispiele, Braunschweig

und Wiesbaden: Vieweg Balzer, W, Moulines, C. U., and Sneed, J. D. (1987) An architectonic for science,

Dordrecht: Reidel Blaug, M. (1997) (5th edition) Economic Theory in Retrospect, Cambridge: CUP Bleicher, H. (1890) ‘Gegenwart und Zukunft in der Wirtschaft’, in: Conrads Jahrbü-

cher, neue Folge, XXster Band, pp. 337-361, Jena: Gustav Fischer [in II] Böhm-Bawerk, E. von (1886) ‘Grundzüge der Theorie des Wirtschaftlichen Gü-

terwerts’, in: Conrads Jahrbücher, n.F., XIIIter Band, pp.1-82, 477-541, Jena: Gu-stav Fischer

Böhm-Bawerk, E. von (1889) Kapital und Kapitalzins, 2te Abteilung: Positive Theo-rie des Kapitales, Innsbruck: Wagner

Böhm-Bawerk, E. von (1890a) ‘Ein Zwischenwort zur Werttheorie’, in: Conrads Jahrbücher Jahrbücher für Nationalökonomie, n.F., XXIster Band, pp.519-522, Jena: Gustav Fischer

Böhm-Bawerk, E. von (1890b) ‘Zur Litteraturgeschichte der Staats- und Sozial-wissenschaften’, Buchkritik von der Festgabe zur Wilhelm Roscher’s Doktor-jubiläum, Gustav Schmoller (hrsg.), Conrad's Jahrbücher für Nationalökonomie, vol., pp. 75-95

Böhm-Bawerk, E. von (1921) Positive Theorie des Kapitales, Vol. 1, 4th ed.; first ed. 1889; third revised ed. 1914; (Volume 2 of the (1921) three volume Kapital und Kapitalzins), Jena: Gustav Fischer. Vol 1 referred to as PTK

Böhm-Bawerk, E. von (1921) Positive Theorie des Kapitales, Vol. 2, (4th ed.; 1st ed. 1889; Volume 3 of the three volume Kapital und Kapitalzins), Jena: Gustav Fischer; (Vol. 2 referred to as ‘PTK2’)

Böhm-Bawerk , E von (1959) Capital and Interest, three volumes transl. by Huncke, G. D. and Sennholz, H. F. of (1914, 3rd ed.) Kapital und Kapitalzins, Illinois: Lib-ertarian Press.

Boianovsky, M. (1998) ‘Wicksell, Ramsey and the theory of interest’, The European Journal of the History of Economic Thought, 5:1, pp.140-168.

Bortkiewicz, L. (1906) ‘Der Kardinalfehler der Böhm-Bawerkschen Zinstheorie’, in: Jahrbuch für Gesetzgebung, Verwaltung und Volkswirtschaft im deutschen Reich, Jhrg. 30, (Vol. 30), Berlin: Duncker & Humblot

References

Boumans, M. (1999) ‘Built-in Justification’, in Morgan, M. and Morrison, M. (eds.), Models as Mediators. Perspectives on Natural and Social Science, Cam-bridge: CUP; pp.66-96

Brown, J. (1998) ‘Natural Kind Terms and Recognitional Capacities’, Mind 107, pp. 275-303

Brown, M., Falk, A., and Fehr, E. (2004), ‘Relational Contracts and the Nature of Market Interactions’, Econometrica 72-3, pp.747-780

Caldwell, B. J. (2004) Hayek’s Challenge. An Intellectual Biography of F. A. Hayek, Chicago: UCP

Camerer, C. F., Babcock, L., Loewenstein, G., and Thaler, R. (1997) ‘Labor supply of New York City cabdrivers: One day at a time’ Quarterly Journal of Economics 112, pp.407-441

Cartwright, N. (1989) Nature’s Capacities and their Measurement, New York & Ox-ford: Clarendon press

Cartwright, N. (1999) ‘Quantum Hamiltonians and the BCS model of supercon-ductivity’, in Morgan, M. and Morrison, M. (eds.), Models as Mediators. Perspec-tives on Natural and Social Science, Cambridge: CUP; pp.241-281

Cartwright, N. (2001) ‘Ceteris paribus laws and socio-economic machines’, in Mäki, U. (ed.) The economic world view, Cambridge: CUP; pp.275-292

Charland, L. C. (2002) ‘The Natural Kind Status of Emotion’, The British Journal for the Philosophy of Science, 53 : 4, pp. 511-537

Commissie Structuur en Financiering Gezondheidszorg (1987) Verandering verze-kerd. Bereidheid tot verandering. (Rapport commissie Dekker) Den Haag

Cools, K., Kuipers, Th.A.F., and Hamminga, B. (1994) ‘Truth Approximation by Concretization in Capital Structure Theory’, in: Hamminga and DeMarchi (eds.), Idealization IV: Idealization in Economics, nr. 38, Amsterdam: Rodopi, pp.205-228

Defoe, D. [1719] Robinson Crusoë, Amsterdam: v. Holkema & Warendorf NV; no year of publication available; Dutch translation of The Life and Strange Surprising Adventures of Robinson Crusoe of York, Mariner: who lived Eight and Twenty Years, all alone in an uninhabited Island on the coast of America, near the Mouth of the Great River of Oroonoque; Having been cast on Shore by Shipwreck, wherein all the Men perished but himself. With An Account how he was at last as strangely deliver'd by Pirates. Written by Himself.

DeMarchi, N. (1988) ‘Popper and the LSE Economists’, in De Marchi, N. (ed.), The Popperian Legacy in Economics, Cambridge: CUP

Eichenbaum, M. (1997) ‘Some Thoughts on Practical Stabilization Policy’, AEA Papers and Proceedings, vol.87, nr.2

Fehr, E., Gächter, S., and Kirchsteiner, G. (1997) ‘Reciprocity as a Contract En-forcement Device: Experimental Evidence’, Econometrica, 65 (40), pp.833-860

Fisher, I. (1907) The Rate of Interest, New York: Macmillan Folmer, H. (2000) ‘Waarom economen vaak miskleunen’, Economisch Statistische

Berichten, 3 November, 2000 Friedman, Michael (1974) ‘Explanation and Scientific Understanding’, Journal of

Philosophy 71, pp. 5-19 Friedman, Milton (1953) ‘The methodology of Positive Economics’, introduction

of Essays in Positive Economics, Chicago: UCP Gamut, L.T.F. (1991) Logic, Language, and Meaning, Volume 2 ‘Intensional Logic

and logical Grammar’, Chicago and London: The University of Chacago Press ‘Good food?’ (2006), The Economist, 9 December, p.9 Haavelmo, T. (1944) ‘The Probability Approach in Econometrics’, Econometrica 12,

supplement, pp.1-115 Hahn, F. (1973) On the Notion of Equilibrium in Economics, Cambridge: Cambridge

University Press Hamberger, K. (2001) ‘Von Böhm-Bawerk, Jevons and the “Austrian” theory of

capital: “a quite different relation”’, European Journal of the History of Economic Thought, 8:1, pp.42-57

Hamminga, B. (1983) Neoclassical Theory Structure and Theory Development, Berlin: Springer

Hamminga, B. and De Marchi, N. (1994) ‘Idealization and the Defence of Eco-nomics: Notes Toward a History’, in Idealization VI: Idealization in Economics, Poznán Studies nr. 38, Amsterdam: Rodopi; pp.11-40

Hammond, J. D. (1990) ‘Realism in Friedman’s Essays in Positive Economics’, in Donald E. Moggridge (ed.), Perspectives on the History of Economic Thought, Vol. 4: Keynes, Macroeconomics and Method, Aldershot: Edward Elgar

Hausman, D. (1992) The Inexact and Separate Science of Economics, Cambridge: CUP Hees, M. van (1997) ‘Explaining institutions: A defence of reductionism’, Euro-

pean Journal of Political research 32, pp.51-69 Hennings, K. H. (1997) The Austrian Theory of Capital, Cheltenham UK: Edward

Elgar. (With letters from Böhm-Bawerk to Wicksell in the appendix, translated by Christian Gehrke)

Hoover, K. D. and Dowell, M. E. (2001) ‘Measuring Causes: Episodes in the Quantitative Assessment of the Value of Money’ in Klein, J. L. and Morgan, M. S. (eds), The Age of Measurement, Annual Supplement to Vol. 33, of History of Political Economy, pp.137-161

References

Hume (1890 [1739]) A Treatise of Human Nature and Dialogues Concerning Natural Religion, Volume 1, (ed. by Green & Grose), London: Longmans, Green and Co.

