Knut Wicksell: Selected Essays I

Citation preview

KNUT WICKSELL

‘Usually when a great economist is translated into English, reputation is deflated. Not so with Wicksell.’ Paul A.Samuelson ‘No finer intellect and no higher character have ever graced our field’ wrote Joseph Schumpeter about the Swedish economist Knut Wicksell (1851–1926). Wicksell made lasting contributions to capital theory, monetary theory and fiscal policy. However, whilst most of his books, originally published in German or Swedish, have long been translated into English, only a few of his more than eight hundred articles are available in English. This volume fills part of that huge gap by presenting a wide range of hitherto untranslated material. This volume includes: • contributions to neoclassical marginalism and Austrian capital theory; • essays on income, taxes and duties, including Wicksell’s doctoral thesis On the Theory of Tax Incidence; • one of Wicksell’s few articles on unemployment. A second volume contains essays on monetary questions and population, as well as a number of Wicksell’s book reviews. Bo Sandelin has been at the Department of Economics, University of Gothenburg, Sweden since 1973. His publications in international journals include articles on the economics of housing, the economics of crime, capital theory and the history of economic thought, with special reference to Wicksell’s capital theory. He has published several books in Swedish, and edited The History of Swedish Economic Thought (Routledge, 1991).

KNUT WICKSELL Selected essays in economics

Edited by Bo Sandelin VOLUME I

London and New York

First published 1997 by Routledge 11 New Fetter Lane, London EC4P 4EE This edition published in the Taylor & Francis e-Library, 2003. Simultaneously published in the USA and Canada by Routledge 29 West 35th Street, New York, NY 10001 © 1997 Bo Sandelin All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloguing in Publication Data Wicksell, Knut, 1851–1926. [Essays. English. Selections] Knut Wicksell: essays in economics/edited by Bo Sandelin. p. cm. Translated from German or Swedish. Includes bibliographical references and index. ISBN 0-415-15512-6 (alk. paper) 1. Economics. I. Sandelin, Bo, 1942– . II. Title. HB34.W48213 1996 96–9038 330–dc20 CIP ISBN 0-203-44354-3 Master e-book ISBN

ISBN 0-203-75178-7 (Adobe eReader Format) ISBN 0-415-15512-6 (Print Edition)

CONTENTS

Preface Introduction

vii viii

Part I Marginalism and capital theory 1 IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

3

2 KAPITAL—UND KEIN ENDE! (Reply to Docent Brisman)

15

3 LEXIS AND BÖHM-BAWERK

26

4 ON THE THEORY OF INTEREST (Böhm-Bawerk’s ‘Third Ground’)

41

Part II Income, taxes and duties 5 ON THE THEORY OF TAX INCIDENCE

57

6 THE INHERITANCE TAX

116

7 TARIFFS AND WAGES

153

8 HIGH PRICES, TARIFFS AND WAGES

158

9 VOLUNTARY OR FORCED SAVINGS?

166

10 THE HISTORICAL DEVELOPMENT OF THE CONCEPT OF INCOME

170

11 THE CONCEPT OF INCOME AS REGARDS TAXATION, AND SOME ASSOCIATED TAX ISSUES

213

12 A FEW COMMENTS

240 v

CONTENTS

Part III Unemployment 13 WHY ARE FACTORY OPERATIONS BEING CURTAILED? Index

253 261

vi

PREFACE

This is the first of two projected volumes of previously untranslated writings by Knut Wicksell. The publishers have already responded to interest in Wicksell by publishing three volumes of articles on him in their Critical Assessments series, edited by John Cunningham Wood. Altogether, then, the availability of material both by and about Wicksell has now been improved for Englishlanguage readers. The texts have been translated by Dr Timothy Chamberlain (1, 4–6, 8– 13) and Julie Sundqvist (2–3, 7). Working with the translators has been very stimulating, and I have been impressed by their skill in interpreting Wicksell’s frequently complicated way of expressing himself. They deserve special thanks for their efforts. The translation has been financed by the Swedish Council for Research in the Humanities and Social Sciences. I should like to thank the Council for its support, which has made this project possible. Bo Sandelin

vii

INTRODUCTION

‘Einige werden posthum geboren’—some are born posthumously—says the German philosopher Friedrich Nietzsche in his autobiography; he has himself in mind. From the perspective of English-language readers, the same claim can be made for the Swedish economist Knut Wicksell (1851–1926), since his most important writings only began to be published in English after his death. Wicksell’s best-known work, the two volumes of Lectures on Political editions had already appeared, the first in 1901–6. In addition, a German ediEconomy, was published in English in 1934–5. By this time, three Swedish tion had come out in 1913–22.1 The Lectures on Political Economy were followed almost immediately by the translation of Interest and Prices. A Study of the Causes Regulating the Value of Money (1936).2 The original edition (in German) had been published in 1898. Wicksell’s first book, Value, Capital and Rent, was published in German as early as 1893, but had to wait until 1954 before appearing in English translation.3 Wicksell’s main work on the theory of public finance, Finanztheoretische Untersuchungen nebst Darstellung und Kritik des Steuerwesens Schwedens (1896) is still not available in its entirety in English. The main part of the section ‘A New Principle of Just Taxation’ has, however, been translated into English by James Buchanan.4 The preceding section, ‘On the Theory of Tax Incidence’, is published in English for the first time in this collection. Wicksell himself published very little indeed in English—barely a dozen short articles, about half of them on population. He published a couple of brief contributions on the gold problem in the Economic Journal and an article on international freights and prices in the Quarterly Journal of Economics. All his more significant essays were published in Swedish instead, and to some extent in German. Ekonomisk Tidskrift, which was renamed the Swedish Journal of Economics in 1965 and the Scandinavian Journal of Economics in 1976, was his most important scholarly forum in Swedish. Wicksell’s German articles were spread among several different journals, most prominently Jahrbücher für Nationalökonomie und Statistik. Fourteen articles viii

INTRODUCTION

have previously been translated for a volume edited by Erik Lindahl, entitled Knut Wicksell. Selected Papers on Economic Theory (1958). A bibliography of works by Wicksell printed between 1868 and 1950 lists 889 titles.5 In addition, he left about one hundred unpublished manuscripts, which are now preserved in Lund University Library.6 As stated above, very few of Wicksell’s works were originally published in English, but some have been translated. All things considered, a very small proportion of Wicksell’s work has been available in English.7 The present collection is the first of two projected volumes containing further translations from Wicksell’s ample production.

PRINCIPLES OF SELECTION The selection has been guided by the following principles. First, only material that—as far as I have been able to establish—has not previously been published in English is included. Second, only writings on economics have been taken into account. (A few works on the closely related issue of population are included in the second volume.) The emphasis therefore lies on Wicksell’s role as a scholar of economics, while his contributions to general public debate are passed over. However, it is worth pointing out that in Wicksell’s lifetime it was his activity as a provocative participant in public debate that made him known to the public at large. Here, no topic was beyond his range. He wrote articles or gave lectures on subjects as far apart as defence, euthanasia, spiritualism, marriage, prostitution, foreign policy, issues of law and religion. On one occasion, he dealt with the latter topic in a lecture in so deliberately provocative a manner that as a reputable professor in his late fifties he spent two months in prison in 1909 for ‘blaspheming and mocking God’s holy word in such a manner as to occasion public offence’. Wicksell’s many contributions to public debate and letters to the press undoubtedly retarded his career as an economist. But he liked journalism and, in addition, for many years this kind of activity provided a necessary contribution to his livelihood. He had no regular income until he was fifty, when he became Professor of Economics and Fiscal Law at Lund University. Nevertheless, his contributions to the daily newspapers are not represented in this collection. A third criterion is that, to be included, an article should throw light on some principle. I have tried to avoid articles that focus too narrowly on a specific case and have little general interest. Fourth, Wicksell published a number of articles that were draft versions of chapters of his books, or that summarized central sections of his books. I have avoided those articles because they do not add much to what the reader can find in the books themselves. ix

INTRODUCTION

Fifth, Wicksell, however, also wrote articles in which he clarified points that may have seemed obscure in his books or in earlier articles, or in which he concurred with critical comments, or, on the contrary, defended his propositions against attacks. A few articles of this kind are included. One example is the first article in this collection, ‘In Defence of the Theory of Marginal Utility’, which contains a critique of ideas that Gustav Cassel had put forward in his article ‘Grundriß einer elementaren Preislehre’. Among other things, Cassel censures the concept of marginal utility in price theory used by Wicksell in Value, Capital and Rent, as well as by other authors. Cassel prefers to go directly to the relationship between price and quantity demanded. Another example is the second article (‘Kapital und kein Ende’), which deals with the concept of capital and the return to capital and is mainly a response to criticism levelled by Professor Sven Brisman against Wicksell’s views after the publication of the first part of the second edition of Lectures on Political Economy. WICKSELL’S AREAS OF RESEARCH AND THIS COLLECTION Most of Wicksell’s scholarly writings can be classified in three broad groups. The first comprises his work on microeconomic price theory and capital theory, the second his work on fiscal theory and fiscal policy, and the third his monetary or, to use a term more adequate today, macroeconomic writings. His best-known publications on price theory and capital theory are his books Value, Capital and Rent and the first volume of Lectures on Political Economy. Even if his achievement here largely consisted in clarifying and refining the thoughts of earlier authors—particularly Böhm-Bawerk, Jevons and Walras—he also made enduring original contributions. His account of the phenomenon in capital theory that has come to be known as the ‘Wicksell effect’ played a major role during the so-called Cambridge controversy in the 1950s to 1970s. Wicksell demonstrated here that problems can arise if capital is treated like any other factor of production. It is an irony of fate that the ideas of the neoclassicist Wicksell were used to support the position of the English Cambridge School, which opposed neoclassicism. Joan Robinson was even of the opinion that the Wicksell effect not only obstructs the construction of an aggregate production function and challenges orthodox marginal propositions, but it ‘is the key to the whole theory of accumulation and of the determination of wages and profit’.8 In this collection a few articles in the first research category are brought together under the heading Marginalism and Capital Theory. In addition to the two articles already mentioned, ‘In Defence of the Theory of Marginal Utility’ and ‘Kapital und kein Ende’, there is first an obituary tribute to Lexis and Böhm-Bawerk. Here Wicksell explains the enormous significance Böhmx

INTRODUCTION

Bawerk’s Positive Theory of Capital had for him; the book hit him ‘like a revelation’. The final essay in this section, ‘On the Theory of Interest’, was originally published posthumously and contains a critical discussion of BöhmBawerk’s grounds for the rate of interest, especially the third ground. Wicksell’s central work on fiscal policy is his book Finanztheoretische Untersuchungen nebst Darstellung und Kritik des Steuerwesens Schwedens (1896). What has attracted most attention here is his argument for the benefit principle in taxation, i.e. that the tax imposed on a given individual should be in proportion to the benefit that person receives from the tax. Further, in Wicksell’s version, the principle implies ‘approximate’ unanimity and voluntary consent in taxation. Wicksell’s views have been made known to the English-speaking world by James Buchanan’s translation of the relevant section of the book, but they had previously been further developed by Erik Lindahl in his doctoral dissertation Die Gerechtigkeit der Besteuerung. Eight pieces on fiscal policy and related areas make up the second part of our collection, entitled Income, Taxes and Duties. It begins with ‘On the Theory of Tax Incidence’, which was Wicksell’s doctoral thesis, defended in 1895 and included as the first part of Finanztheoretische Untersuchungen. ‘On the Theory of Tax Incidence’ was a short thesis, only seventy-five pages, the reason being that Wicksell wanted to keep down the cost of the 360 complimentary copies he was obliged to submit to the university. In his thesis, Wicksell sorted through existing ideas on the effects of taxes on monopoly profits, indirect taxes and income taxes. The second piece on taxes is an article from Ekonomisk Tidskrift on the inheritance tax. In Wicksell’s time, due attention was paid to fundamental conceptual issues, and this informs the whole essay. Thus, should what is called inheritance tax be seen as essentially a fee or a tax, or should it be seen as an expression of the view that the state and municipality also have a right of inheritance when one of their members dies? And where do considerations of social policy lead? Wicksell’s policy conclusion is that the inheritance tax ought not to be used in its entirety for current expenses but that most of it should go to capital formation on the part of the state and the municipality. After a few shorter articles on various issues, including tariffs, we have two lengthy articles on the concept of income. They derive from a government report written by Wicksell and David Davidson. (One of Wicksell’s contributions was also published in Ekonomisk Tidskrift.) It is worthy of note that Wicksell and Davidson were unable to agree on the appropriate concept of income in connection with taxes.9 Davidson wanted to view all revenues contributing to increased wealth or employed in consumption as taxable income. Consequently he included, for example, inheritances, lottery winnings and increases in the value of land. Wicksell wanted to restrict taxable income to the individual’s contribution to the national income. In his view, inheritances, increases in the value of land and similar revenues should be xi

INTRODUCTION

treated separately and be confiscated by the state to the extent that they were unearned; they were not part of taxable income. The final, short article in the second section is entitled ‘A Few Comments’. This article responds to views on the issue of taxation advanced by David Davidson, in part polemically against Wicksell, and which are based on the so-called principle of ability to pay. Wicksell himself, as I have said, is generally cited as an important advocate of the benefit principle. Here, Wicksell tones down the difference between the two principles, and he clarifies his own position in a way that some of his interpreters may find surprising. His intention was not to transform taxes into voluntary payments. ‘The voluntarism I talked about only applied to the actual passing of tax bills’. Furthermore, ‘These days, taxation is actually exercised by and by means of the taxpayers themselves [because of the universal franchise in parliamentary elections and the right of parliament to decide on taxes]. That the result must then be taxation according to perceived benefits, as far as these are capable of making themselves felt, seems to be not merely, as Davidson would put it, ‘a postulate’, but a real axiom’. Thus, in a democratic, parliamentary system, the benefit principle is de facto the main principle applied. Wicksell did not write much about unemployment. 10 Nor did his contemporaries. As is evident from the single article (‘Why are Factory Operations being Curtailed?’) in the last section of this volume (‘Unemployment’), unemployment was an ‘irrational, almost incomprehensible’ phenomenon. For a rational economist it seemed like a paradox that it could ever pay to do nothing, to leave the available productive forces unused. The fundamental question in economics was how to manage an economy with scarce resources, but the question of unemployment was of the opposite nature. This was undoubtedly one of the reasons why the neoclassical economists of the time, including someone as engaged in social policy as Wicksell, did not really feel that unemployment was a central field for their discipline. I have argued that most of Wicksell’s scholarly writings can be classified into three broad groups, the third of which comprises his monetary writings. Two of his books have to do with this area, namely Interest and Prices (1898) and the second volume of Lectures on Political Economy (1906). The element that has attracted most attention in this context is undoubtedly Wicksell’s cumulative process, which he presents in both books, though with minor differences.11 The main idea here is that if the rate of return on investments is higher than the lending rate, it will be profitable to increase production, which will lead—via increased demand for the factors of production, higher incomes and greater demand for consumer goods—to inflation. This process will continue as long as the difference between the rate of return on investments (the ‘natural rate of interest’) and the banks’ lending rate persists. There was no space for Wicksell’s monetary articles in this volume, but a selection will make up the first part of a second volume that is in preparation. The second volume will also include writings on the population question, xii

INTRODUCTION

and a number of reviews written by Wicksell of books by his contemporaries Walras, Clark, Seligman, Mises, Knapp and others. Bo Sandelin NOTES 1 The English edition was followed by editions in Japanese (1938–9), Spanish (1947) and Italian (1950). A new Japanese translation of Volume 1 was published in 1984. 2 Also published in Japanese (1939 and 1984). 3 A Japanese edition was published in 1937. 4 It was published in Classics in the Theory of Public Finance, ed. R.A.Musgrave and A.T.Peacock, London: Macmillan, 1958. The same section and the preceding section (‘On the Theory of Tax Incidence’) were published in Japanese in 1995. 5 Erik J.Knudtzon, Knut Wicksells tryckta skrifter 1868–1950, ed. Torun HedlundNyström. Lund: CWK Gleerup, 1976. 6 These were assembled for publication by Torun Hedlund-Nyström and Lars Jonung in 1985–6, but have not yet appeared in print. Parts of Wicksell’s correspondence, especially letters to Wicksell, have been compiled by Hitoshi Hashimoto and published (in Swedish) in 1992–5 in numbers 19–22 of Economic and Business Review, published by the Society of Economics and Business Administration at Kyoto Sangyo University, Japan. Hashimoto has also put together a catalogue of 4,007 letters to Wicksell, published in The KeizaiKeiei Ronso, The Economic and Business Administration Review, vol. 28, no. 3, Dec. 1993. 7 There is a good deal of material in English by other authors on Wicksell. The classic biography is Torsten Gårdlund, The Life of Knut Wicksell, Stockholm, 1958. A comprehensive survey of Wicksell’s economic thought is given by Carl G.Uhr, Economic Doctrines of Knut Wicksell, Berkeley and Los Angeles: University of California Press, 1962. There are two major collections of articles on Wicksell: Knut Wicksell (1851–1926), Pioneers in Economics 28, ed. Mark Blaug, Aldershot, Hants.: Edward Elgar, 1992, and Knut Wicksell, Critical Assessments, ed. John Cunningham Wood, London and New York: Routledge, 1994. The Scandinavian Journal of Economics, vol. 80, no. 2, 1978, is a special Wicksell issue. 8 Joan Robinson, The Accumulation of Capital, London, 1956, p. 396. A comparison of Wicksell’s approach with that of later economists is given in Bo Sandelin, ‘Wicksell’s Wicksell Effect, the Price Wicksell Effect and the Real Wicksell Effect’, in Perspectives on the History of Economic Thought, vol. 6, ed. William J.Barber, Aldershot, Hants.: Edward Elgar, 1991. 9 A brief survey is provided in Lars Söderström and Bo Larsson, Svensk skatteforskning 1919–1979. En annoterad bibliografi, Riksbankens Jubileumsfond, 1981. 10 Lars Jonung has written a survey article which shows that Wicksell’s few works in this field mainly date from the years immediately following the First World War, when unemployment was high in Sweden. See Lars Jonung, ‘Knut Wicksell on Unemployment’, History of Political Economy, vol. 21, no. 1, 1989, pp. 27–42. These articles, which seem to have numbered fewer than ten all told, were published in very diverse forums. Wicksell’s article on ‘Ricardo and the Present Unemployment’ was sent to the Economic Journal, but was rejected by the editor, Keynes. It was, however, published in the same journal in 1981 with an introduction by Jonung. 11 C.-H.Siven describes in ‘The Early Swedish Debates about the Cumulative Process’ (manuscript, 1995) how Wicksell changed his mind. xiii

Part I MARGINALISM AND CAPITAL THEORY

1 IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

In the third number of this journal for 1899, my fellow-countryman and friend Dr G.Cassel published an essay ostensibly intended as a synthetic presentation of the theory of prices according to Walras, but actually comprising in the main a critique of modern theories of value.1 This critique boils down to the judgement that ‘absolutely nothing is left of the formulas presented by the theorists of marginal utility that is capable of standing up to rigorous criticism’. However, Cassel’s essay includes a substantial number of arguments whose correctness must be challenged. Moreover, since I (along with many other writers) am directly attacked by Cassel, I take the liberty of making a brief response. Cassel first criticizes the theorists of marginal utility on the grounds that it is impossible to measure in real terms and make a direct comparison between the various needs either of a single person or of a number of people compared with one another. For any measurement, he claims, demands a unit of measurement, and the theorists of marginal utility have never established and can never establish a unit of this kind. Rather, we only gain a measurable sign of psychological processes, of the varying intensity of our feelings, by observing some kind of outward effect they have, i.e. in the case in question, by evaluating the goods to be bought or sold in terms of some conventional yard-stick, most simply in monetary terms. ‘In money’, says Cassel, ‘the individual possesses a scale of value by the aid of which he is able not only to classify his needs, but also to express their relative intensity in numerical terms. If need be (i.e. if I cannot get it more cheaply) I am prepared to pay 10 marks (but no more) for a certain good. For another good I might perhaps pay up to 20 marks. This means that not only is the second good more important to me than the first, it is also, over and above that, precisely twice as important.’ In passing, in conventional linguistic usage this is only true when relatively small portions of my assets or income are involved—in other cases it is false. For example, I should if necessary spend half my income on accommodation Originally published as ‘Zur Verteidigung der Grenznutzenlehre’, Zeitschrift für die gesamte Staatswissenschaft, 1900.