Janssen, M. C. W. (1993) Microfoundations, London: Routledge Kauder, E. (1957) ‘Intellectual and Political Roots of the Older Austrian School’,

Zeitschrift für Nationalökonomie, 17, pp.411-25

Keynes, J. M. (1973) The General Theory of Employment, Interest and Money, re-printed from the first edition, 1936. Cambridge: Macmillan/CUP

Kim, J. (1973) ‘Causes and counterfactuals’, Journal of Philosophy, 70, pp.570-2 Kincaid, H. (1997) Individualism and the Unity of Science, Lanham, Maryland:

Rowman & Littlefield Publishers Kitcher, P. (1989) ‘Explanatory Unification and the Causal Structure of the

World’, Minnesota Studies in the Philosophy of Science, Vol. XIII, Scientific Explanation, Minneapolis: UMP

Klamer, A. (1983) Conversations with economists, Savage: Rowman & Littlefield Klant, J. J. (1987) Filosofie van de economische wetenschappen, Leiden: Martinus

Nijhoff. Koningsveld, H. (1973) ‘Empirical Laws, Regularity, and Necessity’, Mededelingen

Landbouwhogeschool Wageningen, 73-2, Wageningen: Veenman & zonen BV, pp.1-29

Kleinknecht, A. (1994) ‘Heeft Nederland een loongolf nodig? Een neo-Schum-peteriaans verhaal over bedrijfswinsten, werkgelegenheid en export’, Tijd-schrift voor Politieke Ekonomie 17, pp.5-24.

Kuipers, T.A.F. (2000) From Instrumentalism to Constructive Realism, Dordrecht: Kluwer, Synthese Library 287

Kuipers, T. A.F. (2001) Structures in Science, Dordrecht: Kluwer, Synthese Library 301

Kurz, H. D. (2000) ‘Wicksell and the problem of the “missing equation”’, in His-tory of Political Economy 32-4, pp.765-788

Lachmann, L. (1976) ‘From Mises to Shackle: An Essay on Austrian Economics and the Kaleidic Society’, Journal of Economic Literature, Vol. 14, nr.1, pp.54-62

LaPorte, J. (2004) Natural Kinds and Conceptual Change, Cambridge: CUP Lawson, T. (1997) Economics and Reality, London and New York: Routledge. Lewis, D. (1973) Counterfactuals, Oxford: Blackwell Lindenberg, S. (1991) ‘Die Methode der abnehmenden Abstraktion: Theorie-

gesteuerte Analyse und empirischer Gehalt’, in: Hartmut Esser und Klaus Troitzsch (hrsg.), Modellierung sozialer Prozesse, Bonn: Informationszentrum Sozialwissenschaften, pp. 29-78

Lucas, R. (1976) ‘Econometric Policy Evaluation: A Critique’. Carnegie-Rochester Conference Series on Public Policy 1, pp.19–46

Lynch, M. P. (1998) Truth in Context, Cambridge, MA and London: MIT press Kincaid, H. (1997) Individualism and the Unity of Science. Essays on Reduction, Ex-

planation, and the Special Sciences, Lanham, Maryland: Rowman & Littlefield Mäki, U. (1990) ‘Scientific realism and Austrian explanation’, Review of Political

Economy 2.3 pp.310-344 Mäki, U. (1991) ‘Practical Syllogism, entrepreneurship, and the invisible hand’, in

Lavoie, D. (ed.) Economics and Hermeneutics, London: Routledge Mäki, U. (1992a) ‘The market as an isolated causal process: a metaphysical

ground for realism’, in Caldwell, B. and Boehm, S. (eds.), Austrian Economics: Tensions and New Directions, pp.35-59, Dordrecht: Kluwer

Mäki, U. (1992b) ‘On the Method of Isolation in Economics’, in Dilworth, C. (ed.) Idealization IV: Intelligibility in Science, nr. 26, Amsterdam: Rodopi, pp.317-351

Mäki, U. (1994) ‘Isolation, Idealization and Truth in Economics’, in: Hamminga and De Marchi (eds.), Idealization VI: Idealization in Economics, nr. 38, Amster-dam: Rodopi, pp.147-168

Mäki, U. (2001) ‘The Way the World Works (WWW): Towards an Ontology of Theory Choice’, in Mäki, U. (ed.) The Economic World View, Cambridge: CUP

Mäki, U. (2002) (ed.) Fact and fiction in economics: models, realism and social construc-tion, Cambridge: CUP

Marelich, W. D., Mann, T., Erger J., & Grusky, O. (1998) ‘A “satisficing” model of HIV health-care provider decision making’; (paper presented at the 1998 Cen-ter for HIV Identification, Prevention, & Treatment Services (CHIPTS) Confer-ence: HIV Research: The Next Generation, University of California), Los Angeles

Mayer, T. (1993) Truth versus Precision in Economics, Aldershot: Edward Elgar Menger, C. (1871) Grundsätze der Volkswirtschaftslehre, Wien: Wilhelm Braumüller Menger, C. (1883) Untersuchungen über die Methode der Sozialwissenschaften, und der

Politischen Oekonomie insbesondere, Leipzig: Verlag von Duncker & Humblot Milford, K. (1989) Zu den Lösungsversuchen des Induktionsproblems und des Abgren-

zungsproblems bei Carl Menger, Veröffentlichungen der Kommission für Sozial- und Wirtschaftswissenschaften, nr.27, ed. W. Weber, Vienna: Verlag der Österreichischen Akademie der Wissenschaften

Milford, K. (1997) ‘Hufeland als Vorläufer von Menger und Hayek’, in Priddat, P. (ed.) Wert, Meinung, Bedeutung, Marburg: Metropolis-Verlag, pp.89-160

Ministerie van WVC (1988) Verandering verzekerd. Tweede Kamer, vergaderjaar 1987-1988, 19945, nrs 27-28, Den Haag (Het plan Dekker)

Ministerie van WVC (1990) Werken aan zorgvernieuwing. Tweede Kamer, ver-gaderjaar 1989-1990, 21545, nr 2, Den Haag (Het plan Simons)

References

Mises, L von, (1949) Human action: a treatise on economics London: Hodge and Company limited

Modigliani, F. and Miller, M. H. (1958), ‘The Cost of Capital, Corporation Finan-ce, and the Theory of Investment’, American Economic Review, 48, pp.261-297

Morrison, M. (1999) ‘Models as autonomous agents’, in Morrison, M. and Mor-gan, M. (eds.) Models as Mediators. Perspectives on Natural and Social Science, Cambridge: CUP; pp.38-65

Niiniluoto, I. (1990) ‘ Theories, approximations, and idealizations’, in Brzeziński, Coniglione, Kuipers, and Nowak (eds.), Idealization I: General Problems, Am-sterdam & Atlanta: Rodopi

Nowak, L. (1989) ‘On the (Idealizational) Structure of Economic Theories’, in Balzer, W., & Hamminga, B. Philosophy of Economics, Erkenntnis 30, ns. 1-2, Amsterdam: Kluwer

O’Neill, J. (2001) ‘Essences and markets’, in Mäki, U. (ed.) The economic world view, Cambridge: CUP

Open Europe bulletin: 30 November – 20 December 2005: http://www.openeurope.org.uk/media-centre/bulletin.aspx?bulletinid=27

Pietrosky, P. and Rey, G. (1995) ‘When Other Things Aren’t Equal: Saving Ceteris Paribus Laws from Vacuity’, British Journal for the Philosophy of Science, 46, pp.81-110

Psillos, S. (2004) ‘A Glimpse of the Secret Connexion: Harmonising Mechanisms with Counterfactuals’, Perspectives on Science, 12-3, pp.288-319