3

MARGINALISM AND CAPITAL THEORY

and dress appropriate to my social class, assuming the prices of other goods remain unchanged; for food I cannot, of course, spend more than my whole income, even if I need to. Does that mean that food, for me, is only twice as important as accommodation and dress appropriate to my social class? ‘Thus’, Cassel continues, ‘money is a scale of value for the individual, and by means of trade it becomes a shared, public scale of value, too. For if A and B are prepared at any time to exchange their units of value, the one mark coin, for the same goods, this proves that A’s and B’s scales of value are in fact identical…. It is of course impossible a priori to compare the intensity of A’s needs with B’s. At least such a comparison lies completely outside the domain of economics. But if I make the assumption that A’s and B’s needs are of equal intensity as soon as they both evaluate these needs at the price of one mark, then I have derived from the psychological presuppositions all that is of significance for the economic side of the matter.’ No more precise justification for these claims is to be found in Cassel’s essay. In response I should now like to observe the following. To take first the case of a single individual, a comparison, and a direct comparison at that, between the intensity of our various needs is not only possible, it takes place, as it were, every minute of our life. According to circumstances, each of us prefers to satisfy one of our needs—for sleep, reading, exercise in the fresh air, etc.—rather than the others, without any need to undertake a prior valuation of these needs in monetary terms. Of course, in this process our judgement of value generally goes no further than to rate two different needs as approximately equal in importance to us, or to rate one of them somewhat higher, or even markedly higher than the other, but it is really only a step from this to a precise numerical evaluation. For example, let us suppose a boy already in possession of a stock of apples and nuts is prepared to exchange one of his apples for ten more nuts, but is also, on the other hand, prepared to give up nine nuts from his stock in order to acquire one more apple. Then, obviously, he values one apple above nine, but below ten nuts, that is, at approximately 9 nuts. But not only is that the case: under the given conditions he values the nuts to be in turn given up or acquired almost equally, and therefore considers the value of an apple 9 times that of a nut. Now for those around him, to be sure, the boy’s judgements of value are perceivable only through their ‘symptoms’, their ‘economic expressions’, but for the boy himself the unmediated intensity of his feelings, the amount of pleasure he anticipates from the apples on the one hand and the nuts on the other, is obviously decisive. It is clearly this, and nothing else, that Böhm-Bawerk intends to emphasize in his polemic against Dietzel, mentioned by Cassel (p. 399). Böhm-Bawerk ‘does not (as Cassel believes) claim to know how much Robinson’s hut is worth’. He merely asserts that Robinson himself knows it, in other words, that he has an unmediated sense of the utility of his hut and of his stock of provisions, and that it is just this sense that forms the basis of his judgement of their value. 4

IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

Of course, it is a rather different matter when dealing with different people, or with the same person in different circumstances. A direct comparison between the intensity of the emotions felt by different individuals is of course out of the question, but this by no means prevents a comparison being possible—not as might be supposed by means of money, but rather via induction and analogy. When dealing with people of the same age, the same sex, possessing the same degree of education, etc., one can reasonably assume that their elementary needs—precisely those that are significant for the science of economics—are virtually the same; and if in addition they have equal wealth, then the extent to which they can and will satisfy these needs is surely probably identical. If on the other hand all these circumstances or even just some of them are different, then of course the comparison will involve the greatest difficulties; it could really only be made by an individual who himself had lived in all these circumstances and preserved a vivid memory of his impressions. Here indeed the imperfection of the standard is commonplace: the old forget they were once young, the rich they were once poor (if they ever were poor at all), the rulers are incapable of comprehending the ruled as creatures of their own ilk, etc. But this does not mean giving up hope that the technique of measurement may be perfected. On the contrary, progress in psychophysics towards this end promises well. Fechner’s well-known psychophysical law—first proposed, if I am not mistaken, by E.H.Weber for sensations of pressure—according to which the minimum distinguibile, the just perceptible change in an impression on the senses within certain limits, always demands the same quota of the strength of the stimulus used in each case, constitutes in fact a kind of confirmation of D.Bernoulli’s and Laplace’s earlier speculations about the relation between fortune morale and fortune physique. To the best of my knowledge, no theorist of marginal utility doubts that in the case of real—not just theoretical—measurements of this kind one necessarily requires a unit determined in advance, and that this unit can only be some use or another, and a concrete use at that, taking effect under quite specific conditions—let us say, for example, the use that a pair of work boots affords a middle-aged agricultural labourer living in Brandenburg over the course of a year. The only one who seems to overlook this fact is Cassel himself when he recommends money as a general standard even for subjective valuations, for like other things, money to be sure is useful, but this usefulness is obviously completely different in differing circumstances. It is like wanting to use the length of the day at the winter solstice as a measure of time, or the length of the seconds-pendulum’s swing, which also varies with geographical latitude, as a unit of length. I hardly need to remind Cassel that the element assumed as fixed in measurements and similar processes can never be something purely conventional. It only became possible to make the rate at which the earth turns the basis for all measurements of time when a process of induction had demonstrated that it possesses in the highest degree the quality that we understand as uniformity of motion. As we have seen, Cassel 5

MARGINALISM AND CAPITAL THEORY

himself terms the process in question ‘an assumption’, at least when dealing with different individuals, but he has been unable to convince me of the usefulness or even the admissibility of this assumption. In the theory of prices, assuming free competition, it is harmless, to be sure, but only because there it never comes to be employed in practice, since, as Cassel himself emphasizes, when determining prices under conditions of free competition there is never any need at all to compare the varying utility or marginal utility a good has for different people, but rather, only its relative significance over against the numeraire good for one and the same person. However, price formation under free competition far from exhausts the domain of economics, and as soon as one progresses from purely individualistic to altruistic or social points of view, the concept he proposes, or rather the way in which he confuses concepts, turns out to have dire consequences. Cassel himself later admits this, moreover, even if with a rather dissatisfied air! Oddly enough, Cassel later attributes to other writers his own definitions of subjective value and marginal utility in place of those they use themselves, as I shall now show. In order to determine the price ratio of two commodities on the market Walras, as is well known, assumes this ratio itself as an independent variable and the quantities of goods given and received in exchange as variables dependent on it. Jevons, in contrast, had viewed precisely these quantities as the independent variables in the problem, but obviously that means taking into account as many (or twice as many) unknown quantities as there are exchanging persons. If on the other hand we restrict our attention to just two exchanging persons, which may have been Jevons’s original notion, then again the law of competition (the law of indifference according to Jevons), which entails that there can only be a single uniform price for each commodity on the market, is deprived of its effect. For this reason Jevons has hit on the idea of bringing commodity owners together into two groups, so-called trading bodies, but now has to operate with the unclear and in fact undefinable concept of a marginal utility valid for each of these groups in its entirety. Cassel considers the observation I make on this (Über Wert, Kapital und Rente, p. 47) ‘incorrect’. ‘If I know’, he says, ‘that I can sell to a group of buyers who have already acquired x kg of a commodity an additional 1 kg for at most y marks, then I am surely justified in designating y this group’s marginal utility. When I use this term, it must simply be ignored for a moment who buys this final kg; for this buyer in any case the marginal utility is equal to y, and the marginal utility of the group coincides with his.’ I could of course simply respond that this concept of marginal utility is actually not the one Jevons had in mind; but in addition I must emphasize that the method indicated by Cassel is just as impracticable, unless it is a matter of the purchase and sale of a few discrete items of the same size and quality, as in Böhm-Bawerk’s well-known example of the horse market. For the level of the marginal utility in Cassel’s sense depends not only on the 6

IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

size of the stock of goods acquired, but also on its distribution among the buyers (and similarly with the seller’s marginal utility). But in order to ascertain this level—assuming the personal allocations of the exchanging individuals as given—one has to solve the original problem or a very similar one at every step, over and over again, or, to put it differently, one has to investigate by whom and at what maximum possible price the first, second, third, and so on up to perhaps the hundred thousandth kilogram of the commodity might be acquired. Certainly this is a method, too, but one that functions very slowly. In this connection Cassel upbraids me for ‘not properly understanding the significance of a common measure of value’, and consequently ‘also failing to lay the axe to the root of all the confusion that Launhardt has brought forth in this area under the cover of mathematics’. The answer to this reproach is already given in the preceding passage. I do not consider incorrect in principle Launhardt’s treatment of the marginal utilities of different persons as commensurable quantities, and in my opinion he has caused no confusion by this procedure. In general Cassel could have treated this author somewhat more kindly, for on the very next page (414) he reaps the benefits of Launhardt’s work. For the elegant method he employs there, ‘diverging from the customary manner of presentation’, by treating all consumers of a commodity uniformly, whether with prices at a given level they be buyers or sellers, originally comes, to the best of my knowledge, from Launhardt. At least I adopted this method from Launhardt, and Cassel probably found it in my work (Über Wert, Kapital und Rente, pp. 52f). The section of Cassel’s essay that might be expected to arouse most interest among theorists of marginal utility is probably the third, where he endeavours to show that the equality or rather proportionality they assert between marginal utility and commodity prices in general does not obtain. This assertion, according to him, rests on ‘two assumptions, namely, first, that consumer goods are divisible at will, and secondly, that our estimations of value represent continuous functions of the quantity previously possessed’. These assumptions, he says, are ‘both fundamentally incorrect’. Now, concerning the first of these assumptions, surely nobody has advanced it as a generally valid fact, and to this extent of course the strict law of marginal utility suffers certain exceptions, as also is generally recognized. On the other hand, there is a fact that is regularly overlooked by Cassel, which entails that those goods which by their very nature can only be employed as discrete units are subject to the laws of marginal utility no less than those divisible at will. This is the fact that in reality they almost always occur in numerous different qualities or nuances. Let us hear how Cassel argues his case. Cassel ‘lives at present in Berlin and rents a room there’. The room costs 30 marks, and he ‘assumes for the sake of simplicity’ that in Berlin there 7

MARGINALISM AND CAPITAL THEORY

exist ‘only similar rooms and, in fact, all at the same price. If the price rose to 100 marks per month, Cassel would ‘probably pay it’; but for an additional room he is only prepared to give at most 5 marks. Therefore, he argues, there is no question of the marginal utility according with the price, neither on one side nor on the other. Now it is clear, however, that the assumption that Cassel makes ‘for the sake of simplicity’ in reality begs the question and decides the entire issue in advance in the way he desires. In fact not just one, but a hundred kinds of room exist in Berlin; if Cassel were prepared to invest even 35 marks he would certainly be able to obtain a more spacious, brighter, more favourably located or in some other way more convenient room in the same part of the city, perhaps even in the same house. If he is not prepared to pay this price, then that of course shows unambiguously that he does not value the better room and therefore a fortiori not the worse room he currently lives in at quite 35 marks. Similarly, he could certainly obtain a simpler or more inconveniently located room for 25 marks; but in fact he prefers the one he lives in now for 30 marks. What could show better that he raises or lowers his demands for space, comfort, location, etc. to the point at which a further addition to or subtraction from these qualities, i.e. what one could designate in the real meaning of the term as the marginal utility of Berlin rooms valid for him, is approximately equivalent to the price demanded for it? But even in the case of goods which he himself conceives as divisible at will, the proposition mentioned above does not in general hold, according to Cassel. ‘Let us take’, he says, ‘my consumption of sugar as an example. I do not consider it particularly healthy to eat sugar, yet on account of its pleasant flavour I consume, say, 10 kg of sugar a year. I should quite certainly consume this same quantity even if sugar cost 2 marks per kg, and it is just as certain that I should eat not a bit more if the price of sugar went down to 10 pfennigs.’ The latter statement is certainly possible, for since he considers eating sugar unhealthy, even a moderately increased consumption might soon bring him more qualms of conscience on account of the danger to his health than pleasure. But this makes me even less able to accept that a price increase would cause no change in his consumption of sugar. For it is quite obvious that if he spends, say, 15 marks more on sugar, a corresponding deficit must arise somewhere in his annual budget; at the end of the year he may have to cut down on the Christmas presents for his wife or the savings put by for his children by 15 marks. Is he willing to do that? Probably not. Rather, he will attempt in future to distribute the sum in question more or less equally over all his needs, and why then should precisely his consumption of sugar, the actual source of the evil, be spared reductions? Here, incidentally, we have a point that does not have to depend exclusively on theoretical speculations, but could instead perfectly well be settled by statistical enquiries. Cassel claims that ‘every even moderately prosperous man satisfies a large part of his needs fully, and is prepared to pay far more 8

IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

for this degree of satisfaction, if need be, than he has to in reality. However, he is not in the least disposed to pay anything for an additional degree of satisfaction.’ If this is true, then so remarkable a fact must of necessity be reflected in consumption statistics. These statistics admittedly do not generally distinguish between poor and rich; with a little ingenuity, however, it would surely not be difficult to establish that in the case of several important articles a price reduction leads to absolutely no expansion in consumption on the part of moderately prosperous people, but at the same time a price increase causes no decline in consumption on their part—if this really is the case in fact.2 In accordance with the rule that the burden of proof is on the one making the claim, I should like urgently to recommend to him statistical studies of this kind. However, in my experience this is not the way things are; on the contrary, I remember from my youth that at the time, when sugar prices were high, even quite prosperous people treated this article far more economically than happens today. For the present, therefore, I should be inclined to assume that the law of marginal utility retains its validity not only, as Cassel claims, for one group of individuals, the least well-off, but rather, if correctly understood, more or less for all people and all commodities. This does not of course prevent its working being in part counteracted or obstructed by other forces, above all the power of habit, human inertia, etc. What Cassel says about the impossibility of reducing the problem of price formation to just two variables, since in fact of course all prices at the same time influence the demand for every single commodity, is perfectly right and correct. However, when he says that it ‘has become customary in economics’ to overlook this, that the ‘invalid argument’ in question ‘comes up again and again in price theories’, and so on (pp. 422ff), he should in justice at least have excluded from this general judgement those writers influenced by Walras. He does not do this, and occasionally accuses Walras himself of having committed this elementary error, which to my mind sounds more or less like trying to accuse Newton of ignorance of the binomial theorem. Cassel is mistaken in claiming that Walras ‘regards the occurrence of competing goods as an exception’, for on the contrary, in the very passage that Cassel cites Walras calls it an everyday phenomenon (‘c’est ce qui se voit tous les jours’). Also, Walras has never said that ‘the demand for each of the previous goods is approximately equally influenced by the appearance of the new commodity’. On the contrary, he says that those changes in demand for the other goods would in certain circumstances be so small that the ratio of the aggregate demand quantities to one another remains virtually unchanged, which means something completely different. I now turn to that section (VI) in which Cassel subjects the view of some theorists of marginal utility, or rather the dogma of the old free-trade school, 9

MARGINALISM AND CAPITAL THEORY

which they accept, that free competition would realize ‘the old Benthamic maxim of the greatest possible happiness for the greatest possible number of individuals’ to a critique which in itself is fully justified. In passing, as far as I know Bentham is not to blame for the illogical formulation of the maxim in question; this formulation derives rather from Beccaria. Bentham, on the other hand, at least in his Traité de législation, always speaks of the greatest possible sum of happiness, an expression that can certainly not be attacked on formal grounds. Here the critique is, as I say, perfectly justified. But the point is by no means novel. A critique of this dogma that essentially coincides with Cassel’s is already given in Böhm-Bawerk’s well-known treatise in Conrad’s Jahrbücher (1886), and subsequently in my work Über Wert, Kapital und Rente. Later I went further—further, too, than Cassel goes now—and showed that free competition only has the quality of bringing about the greatest possible total satisfaction in a special case—e.g. when all traders are in precisely the same economic position (in the broadest sense of the word).3 In the general case, on the other hand, it is always possible to set up another combination of uniform prices without otherwise changing the current distribution of property that must necessarily give rise to greater total satisfaction than can be attained under free competition. And in arguing this I, just like the defenders of the proposition in question, took as my starting point precisely the constancy of marginal utility functions. Cassel is therefore in error when he believes that this assumption could serve in any way to support that proposition (p. 431). On the other hand, precisely in this area the whole inadequacy of Cassel’s own definition of subjective value or marginal utility becomes apparent. For if it were correct, then in fact that proposition concerning the greatest possible total satisfaction under free competition would indeed be impossible to refute. For if the subjective value of the last portion of goods given and received in purchase is equal for both the buyer and the seller—and of course according to Cassel this would be the case, at least for the groups of ‘marginal buyers and marginal sellers’—then, since one can further assume that each unit of goods acquired earlier or sold later is worth somewhat more, while each unit of goods acquired later or sold earlier is worth somewhat less, the necessary consequence must be that both a reduction in consumption on the part of the purchaser together with an increase on the part of the seller, and an increased consumption on the part of the purchaser together with a decrease on the part of the seller, would cause a loss in total value—and this is of course a genuine and correct maximum condition. Rightly anticipating this unwelcome consequence, Cassel here deviates without noticing it from the path he has followed thus far, and at the end of the section arrives at the statement: ‘If one has the least ambition to invest the term total satisfaction with something more [than he himself has done], something that cannot be defined in conceptual and still less in numerical 10

IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

terms, but that must be understood and felt by anyone engaged in making social policy, then all one can do is to strike out at once all theories about maximum satisfaction on the free market.’ Indeed, but that clearly also means striking out his own definition of value, at least in the area of social policy. Finally, I must make a few comments on the final two sections (VII and VIII), which deal with price formation taking into account production. Here Cassel first summarizes what Walras has elaborated in lesson twenty of his Éléments. The fundamental thought is that under free competition the actual profit from operations tends towards zero, so that the purchase price of the productive services or, in Cassel’s terms, the ‘raw materials’, reappears without remainder in the sales price of the finished goods. Sensibly, Cassel excludes for the time being the part played by capital, the treatment of which was precisely the vulnerable point in Walras, by assuming that production demands no time, and this is undoubtedly permissible as a first approximation. But when he proceeds to draw up equations for the demand for finished goods, he forgets that this demand depends not only on the prices of these goods, but also on the purchasing power of the individual buyers, and that this in turn depends on the prices of the ‘raw materials’; for of course, in the long run, each person receives only so much in income as is given for the producer goods he has in his possession. In other words, the demand functions (system 4) become dependent not only, as Cassel thinks, implicitly, but also explicitly on the price of the ‘raw materials’. But over and above this, these functions are not all independent of one another. Rather, if equilibrium is to continue to prevail on the market, they have to be such that the sum of the exchange values of the producer goods corresponds to that of the consumer goods. An equation to this effect must therefore be satisfied identically, or one could, as Walras does, replace one of the equations in system 4 with the equation just mentioned. But then it becomes evident that in system of equations 6, too, by which Cassel intends to determine definitively the prices of the producer goods, not all the equations can be independent of one another. For if one multiplies them each in turn by the prices of the ‘raw materials’, q1…qr, and then adds the results, this yields precisely the equation that must be identically satisfied. In reality, therefore, for the determination of the unknowns we have not as Cassel claims r, but rather only r—1 independent equations. And this has to be the case by the nature of things, for in this connection money is treated only as a means of exchange or even only as an arithmetical quantity; in the final analysis consumer goods are paid with ‘raw materials’ and ‘raw materials’ with consumer goods. But then of course it is impossible to gain any kind of information about the concrete money prices of the goods; all that can be determined are the relative prices, the exchange ratios of these goods to one among them, and there are precisely only r—1 such ratios for 11