Putnam, H. (1975) Mind, Language and Reality, Philosophical Papers, vol. 2, Cam-bridge: CUP

Radder, H. (2006) The World Observed / The World Conceived, Pittsburgh, Pa: Uni-versity of Pittsburgh Press

Regt, H. de (2004) 'Discussion note: Making sense of understanding', Philosophy of Science 71 (2004) 98-109

Ricardo, D. (1996 [1817]) Principles of Political Economy and Taxation, New York: Prometheus books; (this edition originally published in London: Dent & Sons, 1911)

Robbins, L. (1930) ‘On a Certain Ambiguity in the Conception of Stationary Equi-librium’, The Economic Journal, Vol. XL pp.194-214

Rol, M. (1998) ‘Zieke markten’, Tijdschrift voor economisch onderwijs, 6, p.240 Rol, M. (2005) ‘Abstracción por idealización en economía’, Revista Asturiana de

Economía, 28, pp.33-40 Rol, M. (2005) ‘Abstractie in het economisch denken. Relevantie voor beleid’,

Algemeen Nederlands Tijdschrift voor Wijsbegeerte, volume 97 nr. 3, pp. 224-241

Rol, M. (2007) ‘Idealization, abstraction, and the policy relevance of economic theories‘, Journal of Economic Methodology, 14:4, forthcoming

Schliesser, E. (2006) Friedman, Positive Economics, and the Chicago Boys, mimeo Schumpeter, J. A., (1954) History of Economic Analysis, Oxford: OUP Simon, H. (1957) Models of Man, New York: John Wiley & Sons, Inc. Smith, B. (1986) ‘Austrian Economics an Austrian Philosophy’, in W. Grassl and

B. Smith (eds.), Austrian Economics: Historical and Philosophical Background, London[etc.]: Croom Helm

Smith, B. (1990) ‘Aristotle, Menger, Mises. An Essay in the Metaphysics of Eco-nomics’, in B. Caldwell (ed.) Carl Menger and His Legacy in Economics, Durham[etc.]: Duke University Press

Ullman-Margalit, E. (1978) ‘Invisible-hand explanations’, Synthese 39, pp.263-291 Wicksell, K. (1893) Über Wert, Kapital und Rente, Jena: Gustav Fischer; referred to

as WKR

Wicksell, K. (1898) Geldzins und Güterpreise, Jena: Gustav Fischer Wicksell, K. (1907) ‘The Influence of the Rate of Interest on Prices’, Economic Jour-

nal, XVII, pp.312-220 Wicksell, K. (1928) ‘Zur Zinstheorie (Böhm-Bawerks dritter Grund)’, in: Einkom-

mensbildung, volume 3 of Hans Mayer’s Die Wirtschaftstheorie der Gegenwart, 4 volumes; Vienna: Springer

Wicksell (1951, [1901]) Lectures on Political Economy, (translated from Swedish by Classen, E.), edited and introduced by Robbins, L., London: Routledge & Ke-gan

Wicksell, K. (1984, [1913]) [Swedish original 1901] Vorlesungen über Nationalöko-nomie I, (translated from Swedish by Margarethe Langfeldt), Neudruck der Ausgabe Jena, Aalen: Scientia Verlag

Wicksell, K. (1954) Value, Capital, Rent, London: Allen & Unwin; translated by S.H. Frowein

Winkel, H. (1985) Die Volkswirtschaftslehre der neueren Zeit, Darmstadt: Wissen-schaftliche Buchgesellschaft; (Erträge der Forschung, Bd. 18)

Woodward, Jim (2002) ‘What Is a Mechanism? A Counterfactual Account’, Phi-losophy of Science, 69, pp.S366-S377

Yablo, S. (1992) ‘Cause and Essence’, Synthese, 93, pp.403-449

Summary in English

Philosophy of science is traditionally a normative enterprise. But without historical

adequacy no metatheory is worth its salt. For instance, one cannot but notice that

economics has been developing without any predictive success for a century and a

half. The sort of progress it has been making is explanatory. An instrumentalist ac-

count of the development of economics must be inadequate, for the following rea-

son. A description of the dynamics of a discipline in terms of explanatory progress

implies a commitment to ontologies; instrumentalism is the rejection of such a

commitment. It may well be possible to picture the predictive success of a theory

while being agnostic about what the terms of this theory refer to. Black boxes can

be compared in their predictive record. But to say that a scientific theory explains

better (than yesterday; than the competing theory) is to indicate one’s (deeper)

understanding of the explanandum. Black boxes do not give an understanding;

instead they conceal the very nuts and bolts that are to provide such an insight.

Hence, seen from the perspective of the meta-enterprise that philosophy of science

is, it seems to me that in trying to give historical adequacy one is forced to be a

realist about theories.

Besides, from the object perspective of the scientist, it is impossible to inves-

tigate the world in order to gain explanatory insight in it without prior ontological

convictions. Pre-scientific beliefs are supposed when research is to generate scien-

tific knowledge. Probably, most pre-scientific beliefs are implicit and even uncon-

scious. For example, Bert Hamminga found that economists who work in Interna-

tional Trade Theory, as he dissected their reasoning, actively deploy ‘Systems of

Elementary Plausibility Convictions’ in order to assess the acceptability of interest-

ing theorems.

Ontological commitments are by no means limited to International Trade

Theory, my intuition is that all science presupposes such commitments. Pre-

scientific beliefs about what the world is like and how it works imply ontological

realism, but also epistemological realism, for their being beliefs. Böhm-Bawerk’s

endeavour, as analysed in chapters I and II, is an example of this realism. He pro-

vides explanatory unification, defining concepts anew and applying Mengerian

Value Theory to the theory of interest. Of the possible epistemological positions

running from inductive scepticism, via instrumentalism, and entity realism to con-

structive and essentialist realism, Böhm-Bawerk seems to adhere to the last posi-

tion. As he is convinced that the ‘Austrian toolkit’ fits economic reality, as it is, best,

his intuition can be qualified as essentialist. Indeed, his own unifying conceptual

apparatus appears to be more explanatory powerful to him than its competing

Summary

conceptualizations. In any case, Böhm-Bawerk would be ready to claim that, with-

out it, many essential nuts and bolts of the market mechanism remain out of sight.

I present his intuition, that economic mechanisms make up the correct way

of explaining in economics, as a search for social kinds. However, two caveats per-

tain. First, Böhm-Bawerk does not hesitate to dish up the customs of the physicist,

who ‘greift die charakteristischen Typen in solcher Zahl und Auswahl heraus, wie die

Natur seiner allgemeinen oder besonderen Erklärungsaufgabe es ihm zweckmäßig erschei-

nen läßt.’1 In other words, the research aims codetermine, together with the ‘way

the world works’2, the choice for what is taken to be the essential cut. Second, es-

sentialism is a philosophically difficult stance to defend. Essentialism seems to

claim that good science fixes the explanandum, and disallows both vagueness of

terms and fundamental changes in the world. Emancipatory action or discussion

about meanings appear to be impossible. Hence, there seems to be a tension be-

tween historical and normative adequacy: although an essentialist attitude con-

cerning the world and the knowledge about it merely is natural, it is philosophi-

cally difficult to be upheld.

Two questions arise. Which objects (or properties, or relations) in the eco-

nomic environment are the credible candidates for social kinds? And how can such

an essentialist interpretation of economics be defended? As to the first question, in

Austrian economics these kinds are structures that determine the degrees of free-

dom of economic market mechanisms. These structures include market forms with

intentional agents, endowed with preference orderings. The market mechanism,

rooted in this socio-economic structure, is represented by what I label the Law of

Marginal Agents. This is the supposed regularity describing how the subjective

marginal utilities of the market parties, who operate ‘in the margin’, determine the

objective market price.

As to the second question, the philosophical tenability, essentialist interpre-

tations of knowledge acquisition rest on the specification of rigid designators. The

Causal Theory of Meaning holds that instances of a kind help specify a kind term

by a sort of causal baptism. Joseph LaPorte explains that it would be too much to

ask from such a theory of language to weed out all vagueness in our speech and all

change in the world. What such a theory may be asked to do is help recognise

vagueness of speech and change in the world. The labelling of a kind term is no

more than a proposal to carve the (economic) world up in the supposed unique best

way. The rigidity involved in the specification of a kind term is a mere rigidity de

jure, not de facto.