MARGINALISM AND CAPITAL THEORY

the r ‘raw materials’. Perhaps Cassel will regard this observation of mine as petty, but since he himself is of the opinion that ‘Walras is almost always misunderstood’, he really should have taken care not to encourage new misunderstandings himself. Now over against this method of ascertaining, or rather of explaining, the prices of producer goods and with that the part played by the factors of production in the results of production, there stands according to Cassel another, which he would like ‘to designate with the name maximum method’, and whose main representative he names as myself, among others. This statement caused me some surprise, for I should never have wanted to regard the procedure I employ as the opposite of Walras’s. If one disregards the divergent treatment of capital goods (which indeed Cassel does not take into account at all), then the two methods are rather, in my opinion, in essence one and the same, or differ at most somewhat in their starting-points. But my astonishment increased when I read in Cassel that my method takes as its point of departure ‘the principle that the economy tends to achieve the greatest possible total utility’. If I had claimed such a thing, I should in fact have made myself guilty of the most gross inconsistency, for otherwise I have of course explicitly emphasized everywhere in my work that the economy as such under the system of free competition precisely lacks any such inherent tendency. In fact I have never said anything of the kind. On the contrary, I talk about the tendency present in every entrepreneur by the nature of economic activity to maximize his profits, to attain the greatest possible total earnings from his production.4 By means of these competing efforts of individual entrepreneurs a condition of equilibrium will finally be reached, in exact analogy to the process of exchange dealings (and in reality under the simultaneous influence of the exchange trade), and in this equilibrium the parts played by the factors of production under the given circumstances can be neither raised nor lowered. Let me add that the few hints to this effect in the preface to my book, which Cassel praises as ‘clear and concise’, but which he has gravely misunderstood, are elaborated very thoroughly in the book itself. If Cassel has discovered any flaw in my reasoning in that presentation, he should have drawn attention to it; if he has not, or if he has not read the chapters in question, he should really have held back for the time being from judgements such as ‘unsolvable problem’, ‘a priori false’ and the like. On the basis of this brief and succinct discussion Cassel, however, considers himself in a position to dismiss as ‘completely off the mark’ the critique provided in my preface of Wieser’s attempt to ascertain the parts played by the factors of production by means of a system of simultaneous, conventional equations. For according to Cassel, ‘Wieser’s account shows quite clearly that he regards the technical ratios as given from the start’, and ‘there is therefore no maximum problem at all to be solved here’. But first, if he read Wieser’s work carefully, Cassel could easily satisfy himself that Wieser does not in the least assume the manufacturing ratios as given. On the contrary, a 12

IN DEFENCE OF THE THEORY OF MARGINAL UTILITY

few pages previously (Der natürliche Wert, p. 71), Wieser emphasizes quite explicitly that every entrepreneur faces the task of deciding whether under given circumstances he should ‘make savings’ on one producer good or another, ‘on labour or capital, machines or raw materials’, or whether he should ‘on the contrary spend more on them’, and this would of course be impossible if their proportions were already ‘fixed once and for all’ by the technology. Of course it is rather inconsistent of Wieser to go on to reckon those ratios among the known instead of the sought quantities in the problem; my observation that his equations only state the facts of the case, not the how and why of the problem of ‘attribution’ or distribution, is probably justified at least in this connection. But even if one assumed as fixed each and every one of the technical ratios, then too the marginal law and the maximum method of course come into full application, only in a much simpler way, as labour, land, etc., are transferred from the less profitable or loss-making businesses to the more profitable, until equilibrium is reached throughout. To be sure, in this special case the necessary conditions of equilibrium are already included in Wieser’s equations, if one goes along with him in extending them to cover the entire area of production. I am happy to admit this, and perhaps should have emphasized it explicitly. Cassel’s own criticism of Wieser for taking into account only the cost equations and not at the same time those governing exchange I do not consider substantial. In the same way as stocks of goods or rather the periodic supply of goods is in the first instance assumed as given in the problem of exchange, one is doubtless justified in assuming as given the exchange ratios or prices of consumer goods as a first approximation when discussing production and distribution. This approach even has certain advantages from a pedagogical point of view, for later of course it is relatively easy to unite the two systems of equations in a single one. Whether after all this very much remains of Cassel’s critique that deserves serious consideration, the knowledgeable reader may decide. Upsala, December 1899

NOTES 1 ‘Grundriß einer elementaren Preislehre’ [Outline of an elementary theory of prices]. 2 However, let me point out that a consequence of this kind could sometimes occur even in full accord with the law of marginal utility. A reduction in the price of certain foods, e.g. corn or potatoes, could very well cause certain people to give up in part their predominantly vegetarian consumption, i.e. to reduce their demand for corn, etc., in order to go over to more expensive means of subsistence, e.g. meat. 3 See my review of Pareto’s Cours d’économie politique, vol. I, in the Austrian Zeitschrift für Volkswirtschaft, 1897. 13

MARGINALISM AND CAPITAL THEORY

4 Incidentally, it really can be claimed, if commodity prices are assumed fixed, that the sum of exchange values of the goods produced and to that extent the total earnings from the total production of the economy (but not necessarily its total utility) must become a maximum under free competition, for if it could be increased at any point with the available means of production, the entrepreneur involved would of course necessarily pocket a profit.

14

2 KAPITAL—UND KEIN ENDE! (Reply to Decent Brisman)

The controversy about the modern, Böhm-Bawerk capital theory appears highly reluctant to come to an end. Of course, it is not surprising that older economists, brought up as they were in completely different systems, have only unwillingly and hesitatingly acquired this new approach. But one would think that the younger generation of economists might be more interested in elaborating on the grounds already established—which is certainly not the same as uncritically accepting all of the details in Böhm-Bawerk’s presentation—than in wasting time and energy on rather futile attempts to overthrow their very foundations. A priori it is still somewhat improbable that when an author who is generally recognized, even by his opponents, as one of the most astute thinkers of our time, devotes years of his life to studying a particular scientific issue, the result of his work is such that its principal points could justifiably be regarded as ‘meaningless’, ‘unsuccessful’, ‘of little value’, etc. Nevertheless, the criticism continues, also from the younger generation, with undiminished intensity or at least vehemence. For every single upholder of B.-B.’s foundations, there are readily ten subversives. These reflections are provoked perhaps mostly by certain phenomena in the recent German economics literature (Schumpeter, Liefmann, et al.). More specifically, however, I refer to a couple of essays in this Journal by Sven Brisman.1 The latter of these primarily contains some criticism of my own Lectures (Vol. 1, 2nd edition)—remarks whose legitimacy I acknowledge in several instances, and for which I am anyway grateful—as well as, of course, for his otherwise kind recognition of my work. As an introduction to his analysis, however, he has interposed a general criticism of the newer capital and interest theories, which to me hardly seems justified or in any case considerably overshoots the mark. Indeed, the author concludes by saying that ‘as a nonspecialist in the theoretical field, [he] has embarked on these questions only with hesitation’ and he ‘has a feeling that the remarks he wanted to make should have been question marks instead’. However, opinions

Originally published as ‘Kapital—und kein Ende’, Ekonomisk Tidskrift, 1912.

15

MARGINALISM AND CAPITAL THEORY

such as those I cited above, which he levels against Böhm-Bawerk, resemble exclamation points more than question marks. I shall now briefly counter Brisman’s critical remarks, more or less in the order in which he offers them. Brisman begins by pointing out the ways in which economists use the term capital to denote several quite different concepts; he then gives a detailed account of some meanings of the word. Among these distinctions, however, he does not appear to have an eye for the one which I regard as by far the most important, i.e. the distinction between that which comprises capital in the context of the national economy and that which is merely capital in a private economic sense.2 For example, the latter category includes everything associated with assets which, of course, are frequently an integral part of private individuals’ capital resources. But if they were regarded as a component of social or national wealth, then all liabilities would necessarily have to be incorporated as negative items, and since the algebraic sum of assets and liabilities is obviously equal to zero, the simplest solution would be to exclude both altogether.3 Such assets cannot be treated as a component of national wealth unless a country has a credit surplus with the rest of the world, but they disappear again if we adopt a worldwide-economy perspective. This also applies to capital in the form of cash. Böhm-Bawerk does include ‘money as an instrument of exchange’ in social wealth, but this is probably inappropriate. For society, only the function of money is important, while its volume or substance is again altogether trivial—at any event from a world-economy point of view. If the world supply of coins and banknotes were reduced by half, then apart from some disturbances during a transitional period as well as in the technical use of precious metals, this would not cause the slightest change in real economic life or the welfare of nations. Admittedly, an individual would again be much poorer if half his money were taken from him. Therefore, it is indeed of the utmost importance that a careful distinction is made between real capital and means of exchange (money). It may be argued that it was precisely because this distinction was established, and rigorously implemented by the Physiocrats and by Adam Smith and Ricardo, that economics became a bona fide science for the first time. It is certainly true, as the author points out, that most later economists (even a Mill or a Jevons) were not always able to maintain this distinction; sometimes, against their better judgment, they fell back on an unscientific, mercantilistic approach. But such a procedure merely shows that although this distinction is a difficult matter, the application of which requires rigid consistency and not a little mental acumen, it would by no means be impossible to execute, as the author claims (p. 112). The fact that the return on real capital determines the level of interest, and not vice versa, is undoubtedly one of the most important and best certified truths in economics. The return on tangible capital assets in relation to their cost may fluctuate, as the author asserts, ‘between 0 and 16

KAPITAL—UND KEIN ENDE!

100%’; however, this variable return is surely concentrated within the first 10 per cent, and the money rate of interest adjusts to this level. How this dependence arises and what happens when the money rate of interest is temporarily too high or too low in relation to the real rate of interest are, in principle, very interesting problems; but they belong to monetary theory and the theory of changes in the general price level of commodities, not to the foundations of capital theory. Although an accurate distinction between these concepts can only be achieved on an abstract level, without such abstractions theoretical economics would simply be in hopeless chaos. Even after making these distinctions, which the author has completely neglected or does not want to acknowledge, the fact remains, however, that the word capital in economic parlance, as in everyday language, has several different meanings. In some instances, this occurs intentionally, and should seldom give rise to misunderstandings. In discussions of social issues, where ‘capital and labour’ represent opposing class interests, it would be pure pedantry to separate, for example, ownership of land from capital wealth. As regards production and distribution, once again, and particularly in population theory, ‘capital and labour’ are instead often united in relation to natural resources. On the other hand, land or agricultural output and labour which, among themselves are analogous, original forces of production, form a sharp contrast to the ‘producing means of production’, capital. The old trinity: land, labour and capital, is self-evident and can hardly be replaced by some shorter formula in a purely economic context. But once this classical classification has been accomplished, I do not think anything can be gained scientifically by further decomposing the concept of capital. Of course, some gradations, especially important in a practical respect, have to be taken into account, such as when distinguishing between fixed and circulating capital; but this is of no importance to the principle—all capital is more or less circulating, etc. Since the days of Adam Smith, however, it has been customary to use an additional classification, i.e. into that which Brisman, in undeniably improved terminology, calls productive and gainful capital, where the latter is then a more general concept and includes objects which provide their owner with income without actually participating in production themselves. For example, a dwelling house would belong to this category if it were rented out; but when inhabited by its owner, according to Adam Smith, it would be considered a pure consumption good and should not be classified as capital. Brisman approves—this time in agreement with Böhm-Bawerk, but contrary to the authorities for whom he otherwise has the highest esteem—of this distinction; as far as I am concerned, almost twenty years ago4 I tried to show that it is unwarranted and scientifically unproductive. Why should a dwelling house not be perceived as a means of production just like a factory building, such as a bakery or a sugar factory? The latter produce food—with an admixture of human labour and additional capital; the former produces, primarily quite 17

MARGINALISM AND CAPITAL THEORY

spontaneously, another, almost equally important consumption good, i.e. that which we call a ‘roof over our heads’, protection from the outside world, in other words, a place to live. But the house itself is not consumed in a literal sense (unless, like Strindberg’s bohemians, someone takes up the floorboards to be used as fuel, etc.); the gradual deterioration of a house is caused mainly by the elements and is on the whole of secondary importance. Similarly, it is inappropriate to say that someone who rides the streetcar consumes the streetcar or the tracks. He probably does so to a small extent, but he primarily consumes a streetcar ride, which the streetcar system and its staff produce for him, and which can be either a final consumption good if he travels for pleasure or, alternatively, a means of production if he travels to his business premises or workplace. The same approach can be used only too well for all durable so-called consumption goods: books, furniture, even clothing. As for food and other immediately nondurable means of consumption, even Brisman admits that they should be treated as capital, as long as they are still in the hands of the producer or merchant. But once they are in the hands of the consumer, their remaining visible existence—at least as regards the consumption of the urban population—is usually reduced to a few hours, and when dining at an inn no more than a few minutes. Whether or not they continue to be thought of as capital during this short period of time is altogether trivial—de minimis non curat lex [the law does not concern itself with trifles]. For these reasons, I have always been of the opinion that the question of whether or not workers’ basic necessities in particular should be treated as capital fundamentally amounts to nothing more than quibbling and could easily be deleted from the agenda. Böhm-Bawerk’s painstakingly constructed distinction—retained in the third edition of Pos. Theorie d. Kap, according to which these necessities, when advanced directly by the employer, are to be considered solely as gainful capital, but when they are in a granary or other merchants’ storehouses, where they are purchased by the workers themselves, the term is productive capital—seems undeniably as contrived and far-fetched as it could possibly be. This long-winded discussion about the dimensions of the concept of capital would probably have been over a long time ago if we had been initiated into Böhm-Bawerk’s really epoch-making idea right away, even though, regrettably, he himself did not execute it consistently. This idea was to recognize in the capitalistic production process itself—i.e. production to the extent, or from the standpoint, that it always more or less applies to the future—the simple, primary concept, of which capital in all its guises is then only derived or secondary. If so, it soon becomes clear that there is nowhere to draw a boundary line successfully; rather, everything produced but not yet consumed, and which has some exchange value, becomes capital.5 Production without capital is not really production; in any case, it does not yield any products. A tribe of monkeys climbing about in the bush has no capital, but it does not 18

KAPITAL—UND KEIN ENDE!

have any products either; it sustains itself literally from hand to mouth (or perhaps from foot to mouth) on what it harvests from the trees. Hence, nor is there basically any real difference between productive and gainful capital. All gainful capital produces, and all productive capital (except for failed capital investments) yields a return. Oddly enough, Brisman does not even want to accept this last thought: as examples he cites ‘streets and roads which merely generate costs for the owner without providing any income’. But it would not be difficult to show that streets and roads actually do furnish income to the corporations that could be regarded as owning them; if they are privately owned then, of course, the owner is free to charge travellers a toll, as is still sometimes done. Moreover, the capitalistic production process is not finished in so far as there is a finished product. The labour and land that are utilized naturally remain invested in it and mature only gradually, perhaps over many years, into immediately consumable products or definitive consumption activities. Therefore, later on, when Brisman rejects my definition: capital is (in the perspective of economic theory) saved labour and saved land, and instead proposes the expression: capital is the product of saved labour, this is hardly any improvement. A dwelling house is usually the product of, let us say, two years’ work, but many decades will elapse before this labour as a whole can be said to have been used consumptively or consumed, and in this sense, of course, it remains saved. Something peculiar now occurs. While Brisman’s entire train of thought, in so far as I have understood it correctly, gravitates towards precisely this conception of capital, he is almost completely unsympathetic towards the concept when he encounters it in the guise imparted by Böhm-Bawerk: the profitability of prudent roundabout methods of production. He calls this aspect ‘a pure truism, not worth the effort of wasting any words on it’ (p. 103). A roundabout method of production would not be wisely chosen if it did not give rise to a technical surplus, so he says. But the emphasis, of course, is on the fact pointed out by B.-B. that such a surplus always, or at least as a rule, can be attained, although at the cost of prolonging production—which obviously also has to be chosen discriminately.6 This theory, disputed so ardently and perhaps not initially formulated with sufficient rigour by B.-B., may be explained in the following simple way. It may safely be assumed that all, or at least most, production processes could be changed in a number of ways. Since every change implies either prolongation or shortening of the production process, while either increasing or decreasing output, all possible and conceivable changes in production necessarily fall into the following four categories:

19

MARGINALISM AND CAPITAL THEORY

1 those which prolong the production process and simultaneously increase output (per unit of labour); 2 „ prolong „ decrease „ ; 3 „ shorten „ increase „ ; 4 „ shorten „ decrease „ . In practice, the second category should obviously be eliminated as uneconomical. Nor does the third category have to be taken into consideration for the simple reason that every rationally managed production company can be assumed to have implemented such changes already. That leaves categories 1 and 4 as the only practical alternatives. Except for new inventions, in order to increase output, one always has to submit to prolongation of the production process; again, it can only be shortened by relinquishing some of the output. But why is it, someone might ask, that all the changes in category 1 have not yet been carried out? The answer varies, depending on whether one takes a private or a national economic standpoint. These changes are not implemented by the private entrepreneur because, although always ‘technically’ worthwhile, they are not always economically profitable. Production profits have to be acquired by prolonging the period of capital circulation and this can easily result in a lower return on capital than that already achieved. In a national perspective, again, such changes cannot take place because there is not enough social capital to support them, unless it is deducted from other branches of production, which would then be left with shorter periods of production and lower returns. If capital is increased, however, then production processes are necessarily prolonged, even without all kinds of new inventions, as new capital would compete with the old for labour and land; wages and rents would rise, and hence, even from a private economic point of view, prolongation of the production process would be profitable. Böhm-Bawerk has, on the whole, irrefutably demonstrated all this, whereas Jevons, for example, by no means had a clear idea of this causal relation. This is also why it is unfair of Brisman (p. 120) to assert that B.-B. had derived the essential thought content of his theory from Jevons.7 There are further circumstances which are frequently overlooked. In the case of the individual capitalist, when capital assets are rising, prolongation of the production process can very well serve as a means of partially neutralizing the unavoidable fall in profits but cannot increase them. From the point of view of society, prolonged production brought about by an increase in the capital stock can raise the social output, but not specifically interest and, at any event, not the rate of interest. On the contrary, the latter declines, while wages and rent (either or usually both) more or less soar. So when Brisman (at the bottom of p. 120) remarks that a war or, ‘as Cassel pointed out’, a population increase and economic progress could have a strong effect on interest ‘without being associated with prolongation 20

KAPITAL—UND KEIN ENDE!