1 PTK, p.262. 2 Mäki’s phrasing, see his (2001).

The last chapter of this dissertation heavily depends on LaPorte’s new view

on language and philosophy of science. One attractive aspect of it is that the idea of

unique right conceptualizations does not preclude fallibilism. We may be wrong

but whatever we think is the correct theory to explain the world and its workings

is also a theory that usefully retrieves abstract kinds. Just like “water=H2O” was the

discovery of a necessary truth – and although the extension of a kind term differs

across possible worlds – in every possible world where we find the kinds as we

happen to have specified them, the kind terms refer to the same abstract kind. In this

rather technical sense science is involved with (natural) necessity.

Although there is a certain deliberate weakness to this way of explicating es-

sentialist explanation, kind terms do help explain. Just as the peculiar behaviour of

a polar bear (LaPorte’s example) is explained by denoting the kind ‘polar bear’,

Böhm-Bawerk’s case shows that many market phenomena can be explained by

constructing the kind ‘market’. In the Austrian view, a ‘market’ gives rise to causal

mechanisms and the dissection of a fundamental structure, in which this mecha-

nism is supposed to be rooted, has explanatory power indeed.

Another question remains. What sort of epistemic operation characterises

the specification of kinds? Chapter III answers this question, chapter IV gives some

examples. Science proceeds by building theories. Theoretical knowledge is knowl-

edge of abstract kinds. So, apparently, abstraction is the key operation in the con-

struction of theories. A confusion as to what sort of operation this is seems to be

ubiquitous. In consequence of the influence the Poznań School has had on some of

them, philosophers often speak of ‘Idealization and Concretization’ as a term for

the subsequent introduction and removal of limit values in theorising. It is also

common to hear economists say that ‘the ceteris paribus clause helps to give an

abstract representation’ of the economic realm. But hedging a general hypothesis

against expected or unexpected interferences is not the same as abstraction. This

becomes clear, for instance, when a clause gets in the way of a useful description of

the actual state of affairs. In order to remove a clause, one has to be aware of what

it is that is being specified by it. This means that the details summed up in the

clause are actually being referred to. In case of abstraction, these details are un-

known – hence they cannot be referred to – which is precisely the reason to ab-

stract from them.

The difference is important, because abstraction does not introduce falsity,

while idealization, in a sense, does. Clauses that require a ceteris paribus, a ceteris

absentibus, or a ceteris neglectis are in fact deliberately false claims. They specify a

state of affairs – that friction is absent, that transaction costs are close to nil – which

clearly is not a property of the actual world. Such a non-actual state of affairs is,

however, a property of a conceptually possible world. The question now arises

how scientists (and, above all, economists) can think of themselves as seeking true

Summary

theories while inserting deliberate falsity in their claims. Self-proclaimed ‘instru-

mentalism’ does not immunize economics against this question. The answer is

given in chapter III. Here I introduce the idea that such a clause is part of the ante-

cedent of a counterfactual proposition, i.e. a proposition about conceptually possi-

ble but not actual worlds. According to a Kripke-Lewis semantics, the counterfac-

tual can be true even if the clause is false of the actual state of affairs, that is, if the

clause does not count the actual world among its models. This is important, be-

cause idealizational strategies in social scientific explanation are inevitable. The

semantics developed shows that the falsity of the clauses, which hedge the expla-

nations of real phenomena, is no threat to the truth of theories. Idealizational con-

ditionals can themselves be true, qua counterfactuals. Note that the falsity of their

clauses can nevertheless be an obstacle for policy intervention, when the false as-

sumptions involved have to but cannot be dropped.

Abstraction, in contrast, is capable of leading to truth in a more straightfor-

ward way. I present abstraction as an operation by which an unknown number of

spatio-temporal details are omitted from consideration, possibly unconsciously, so

as to focus on some aspects only. These isolated aspects are rendered interesting.

The logical form of it is existential generalization. Thus, abstract and concrete are

matters of degree. Starting from a more concrete but true proposition, the existen-

tial generalization leads to a more abstract but still true proposition. But note the

greater extension of abstracted claims relative to their more concrete counterparts.

Due to this, starting from the more concrete but false proposition, the more abstract

proposition may absorb the actual world among its models.

One problem, however, is that abstract claims can be too weak to be interest-

ing. This is a feature that follows from such claims being generalizations. Stronger

claims are those that exclude more possibilities, while abstraction, conversely, leads

to the inclusion of more possibilities. The fruitfulness of an abstract proposition

depends on our background knowledge and our interventionist interests. I hold, in

chapter III, that it is obvious that there are abstract but interesting claims also in

economics. One candidate is: ‘the peoples who engage in trade instead of in au-

tarky will enjoy a greater welfare than those who do not.’ Another perhaps is:

‘there is a natural rate of unemployment, which is insensitive to budget policy’. I

believe that these claims are true; they are in any case generally subscribed to by

the core of economists and clearly relevant. Admittedly, there is an ideological

component to this belief, which is excluded in cases of physical or chemical engi-

neering. So, it is a political matter precisely how interesting these claims are. But it

is a matter of scientific research whether they are true.

Thus, the verdict that economic theories are abstract in itself cannot sustain

claims to policy irrelevance. At most, the verdict of theories being idealizational

Name index

Anderson, C. 156 Janssen, M. 14

Ankersmit, F. 79-80 Jevons, W, 29

Balzer, H. 111 Kauder, E. 62

Blaug, M. 41, 143-144 Keynes, J. 149-154

Bleicher, H. 5 Kim, J. 187

Boianovsky, M. 33 Kincaid, H. 62, 213

Bortkiewicz, L. 72 Kirzner , I. 64, 170

Boumans, M. 92, 102 Kitcher, Ph. 61

Boyle , R. 105-107 Klamer, A. 96, 98

Brown, J. 173-177 Klant, J. 29

Brown, M. 148 Kleinknecht, A. 45

Caldwell, B. 62, 64, 170, 212 Kneale, W. 61

Camarer, C. 156 Koningsveld, H. 164

Cartwright, N. 111, 117, 133, 167 Krabbe, E. 110

Charland, L. 199 Kuipers, Th. 4, 87, 94-95, 117, 183

Clark, J.-B. 21 Kurz, H. 143-146

Cools, K. 119 Lachmann, L. 74, 81

Defoe, D. 16 Landry, A. 21

DeMarchi , N. de 86 LaPorte , J. 171, 178-182

Eichenbaum, H. 125 Lawson, T. 93

Fehr, E. 99 Lewis, D. 111, 114

Fisher, I. 21, 72 Lindenberg, S. 4

Folmer, H. 129, 157, 199 Lipsey, R. 86

Friedman, Michael 61 Lucas, . 96

Friedman, Milton 97-98, 115 Lynch, M. 184-186

Haavelmo, T. 104 Mäki, U. 12, 13, 57, 74, 97, 102, 104,

123, 168-169, 176-177, 207-211

Hahn, F. 169 Marelich, D. 156

Hamberger, K. 10, 29 Marshall, A. 21, 71,

Hamminga, B. 87-91, 93, 124 Marx, K. 17, 30

Hammond, D. 98 Mayer, Th. 92

Hausman, D. 126, 134 Menger, C. 12, 21, 23, 171

Hayek, F. 74, 75, Metzler, L. 90

Hennings, K. 71 Mises, L. von 75, 210

Hoover, K 128 Modigliani, F. 118

Hume, D. 56 Morrison, M. 116

Name index

Niiniluoto, I. 110

Nowak, L. 4

O’Neill, J. 98, 186

Pietrosky, P. 115

Psillos, S. 93, 187-188

Putnam, H. 172

Radder, H. 164-167

Ricardo, D. 53, 139, 193

Robbins, L. 145

Rodbertus 18

Rol, M. 155

Salmon, W. 168

Schmoller, G. 58, 212

Schumpeter, J. 1, 8, 27, 29, 31,

Sienra, A. de la 104

Smith, A. 17, 21, 193

Smith, B. 62

Thünen, J. von 139

Ullman-Margalit, I. 74

Waals, J. van der 106

Weber, M. 157

Wicksell, K. 2, 21, 41, 46, chapter II, 138-147

Woodward, J. 189-192

Yablo, S. 187

Milford, K. 210-214

Hees, M. van 212

Roscher, W. 212

Begripsmatige vooruitgang in de economie

Abstractie van sociale soorten versus idealisering

Een Nederlandse samenvatting voor groter publiek

‘Voor een econoom is het echte leven een speciaal geval.’ Tik ‘jokes economics’ in bij Google en je krijgt deze grap op verschillende van 1.390.000 sites. Hier is er nog één: ‘Hoeveel economen zijn er nodig om een peertje in te draaien? Acht. Één draait het peertje in, de zeven anderen houden de overige omstandigheden gelijk.’