of production’, it is really not easy to understand his way of thinking. If the war or population increase had not taken place, then during capital accumulation, the result would be both prolonged production and a reduction in the interest rate; now perhaps neither or the opposite of both occur, but this would not even remotely contradict the Jevons-Böhm-Bawerk theory. Meanwhile, of course, economic progress causes more or less of a shift in the overall economic scenario or its background. It could have either a favourable or an unfavourable effect on each product share of the productive factors, as well as on interest. Inventions which shorten roundabout methods of production have an unconditional tendency to lower the interest rate—but they do not invalidate the basic rules for solving distributionary problems. Therefore, during an initial orientation, it seems wisest to assume that the technical standpoint remains unchanged and, in general, that all conditions remain stationary—except for changes in the capital stock. Brisman’s lengthy enumeration on p. 105, which is intended to show that even factors other than capital, such as inventions, exchange and rational economic organization, could increase the productivity of human labour—a conviction which no one would deny—seems to be more devoted to obscuring than to analysing the key issue. At the end of the first essay, Brisman occupies himself, in particular, with the origin of interest. I have already conveyed some of his reflections in this context; it seems to me that the others are also rather insubstantial. For instance, Brisman could hardly imagine that the analysis of these issues is advanced by set phrases such as ‘interest is a common phenomenon of price formation’ or that ‘the reason why savings have a price is exactly the same as why wood, iron, meat or any other good has a price’. If it were this simple, then it is indeed difficult to comprehend how ancient and medieval thinkers could have been so totally unsympathetic towards this ‘common phenomenon of price formation’. It is no doubt somewhat easier to understand that wood and meat cost money, than that money costs more money, or that ‘savings have a price’—besides, the latter expression is quite unclear and somehow invites contradiction. There is, of course, an analogy; the difficulty lies in showing what it contains and how far it reaches. Here, as in a couple of other matters, Brisman refers to Cassel’s authority and appears to have a very high opinion of his work, Nature and Necessity of Interest. For my part, however, I am unable to grant that Cassel’s presentation of any problem either critically or positively surpasses Böhm-Bawerk’s; in most cases it probably lags far behind. The relatively most original idea which can be attributed to Cassel is his attempt to show that a reduction in interest to between 2 and 1 1/2 per cent would cause all further capital accumulation to come to a standstill. But here, Brisman wholly disagrees with Cassel and maintains instead that almost all savings would be saved even if there were no interest—which is probably a contradictory exaggeration. 21

MARGINALISM AND CAPITAL THEORY

All in all, I fear that in his first essay Brisman has not made any noteworthy contribution to the development of capital theory or to criticism of Böhm-Bawerk’s theory. Sometimes, it may even be asked whether he seriously examined, in particular, the later parts of B.-B.’s work, where the relation between the size of capital and interest is explored for the first time in the literature in a completely clear and convincing, albeit schematic, way. BöhmBawerk summarizes his analysis by the familiar cliché that it is the ‘last, economically permissible prolongation of production’ which determines the interest rate. Even the expression ‘permissible’ is, in my opinion, somewhat unfortunate; ‘economically necessary’ would have been more appropriate. The statement is nevertheless clear from the context, and the fact that someone could object that it is instead ‘the prevailing interest rate’ which determines the duration of roundabout methods of production, is surprising. The task, of course, is to explain how the prevailing interest rate arises or is maintained. Given are: social capital of a fixed size, a fixed stock of labour and a fixed productivity of labour which increases with the duration of the production process. In equilibrium, how will productive output be distributed between workers and capitalists or, in other words, how high will wages and interest be? This is the question, and B.-B.’s answer is that, under such circumstances, the production process will be extended to a specific duration, which he carefully defines, whereby the marginal return on the last prolongation of production will determine the interest-rate level. Can words like ‘unreasonable’, ‘incorrect’ and ‘meaningless’ be justified when describing this argument? The fact that the size of the interest rate thus determined in turn affects capital formation and thereby the duration of roundabout methods of production is another matter, which B.-B. has by no means overlooked (and which Brisman, by the way, hardly mentions). But when Brisman himself explains the ‘interaction’, which in his view takes place between these phenomena, he takes refuge in ‘new inventions’ and thus introduces a new element which is alien to the argument. This amounts to nothing other than confusion. As regards the criticism of my own book, I willingly admit, as already mentioned, that some of it is warranted. Regrettably, I have not always expressed myself with sufficient lucidity and stringency; however, I hope that most of the obscurities are due to the way I have expressed myself rather than to unclear reasoning. Thus Brisman reproaches me for having ‘combined two concepts as completely different as physical productivity and value productivity under one hat’ (p. 162). This criticism is justified to some degree (particularly with respect to a few subsequent words which Brisman does not cite). Allow me to point out, however, that above, when treating production and distribution, I distinctly declared my intention to abstract from the phenomenon of exchange. I do this, in principle, by considering—in the two respects indicated—a closed society which, due to 22

KAPITAL—UND KEIN ENDE!

technical conditions, is limited to producing one or a few staples whose selling price is already determined by the foreign market. Under this condition, the concepts of physical and value productivity obviously coincide. Moreover, in the general case, when commodity prices are also variable, I strongly suspect that the two concepts can never be defined and separated in an unambiguously clear way: the different goods evidently cannot be compared in a physical sense or added together. Again, the sum of their exchange value depends on the more or less conventional choice of a standard of value and hence is of no immediate or primary importance to economic life. In one instance, however, I am in fact to blame for a vague classification, and this is the reason for several of my expressions which Brisman criticizes as too imprecise. The issue is whether some of nature’s treasures: mineral deposits, newly accessible natural forests, etc., should be classified as ‘capital’ or ‘land’. They are not the fruit of any labour or even of valuable, already occupied, natural resources. But during their subsequent use in the service of production, they undeniably bear greater similarity to real capital than to natural resources that are renewed from year to year, i.e. land. In the fineprint paragraph on p. 157 of my book (2nd edition),8 I have tried to solve this difficulty by precluding the supply of such immediately accessible natural resources of limited size from the scheme of the stationary society, thereby disregarding them in an initial orientation. In a ‘dynamic’ approach, they obviously have to be incorporated, but then there is so much else to be taken into account which would more or less modify the preliminary conceptual scheme. On other occasions, again, Brisman’s criticism is probably due to a misunderstanding of my intentions—for which I myself might be partly to blame—or of the case in point. Thus, it is obvious that he has completely misunderstood my remarks regarding Thünen’s theory of adjusting interest on the basis of the return on ‘the last share of capital’; yet it would take me too far to demonstrate this in detail here, and I think I have examined this question sufficiently in my book. Nor do I need to address Brisman’s own objections to this theory—which, formulated with appropriate care, is undoubtedly correct and has by no means, as the author claims earlier (p. 119), been abandoned by economists—since they have already been implicitly refuted above. Moreover, I believe I have already provided a sufficient explanation for another seeming contradiction, pointed out by Brisman in this context. It concerns my statement that interest is no doubt a more general concept than productive capital itself, which in his opinion would imply ‘that interest can be obtained from something which is not capital’. This refers to interest on pure consumption loans. In a private economic sense, the lender’s claim is capital, and thus far, as well as in everyday language, his income can, of course, be called interest. But it is not part of the return on social capital; instead it only parasitizes, so to speak, on one of the large social income 23

MARGINALISM AND CAPITAL THEORY

categories: wages, rents and (with respect to the whole economy, as derived from social production) interest. Another strange and hardly correct view held by Brisman is that owners of fixed and circulating capital have different, and even conflicting, interests in regard to an increase or decrease in interest. This may be true with respect to a temporary change in the interest rate, but certainly not in the long run. For example, if mortgage interest has declined due to increased capital accumulation, it is undoubtedly to a property owner’s advantage to replace his old mortgage debt with new loans at lower interest. Later, however, low interest is likely to encourage increased housing construction, so that his own rental income will decline in the same proportion as the decrease in interest; then, in sum, he has obviously made a loss and not a gain. Only that share of the rental which corresponds to interest on the site will be unaffected (or rather, favourably affected) by the fall in interest. But the interest on the site is rent, and the concerns of property owners are unquestionably in conflict with those of (all) capitalists with respect to the interest rate level. I herewith conclude my countercriticism. I hope it will not be regarded as stemming from an author’s wounded pride, which would be unwarranted since, on the whole, Brisman is very favourably inclined towards my book. Besides, I am grateful for all objective criticism, including these critical aspects. Nor do I rule out the possibility that the idea which Brisman appears to prefer: capital perceived as saved purchasing power or postponed consumption—because this is the real meaning of his, not fully defined, expression ‘savings’—could serve as a basis for 1a very useful theory of capital and interest. Under the condition that the time element is given its proper place in the theory of capital, I think the point of departure could be almost anywhere, even in Hermann’s utility theory. (Only the theory of ‘exploitation’ would have to be eliminated as unscientific because, in principle, it neglects the time element.) But I am also strongly convinced that any such undertaking, consistently carried out, can only be a parallelism, a more or less faithful counterpart to Böhm-Bawerk’s theory. This is not to imply that it would be superfluous, however. On the whole, this anti-critique of mine is by no means intended to discourage Brisman from further visits to the theoretical realm, where I rather bid him welcome.

NOTES 1 Editor’s note: Wicksell’s article is a response to two articles by Brisman which appeared in Vol. 14, 1912, of Ekonomisk Tidskrift (now the Scandinavian Journal of Economics). The first is entitled ‘Kapital och kapitalräntan’ [Capital and Interest], (pp. 89–121) and the second, ‘Prof. Wicksell’s framställning af kapitalet och kapitalräntan’ [Professor Wicksell’s Interpretation of Capital and Interest], (pp. 157–70). After Wicksell’s critique, Brisman replied in a third article (also 24

KAPITAL—UND KEIN ENDE!

2 3

4 5

6 7 8

published in Vol. 14) entitled ‘Ännu några ord om kapitalet och kapitalräntan’ [A Few More Words on Capital and Interest], (pp. 399–416). Not to be confused with the, even in my view unfortunate, term ‘private capital’ used by B.-B. et al. Of course, this is by no means intended to deny that even in the context of the national economy, credit is of considerable importance; but it belongs on another level, in monetary theory and applied economics, not in the basic theoretical foundations. In Über Wert, Kapital und Rente (pp. 74ff) [Value, Capital and Rent (pp. 99ff)]. Brisman has not fully understood me, when he discusses my implicit agreement with Böhm-Bawerk’s concept of capital called intermediate products (p. 159). At any event, in the portion of my text referred to, this concept by no means designates the opposite of ‘finished products’—on the contrary, I think this distinction is irrelevant. I merely mention capital assets, which are thought of as secondary in relation to the capitalistic production process. This is what, in my opinion, ‘makes all further discussion concerning the essence and extent of the concept of capital superfluous’. On the other hand, I have no objection to conceiving of the production process as extended all the way to the point in time of consumption, when all products basically become ‘intermediate products’. As an old ‘mathematician’, I am more inclined to generalize and schematize than to classify. Brisman’s objection that, according to B.-B., ‘in exceptional cases, prudent roundabout methods of production would exist which do not yield any return’, is nothing more than a so-called quaternio terminorum [a fallacy of four terms]. Moreover, immediately afterwards, Brisman explains that ‘all similar attempts… failed’, and then, at least, there would not be much argument about whom to ‘credit’ with the invention. Editor’s note: Page 151 in the English edition.

25

3 LEXIS AND BÖHM-BAWERK

I This past summer, two of the most prominent men in German economics, Wilhelm Lexis and Eugen von Böhm-Bawerk, were carried off by death. Lexis had reached the respectable age of seventy-seven and, although active until the very end, it may be said that his lifework had essentially been completed. Böhm-Bawerk was almost fifteen years younger; all things considered, he would still have had occasion to afford us many products of his incomparable acuity; his death was totally unexpected, and it elicited, not least for the author of these lines, an immensely painful shock. In many respects, these two competent men were opposites of each other. Indeed, Lexis’s most distinguishing quality was, as indicated, his incredible versatility. He can hardly be regarded as having produced any great or monumental oeuvre, but in nearly all areas of economics and related disciplines, statistics, finance, sociology, he has written monographs which belong among the most valuable of their genre. It is enough merely to read the list of his contributions to the vast Handwörterbuch der Staatswissenschaften (Conrad’s Hwb.): approximately ninety articles, most of them comprehensive and dealing with specific issues in all areas of economics and statistics. In addition, we have his two excellent essays in Schönberg’s Handbuch, on consumption and on trade (a summary of the latter has been translated into Swedish in Ekonomiska Samhällslifvet [Economic Society], although with some amplification which is of the translator’s own doing). Further, there are numerous articles in journals and independent treatises on widely different topics. In the forthcoming collection, Die Kultur der Gegenwart [Modern Culture], he has written not only an entire volume on general economic activities (at the age of seventythree), but also a second volume—Das Wesen der Kultur… [The Nature of Culture].1 In several of these areas, particularly statistics, the monetary system and trade policy, he has acquired a name and reputation as one of the most eminent scientific authorities of our time. Originally published as ‘Lexis och Böhm-Bawerk’, Ekonomisk Tidskrift, 1914.

26

LEXIS AND BÖHM-BAWERK

Böhm-Bawerk, on the other hand—if I disregard his purely practical work as a civil servant in the Austrian Ministry of Finance, over which he presided several times—concentrated his entire lifework on a single question: the modern theory of value in general and the theory of interest in particular. It should be kept in mind, however, that he devoted ten or more years in the prime of his manhood exclusively to his activities as a civil servant and statesman. Otherwise, his literary production would probably also have been somewhat more versatile—his splendid critical essay Zum Abschluss des Marx’schen Systems2 (in Festgaben für K.Knies) gives us an idea of what he could have accomplished even outside the area which literally became his own. But in this area he has written a study which, if I am not mistaken, will someday belong to the, regrettably few, classical works in our scientific literature, and place him beside Adam Smith, von Thünen, Malthus, Ricardo, Mill, Jevons and a handful of other truly great economists. He has not quite reached this point yet. His theory of capital has certainly remained at the absolute forefront of theoretical discussion in recent decades, but the declared proponents of Böhm-Bawerk’s theory are, at least ostensibly, still rather few in number—I count myself among the earliest and staunchest—while most modern authors in this area have more or less polemicized against him. Even a man like Alfred Marshall, who should have all the requisites for a rightful appreciation of this theory, has on the whole been rather brusquely unsympathetic towards it, without realizing that he has thereby denounced his great countryman Jevons, whose theory of capital, although inexplicit and fragmentary, in its very essence fully corresponds to Böhm-Bawerk’s. Opponents would like to replace the theory they attack with unequivocal propositions. On closer examination, however, one can hardly avoid discovering that, to the extent they have any content and consistency, these propositions are in fact merely—the Böhm-Bawerk theory, more or less thinly disguised; of course, according to the old adage that imitation is the highest form of praise, this seems to vouch for its enduring success. Lexis and Böhm-Bawerk both shared perseverance and a tireless capacity for work, which are, so to speak, the professional secrets of German scientists. Compared to most of their countrymen, they excelled in their remarkably clear and easily comprehensible style of writing—in the case of Böhm-Bawerk this quality not infrequently ascended to true artistry. As regards Lexis the economist, however, I have not a few misgivings; it would be unnecessary to recount them here were not for the fact that he enjoyed a reputation for authority on so many topics, which imposes some sort of scientific obligation. No matter how praiseworthy it might be, versatility is almost always accompanied by the danger of superficiality. But this is not to be found in the way Lexis assembled or organized the material itself—as far as I can tell, he was exemplary in this respect. When subjected to intensive scrutiny from a strictly theoretical point of view, however, this 27

MARGINALISM AND CAPITAL THEORY

material was not always on the same level. In monetary theory, which after all was one of his most distinguished fields, it seems to me that his own theoretical stand-point was rather wavering and uncertain. For example, for a long time—if I am not mistaken—he was sceptical towards the Indian experiment of appreciating the value of the rupee by restricting the free coinage of silver, although this experiment turned out to be highly successful. Its potential had already been anticipated and established theoretically by L.Walras; moreover, a similar experiment had been carried out, also quite successfully, in the Netherlands twenty years earlier.3 Lexis’s criticism of Ricardo’s theory of so-called comparative advantage, which regulates international trade, is also well known. This criticism has appeared unchanged in all four editions of Schönberg’s handbook (Ekonomiska Samhällslifvet II [Economic Society], pp. 477ff) and has finally begun to receive the rank and dignity of authentic refutation of this theory. As a matter of fact, however, Lexis’s entire argument is based on a petitio principii. He assumes that in a country which is in all ways economically inferior to another country, money wages would, under the protection of tariffs, nevertheless be equally high and commodity prices would be consistently higher than in the latter country. Hence, after tariffs were lifted, this country would, at least for some period of time, undersell the other country in all commodity sectors. This assumption of his is absurd, however. Regardless of how high the tariff barriers surrounding a country might be, there is always some good or commodity group whose price is lower than in other countries; otherwise, the country in question would not be able to satisfy its demand for gold from abroad. The only possibilities would be if the country itself produced an over-abundance of gold—in which case gold would be its obvious export good—or if it increased its gold supply excessively by some artificial means, through economic sacrifice. But Lexis does not mention either of these alternatives. Wages in the economically inferior country would thus of necessity be so much lower than in the favoured country that exports from the former, after the introduction of free trade, could begin immediately. Concerning modern theories of value and capital, Lexis has made a number of statements which reveal that he had not become very carefully acquainted with them; even in this instance, his high scientific reputation has probably prevented these theories from gaining widespread acceptance, at least in Germany. One such statement (regarding the prolongation of ‘round-about methods of production’ as synonymous with an increase in capitalist production) may be found in an otherwise favourable review of my book Uber Wert, Kapital etc. which appeared in Schmoller’s Jahrbuch. This prompted Böhm-Bawerk to write a very exhaustive countercriticism (Einige strittige Fragen der Kapitalstheorie [Some Controversial Issues in Capital Theory]—formerly in the Austrian journal), where Lexis may definitely be said to have lost out; nor did he, to my knowledge, engage in any defence. 28

LEXIS AND BÖHM-BAWERK

As a matter of fact, Lexis’s contributions to statistical theory are probably of greater and more enduring importance than his authorship in economics. To be sure, he did not come up with anything positively new in this area either: now currently in use, the catchwords he coined, such as ‘normal’ as well as ‘below-normal’ and ‘above-normal’ dispersion (distribution) in statistical series, are simply facts which have been well known in probability theory for a long time. But Lexis is credited with being the first to have pointed out the crucial importance of these concepts for statistics, where experience has shown that, with few exceptions, an above-normal dispersion almost always has to be taken into account. The following modest attempt to explain what this is all about may be of interest. If we shake a number of coins in a box and scatter them on a table, experience shows that approximately half of the coins will fall with the obverse (script) side up and the other half with the reverse (weapon or portrait) side up. An excess of even a few coins of either kind above half occurs relatively less often, large deviations almost never. Under the assumption that none of the coins has any inherent tendency to fall on one side more frequently than on the other, in principle, a very simple calculation— in fact merely an application of Newton’s binomial theorem—can be used to determine the probability that the different combinations will occur and, for instance, prove that under repeated experiments with, say, 100 coins, the number of obverse sides below 47 or above 53 should only occur every other time. Moreover, the distribution of coins should be outside the margin 38–62 for either occurrence in only about one case in a hundred. If we instead have 1,000 coins, then the most frequent deviations in absolute numbers will be approximately three times larger, but relative to the number of coins about three times smaller; that is, they vary by the square root of the number of coins. If the coins had a predominant tendency to fall with either (e.g. the obverse) side up, then the mean of a sufficient number of experiments would infallibly yield somewhat more obverse sides, for example, half plus one or two per cent; but during the various experiments, the most frequent deviations around this mean would crowd together in the same way. This is the ‘law of large numbers’; and a ‘misdistribution’, which in this way corresponds to the distribution stipulated by the calculation, is called normal dispersion. It could happen, however, that the mass of coins is not completely homogeneous: some of the coins might have a predominant tendency to fall with the obverse side up, others a similar tendency in the opposite direction. If this is the case, the result is reinforced, i.e. the deviations will on average be smaller than expected. This is easily understood if we consider an extreme case where each half of the coins exhibited a very strong tendency in one direction or the other. Then the ultimate result for each new experiment would invariably be 50 (and 500, respectively) coins of each kind. This is called below-normal dispersion. As a further eventuality, the mass of coins could also vary from one experiment to the next. In this case the deviations from the mean would, on the contrary, be larger than those stipulated by the ‘theory’, i.e. those calculated under the assumption of an invariable (and homogeneous) mass. This is also easy to grasp if we imagine an extreme case. For instance, if all or most of the coins used in half of the experiments had a strong tendency to fall with the obverse side up, but an equally strong tendency to fall with the reverse side up in the other experiments, the mean of all the experiments taken together would, of course, still yield an approximately equal number of obverse and reverse sides. However, the separate experiments could deviate from this mean to 29

MARGINALISM AND CAPITAL THEORY

a vast extent and in any case much more than under the assumption of an invariable mass of coins. This therefore becomes what is termed above-normal dispersion. As a rule, both of these sources of error occur in statistical analyses; the material under study is almost always heterogeneous in itself and variable from one ‘experiment’ (e.g. one year) to the next. But the effect of heterogeneity is generally of little importance because statistical data are usually quite extensive, covering tens or hundreds of thousands of individuals, so that the relative deviations from the mean would be very small even for completely homogeneous data. The deviations based on variations in the data from year to year (or even month to month) are usually much larger. Even if such deviations are, on the whole, more temporary (for instance, variations in mortality due to favourable or unfavourable weather conditions, epidemics, etc., or the incidence of marriage which rises or declines with economic conditions), they are often several times larger than those calculated according to the simple probability framework above, which, as mentioned, assumes that the basic data are invariable.—Obviously, there are additional effects of unilateral, constant changes in the same direction (such as a continuous decline in mortality due to sanitary improvements, lower infant mortality through voluntary limitations on the number of children, etc.), which could make all calculations of probability illusory. There are only a very few instances of almost complete agreement between the basic framework and reality, among which the sex ratio of newborns is perhaps discussed the most. In other words, in this respect statistical data remain—for reasons unknown to us—more or less the same year after year.