Economen zouden natuurlijk best willen dat alleen de verschijnselen waar-van ze de gevolgen willen onderzoeken veranderen terwijl de rest van de wereld constant blijft. Het effect van maatregelen tegen schooluitval onder allochtone jon-geren op de werkloosheid is niet meer goed na te gaan door de emancipatie onder allochtone meisjes. Dat mag een sociaal gewenst verschijnsel zijn, maar als allerlei andere oorzaken dan die waarvoor onderzoekers nu net belangstelling hadden de uitkomst beïnvloeden, is de oorspronkelijke vraag niet meer te beantwoorden. Daarom kennen wetenschappelijke theorieën clausules, waarin opgesomd staat welke ‘overige omstandigheden’ constant of afwezig verondersteld worden. Eigen-lijk zegt de wetenschapper dus: ‘die meisjes emanciperen zich niet en elke veran-dering van het werkloosheidscijfer wordt veroorzaakt door onze maatregelen te-gen schooluitval’. De bedenkers van die theorieën weten ook wel dat die clausules vol onware uitspraken staan, maar daarzonder kunnen ze blijkbaar niet. De clausu-les beschermen als het ware de theorie.

Een vraag die filosofen bezighoudt is hoe wetenschappers nu eigenlijk ware theorieën kunnen bedenken als die slechts houdbaar zijn onder bescherming van onware clausules. Ik stel deze vraag in dit proefschrift vooral met betrekking tot de economie, maar ik kom met conclusies die voor alle sociale wetenschappen belang-rijk zijn.

De grappen bespotten de economische wetenschap omdat ze onbelangrijk zou zijn voor ons dagelijks leven. Er is blijkbaar aanleiding te twijfelen bij de rele-vantie van het vak voor (economisch) beleid. De veelgehoorde klacht dat economi-sche modellen te ver af staan van de werkelijkheid kan op verschillende manieren worden opgevat. De eerste grap hierboven gaat bijvoorbeeld over het abstracte karakter van de economie, terwijl de tweede grap over idealiseringen gaat. Ab-stractie en idealisering zijn twee heel verschillende technieken.

Invloeden isoleren

In fundamenteel onderzoek proberen wetenschappers tot algemeen geldige uit-spraken te komen over ‘de’ – of, bescheidener, ‘een’ – structuur van de werkelijk-heid. Voor allerlei praktische toepassingen van kennis is het handig als je zulke

Samenvatting Nederlands

algemeen geldige uitspraken kunt doen, want dan valt min of meer te voorspellen wat er zal gebeuren als je ingrijpt. Dat is niet uniek voor wetenschap. Iedere leek weet dat marktkooplui op zaterdagmiddag hun waar kwijt moeten en dat je op die dag na vieren dus goedkoper aan je tomaten komt. Sommigen van ons vinden het financiële voordeel dat daarvan uitgaat groter dan het wellicht praktische nadeel van laat in de middag kopen; zij handelen ernaar. Dit is een structureel aspect van de sociale werkelijkheid, vatbaar voor een algemene uitspraak. Wetenschappers zijn blij als ze zo’n algemene uitspraak kunnen doen die weinig uitzonderingen kent, zoals dat de versnelling van een object met massa recht evenredig is met de som der krachten die op dat object werken (Newton’s tweede wet). De economie en de andere sociale wetenschappen lijken in mindere mate op zulke successen te kunnen bogen dan de natuurkunde.

Het is duidelijk dat de tweede wet van Newton nogal ver van de dagelijkse praktijk afstaat. Een nadeel daarvan is dat je er niet direct uit kunt aflezen met welke snelheid je te pletter slaat als je van een flatgebouw springt. Blijkbaar is er een vertaling nodig van deze abstracte wet naar specifieke omstandigheden. Maar die abstractie heeft ook een reusachtig voordeel. Zij helpt ons doorzien dat een sprong in de buurt van het aardoppervlak andere gevolgen heeft dan een in de buurt van het oppervlak van een andere planeet. Als de aarde de massa had van Jupiter, dan zou een sprong van twintig centimeter mijn benen al breken. Abstracte theorieën geven, als ze waar zijn, potentieel veel kennis, maar de toepassing van die kennis wordt wel vaak lastiger naarmate het niveau van abstractie hoger is. Een praktische ingenieur is dan ook geïnteresseerd in wat minder abstracte verha-len. Anderzijds zei Vincent Icke eens in NRC-Handelsblad dat een theoreticus ie-mand is die uitzoekt of wat in praktijk werkt ook in theorie werkt. En zo is het.

Een clausule sluit dus oorzakelijke invloeden, buiten de oorzaak die object is van onderzoek, uit. In experimenten gebeurt dat feitelijk. Dat wil zeggen, in de experimentele opzet probeert men de te onderzoeken invloeden materieel te isole-ren van de invloeden die men buiten beschouwing wil laten. Maar in zuivere theo-rievorming – wetenschap zonder dat er direct een echt experiment aan te pas komt – is de isolerende clausule minstens zo belangrijk. Elk leerboek economie voor middelbare scholieren, bijvoorbeeld, somt de voorwaarden op voor een functioneel verband tussen vraag naar een product en te betalen prijs zoals de eis dat de voor-keuren van de vragers gelijk blijven. Het lijkt dus wel of de economie als weten-schap vooral niet-reële situaties beschrijft. Als dit waar is, lijken de grappen over economie volstrekt terecht. Ik geloof echter dat de meeste grappen (en ook veel serieuze kritiek) de kwestie volledig misvatten.

Idealisering

Ik stel idealisering voor als een redeneerstrategie waarbij zo’n clausule wordt ge-bruikt. Hoe kunnen economische theorieën ooit (bij benadering) waar zijn gegeven

dat talrijke idealiserende clausules per definitie op de één of andere manier onware uitspraken in de theorieën betrekken?

Het antwoord hangt samen met de manier waarop mensen zich in het dage-lijks leven uitlaten over hypothetische situaties, dat wil zeggen, situaties die niet actueel zijn. Zo wordt het gewoonlijk als zinvol ervaren te spreken in hypotheti-sche zin, zoals wanneer iemand zegt ‘als het nog harder waaide, zouden de dak-pannen van mijn huis vliegen’. Toch waait het niet harder dan het feitelijk doet. Waarom zouden we deze uitspraak wél accepteren, en niet de uitspraak van een econoom dat bij gelijkblijvende preferenties, prijzen van andere goederen en aantal kopers de vraag naar een goed stijgt bij dalende prijs? Ondanks dat al die factoren op de markt nooit gelijk zullen blijven, geeft de economie toch een vruchtbare visie op het vraaggedrag van marktpartijen, net zoals de alledaagse beschrijving van onwaarschijnlijke hypothetische situaties soms informatief kan zijn. Dit komt doordat de structuurovereenkomsten tussen zo’n niet-actuele denkbare wereld en de actuele wereld ons veel kan leren, vooral als die niet-actuele wereld een prettig overzicht biedt van een veelheid van mogelijke verschijnselen.