The practical detection of this circumstance has given rise to two important consequences. It has modified the notion that even statistical phenomena could a priori be subjected to applications of simple Gaussian formulae, which have served physics (as well as biology) so admirably. On the other hand, it has also eliminated the unwarranted suspicion, i.e. regarding the overall importance of mathematical statistics, which had begun to evolve precisely because of the hypothetical discrepancy between ‘theory and reality’. To have achieved this is certainly not a trivial service.

II As if it were only yesterday, I recall the day twenty-five years ago in Berlin— while visiting there on a scholarship—when I looked in the window of a bookshop and read for the first time the title: Positive Theorie des Kapitales4 by Eugen von Böhm-Bawerk. I was already familiar with the author’s Geschichte und Kritik der Kapitalzinstheorien5 and, in particular, I had studied and greatly admired his exceptionally clear and convincing presentation of the theory of marginal utility in Conrad’s Jahrbücher from 1886. In passing, let me mention that in my opinion, these splendid essays should have been published in book form, instead of incorporating excerpts from them in the above-mentioned volume, where they primarily disrupt the course of the exposition and give the general impression of a longueur. I bought a copy and was soon engrossed in a study of the book. My notes, which are still in the margin, testify to the fact that I understood a 30

LEXIS AND BÖHM-BAWERK

great deal of it rather imperfectly; in particular, at the time I did not succeed in absorbing all of the last section: ‘Die Höhe des Kapitalzinses’ [The Level of Interest], although later I came to appreciate that part more than the rest. Nevertheless, this volume was a revelation to me. Earlier, I had tried—on my own, but with little success—to penetrate theoretically the phenomenon of interest and the overall problem of economic distribution when it is complicated by the presence of capital (in addition to labour and the forces of nature). Not even the ingenious ideas in Jevons’s Theory of Political Economy had helped me on my way; yet they do in fact contain exactly the same basic concepts, but in an overly aphoristic form. All of a sudden I saw, as if before my eyes, the roof being erected on a scholarly edifice. Ever since the days of Ricardo, economists had only managed to construct its lower floors and otherwise had to be content with collecting more or less useful building materials. In order to make an impartial evaluation of the significance of BöhmBawerk’s achievement, his book should be considered in the context of the standpoint on these issues which prevailed in economics at the time. Not too long ago, the wages fund theory, according to which the old-time economists thought they had the key to solving the problem, was explicitly repudiated, even by its most distinguished former advocate, John Stuart Mill. Regrettably, however, despite considerable effort, nothing had yet been found to replace it. Everything was in chaos. Mithoff’s vast monograph in Schönberg’s Handbuch (2nd edition, 1885) on Die volkswirtschaftliche Verteilung [National Economic Distribution] may serve as a good exponent of the situation in those days. Other topics are presented in a fairly satisfactory way, but when discussing the distribution of national product between labour and capital, he fails completely. After having shown—which, of course, is not difficult—that wages necessarily have to be enclosed between the worker’s most indispensable cost of living as their minimum and the value of aggregate output as their maximum, he proceeds (in section 32) to study how the real wage rate is determined within these limits. According to Mithoff, ‘such determination’ should be ‘based on the ability to pay of those who employ labour’. Entrepreneurs, he says, pay wages out of capital which is either their own or borrowed from others, but they have to be compensated for this capital (along with the profit on it) by consumers. As a result, the demand for labour ultimately emanates from consumers and the total amount of capital which can be used to purchase labour thus depends on their ability to pay. On the other hand, of course, consumers’ purchasing power depends on the size of national income; the share of this income used to pay for labour is therefore the source of wages. ‘This capital stock or the so-called wage fund’, however, is not of any fixed or predetermined size; in reality, it can only be determined as ‘the sum of wages actually paid at each point in time’(!). If wages rise, then the wage fund also increases, because ‘if entrepreneurs wish to or could be satisfied with lower profits, or else have reason to expect 31

MARGINALISM AND CAPITAL THEORY

consumers to pay a higher price for their products’, then they could also obtain the requisite capital ‘by enlarging the share of their wealth thus far allocated to pay wages, at the expense of the other share, and by raising their credit’. (As regards this ‘other share of their wealth’ and the capital still unborrowed, M. gives no indication of what their earlier purpose might have been.) But let us go on. Workers themselves, M. remarks, occupy an important position among consumers. ‘And when national income is designated as a source of wages, then it is in large part created by the workers themselves.’ (As a matter of fact, the ‘part’ here is probably equal to the whole, because production generally cannot, of course, take place without labour.) ‘As a result’, M. closes the circle of his observations, ‘total output can also be designated as the source of aggregate wages.’ In other words, he arrives at the peculiar result that total output, which was initially specified as the upper limit for wages, now becomes the determining factor for their actual level below this limit.6 Against these and other ‘explanations’, which with good reason could be regarded as a scientific declaration of bankruptcy on this highly significant point—now and then disguised by high-sounding nonsense about ‘historicalethical-psychological’ viewpoints, etc.—Böhm-Bawerk has, as we know, taken up the classical economists’ train of thought and implemented it in a profound, fully motivated wages fund theory. In order to get a clear idea of the relative shares of labour and capital in production output, he says, total social capital or—as he calls it, with a subtle distinction which, in my opinion, makes no difference one way or the other—the social subsistence fund, has to be set in relation to the aggregate labour force, i.e. the annual ‘labour donation’. It should be kept in mind, however, that this ‘fund’ in its entirety is merely a figment; it is gradually transformed into consumable necessities, only to be constantly reinvested in production. Like Jevons, he maintains that even socalled fixed capital has to be replaced sooner or later, inasmuch as all capital can be called circulating—a fluid, although more or less sluggish, part of the general flow of capital. Hence, in getting to the bottom of this approach, capitalistic production becomes intrinsically synonymous with roundabout methods of production; the progressively capitalistic nature of production straightforwardly implies that, on average, an increasingly longer period of time is interposed between labour input and mature products, i.e. products immediately available for consumption. This fiercely controversial proposition is in fact a cornerstone, which I am convinced cannot be overturned, in both Böhm-Bawerk’s and Jevons’s capital theory. Evidently, in order to be profitable, these roundabout methods of production have to be chosen wisely and they are certainly not intended to be available at every stage of production. In general, however, the fact that the capital stock per capita also increases continuously, without bringing interest 32

LEXIS AND BÖHM-BAWERK

down to zero or the like, should be regarded as validating the emergence or availability of such ‘profitable roundabout methods’, with or without the assistance of technical progress. On the other hand, under otherwise equal conditions (i.e. if the total amount of capital remains unchanged), the average duration of the roundabout method of production and thus of capital investment or— according to Böhm-Bawerk’s more lucid, but less exact term—the production period, would now appear to be inversely proportional to the share of this capital which is freed annually and takes the form of consumable necessities, and thereby also to the average wage rate. If, for simplicity, we begin by disregarding agricultural output and some similar elements of production, whose compensation is in reality also advanced by capital in its free form, then every öre of this capital has to pass through the stage comprised of labour and wages; otherwise, of course, it could neither be reinvested nor yield any return. Needless to say, the length of different production processes depends to a large extent on technical conditions. Technical progress which cannot be implemented without increasing amounts of fixed capital has an inherent tendency to reduce the future share of capital freed annually, thereby depressing wages. At the same time, however, technical progress also encourages and facilitates new capital accumulation, which to a greater or lesser degree can counteract this unfortunate effect on labour. Meanwhile, at every given technical level, the wage level also affects the length of the most economically profitable production period. This circumstance had not been fully understood until it was clarified by BöhmBawerk. It is a commonplace that high wages dispose entrepreneurs to look for mechanical expedients; but the usual explanation for this, given even by professional economists, is superficial and basically without meaning. If labour becomes relatively costly, so they say, then machinery becomes relatively cheap; but they overlook the fact that machines are also the product of labour and should therefore—so it seems anyway—increase in price along with labour. Ricardo had already implied the correct explanation, but economists have since forgotten about it. Böhm-Bawerk definitely provided it in the general proposition that a spontaneous rise in wages (e.g. due to a diminished supply of labour or an increase in demand, or if workers succeed in enforcing a general pay rise through strikes, etc.) has to make it economically advantageous for each individual employer to increase the length of the production period which he had previously maintained. This ultimately depends on a purely mathematical relation that is certainly simple enough, but nevertheless sufficiently complicated to be neglected in most economic analyses which, as we know, are not generally distinguished by any high degree of mathematical stringency.7 All the more estimable in that it was a non-mathematician who showed this clearly for the first time. 33

MARGINALISM AND CAPITAL THEORY

Assume, for simplicity, that a productive firm has reached such a high degree of ‘integration’ that it is able to produce all the machines and equipment necessary for its operations. Suppose the entrepreneur has to choose among, say, three different systems or degrees of mechanical production, of which the first would require his total capital to be invested for, on average, four years, the second, five years and the third, six years, i.e. this is the average length of time that will elapse between each unit of labour expended and the completion of the resulting finished product, hence also between advance payment of wages and compensation from selling the products. Of course, the latter two production systems are completely out of the question unless they are technically superior to the first system. We therefore assume that the entrepreneur’s net profit for the entire production period as regards the three systems will be 400, 510 and 600 kronor, respectively, per unit of capital. We also define a worker’s average yearly wage at the time as one unit of capital; the amount does not even have to be specified as long as interest is not compounded. The third and highest of these systems would appear to be the most profitable to the employer, but this is not the case because his objective is naturally to earn the largest possible annual net profit on his (own or borrowed) capital. Therefore, the first profit figure has to be divided by four, the second by five and the third by six. The quotients are 100, 102 and again 100 kronor, respectively, indicating that the middle system is more profitable than both the first and the third. If, however, the wage is increased by, say, 100 kronor, then ceteris paribus the employer’s profit (for the entire production period) is reduced by the same amount and would be 300, 410 and 500 kronor, respectively, for the three systems. If these figures are then divided by four, five and six, respectively, the quotients will be 75, 82 and 83 , i.e. the highest system is now relatively most lucrative. If the wage had instead been reduced by 100 kronor, the quotients would be 500/4=125, 610/5=122 and 700/6=116 kroner. Hence the lowest system is now preferable. As is easily seen, these arguments are completely general; the geometric proportion between two numbers cannot remain unchanged when they are both reduced or increased by the same quantity, but is displaced, in the former (latter) case to the advantage of the larger (smaller) number. (The same result is obtained when calculating compound interest.) So the secret is really no bigger than this, reduced to its simplest expression; but it was big enough before it was properly uncovered.

Under otherwise given conditions, this reciprocal effect between the wage rate and the degree of capitalistic production gives rise, in ways which we do not have to analyse here, to an equilibrium where the share of annually freed capital barely suffices to pay actual wages, while the actual length of the production or capital investment period is the most economically profitable for each entrepreneur at exactly this wage rate.8 It does not have to be emphasized how extremely abstract such reasoning is. It neglects (intentionally) all elements of production other than labour and capital (in a strict sense); this shortcoming has to be amended without fail before there can ever be any question of verifying the hypothesis in practice. It does not take into account the increase or decrease in labour efficiency which a rise or reduction in wages is apt to evoke (which, as pointed out by Brock, can sometimes lead to interesting complications). Moreover, it presupposes not only rapid and smooth adjustment of production technology, which far from corresponds to reality, but also extensive, ‘atomistic’ free 34

LEXIS AND BÖHM-BAWERK

competition. If the last condition does not hold, the formal solution will also be considerably different. Finally, this hypothesis makes use of the tricky concept of mean or average, which unfortunately does not have any concrete meaning except in a few, extremely simple cases, whereas it is merely a mathematical concept in more complex cases. Furthermore, the definition itself involves precisely the concept it is intended to determine, namely the level of interest (and hence of wages).9 But with all its shortcomings or, rather, imperfections (which, of course, could subsequently be amended), as compared to all previous attempts to explain the distribution problem, Böhm-Bawerk’s solution has a distinct advantage in that it is neither circuitous nor self-contradicting. It may be regarded as containing not only an ‘essential truth’, but probably also the essence of the truth itself in this area. The parts of B.-B.’s work considered so far are those which I value most highly, even though criticism of his theory has not dealt with them at all or else has understood them rather poorly. In addition to these aspects, his book also contains a chapter which initially gave rise to such criticism. Here, he tries to derive the essence and necessary existence of interest—in reality a superfluous undertaking because if it has been shown that the level of interest under given conditions has to be such and such, then the proof of its existence is thereby provided. As we know, he explains that interest is an agio between ‘the value of present and future goods’ on ‘three main grounds’: potentially more abundant satisfaction of needs in the future, unavoidably subjective underestimation of future needs and overestimation of future resources, and technical superiority of today’s productive resources to those which will not become available until some future date. Elsewhere in this Journal, I have already commented at length on BöhmBawerk’s presentation, and will not pursue the matter here. Lately, most of the attacks have focused on the notion that only the second of his three grounds would in fact be conclusive with respect to the existence or level of interest. The first ground cannot be considered universal: in a stationary society—if not for the individual, then for society as a whole—satisfaction of needs is the same at present and in the future. But the third ground, or what is now usually called the marginal productivity of capital (or rather of ‘waiting’), would only be secondary or temporary. If it allowed a return on capital higher than the above-mentioned subjective underestimation of future needs, the result would be rapidly progressive capital accumulation, so that interest would soon decline to the latter level—and likewise in the opposite case. The real reason for interest thus lies exclusively in the human psyche which, although it can be changed by cultural progress, legal rights, etc., by and large remains such as it is at any given point in time—so say, in particular, American authors such as Irving Fisher, Frank Fetter, etc., and among the Germans, Bortkiewicz. 35

MARGINALISM AND CAPITAL THEORY

The formal accuracy of this objection should probably not be disputed, that is, under the assumption of fully stationary conditions. It should nevertheless be pointed out that even if the productive properties of capital are not the sole determinants of the interest rate level, they obviously prescribe the role of interest as the important social income category it is in reality. But as soon as it is recognized that during the past hundred years we have lived, and still live, under conditions which are anything but stationary, it seems to me that the objection loses its factual implication. So a ‘static’ economic equilibrium is entirely out of the question; only a ‘dynamic’ one is relevant. Accordingly, the marginal productivity of capital becomes the primary determinant of interest; indeed, even B.-B.’s first ground would then serve as a determining factor. Since capital accumulation still occurs today, and on a very large scale, it confirms that a permanent difference remains between the interest rate level determined by the return on productive capital and the more frequently mentioned subjective measure of underestimation in the future. The gap between the two is invariably filled by means of the first ground. In other words, capital accumulation and saving are constantly pushed to the limit where the underweight of present supply, and hence the overweight of present marginal utility, exactly correspond to this difference. Here, even the ‘supply ground’ comes into favour and, precisely as B.-B. wished, cumulates its effects with those of the second ground. For example, if the expected return on free productive capital is 5 per cent and the subjective underestimation of future needs is approximately 3 per cent (by the way, I doubt whether it is so high even in our country), then saving does not stop until the level at which present marginal utility is 2 per cent higher than the future level. The dynamic equilibrium will be expressed by the equality 5=3+2; but the actual interest rate will thus be 5 per cent, by no means a mere 3 per cent.

Böhm-Bawerk himself defended his approach very astutely in No. XII of his ‘Excursuses’ (to the 3rd edition). However, he might have made the mistake of concentrating on individual cases, whereas his opponents, especially Fisher, focused more on an overall social context. The Positive Theory is not Böhm-Bawerk’s only principal undertaking; five years earlier he had published the above-mentioned Geschichte und Kritik der Kapitalzinstheorien, which later appeared in an extensively enlarged second edition and now, if I am not misinformed, will be published posthumously in a third, which he had lately been preparing. Some regard this book as his best. But I cannot quite concur with this opinion; in any event, the positive part of his analyses has to some extent lessened my interest in the historical-critical aspects. As long as a fully accepted or completed theoretical scientific foundation does not yet exist in a certain area—compare, for example, the theory of crisis—the screening process within historical criticism remains highly important. But no sooner have we acquired a theory that meets all reasonable demands—as, in my opinion, thanks to Jevons and 36

LEXIS AND BÖHM-BAWERK

Böhm-Bawerk, now appears to be the case—than historical interest turns more to the germs and inception of the theory, which can be traced in earlier writings, than to disproving false statements and failed attempts. In this respect, my impression is that Böhm-Bawerk’s work, despite its breadth, has all too little to offer. For the most part he proceeds negatively, often with such fervour and ceremoniousness as to call to mind the German proverb: ‘allzu scharf macht schartig’ [excessive severity overshoots the mark]. What is new in his own theory, of course, is that the time element in consumption and production becomes the exclusive content in the concepts of both capital and interest.10 In their explanations of interest, many authors either totally neglect or merely touch on this viewpoint. Böhm-Bawerk’s criticism of them—which regrettably applies to most of the authors he discusses, old as well as young—is, in my opinion, entirely decisive. But the really great and original theorists, von Thünen, Ricardo, Senior and, in particular, Jevons, adopt instead precisely this viewpoint as the core of their own analyses of the phenomenon of interest, even though they have not succeeded as far as is necessary in forming a systematic whole. Therefore, so it seems to me, Böhm-Bawerk’s criticism in this respect is often rather forced and strained, sometimes even misleading. For instance, when admonishing the so-called production theorists, von Thünen, etc., he makes the—in this context—meaningless remark that ‘one more of a product is not necessarily one more [unit] of value’. Or, when commenting on Jevons’s familiar interest formula: ‘the rate of increase of produce, divided by the whole produce’, he asks for proof that the increase of produce which takes place ‘by means of the last increment in capital’, also comprises ‘an increment in value in addition to the capital spent on investment’.11 However, Jevons’s formula does not refer to any increment in capital, but solely to a (marginal) prolongation of the investment time for a given amount of capital. Therefore, the ‘concrete example’ B.-B. uses to illustrate Jevons’s supposed error is misdirected and on the whole of little value. In my view, when he initially wrote these and similar remarks, Böhm-Bawerk appears to have had no clear idea of all the details of his intended theory, and thought he could provide it with perfection which, according to the nature of things, cannot be achieved.12 Otherwise, there is no reason to doubt that Böhm-Bawerk, as he himself asserts, worked out his theory independently of Jevons; besides, their approaches are much too different. But if he had tried harder to investigate and emphasize the similarities between his own and Jevons’s theory, instead of pointing out the—often only illusory—dissimilarities, he would no doubt have greatly facilitated the establishment of the new theory in the world of economics. I have already mentioned Böhm-Bawerk’s brilliant style, which has turned large parts of his principal writings into true models of the art of presentation and many of his critical essays into veritable dialectical pearls. But more than 37

MARGINALISM AND CAPITAL THEORY

anything else, in an oeuvre whose essence was predominantly polemic, I have to admire the unwavering urbanity with which he treated his opponents. Such a quality, unfortunately so uncommon in the scientific literature, is all the more praiseworthy in that it was exhibited to a far lesser degree by most of his antagonists. It was only on rare occasions, towards an unusually discourteous adversary, that he allowed himself to take the small revenge of disclosing—in a particularly incisive expression of his consistently objective argumentation—the weak intellectual powers of his opponents. So, in the Annalen des deutschen Reiches (1906), a certain Schade had undertaken, in a manner as arrogant as it was superficial, to refute Böhm-Bawerk’s basically self-evident proposition that an increment in a country’s capital wealth per capita of the population necessarily has to be associated with an increase in the average length of production processes. In three pages13 B.-B. dismissed this endeavour by an author who ‘probably took little pains to analyse his propositions as to their perceptive content’, and mercilessly exposed Schade’s faulty logic. He concluded with an edifying summation: ‘Followed to its logical conclusion, Schade’s argument results in the double impossibility of logic that an increase of capital per head brings about employment of more people than are available, and that this increase per head, because of his arbitrary increase of people, no longer signifies capital growth per head!’

The kindness he bestowed on all novices in our science, whenever he thought he had perceived even the smallest signs of productive future work, was unlimited; in this respect, the writer of these lines owes him an ineradicable debt of gratitude. During the past few years, Böhm-Bawerk had been president of the Akademie der Wissenschaften in Vienna. Our own Academy of Sciences had also elected him among its very few foreign members in the economics class.

NOTES 1 Lexis had studied in Paris and was said to have mastered French as if it were his native language. He was a mathematician in his youth and wrote a thesis in theoretical mechanics—in Latin! It concludes with eight propositions, of which the last is: ‘Lingua Latina ad modernam scientiam physicam tractandam apta non est’ [The Latin language is not well suited to treatises in the modern physical sciences]. 2 Editor’s note: Translated as Karl Marx and the Close of his System, London: Fisher Unwin, 1898; reprinted in P.M.Sweezy (ed.), Karl Marx and the Close of his System by Eugen von Böhm-Bawerk and Böhm-Bawerk’s Criticism of Marx, New York: Augustus M.Kelley, 1949; also translated as ‘Unresolved Contradictions in the Marxian Economic System’, in Shorter Classics, South Holland, IL: Libertarian Press, 1962. 38

LEXIS AND BÖHM-BAWERK

3 At the same time, however, on at least one occasion—in a book review, if I remember correctly—Lexis expressed his belief in the international paper standard as the medium of exchange of the future. In this respect he thus went further than, for example, K.Helfferich. 4 Editor’s note: First edition translated as The Positive Theory of Capital, London: Macmillan, 1891; fourth edition translated as Capital and Interest, vols 2 and 3, South Holland, IL: Libertarian Press, 1959. 5 Editor’s note: First edition translated as Capital and Interest, London: Macmillan, 1890; fourth edition translated as Capital and Interest, Vol. 1, South Holland, IL: Libertarian Press, 1959. 6 In passing, however, M. also reflects that wages should rise or fall as consumption is channelled towards goods whose production requires relatively much or little labour. But this idea, which in itself is correct and could undoubtedly have been used as a basis for a real theory of wages, is not pursued. 7 However, the fact that mathematics alone is of no help here may be exemplified by the following bizarre case. In the first volume of the Danish journal Nationaløkonomisk Tidsskrift (1883), two well-known mathematicians, F.Bing and (later Professor) J.Petersen, tried to examine the problem of the interrelation between interest and wages. They had been dissatisfied, and justifiably so, with the way this relation had thus far been treated in the economics literature. Their own proposal: compensation for labour and machinery in proportion to—what would now be called—their relative marginal productivity is altogether correct and, on this basis, the authors also manage to advance the discussion somewhat. But, incapable of penetrating the crux of the problem, they fall upon the unfortunate idea of representing the unit of measurement for the flexible wage by a mathematical expression which, on closer inspection, is synonymous with this flexible wage itself (merely enlarged by one year’s interest). After numerous calculations, they arrive at the inevitable result that a change in the capital stock and interest level has next to no effect on wages—naturally, because anything which measures itself always remains unchanged! 8 A mathematician is inevitably reminded of the way in which equilibrium is determined according to the principle of virtual movements in mechanics. 9 If capital invested once and for all in, for example, a machine, a house, a ship, etc., over a number of years initially yields a certain return, but then nothing, the average duration of this investment will depend on the size of the return itself, i.e. on the interest rate according to which the capital bears interest. 10 Six years before Böhm-Bawerk, [the Swedish economist, David] Davidson published his praiseworthy little treatise entitled De ekonomiska lagarna för kapitalbildningen [The Economic Laws of Capital Accumulation], which is solely concerned with capital accumulation, and not directly with interest. While arguing against Herrmann and to some extent against Menger, he emphasized this very same idea in an equally conclusive and intelligible way. 11 Geschichte und Kritik, 2nd edition, p. 568. 12 As is well known, in Positive Theorie, he tries to resolve the hypothetical problem by comparing two production elements from different points in time, e.g. two ‘working months’ of different vintages, where the older of the two would always be more productive than the younger, regardless of the future point in time for which production is intended. Subsequently, however, he does not make any use of this idea; instead, he continues on a path parallel to the reasoning of Jevons and von Thünen. As a general proposition, taken litreally, it can hardly be approved—as is also pointed out by Bortkiewicz. 39

MARGINALISM AND CAPITAL THEORY

On the only occasion I met B.-B. personally (autumn 1911), I asked him why it was that the different parts of his Positive Theorie seemed to originate from different premises. His reply was that, due to various circumstances, the first half of his study had already been published before he had written the latter half. As a result, up until the last minute he had encountered difficulties with respect to certain details, which he had not been able to overcome without further deliberation. 13 Exkurse zur ‘Positive Theorie des Kapitales’ [Excursus…], pp. 158ff [Editor’s note: Capital and Interest, Vol. 3, pp. 71ff.]

40

4 ON THE THEORY OF INTEREST (Böhm-Bawerk’s ‘Third Ground’)

On the first and only occasion I was fortunate enough to meet Böhm-Bawerk in person—in Vienna in the autumn of 1911—I asked him why it was that his positive theory made the impression, at least on me, of not forming a single perfect whole, but rather of arising out of several parallel strands of thought. When I asked this I was thinking of the differing treatment of the concept of capital at the beginning and end of his work. At the beginning, ‘capitalist’ production focused on the future is presented as the primary phenomenon, capital itself as secondary—as ‘the embodiment of the intermediate products that arise at the different stages of the roundabout path’. At the end, however, he recurs instead to the older concept of a subsistence or wage fund, which means that capital is again conceived of as the primary phenomenon and capitalist production in contrast as secondary. But I was thinking above all of the treatment of what is at bottom the same problem, the origin of interest on productive capital, which also varies greatly. It is first discussed in the ‘Third Ground’, which is very controversial and far from easily understood, and then again, quite differently, in the section on ‘The Rate of Interest in Market Trading’,1 in whose brilliant composition and cogency only very few critics have found anything to fault. My question did not seem to surprise him, but his answer certainly surprised me greatly. For he said simply that for extrinsic reasons he had had to hurry so much with the printing of the original edition of his book that the first half of the manuscript was already in press before he had finished writing the second half. This had in fact meant that he had found himself confronted at the last minute with difficulties of a theoretical nature. For example, in the well-known tables in the section already mentioned, ‘The Rate of Interest in Market Trading’, the numbers in certain columns kept turning out twice as high as he felt they should, until at last the happy idea of a so-called staggering of production brought everything into place. (The part in the third edition that has to do with this is a later insertion.) Originally published as ‘Zur Zinstheorie’, in Die Wirtschaftstheorie der Gegenwart, Vienna: Hans Mayer, Verlag von Julius Springer, 1928.

41

MARGINALISM AND CAPITAL THEORY

If I have understood and reported this statement of his correctly, it might explain a lot, for it is probably inevitable when working in such a way that discrepancies and contradictions of greater or lesser significance creep in here and there against the author’s will. But all the indications are that in the beginning Böhm-Bawerk did not regard the original edition of the book as definitive. In the brief preface to the (unchanged) second edition he still speaks of the future ‘resumption’ of his plan ‘to subject the positive theory to thorough revision in terms of its composition’, and declares that he is determined not to give up this plan.2 But nothing came of it. For many years he was prevented by his duties as a civil servant from taking his work in hand at all—as is well known, the second edition is merely an unchanged reprint of the first—and when at last towards the end of his life he set about composing the definitive text, his book had already stood at the very focus of scholarly discussion for so long that I suspect it had become a kind of point of honour to change nothing in it, or only the most necessary things, but otherwise to let the book stand just the way it was written, for better or worse. He went so far in this as to leave the much-contested tables of the ‘Third Ground’, more specifically those on pp. 463ff (in the third edition) unchanged,3 although he has to confess in his polemic against Fisher (excursus XII, p. 374) that in their original form they ‘lack conclusiveness’. Perhaps, as he claims, he was aware of this fact from the very beginning and only set up these tables at the time as ‘highly illustrative’. Incidentally, the latter claim is inaccurate too; on the contrary, the numbers are poorly chosen. For if one completely disregards for the time being the added productivity of the roundaboutness of production, and with that the third ground, in other words, if one allows all the numbers in the table’s second column to become equally large, then by the sole effect of the first two ‘grounds’ it would obviously be possible to express the superiority in value of a month of labour this year over a month of labour next year by the ratio 5:3.8. However, if the third ground is added, then this superiority in value will instead be equal to the ratio of the two maximum values 840:720=7:6. But 7:6 is less than 5:3.8. Consequently—I do not know if any critic has observed this—the appearance of the third ground in itself would even cause a reduction of the premium that had been produced by the other two grounds—and that certainly seems completely paradoxical. Of course it is very easy to alter the numbers in such a way that the ratio of the two maximum values instead turns out larger than the decline in the perspectivally reduced marginal utility during the first year. In that case it is clear that the third ground, as constructed by Böhm-Bawerk, must have contributed at least something to the actual formation of the premium between the value of the two months of labour. As if this were not enough, Böhm-Bawerk has set up new tables in his polemic against Fisher (excursus XII, pp. 377ff), and these tables show that 42

ON THE THEORY OF INTEREST

in the absence of the third ground the two other grounds in certain circumstances would produce absolutely no premium: a premium only begins to be produced when the third ground comes into play. But even in these cases the first two grounds are still present, although they only take effect in the later years; I do not understand Böhm-Bawerk’s new attempts to demonstrate that the third ground is capable of producing a premium even on its own, without the aid of the other two. In my opinion, therefore, Bortkiewicz and Irving Fisher certainly go too far when they deny Böhm-Bawerk’s third ground any independent effect and characterize its formulation simply as a self-deception on the part of the author. On the contrary, it seems to me that Böhm-Bawerk has vindicated his logical honour in excursus XII, painstakingly worked out and filling nearly 100 pages. It is surely not possible to reproach him for circularity, an error that is so extremely frequent in economics literature. The third column [of the table] refers to the so-called First Ground (the necessity or subsistence factor), and the fourth column to the Second Ground (subjective undervaluation of the future), neither of which are to be discussed further here. For a month of labour available in the next year, 1910, a similar pattern applies. The two columns just mentioned remain completely unchanged, the numbers in the second column, on the other hand, are displaced one step downwards (the final number falls away in the process), and the fifth column, finally, which is arrived at by multiplying the second and the fourth, acquires completely different numbers, of which the largest, namely 792, appears in the sixth year, 1914. But the third ground still remains rather unsatisfactory; its effect on the reader is more confusing than enlightening. First, what is its actual relation to the two other grounds? Böhm-Bawerk answers without hesitation that the first two grounds ‘aggregate their effects, while the participation of the third factor in contrast occurs not by means of aggregation but rather by Table 4.1 One month of labour available in 1909 yields

43

MARGINALISM AND CAPITAL THEORY

alternation’ (pp. 475ff). The first point is undoubtedly correct. However, how could a factor that actually only takes effect in combination with one or more other factors be shown to alternate with it or them? When Böhm-Bawerk attempts to demonstrate this, in my opinion he forgets his own presuppositions. He speaks of ‘the adoption of a one-year production method’ with the aid of the present month of labour, while the month of labour in the next year ‘is used for unproductive immediate output’ (p. 476), as if this were already settled. He seems to have forgotten that in both cases the length of the most suitable production period was supposed to be determined by means of the interaction of all three factors or ‘grounds’. The position of the third ground with respect to Böhm-Bawerk’s later statements appears still more mysterious. Here both the length of the production periods and the level of the premium or interest are determined by the relative size of the capital. Already in the immediate continuation of his observations on the third ground (p. 472), Böhm-Bawerk suddenly begins to play this tune. He speaks of people who already own ‘a certain stock of goods’ (which provides them with support for several months or years of labour). ‘If their stock of goods suffices for three years of subsistence, then they can utilize their productive equipment in a production process averaging three years. Now if an additional year’s quota of current subsistence means is put at their disposal, then they can increase the length of the average production period from three to four years’, and so on. That may be true, although ‘can’ is of course not synonymous with ‘will’— if I adopt the most Spartan way of life my means might last for three years, but that will not necessarily prompt me to adopt a three-year production period. Moreover, how is this whole way of looking at things to be reconciled with the result of the tables mentioned above, according to which the present month of labour should be invested in a production process lasting only two years, while the next month available should in contrast be invested in a production process lasting four years, and not a word is said about how much capital may be involved? So some error or another, even if only of a pedagogical or presentational kind, must have slipped in after all. After much reflection I have come to the conclusion that this error lies in a mix-up that is easily committed yet has serious consequences. It seems to me that Böhm-Bawerk mixes up the yield of a production period of a certain length—or rather, the portion of this yield that could be attributed to the month of labour in question—with the highest possible yield that could be attained in any way with the aid of this month of labour in the period concerned. But these two things are not the same, and under certain circumstances can be quite different, as the same space of time is often occupied more successfully with several shorter production periods than with a single longer period. The problem is compounded with Böhm-Bawerk’s peculiar habit of always deferring the enjoyment of the fruits of some saving to a single point of time in the future, 44

ON THE THEORY OF INTEREST

whereas in general, surely, our savings, big or small, are designed to enrich our entire future.4 I shall attempt to show all this and its consequences using two examples, which, though only hypothetical, present the pedagogical advantage of investing the somewhat nebulous concept of the production or investment period with concrete meaning, while the question of ‘the continuation of work’, left very much in the dark by Böhm-Bawerk in his discussion of the third ground, completely falls away. For in both cases it is assumed that the necessary labour is carried out at the very beginning of the production or investment period, and that for the remainder of this period the raw product is then simply left to the beneficial influence of cost-free natural forces. As is well known, the simplest example of this kind is furnished by the way in which certain commodities, above all wine, improve with keeping. So let us assume that a colonist in some region with a suitable climate cultivates some wine every year, but only for his own consumption or that of his family. Let us assume that through ignorance of the fact that the wine would be greatly improved by being kept, he consumes his entire stock in the year after harvesting, as grape-juice or new wine. But now he acquires the necessary knowledge about the method and purpose of storing wine and therefore takes the decision to refrain from drinking a portion, say 10 litres, of his current stock in order to save it for the future. For what future? Let us assume the quality of the wine rises, though at a decreasing rate, for twenty years. According to Böhm-Bawerk he would then prima facie prefer to let the wine lie for twenty years, this being the most productive ‘production period’, and the only thing that might lead him to adopt an at any rate somewhat shorter ‘period’ in reality is the (justified or unjustified) underestimation of future needs in comparison with those of the present (or of the immediate future). Is this reasoning correct? Hardly. I make so bold as to say that even if the man fully appreciates the value of the pleasure to be derived from wine in the future, or at least over the next twenty years, he will still in the circumstances described preserve the 10 litres of raw wine for just one year, and will then drink it up. For if he does this, then the next year he gets a plus of 10 litres of (one-year-old) wine, which will put him in a position to lay aside 10 litres of raw wine from that year’s harvest for the following year without any further sacrifice at all. The same thing applies in the third year, and so on. In other words, by means of the sole sacrifice of the present year, as it were as a continuous fruit of this year or an eternal interest on this capital, he acquires the advantage of being able to drink 10 litres of one-yearold wine instead of only new wine every year for the entire future, and more specifically for the next twenty years. It must after all be assumed that this advantage must far outweigh the single enjoyment of 10 litres of twentyyear-old wine (instead of new wine), no matter how good it tastes.5 The difference between the pleasure given by one-year-old and zero-yearold wine constitutes, as stated, the annual interest for the capital sacrificed at 45