Een hypothetische uitspraak is uit te drukken als een tegenfeitelijke zin, dat wil zeggen, een zin gesteld in de aanvoegende wijs, zoals zin A:

(A) ware X het geval, dan zou Y het geval zijn.

ook al is het eerste deel van de tegenfeitelijke zin onwaar (de antecedent genaamd, waarin vóór de komma gezegd wordt ‘X is het geval’), toch kan deze zin als geheel nog wel waar zijn. Niemand ontkent dat als ik mijn kop koffie los laat, deze op de grond zal vallen, zelfs als ik mijn koffie in werkelijkheid stevig vast blijf houden. Het is nu de wetenschappelijke kunst om hypothetische tegenfeitelijke uitspraken te vinden die tegelijk waar en interessant zijn. Want er zijn gemakkelijk ware te-genfeitelijke zinnen te bedenken die slechts tot schouderophalen aanleiding geven. Maak het spannend, is dus het devies.

Maar hoe spannend? Te zeggen dat de Nederlandse autoriteiten verbaasd zouden zijn als zij geconfronteerd werden met buitenaardse wezens is niet zo spec-taculair. Te zeggen dat de militairen de regie zouden krijgen als er buitenaardse wezens in Nederland gesignaleerd werden is al veel informatiever, maar misschien niet waar. Te zeggen, ten slotte, dat in zo’n geval de Derde Wereldoorlog zou uit-breken is zeer informatief, maar vermoedelijk onwaar. Blijkbaar is er een uitruil tussen de kans op waarheid en het belang van die waarheid.

Een theoretische hypothese beschrijft dus altijd één of andere denkbare we-reld die verschilt van de actuele wereld, de wereld waarvan wij de geschiedenis en het heden al kennen. Wij kunnen ons een ander verloop van de geschiedenis voor-stellen. Daardoor kunnen we met verschillende mogelijkheden voor de toekomst rekening te houden, anders zouden wij niet eens dagelijkse keuzes kunnen maken. Ik zal verder naar de actuele wereld verwijzen met het symbool @.

Voor een econoom is het nu interessant om uit te maken waar de grens ligt tussen economisch mogelijke werelden en werelden die weliswaar logisch mogelijk

zijn, maar niet economisch mogelijk. Dit uit te maken is ‘werken aan de grens van het weten’. Als ik zeg: ‘als ik mijn koffie losliet, dan zou deze op de grond vallen’, dan beschrijf ik in de antecedent een verschijnsel – het loslaten – dat zich in de actuele wereld niet voordoet; ik houd mijn koffie namelijk goed vast. Als ik dat niet doe is de uitspraak niet tegenfeitelijk meer. De antecedent is dus niet waar, of – zoals filosofen dat zeggen – niet waar ‘op @ ‘. Maar hij is wel waar op één op andere fysisch mogelijke wereld, die in vele opzichten lijkt op @. De wetten van de zwaartekracht gelden in die wereld. En ik kán mijn koffie in die wereld wel losla-ten. De nieuwe vraag is nu: kun je ook vruchtbare tegenfeitelijke uitspraken doen met antecedenten die een denkbare wereld beschrijven (die ‘waar zijn op’ zo’n wereld) die fysisch of economisch onmogelijk is? Kun je, om een voorbeeld te noe-men, zinvol hypothesen onderzoeken die er tegenfeitelijk van uitgaan dat vragers altijd goed geïnformeerd zijn over de kwaliteit van het verhandelde product?

Het antwoord op deze vraag luidt bevestigend. Hoewel werelden waarin mensen perfecte kennis hebben nooit actueel zullen worden, kunnen aspecten uit deze werelden toch overeenkomsten vertonen met de actuele wereld. Als over die aspecten iets boeiends beweerd wordt in een theorie, dan kan deze theorie dus erg leerzaam zijn. Het is niet gemakkelijk tevoren te bepalen hoe de verzameling eco-nomisch mogelijke werelden er uitziet. Komen er interessante aspecten van de werkelijkheid (dus van @) in het zicht als je theorieën maakt over werelden die overigens allerlei niet actuele eigenschappen hebben?

Abstractie versus idealisering

Abstractie is iets totaal anders dan idealisering. Het is een operatie waarbij aspec-ten van de werkelijkheid op een bepaalde manier worden belicht, met bewuste of onbewuste weglating van allerlei mogelijk oorzakelijke invloeden.

De meeste theoretische hypothesen in de economie zijn niet voorzien van een eindig aantal keurig opgesomde voorwaarden in de clausule. Vaak houdt de theoreticus niet expliciet rekening met verstorende factoren maar abstraheert hij ervan. Abstraheren is iets totaal anders dan idealiseren. Als je een beschrijving van concrete verschijnselen op een abstracter niveau brengt, dan laat je elementen van die beschrijving weg. Als ik zeg dat een rode appel uit mijn fietstas rolt, dan zeg ik ook dat er een appel uit mijn fietstas rolt. En trouwens ook dat die appel uit mijn tas rolt, en dat hij rolt. De abstractere beschrijving volgt dus direct uit een concrete-re beschrijving, dat wil zeggen, zij geeft op het eerste gezicht geen nieuwe informa-tie. In de wiskunde wordt veel geabstraheerd: een getrouwe beschrijving van het rondje dat ik met mijn passer op papier teken vermeldt concrete details zoals een kleine onregelmatigheid waar het potlood is begonnen en geëindigd en de kleur grijs afkomstig van mijn potloodpuntje. De wiskundige beschrijving van een cirkel abstraheert van zulke details. De wiskundige werkt met abstracte objecten zoals cirkels, niet met rondjes, en liefst met cirkels met een diameter ‘d’ in plaats van een

diameter van 5,45 centimeter. Andere wetenschappers doen dat ook: liever de zin ‘de gevraagde hoeveelheid vertoont een functionele relatie met de marktprijs’ dan de zin ‘tegen een prijs van twee euro per kilo kunnen we eenenzeventig ton toma-ten kwijt’. We zagen het al: een theoreticus heeft andere interesses dan een prakti-sche veilingmeester.

Als je dus een specifiek object met een steeds abstractere term beschrijft, dan verwijst de term naar steeds meer objecten. ‘Rode appel’ verwijst naar alle rode appels, maar ‘appel’ naar alle appels. Dat zijn er meer. Maak je een uitspraak ab-stracter, dan is zij – zoals eerder uitgedrukt – ‘waar op’ een grotere verzameling werelden (al is zij misschien nog steeds onwaar op @).

Het verschil tussen idealisering (het gebruik van een onware clausule) en abstractie (een verwaarlozing van concrete eigenschappen van het onderzoeksob-ject) blijkt uit het volgende voorbeeld. Neem zin P:

(P) Ceteris paribus, de gevraagde hoeveelheid van goed A stijgt als de prijs daalt

waarbij ‘ceteris paribus’ staat voor de idealiserende uitspraak dat alle overige om-standigheden onveranderd blijven. Deze zin is waar op alle werelden waar het genoemde verband inderdaad negatief is voor goed A – dat wil zeggen, waar een stijging van het één gepaard gaat met een daling van het ander – en waar de bijbe-horende ceteris paribus clausule ook geldt. Een geabstraheerde formulering van de uitspraak is zin Q:

(Q) Ceteris paribus, de gevraagde hoeveelheid van goed A heeft een functioneel verband met de prijs

Q is waar op méér werelden, namelijk ook die waar het verband tussen de vraag naar het goed niet negatief is (en waar voorts de bijbehorende clausule geldt). Q volgt dan ook rechtstreeks uit P. Maar merk op dat het weglaten van de ceteris paribus clausule niet tot hetzelfde resultaat leidt. Zin R laat de clausule weg:

(R) De gevraagde hoeveelheid van goed A stijgt als de prijs daalt

Zin R is helemaal niet waar. Omdat R niet verwijst naar enige hypothetische we-reld, maar naar @, is er geen sprake meer van een bepaalde verzameling werelden waarop de clausule waar is. R is immers in het algemeen gesteld. De ceteris pari-bus clausule geldt niet in @, dus is R onwaar.