MARGINALISM AND CAPITAL THEORY

a single time, and this interest corresponds to the premium between the means of production in the present and the next year, embodied in one bottle of raw wine each. By saving one of next year’s bottles of raw wine for the next year but one, he can similarly guarantee himself the same pleasure for the entire future, with the sole exception of the next year. After the same futures have been cancelled out there therefore remains the enjoyment of one bottle of one-year-old (instead of zero-year-old) wine for the next year.6 Finally, the premium in question can be regarded from yet a third point of view. For if we ask why the man only saves 10, instead of immediately saving, for example, 20 or 30 litres for the future, then the answer will of course be that that would mean too great a reduction in the stock still available for his present consumption. In other words, the pleasure of consuming the ‘last’ bottle of new wine will be precisely equal to the total advantages of the ‘last’ of the ten bottles saved for the future. That means it will also have the same superiority in value as the last saved bottle over a bottle of next year’s wine, not because it gives pleasure earlier, but rather because as part of an already diminished stock it gives greater pleasure than the bottle of next year’s raw wine, which in line with our assumptions belongs to a stock of normal size. In this way, Böhm-Bawerk’s ‘first ground’, the subsistence factor, certainly also comes into play, and necessarily so at that, but as a secondary rather than a primary phenomenon, or, as Böhm-Bawerk puts it in his reply to Fisher, ‘this factor marks not the beginning but rather the end (for the time being) of the process that makes the present means of production superior to future means’. But a premium at this level will not last long. For the man is still dependent mainly on the consumption of new wine and can only enjoy the one-year-old wine as an exception. In order to extend this advantage he will surely be prepared by and by, perhaps as early as the following year, to impose upon himself a further sacrifice of, say, 10 litres of raw wine, which will put him in a position to drink 20 litres of one-year-old instead of new wine every year in future, and so on.7 Finally he reaches the point where he only consumes oneyear-old wine and regularly keeps the whole of the current year’s vintage for the next year’s consumption. Then the whole state of affairs suddenly changes. The premium between present and future goods whether of consumption or of production would actually already have sunk to zero, which is hardly surprising, since of course in so far as a production period of only one year has been adopted, the production would already be saturated with capital. In fact if he now receives a proposal from some neighbour to exchange a bottle of raw wine for a bottle of next year’s vintage, then obviously all this will mean to him is that he may have a bottle less for next year’s consumption, and a bottle more for the year after (to be specific, in both cases a bottle of one-year-old wine), and this cannot make any real difference to him. But now the time has come for two-year-old wine. In order to get any further, he now has to save a part of the vintage for two years, i.e. he must 46

ON THE THEORY OF INTEREST

not consume the wine in question until the year after next. This causes the premium to rise again, though not to its original level, representing the difference in quality between zero- and one-year-old wine, but rather, as can easily be proved, only to a level representing the difference between one- and two-year-old wine, which is probably rather smaller.8 At the same time the ‘accumulation of capital’, the successive savings, will proceed more and more slowly as the difference in quality between successive vintages of the wine and therefore the incentive to further savings becomes smaller and smaller. But in addition, in real life the perspectival underestimation of the future, Böhm-Bawerk’s second ground, will come into play and, accumulating with the first ground, will work against further savings and finally bring them to a halt, probably long before the wine has attained its highest possible quality. If the ‘Third Ground’ is understood in this or a similar way, it seems to me it acquires a genuinely independent position vis-à-vis the other two grounds. It is thus even possible to abstract from them—i.e. to retain the first ground only in so far as it is called into being by the activity of saving (to be precise, as a reaction against saving), and defer the effect of the second ground to the more distant future, which is not decisive for the length of the production period—without the premium disappearing. At the same time the results of the closed individual economy acquire a direct analogy to those of the open capital-labour market, as finally described in exemplary fashion by Böhm-Bawerk. The productiveness of the roundaboutness of production provides, so to speak, the material for the capital interest, on true productive capital, that is, in distinction to loan capital in consumer lending. The length of the production processes actually adopted, however, and the level of the capital interest or premium, are regulated in the first place not as might be thought by the influence of the first and second grounds, but simply by the relative size of the capital. The size of the capital in each case and the pace at which it increases or perhaps decreases—this, to be sure, is closely connected with Böhm-Bawerk’s first two grounds, and Böhm-Bawerk of course never denied this. (However, the analogy is not complete.) If the reader’s patience is not completely exhausted, I should like to present another example, which in certain respects is still more instructive, and seems to fit better with Böhm-Bawerk’s presuppositions. I am afraid it is only a Robinsonade, though no use will be made of the ‘net and boat’ which are only too well-known and in any case do not fit properly into Böhm-Bawerk’s scheme. Until now, Robinson has eked out a scanty living on his island from one day to the next by gathering fruits and roots, primitive fishing, etc. But at last he has accumulated enough food supplies to live on at his customary standard for a whole month. Now he is in a position to allow himself the rest he has long desired, but since he has a keen eye to the future he prefers to work during this month too. To be precise, he decides to work on clearing a small 47

MARGINALISM AND CAPITAL THEORY

field near his hut, which he then intends to sow with grains of wild corn that he has found on his expeditions. Let us assume he has at his disposal in this enterprise several varieties of grain, which possess the following qualities. The first variety is ripe after one year. The second variety requires two years to reach maturity, but then delivers a larger (single) harvest. The third variety needs all of three years to reach maturity—I admit I do not know whether such grasses exist in nature—but its yield is then greater than that of the second variety, and so on. We assume in accordance with the numbers in the table printed above that the yield of his labour until now and therefore his customary food supply for one month can be denoted by 100, the yields of the various varieties of grain by 200, 280 and 350. The decisive point is that they increase not in proportion to time and therefore to the length of the production periods, but rather, somewhat more slowly. So which variety of grain will he choose? According to Böhm-Bawerk, if no underestimation of the future intervened, he would necessarily select the longest and most productive production period, i.e. the three-year period, but given the subjective underestimation of future needs he might choose the two-year period. But this is quite wrong! In the assumed position he will rather quite definitely employ the one-year variety. For even if he does not intend to enjoy the fruits of his month of labour until two or three years from now, he will gain more by adopting two or three successive one-year production periods (with the aid of the one-year variety of grain) than by immediately observing a two- or three-year period (using the two- or threeyear variety of grain). In the first case, in the very next year he will get a twomonth supply of food at his customary standard of living, which will free two months of labour, which he would otherwise have had to use in the laborious yet unproductive gathering of provisions. He can then use these two months of labour to cultivate a doubled area, and this will free four months of labour in the third year, not just 2.8 months, as would be the case using two-year corn. In the same way in the fourth year he can get enough food for eight and not just 3.5 months, and so on, always measured by his standard of living up to this point. At first, therefore, he will only employ the one-year variety of grain, or one-year production period. If he wants to, he can of course devote both the months freed in the next year to leisure. Then he will have gone through both the capital and the interest and it will all be over—rather as if the wine grower in the previous example had wanted to consume the plus of 10 litres of wine along with his usual stock in the second year. But if he prefers—like the wine grower—to consume only the ‘interest’ and to reinvest the ‘capital’, then as can easily be seen he can procure himself one month free of labour per year for the entire future, while otherwise having to live at about the same standard as previously. This will mean the annual interest on his single sacrifice of capital will amount to no less than 100 per cent, and that this corresponds to the premium between the present month of labour and a 48

ON THE THEORY OF INTEREST

month of labour next year is immediately obvious from the fact that the former month is as it were transformed into two months next year by means of production. But it may be that he wants to save both capital and interest—and perhaps a little more besides—for the future. Then the process described above will occur: he will gradually—and in fact very soon—completely give up his food-gathering expeditions and go over entirely to agricultural production. He will organize this production so that he either ensures himself his usual amount of food for the entire following year in six months, or else in such a way that he works for eight, ten, or, let us say, still all twelve months of the year, in order to provide himself in the next year with nourishment that is both more plentiful and tastier—in short, twice as much food as previously. At the same time the premium between present and future goods has sunk to zero (which, incidentally, means nothing to him in itself, as in his case the roles of capitalist and labourer are combined, and do not appear separately as on the market of a trading economy). To be sure, one month of labour still yields him 200 units of produce, but since this now denotes his food supply for one month, as long as he keeps up a one-year production period one month of labour this year can only free one month of labour next year, which will also give him just 200 units in the year after that. But now the time has come for the two-year period. If he now sows a (small) part of the area he cultivates each year with two-year grain, then in one month of labour he will produce 280 units for the next year but one, and since his accustomed food supply for one month now amounts to 200 units, he can therefore free 1.4 months of labour in the next year but one. The present premium therefore comes to 40 per cent over two years. Here the one-year premium cannot be directly ascertained, as it is always determined by the new additional capital, which is now applied in a two-year investment. But through competition on the market it would of course work out at 20 per cent, in the first instance. By means of continued savings, our man Robinson sooner or later reaches a position where he only sows two-year grain, in other words, he invests all his labour in the current year for consumption in the next year but one. His capital has now been doubled, in that each time twenty-four months of previously performed work (instead of just twelve) lies in the ground. His monthly food supply reaches the level of 280 units, the premium to be sure has again sunk to zero. But if he now switches to the three-year production period (three-year grain), then the premium rises at once, as is easily seen, to (350–280):280=25 per cent for three years or in the first instance 8 per cent for just one year. Here too, then, the ability of roundabout production processes to create interest or premium seems hard to deny. That this ability at the same time works towards its own demise by leading to expansion of the capital is a 49

MARGINALISM AND CAPITAL THEORY

quite different matter, but one which Böhm-Bawerk did not deny; indeed, he pointed it out most emphatically. In general I have no wish to claim that the thoughts elaborated here are completely new or foreign to Böhm-Bawerk’s system. For the most part they are already presented in his book, though not always in a suitable context, and sometimes mixed in with somewhat superficial arguments, which do more to confuse than to enlighten the reader. There is talk of an individual having at his disposal a ‘month of this year’s labour’, but we learn nothing else about this person’s situation. For example, is this month of labour the very first he is able to use for future production, or has he instead already long been in a position to use his present labour mostly or ‘on average’ for a more or less distant future, while at the present time he lives on the fruits of work done in advance in earlier years? Böhm-Bawerk should have told us this at once; if he had done so, in my opinion he would from the very outset have got on to the track that was later to lead him to ‘The Rate of Interest in Market Trading’. I consider this section the most valuable and most original in the entire book. Admittedly the germ of it is already present in Jevons and even in Ricardo, but here, for the first time in the literature, a well-reasoned attempt is made to present a systematic, truly consistent solution to the old problem of the relationship between capital interest and wages under free competition on both sides. That this solution could only occur at all in hypothetical form under strongly simplifying assumptions, is obvious—among other things the involvement of non-free natural forces and consequently rent is completely eliminated. That Böhm-Bawerk at the same time, as mentioned, had to deviate somewhat from his original conception of capital, and represented capital as a given sum of exchange value instead of as stored labour, is a little troubling, but may be excused since it allows easier comprehension.9 What seems more important is that the ‘simplest hypothesis’ assumed by Böhm-Bawerk is so abstract that it can hardly be regarded even as a first approximation to reality. There are only very few cases where a production process continued over the course of years leads to the output at a single time of products ready for consumption. The typical method of production, especially in our day, consists rather in the collaboration of more or less durable capital goods (machines, buildings, ships, railroads, etc.) and human labour or the products of the soil. The actual production process here regularly becomes rather short, sometimes so short that it is negligible compared with the life of the machines and buildings. Consequently, the relative contributions made by the owners of the capital and the labourers (or landowners) cooperating with them to the final product ready for consumption are determined simply according to the rules of marginal productivity as in capital-free production. But these workers do not make up the entire existing workforce. Other workers are continuously occupied with the maintenance or renewal of the fixed capital, even in a stationary society. The number of these workers 50

ON THE THEORY OF INTEREST

of course depends on the size of the capital, though not just on this, but rather also on the expected life of the individual capital goods. The most profitable life of the individual capital goods depends in turn—more or less like the length of the most profitable production period in Böhm-Bawerk’s scheme—on the level of wages (imagined as equal for all workers) relative to the value in use of the capital objects, and therefore, as mentioned above, on the respective marginal productivity of the labour and the capital items in the ultimate production of products ready for consumption. Even under the simplest thinkable presuppositions the determination of the level of interest at a given size of social capital will therefore be mediated by a whole series of unknown, mutually dependent quantities, which of course demands the setting up and ‘solution’ of a corresponding number of independent logical relations or ‘equations’. Obviously the problem formulated in this way has much closer contact with reality than Böhm-Bawerk’s construction. On account of its complicated nature it has until now only been hinted at quite perfunctorily by most writers (including myself), and never treated in detail. As is well known, Walras’s approach leaves the whole question of the differing or variable life of the means of production out of consideration. It is the achievement of a younger Swedish economist to have treated this difficult problem in all its details, and indeed almost throughout with elementary observations as in Böhm-Bawerk’s work.10 This leads to many very interesting and also surprising results. Among other things it becomes clear that the generally accepted conception of capital as previously performed, stored labour (along with the resources of the soil), a conception also approved by Böhm-Bawerk, cannot be directly applied to fixed capital, since on this view the amortization of the individual capital item remains a more or less conventional matter, whereas in Böhm-Bawerk’s construction the capital in question or the portion of capital involved is always amortized all at once, by means of and simultaneously with the production of final products ready for consumption. If I am not mistaken, this shows that with respect to fixed capital it is not enough to look at ‘static’ circumstances, ‘staggered’ production, and so on: the ‘dynamic’ prehistory of the capital involved must also be taken into account.* NOTES 1 III, section II, of the (present) IVth Book. 2 In a letter to me in 1893 he already speaks of the ‘self-criticism he has never ceased to exercise with respect to his own work’. It would be exceedingly interesting to know which points he had in mind at that time, for it must be admitted that very few traces of actual self-criticism are preserved in his later writings. 3 In order to spare the reader the trouble of looking it up, I reprint the first of the tables in question below, while otherwise assuming familiarity with Böhm-Bawerk’s relevant writings. 51

MARGINALISM AND CAPITAL THEORY

4 In the third edition, Böhm-Bawerk has presented an example that is better in many respects, taking up an idea of John Rae. With a certain sum of labour, let us say A, I can build a house suited to my needs, but of so fragile a construction that it will probably only last thirty years. If after these thirty years I have a new sum of labour A at my disposal, then I can use it to erect a new house, which will also last for thirty years, so that I am assured of housing for a total of sixty years. But if already at the present time I have at my disposal a sum of labour not of 2A, but rather only of 1 A, it is reasonable to assume that I shall be able to give the house so solid a construction that it will last for sixty years. In my present situation, i.e. if I already have at my disposal a sum of labour A, the availability at the present time of an additional A will therefore mean the same to me as the availability of A thirty years from now. In my view it would be pedantry to refuse to admit that here the premium in question arises directly from the technical superiority of the present means of production. This does not of course prevent the size of the capital in each case, the rapidity of its increase, etc., and therefore in the final analysis the rate of interest from being governed by Böhm-Bawerk’s first and second grounds. 5 Logically this must be the case, whatever time period within the twenty years is involved, as soon as it is assumed that the improvements in the quality of the wine constitute a declining series. For n times the first and largest member of the series must necessarily be greater than the sum of the first n members. But it is precisely the former quantity that denotes the enjoyment of one-year-old wine (instead of new wine) during n years, while the latter quantity in turn denotes the single enjoyment of n-year-old wine (instead of new wine) during the nth year. The time after n years, more specifically the time after twenty years have passed, can of course simply be eliminated as it is identical in both cases. 6 It should be observed that here we are dealing with a series of real, successive pleasures, not with a series of imagined, alternatively appearing pleasures, as in Böhm-Bawerk’s scheme. Of course, sooner or later these pleasures are perspectivally reduced from the present point of view; indeed, so much so that they form a converging series, whose final sum then constitutes the value of the means of production. 7 Strictly speaking, subsequent savings will not cost him the same sacrifice as in the first year—they will cost him no real sacrifice at all. For since he finds the enjoyment of 10 litres of one-year-old wine equal to the enjoyment of, say, 15 litres of raw wine, he will for example in the second year be able to lay aside not just 10, but 15 litres of raw wine and yet still retain a normal wine consumption. But in addition to this in the second year, rationally speaking, he will have approximately the same motivation for a new, real savings sacrifice as in the first year. Moreover, our savings plans of course usually encompass not just one year, but several successive years. However, to take this into account would make the problem too complicated and at the same time too poorly defined for it to be possible to analyse it here. 8 The fact that the premium progresses in discontinuous leaps is of course connected with the fact that the production of wine is itself discontinuous, i.e. it proceeds in one-year steps. It should incidentally be observed that our imagined cultivator as labourer and capitalist in one can have no personal interest of any kind in the rise or fall of the premium. 9 In my published lectures, vol. 1 (Jena: Fischer, 1913), I have attempted to pursue Böhm-Bawerk’s original idea of conceiving of capital as an intermediate product (stored labour) somewhat more closely than he himself does. 52

ON THE THEORY OF INTEREST

10 G.Åkerman, Realkapital und Kapitalzins. Inauguraldissertation zur Erlangung der juridischen Doktorwürde (Stockholm, 1923; in German). In the Swedish Ekonomisk Tidskrift (1923) I have both given a thorough review of Åkerman’s work and attempted to approach the same problem using conventional mathematical means. * Editor’s note: This article was typed out after Wicksell’s death which may explain a few obscurities. In Table 4.1 the figures for 1913 and 1914, which were confused in the German version, have been corrected.

53

Part II INCOME, TAXES AND DUTIES

5 ON THE THEORY OF TAX INCIDENCE

I INTRODUCTION The theory of tax incidence or, as it is usually termed by German writers, tax shifting, is not a subject that has received much attention on the part of more recent financial theorists. Among older economists, above all the English writers Smith, Ricardo and Mill, studies of tax incidence still constituted a major part (in the case of Ricardo even the major part) of the theory of public finance. In more modern economic theorists, on the other hand, this subject, which is as important as it is difficult, is mostly discussed with the utmost brevity and with a superficiality that stands in stark contrast to the comprehensive, meticulous scrutiny accorded other details of taxation procedures, details that are often probably far less important. Even though Adolph Wagner fully acknowledges the importance of questions concerning tax incidence, he too deals with them in a way that is surely much too summary, at least in the first edition of his Finanzwissenschaft. This is still more the case in Leroy-Beaulieu, Cohn, Schäffle, and not least von Stein, who polishes off the whole theory of tax shifting in three or four pages, and yet thinks this is enough to achieve a thorough reform of the prevailing views in the field. In one of the most recent German works on public finance, the Grundzüge der Finanzwissenschaft by W.Vocke, the claim is even made that tax shifting is ‘not really a subject that has any place in the theory of taxation’ (p. 210). In making this claim, Vocke, to be sure, wants to include under taxes properly speaking only direct taxes. ‘If’, he continues, ‘excise duties had not been regarded as indirect taxes, nobody would ever have talked about tax shifting’. This is certainly a gross exaggeration. In the case of direct taxes, too, indeed even in the case of income tax, the possibility of shifting cannot necessarily be discounted, at least it has sometimes been claimed that it is possible. However, as long as the greater part of public revenue is raised by indirect Originally published as Zur Lehre von der Steuerincidenz, dissertation, University of Uppsala, 1895; reprinted in Finanztheoretische Untersuchungen, Jena, 1896.