Ik kom tot een belangrijke conclusie. Hoe abstracter de uitspraak, des te gro-ter de kans dat hij waar is. Anders gezegd, abstractie van uitspraken (zoals van zin (P) naar zin (Q) vergroot de kans dat de actuele wereld @ element is van de verza-meling mogelijke werelden waarop die uitspraken waar zijn. Abstractie van uitspra-

ken leidt dus tot waarheidsbenadering. De klacht dat het hoge abstractieniveau van de economie schuld is van haar irrelevantie is in dit licht merkwaardig. Onwaar wordt de economische theorie er in elk geval niet van. Als een abstracte uitspraak onwaar is, dan is de concretere uitspraak die er aan voorafgaat ook al onwaar.

Behalve onschadelijk voor kwesties van waarheid is abstractie ook vaak bijzonder

nuttig. Zonder abstractie zouden we niet kunnen weten dat zowel roesten als

branden concrete verschijningsvormen van oxidatie zijn. We zouden blijven staan

bij details als het roestige oppervlak van een oude spijker en de hitte van een gas-

vlam; details die te weinig op elkaar lijken om er een verband tussen te zien.

Afsluitend kunnen we dus zeggen dat abstractie en idealisering loodrecht op

elkaar staan: al kun je tegelijk abstraheren en idealiseren, ze mogen niet met elkaar

verward worden. De eerste is een ‘verticale’ operatie en de tweede een ‘horizonta-

le’ strategie.

Abstractie in de Oostenrijkse economiebeoefening: essentialisme

In de economie raakt niemand onder de indruk van de gedachte dat abstractie een

belangrijk voertuig van het denken is, maar in de filosofie wel. De vraag rijst name-

lijk hoe voorafgaand aan wetenschappelijk onderzoek al is gekozen voor wat be-

langrijk is en in de abstracte beschrijving opgenomen wordt en van welke concrete

verschijnselen geabstraheerd wordt. Is dat niet een vraag die we pas beantwoorden

nadat we het onderwerp door en door begrepen hebben? We kunnen toch niet in

het wilde weg begrippen (zoals ‘oxidatie’) naar zaken in de wereld laten verwijzen

zonder die wereld eerst, op z’n minst deels, te begrijpen? Kennis van de wereld is

vooraf nodig om te weten welke concepten geschikt zijn om over die wereld te

vertellen, maar kennis van de begrippen stuurt tegelijk de wijze waarop we de

wereld onderzoeken. Op deze manier is de vraag door de Amsterdamse filosoof

Hans Radder gesteld en ik heb zijn werk dan ook nadrukkelijk gebruikt.

Economen uit de Oostenrijkse School hadden een grote methodologische en

filosofische interesse. Zij dachten uit het geschetste dilemma te komen door aan te

nemen dat de economie een betrekkelijk vast structureel karakter heeft. De kunst is

dan nog wel die structuur op te sporen. Maar op grond van deze aanname mag je

verwachten dat sommige begripsmatige gereedschappen een betere verklaring

helpen geven van de economische verschijnselen in hun samenhang dan andere.

Zelfs als je niet experimenteert of theorieën toetst kun je dus al een beetje zien of je

begrippenapparaat goed gekozen is. Zo wist bijvoorbeeld de econoom Eugen von

Böhm-Bawerk (1851-1914) zeker dat zijn ruilconcept van krediet een betere be-

schrijving (lees: ‘ware theorie’) gaf van de markt voor leningen dan het tot dan toe

bekende gebruiksconcept had gedaan. De reden hiervoor is dat hij met zijn concep-

tualisering een heleboel andere verschijnselen onder één noemer kon brengen. Hij

schikte de goederenmarkt, de markt voor kapitaal en de arbeidsmarkt alle onder

dezelfde abstracte idee. De eenheid die zo in zijn abstracte theorieën ontstond

weerspiegelde als het ware de eenheid die ook in de concrete werkelijkheid gele-

gen was.

Ook vandaag de dag wordt conceptuele unificatie nog als een deugd van theo-

rieën gezien. Maar de gedachte dat wetenschap een objectief gegeven structuur in

de werkelijkheid opspoort, met behulp van de enig juiste begrippen als gereed-schap, heet essentialisme. De Oostenrijkse economen hadden een essentialistische visie op hun vak. En essentialisme geeft allerlei moeilijk te kraken filosofische pro-blemen. Een typisch voorbeeld van essentialistisch taalgebruik bijvoorbeeld is als iemand stelt dat ‘de vrouw in wezen een zorgende natuur heeft’. Vrouwelijkheid is dan tot een natuurlijke soort gemaakt. Het gevaar van zo’n benadering is duidelijk: essentialisme ontkent verschillen. Het legt de wereld vast en daarmee is de moge-lijkheid van verandering – en dus ook van emancipatie – weggegeven. Andere manieren van kijken wordt onrecht gedaan. Verandering en verschil in de evolutie is daarmee ook niet meer te begrijpen. De indeling van het gewervelde leven in zoogdieren en beesten die eieren leggen (ook een indeling in natuurlijke soorten) is ontoereikend om een vogelbekdier te duiden (geen baarmoeder, wel zogen). Elke conceptualisering in andere natuurlijke soorten, essentialistisch opgevat, leidt tot nieuwe problemen, omdat er evolutionaire ontwikkeling is en omdat de mogelijk-heid van verschillende, onverenigbare perspectieven ermee is miskend.

Hoe we met begrippen de werkelijkheid vastleggen

Ik herhaal nu de vraag: hoe kunnen we tegelijk abstraheren van de wereld en een abstracte conceptualisering ervan inzetten om die wereld overzichtelijk te maken?

Stel dat twee mensen naar hetzelfde ding in een schoenendoos kijken. De één kijkt door de korte kant naar binnen en ziet een rond object. De ander kijkt door de lange kant en ziet een driehoekig object. ‘Rond’ lijkt voor de één een ge-schikt begrip om te verwijzen naar het object, en ‘driehoekig’ voor de ander. Maar

als een derde persoon de doos openmaakt zal deze een kegel zien. ‘Kegelvorm’ is dus het vruchtbaarste begrip.

Maar wat nu als niemand de kans krijgt om de doos open te maken? Dit is de situatie van de wetenschapper. Volledige kennis van alle dimensies van de werkelijkheid is niet beschikbaar, we kunnen slechts door haar smalle openingen turen om een glimp op te vangen van haar structuur en daarover een hypothese opstellen. De één zegt ‘rond’, de ander heeft een concurrerende hypothese: ‘drie-hoekig’. Deze concepten zijn enerzijds geabstraheerd van wat de twee aantreffen door te kijken, maar anderzijds is kennis van die vormen vooraf nodig om de wer-kelijkheid te kunnen duiden. Abstractie van het ‘juiste’ begrip uit de verschijnselen zoals die zich aan ons voordoen is makkelijk als je het geheim al kent, maar ge-schiedt tastend als je het niet kent.

Hans Radder gelooft niet dat de begrippen waarmee we de wereld begrijpen tot stand komen door dat wat we zien ‘samen te vatten’ zonder een voorafgaande conceptualisering. Kennis van een begrippenapparaat dat onze wijze van kijken vooraf in banen leidt is nodig om te scheiden tussen belangrijke en bijkomstige eigenschappen van de geobserveerde wereld; tussen wat in de samenvatting op-genomen wordt en wat niet. Daarmee verwerpt Radder ook een causale theorie van

betekenis. Erger nog, de causale betekenistheorie is volgens Radder essentialistisch en dus kwetsbaar voor alle bezwaren die het essentialisme aankleven.

De causale betekenistheorie zegt dat betekenis van de begrippen veroor-zaakt wordt doordat we naar een ding Z wijzen en dan zeggen: ’Kijk, dat is Z. Alle dingen die hierop lijken zijn vanaf nu allemaal een Z.’ Door naar een voorbeeld te wijzen wordt de verwijzing gefixeerd. Men spreekt in zo’n geval van een ‘starre verwijzer’, een term die steeds naar hetzelfde wijst, in alle denkbare werelden waarin het betreffende ding bestaat. Zo’n voorstelling van zaken lijkt in strijd met de evidente mogelijkheid om bijvoorbeeld evolutionaire veranderingen te zien of verschillen tussen unieke individuen te onderkennen.

Waarom essentialisme

Toch heb ik geprobeerd een rechtvaardiging van de essentialistische wetenschaps-opvatting te vinden. Waarom dat? Ik denk dat er een reden is om essentialisme een plaats te geven in de wetenschapsfilosofie. Die is gelegen in het feit dat de weten-schapsfilosofie ook rekening moet houden met wat wetenschappers zelf denken over wetenschap. Zij zijn ten slotte de specialisten.

Ook sociale wetenschappers nemen aan dat de wereld een zekere structuur heeft. Er is bijvoorbeeld ruimte en tijd, er zijn oorzaken en gevolgen en die zijn op een bepaalde manier verbonden, de sociale omgeving bestaat. De zogenaamde in-strumentalistische visie op wetenschap zegt heel bescheiden dat de structuren die in theorieën gepostuleerd worden niet meer zijn dan werkbare aannames en dat de theorie een instrument is voor kennisverwerving over de werkelijkheid, niet een

waar verhaal over welke dingen we aantreffen in die werkelijkheid. Instrumenten zijn immers niet, zoals zinnen, waar of onwaar. Deze visie staat tegenover het es-sentialisme.

Wetenschappers die het instrumentalistische standpunt aanhouden, stellen zich in de dagelijkse praktijk een gestructureerde wereld voor. Het woord ‘waar-heid’ komt dan in hun officiële mededelingen niet voor, maar intussen kunnen ze moeilijk anders dan op zoek zijn naar ware verhalen over wat zich in de werkelijk-heid voordoet. Gevraagd of zijn werk een zoektocht naar waarheid is antwoordde de econoom Robert Lucas met een weifelend “Yeah”, “Maar ik weet eigenlijk niet wat dat betekent”. Dit antwoord typeert veel economen. Daarbij maken weten-schappers gebruik van een begrippenapparaat dat ze niet graag van de ene dag op de andere inruilen voor een alternatief. Ze gedragen zich alsof hun concepten de best mogelijke zijn, of dat ze zoeken naar de best mogelijke. ‘Rond’ en ‘driehoekig’ zijn onjuist, ‘kegelvormig’ is juist. Zulk gedrag is natuurlijk heel praktisch en ver-standig. Hoe zij daar, als het ware buiten werktijd, ook over denken, in gebruik is het essentialistisch. Zo gezien is instrumentalisme een luxeproduct. Eerst gaan we immers met essentialistische intuïties werken aan de grenzen van het weten. Als we eenmaal verklarende en voorspellende theorieën hebben, geschraagd door een doorontwikkeld begrippenapparaat, dan permitteren we het ons pas om instru-mentalist te zijn. Het ligt voor de hand zulke essentialistische intuïties als natuur-lijk gegeven te beschouwen (filosoferen met historische adequaatheid) en te kijken in hoeverre deze intuïties analytisch nog te handhaven zijn (filosoferen met norma-tieve adequaatheid). Ik heb geprobeerd beide te doen.

Starheid de jure

Het verband dat Radder tussen de causale betekenistheorie en essentialisme ziet ligt voor de hand. Maar er is een opvatting van de causale betekenistheorie en van starre verwijzers die niet met een rigide vorm van essentialisme geassocieerd hoeft te worden. Deze benadering, van Joseph LaPorte, laat aan de ene kant toe dat men verschillende vruchtbare perspectieven op de wereld kan hebben en aan de andere kant dat natuurlijke soorten bestaan.

In de eerste plaats sluit de causale betekenistheorie geen vaagheid uit, maar kan zij vaagheid juist zichtbaar maken. De vondst van het vogelbekdier is een voorbeeld van een moment in de geschiedenis van talen waarop onzekerheid ont-stond. Er bleek een nieuwe term nodig. Men besloot uiteindelijk om ook het vogel-bekdier een zoogdier te noemen en veranderde daarbij de verwijzing van de term ‘zoogdier’ enigszins. Opnieuw kon nu gezegd worden: ‘kijk, dit hier is ook een zoogdier, net als mensen en walvissen’.

In de tweede plaats kan men naar de dingen verwijzen met een term die star is voor alle denkbare werelden waarin een ding bestaat, als gevolg van een simpel be-

sluit. De term verwijst alleen naar het ding in de verzameling werelden die de ge-

bruiker van de term, als het ware, oproept. Toen nog niet ontdekt was dat walvis-

sen geen vissen waren kon je de term ‘zoogdier’ gebruiken om te spreken over

dieren die alle op het land liepen. En gegeven de toenmalige definitie van ‘vis’

verwees deze term inderdaad in alle werelden waar walvissen en andere vissen

zwommen naar die zwemmende natuurlijke soort die vissen (in de huidige bete-

kenis) én walvissen omvat. Dit zelfs ondanks dat men later de term ‘zoogdier’ her-

definieerde om er ook walvissen onder te laten vallen. Anders dan door tegenstan-

ders van essentialisme vaak wordt gedacht, betekent de starheid van de verwijzing

niets voor de structuur van de wereld maar wel iets voor hoe taal functioneert. Als

gevolg hiervan gaan theorieveranderingen vaak samen met betekenisveranderin-

gen. Von Böhm-Bawerk probeerde zelfs theoretische vooruitgang te boeken door

het stipuleren van nieuwe betekenissen.

In de derde plaats geeft de causale betekenistheorie geen beletsel voor com-

municatie tussen aanhangers van verschillende theorieën. Elk nieuw theoretisch

inzicht gaat gepaard met een wereldbeeld dat geladen is met veronderstellingen

over de aard van de wereld. Zulke veronderstellingen noemen we ‘metafysisch’.

Maar omdat in de termen waarin dat wereldbeeld wordt gevat valt altijd een dun-

nere, metafysisch meer neutrale en minimale betekenis te onderkennen valt, is er

altijd een verstandhouding mogelijk tussen strijdende wetenschappers.

Essentialisme en perspectivistisch essentialisme

Voor essentialisme kan nu een onderscheid aangebracht worden tussen essentia-

lisme zonder meer en perspectivistisch essentialisme. De eerste positie zegt dat pas

als wetenschappelijke theorieën in de enig juiste concepten geformuleerd zijn, deze

theorieën helpen om, als het ware, de scharnieren van de wereld te vinden. Die

dicteren de keuze van het begrippenapparaat, precies zoals Von Böhm-Bawerk

meende dat hij wetenschap moest bedrijven. De tweede zegt dat er voor elke invul-

ling van de gebruikte concepten een metafysisch geladen perspectief op de werke-

lijkheid bestaat. Met zo’n perspectief, eenmaal ingenomen, gaat een bijbehorende

essentialistische wetenschapsopvatting samen.

Erkenning van andere mogelijke perspectieven sluit een essentialistische

opvatting geenszins uit. Von Böhm-Bawerk’s wetenschapsbeeld is zo op te vatten.

University of Groningen Conceptual progress in economics

Documents

University of Groningen Roadmap Paganetti, Harald; Beltran

University of Groningen Dakṣiṇa Kosala Bosma, Natasja

H.Landman.pdf - Rijksuniversiteit Groningen

University of Groningen Expectancy-value antecedents and

University of Groningen Neurolinguistic and psycholinguistic

University of Groningen Ensuring meaningful lifelong learning

Chapter_2.pdf - the University of Groningen research portal

HISGIS Groningen. Opzet, verrijkingsmogelijkheden en prioriteiten (Leeuwarden 2013)

HERINRICHTING OOST-GRONINGEN EN ... - Noorderbreedte

Chapter_9.pdf - Rijksuniversiteit Groningen

DRAFT - A progress resport on the science of conceptual systems

University of Groningen Father Absence Girigori, Odette Janice

University of Groningen - Student theses Faculty of Theology

University of Groningen Immuno-Imaging to Predict Treatment

RIJKSUNIVERSITEIT GRONINGEN TRADEABLE CO2

University of Groningen Websites as Information Hubs Kleinkercher

University of Groningen 9th SC@RUG 2012 proceedings

University of Groningen Between direct and indirect speech

University of Groningen Hymenoptera venom allergy Vos, Byrthe

proefschrift Leonie de Weerd.indd - Rijksuniversiteit Groningen