57

INCOME, TAXES AND DUTIES

means, it can surely never be a matter of indifference, precisely when considering the justice of the taxation, who it is, in the final instance, who has to bear the burden of these impositions—whether they are called taxes or not. Vocke’s warning to legislators not always to count on shifting, i.e. not ‘to rely on the market to make good the transgressions of legislation’, is most deserving of appreciation. None the less, in actual fact the indirect taxes are in place, and cannot be got rid of so easily; for the time being, therefore, the question of their incidence will remain one of the most important topics that the theory of public finance has to consider. That it is at the same time one of the most difficult subjects in the entire field of economic theory can hardly be disputed, but this is really no reason not to pursue further studies in the area. Admittedly, in many cases we shall have to make do, perhaps for ever, with the purely negative result that it is impossible to calculate with any accuracy the relative distribution among the persons involved of the duty or tax at issue. But even a negative result of this kind is important and valuable. A tax of unknown or uncertain incidence is obviously for this very reason a bad tax and should if possible be replaced by some other means of raising the public revenue concerned. Fortunately, however, there are plenty of kinds of tax, even among the so-called indirect taxes, where there can hardly be any doubt as to their relative incidence. Moreover, a closer examination will reveal that the difficulties of tax incidence inhered not so much in the subject itself as in the underdeveloped nature of earlier economic theory. The most recent progress in the field of theoretical economics will in fact enable us to pursue for at least a considerable distance several problems of tax incidence whose resolution earlier seemed as good as unattainable. Coincidentally, particularly in recent years the number of works concerned in a more detailed way with questions of tax incidence has again increased somewhat. In the new edition of Wagner’s work considerably more space is devoted to the question of tax shifting, though without a deeper investigation of its fundamental principles. Similarly, in C.F.Bastable’s work Public Finance (1892) incidence is treated fairly thoroughly, both generally and with respect to each individual kind of tax. Several more recent monographs on the theory of tax shifting are worth mentioning, among others Pantaleoni’s Teoria della traslazione dei Tributi, which, though rather long-winded and in terms of its fundamental conception surely somewhat outdated, nevertheless contains many correct observations; Kaizl’s deserving little work Die Lehre von der Überwälzung der Steuern; and the relevant works of Falck, Fleeming Jenkin and Conigliani. The most recent work in this field, as far as I know, is Seligman’s book On the Shifting and Incidence of Taxation, which is both comprehensive and written in a stimulating manner. This book begins with a very complete history of the 58

ON THE THEORY OF TAX INCIDENCE

development of the subject and then proceeds to analyse the most important cases of tax incidence and to subject older opinions on the topic to a critical examination. Unfortunately this part of the work does not always live up to the standard set by the first, historical part. Here and there even logical errors have crept in, which in my opinion seriously diminish the value of the book. Some of these I shall discuss below, others, e.g. his attacks on Mill among others in connection with the practicability of shifting the burden of a tax on houses,1 will be mentioned here only in passing. (Vocke’s remarks on this subject2 also seem to me to be based in the main on a logical error.) In general, comparing older and more modern writers on this topic, I have become convinced that in terms of formal reasoning the former are frequently superior to the latter.3 Only by a deeper analysis of their fundamental principles can we hope at some time definitively to progress beyond the standpoint they represent. Such an analysis, as is well known, was attempted as early as 1868 by Adolf Held in his much-cited essay in the Zeitschrift für die gesamte Staatswissenschaft, ‘Zur Lehre von der Überwälzung der Steuern’ [On the Theory of Tax Shifting]. However, the results were for the most part purely negative. Held points out, quite correctly in my view, that the classical theory of value, which attempted to make average commodity prices dependent exclusively on the costs of production, actually proceeds in a circle, as the cost elements ultimately cannot be determined independently, but rather depend in turn on commodity prices. But he stops short at this result, without even attempting to lay bare the real connection between these things. What lies behind Held’s actual attacks on prevailing theory, incidentally, is not its circularity but rather the fact that he thinks the effectiveness of the compensating forces demanded by the Classical School is very questionable. That the ‘process of levelling can never attain to rest’, on account of other counteractive forces, must absolutely be conceded to Held. But this is a completely different thing from it having no effect, or having an insufficiently powerful effect, and it is precisely on this effect that the whole theory of tax shifting depends, after all. It does not lie within the scope of this work to respond to Held’s other arguments. Some of them at any rate seem to me highly contestable. For example, with reference to land tax, he claims it is completely impossible to prove whether the increase in land rent (in the usual sense) ever ‘derives’ from anything other than ‘newly raised capital invested in the land’, and this tenet is implicitly extended even to urban land rent! The amortizability of land taxes is—if I understand him correctly—simply denied with the strange claim that ‘a land tax of this kind is obviously borne by nobody, which is an absurdity’. Rather, ‘the next owners will find the remaining inequalities [of taxation] already in place as something given and therefore feel them the less’. According to Held, then, these inequalities have absolutely no influence on the purchase price. In Held’s opinion, moreover, the increase in land rent is not an isolated phenomenon. Rather, something similar is found in all areas of business life, and ‘there would be just as much justification for deducting from labourers in the form of taxes everything they now earn in excess of what they earned 1,000 years ago, on account of the growth of civilization, improvements in production, etc., as there is for taking away [the increase in] land rent’ [!]. 59

INCOME, TAXES AND DUTIES

If Held had lived in Sweden, he would surely have become the hero of our ‘Farmers’ Party’, which as is well known has recently freed itself from all land tax and ‘shifted’ it largely on to the unpropertied classes, presumably because the latter now earn enough, compared with 1,000 years ago, to be able to bear this burden, too. To be sure, a few excellent passages in the essay, e.g. the first-rate remarks on the practicability of shifting the tax on labourers asserted by the Classical School, are rightly highly esteemed. On the whole, however, one surely has to concur with Wagner’s judgement on this work, namely, that it is among its author’s weaker achievements.

Now as is well known, since Held wrote that work a complete revolution has taken place in the theory of value and prices. This revolution began with the studies of Jevons, Menger and Walras, and has found its conclusion, at least for now, in Böhm-Bawerk’s splendid work on the interest on capital. The true connection between value and cost, and the path one has to pursue in order to avoid the circularity of the older economists, to which Held merely draws attention, is no longer a secret to science. It therefore seems appropriate to undertake a revision of the prevailing theory of tax shifting, building on the new foundation laid by these advances, and this in fact constitutes the main purpose of the following investigations. But a revision of this kind also seemed to me desirable in two or three other respects, which I shall here briefly enumerate. 1 First, in my opinion, the very concept of tax shifting suggests a rather too narrow conception of the financial problem at issue, which is why I have used the somewhat more general word ‘incidence’ in place of ‘shifting’ in the title of this thesis. For to shift a tax implies that the taxpayer makes someone else reimburse him for a portion of the tax amount he has paid, but he can only do this indirectly, e.g. by successfully raising the price of his product. Now by this process the consumers in question are forced to make a sacrifice, but it is by no means obvious that this sacrifice will correspond only to the portion of the tax from which the producer has thus freed himself. On the contrary, it is fairly certain that in most cases it will turn out to be much higher. Indeed, as can easily be proved with reference to the taxation of monopoly profits, it can even happen that while the producer is forced by the imposition of the tax to raise his prices a certain amount (because he would otherwise have lost still more), in spite of this he still finds his income reduced by the full amount of the tax. In this case one can therefore hardly talk about any real shifting, and the consumers’ sacrifice is completely in vain, in terms both of fiscal results and of the national economy. The latter point was explicitly emphasized by Pantaleoni, though perhaps not quite clearly; Seligman on the other hand presents a completely wrongheaded view of the matter, as I shall show. 2 Second, more carefulness than is sometimes observed is required if the problem of tax incidence in general is to be formulated with complete precision. If a new tax has to be imposed in order to meet outlays already 60

ON THE THEORY OF TAX INCIDENCE

decided on, or if on the other hand an old tax can simply be decreased on account of a financial surplus, then of course it can be asked without further ado by whom the new imposition will be borne or who will benefit from the alleviation. But if we are investigating the incidence of one of the existing taxes or of the tax system in its entirety the matter becomes rather more complicated. To take the first case, it is impossible to imagine a single tax disappearing without a corresponding portion of government expenditure also vanishing and it cannot be wholly without significance for the question of tax incidence which category of expenditure should then be cut. I am not speaking here of the direct utility of the state activity in question for individual members of society; this may be completely impossible to ascertain and is in any case irrelevant to the theory of tax incidence. But the state as a consumer of goods or personal services or as a producer will certainly exercise a certain influence on other factors of demand and supply and thus also on the incidence of the tax. In the case of the entire tax system one would really have to imagine the absence of all state expenditure, which is impossible. Here it does not even help to suppose (as Held does) the case that the state could meet all its responsibilities completely without the help of material means, for then the state would necessarily have to cease completely being a consumer and a producer, and as before, this fact could not be ignored. In such cases the problem must therefore be understood not in the absolute but rather always in the relative sense, i.e. as soon as the incidence of a certain tax or of a system of taxes is under discussion, it has to be imagined that the amount of public revenue at issue is raised in some other way, e.g. by direct income taxes, and then we have to endeavour to ascertain the change in the tax burden that would thus arise. If this is done, in both cases the state activity in question would occur in a very similar way, and can consequently simply be eliminated, which greatly facilitates the argument. The failure to observe this fact has often led to erroneous or uncertain conclusions, particularly with regard to the treatment of the taxation of labourers. No less a person than John Stuart Mill advances the claim4 that a tax on labourers’ wages will do labourers little or no harm, if the state applies the revenue thus derived in such a way that the demand for labour increases accordingly. Even if this view were correct in other respects, which it certainly is not, it would still always be subject to the objection that the state does not seek labour in order to do the labourers a favour, but rather because it needs their labour, and that the manner in which the revenue in question would be used would probably remain the same even if the taxes were derived from some other source, e.g. by a progressive tax on income or property. In other words, the question of the application of state expenditure, however important it may otherwise be, should in general be carefully kept apart from the question of tax incidence, and this is always possible if the latter is understood in the relative sense mentioned above. 61

INCOME, TAXES AND DUTIES

3 Finally, and this is my main objection to the prevailing theory, shifting is chiefly regarded as a matter between the producer and the consumer, or between the representatives of various branches of production. The traditional theory contains very little about the ways in which taxes are shifted between the various classes of producers—the capitalists, the landlords, the labourers, etc.—and the little there is in my opinion is quite unsatisfactory. This obviously has something to do with the undeveloped state of the theory of income in general. Of course Ricardo’s theory of land rents has long provided a certain foundation, but this theory could not be developed further on account of a lack of insight into the reasons determining the interest on capital. The theory of wages was in even worse condition. The theory of the natural wage for labour held by Turgot and Ricardo and the wage fund theory of Mill and Senior are now generally dismissed, but have never been replaced by a more correct theory, again for precisely the same reason.5 But now this lacuna in the theory has been filled in in the most fortunate way by Böhm-Bawerk’s theory of capital interest, or at least this theory forms a point of departure from which further work can be done, as I shall endeavour to show in what follows. The practical significance of the failure to consider the way in which taxes are shifted between different classes of producers must not be underestimated, for the relative portion of the various taxes borne by the producers as such is undoubtedly very substantial, and actually grows with each extension of the range of taxes on production or duties on consumption. If one imagines a final state in which all products were taxed so evenly that all prices, even money prices, remained unchanged, then of course there would no longer be any question at all of the consumers bearing the taxes. However, for the purposes of a more thorough examination, the distinction between consumers and producers as separate economic categories is actually an unnecessary and even incorrect restriction of the problem. Everyone consumes, almost every adult is also a producer of commodities or personal services on the strength of his or her labour or property. The entire domain of private economic life can be understood as a single process of production, in which, after an exchange of products has taken place, the compensation of the productive factors—labour, the productive powers of the land, capital, etc.—is shown in a certain real income, a certain sum of practical utility. Now the question is, when pressure is exerted by the state on some point or another of this system by means of raised taxes, to what extent will this pressure be transmitted to the various other parts of the system, either in absolute terms or in comparison with an equivalent pressure on some other point? This is the most general formulation of the question of tax incidence and actually the only one that incorporates all the consequences of a given tax measure. 62

ON THE THEORY OF TAX INCIDENCE

Of course it is clear that if one puts the problem on so broad a basis, in the first instance it will only be solvable in an extremely abstract form, bearing little resemblance to real conditions. But that in itself does not mean that an investigation of this kind completely lacks (direct) practical application. In fact in the following we shall encounter several propositions advanced in the theory of tax shifting and in tax policy, find some of them confirmed, but be forced to cast doubt on others. In the former case, to be sure, it can certainly not be claimed that a conclusive proof of the propositions in question has been supplied, but none the less a confirmation of these propositions arrived at by a new method has a certain value. In the second case, in contrast, a proposition that does not hold good even under the simplest thinkable conditions can still less have any general validity and must therefore be considered refuted, provided the argument was sound. Moreover, in itself this investigation will uncover certain hitherto unnoticed facts that appear to have decisive significance for tax shifting from a theoretical point of view. Of course it cannot be decided without closer examination whether this significance will also prove to have practical applications; but none the less the theory here provides a hint as to the lines along which practical (statistical) studies might perhaps successfully be organized. However, before we enter into the main part of this study, it seems appropriate to undertake a brief discussion of a specific part of the subject, namely the taxation of monopoly profits and the manner in which it is shifted. In the course of this discussion, to be sure, the opposition between producer and consumer comes into the foreground again. Here the theory is very simple, relatively speaking, and is therefore especially well suited to serve as an introduction to the treatment of the main problem in an exact form, which is to be attempted subsequently. In addition, the topic is in itself of no little importance. Monopoly prices and monopoly income must in fact be regarded as a direct opposite of and necessary complement to the prices and income that arise on the ground of general competition. The following analysis was incidentally motivated in particular by the fact that the discussion of precisely this subject provided by the most recent scholar to work on the field of the theory of tax incidence, Seligman, is in my view incorrect. II THE TAXATION OF MONOPOLY PROFITS In a manner of speaking, any extraordinary business advantage could be regarded as a monopoly. Here, however, we shall only consider those that lend their owner a dominant position in the production of some kind of good or another, so that within certain price limits he is free of all noticeable competition. If in addition the quantity of the commodity in question can be increased at will, then his profit will be determined simply by the volume of sales he is able to make. Now if one further assumes—though to be sure this is only a theoretical ideal—that the possible level of sales at every 63

INCOME, TAXES AND DUTIES

thinkable price is known exactly and is a constantly variable quantity, then as is well known the theory of monopolistic price formation is extremely simple and can even be treated in an exact mathematical form, as Cournot proved long ago.6 Like any other businessman, the monopolist strives for the greatest possible earnings and therefore raises his prices as long as he can. However, with every price rise demand and sales decrease, and finally there comes the point at which these begin to decrease at a faster rate than the net profit on each unit of the commodity sold increases as a result of the price rise. From this point on, every additional rise in the price would again curtail the monopolist’s total net profit and rationally speaking he will consequently have to refrain from further price rises. Now, first, it can readily be seen that a tax that is imposed either at an invariable rate or purely relative to the monopolist’s total net earnings, whether it be proportional to these net earnings or in a progressive ratio to them, can have absolutely no influence on the level of the most profitable monopoly price and will consequently not cause the monopolist to raise prices at all. His profits may be reduced considerably by the tax, but will none the less still attain the greatest sum at all possible in the given circumstances. But it is a different matter if the tax is relative to the volume of the business itself, e.g. if it is calculated according to the quantity of goods sold. Then it obviously has the same effect as an increase in the production costs of each unit of the commodity, and consequently it will now be to the monopolist’s advantage to raise the sales price by a greater or lesser amount. Of course as a result, according to our presuppositions, his net earnings (inclusive of tax) will necessarily decrease somewhat, since of course they had already reached their highest level, but in theory this decrease will always be more than offset by the savings in taxes which will be attained precisely in consequence of the monopolist’s diminished production. This is a direct consequence of the conditions that must be satisfied in the maximization of a constant quantity, and follows very simply from the mathematical treatment of the problem, which we shall undertake subsequently. But the matter is not so easily appreciated without a mathematical treatment. Seligman has obviously not understood it properly, for he says,7 ‘If the tax is small and the demand is apt to fall off a great deal with an increase of price, the monopolist will be likely to find it profitable to bear the tax himself. If, on the other hand, as is the case in general, the demand is more stable, he will be in a position to pass on more and more of the tax to the consumers’, and so on.8 This gives a thoroughly distorted idea of the actual state of affairs. In the latter case the demand cannot be ‘stable’ at all, otherwise obviously the most profitable monopoly price could not have been reached by a long way. Rather, one has to assume here that the demand will always diminish when prices are raised. More precisely, if the 64

ON THE THEORY OF TAX INCIDENCE

greatest possible profit (without taxes) has already been reached, demand will decrease at the same or at a greater rate than the net profit on each single unit increases. Theoretically speaking, however, such a tax will nevertheless always have to lead to a price rise. If in reality a price rise sometimes fails to occur, the cause probably lies, for the most part, in the fact that when the tax rate is small, the relative profit the monopolist could make by saving on taxes is extremely tiny. For in fact, as we shall see, this profit declines in a quadratic ratio when the tax rate declines. This far from unimportant fact can also probably hardly be demonstrated with sufficient clarity without a mathematical mode of treatment. However, even at this point a very interesting fact may be pointed out, which Seligman seems to have completely overlooked, even though it was explicitly emphasized by Pantaleoni.9 The monopolist’s net earnings necessarily decrease in consequence of the price rise, and out of these diminished earnings he then has to pay the amount of the tax. Of course he saves a portion of the tax sum he was intended to pay, but nevertheless he will actually have to pay out of his own pocket and bear (more than) the whole amount of the taxes that flow into the public treasury. Here, consequently, no real shifting of taxes, properly speaking, takes place, but at most an ‘evasion’ of taxes, in the sense used by Hock and Wagner. Now of course this does not do the consumers much good, for obviously they too suffer a loss. In so far as they continue to consume the commodity in question, they lose because of the raised price, and to the extent that they wholly or partially give up their demand for the commodity they undoubtedly lose a certain utility. The latter quantity, of course, cannot easily be given numerical expression, but even if one completely ignores it, the result in any case remains that through this means of levying the tax both parties suffer an extra loss which taken together under certain circumstances can equal or even exceed the amount of money flowing into the tax-collector’s office. This therefore goes to prove in striking fashion the economic superiority of the taxation of net earnings (income tax). In this respect one could even go a step further, as we shall see later. It seems to me that the above result must also be of significance for the regulation of international trade, in so far as monopoly prices occur there. The home country, to be sure, cannot tax the foreign monopolist directly. But if two countries came to a mutual agreement that each would replace an approximately equal sum of tariffs on goods of the kind in question by direct taxes, then all those involved, both producers and consumers, would undoubtedly be much better off than today, when each country is at pains to give tit for tat.

Our task becomes somewhat more difficult when the tax on monopolies is proportional to the value of the goods produced, instead of to their quantity, in other words, when it is measured according to the volume of the gross earnings. In Seligman’s opinion a tax of this kind could not have any influence 65

INCOME, TAXES AND DUTIES

at all even on the most profitable monopoly price, and would therefore necessarily leave the consumers completely unaffected. However, this is quite wrong. The supposed proof Seligman has provided for this claim of his is based, as every reader of his book will easily convince himself, on a confusion between gross and net earnings, since he implicitly assumes that the monopolist has no production costs at all, or at least only ‘overhead costs’.10 In this case, the greatest possible gross revenue would necessarily give rise to the greatest possible net earnings. But if the monopolist, as is usual, also has running costs (costs on each unit of goods produced or sold), then obviously the greatest net earnings will not be reached at the price that the greatest gross revenues would provide, but rather at a somewhat higher price. But this means that with a tax proportional to gross revenues, too, a price rise cannot fail to occur. The simplest way to understand this is probably by means of a ‘reductio ad absurdum’. For of course the tax could conceivably be assessed at so high a rate that it would swallow up the monopolist’s entire previous net earnings. But at that point it becomes self-evident that he has to raise his previous price in order to earn at least something. It is true, however, that an ad valorem tax cannot cause as great a price rise as would occur if the same sum of taxes were raised by means of a (fixed rate) quantity tax. The former means of raising taxes is therefore more advantageous to the consumer—yet it is also more advantageous to the producer. In both cases, as a result of taxation the producer loses more than the entire sum of taxes that is really to be paid, but his excess loss is somewhat smaller with the ad valorem tax than with the quantity tax. So in every respect the ad valorem tax lies right in between the quantity tax on the one hand and the tax on net earnings on the other hand, this latter being, as already stated, the most advantageous for all parties. For the complete proof of the claims made here I must refer to the appendix at the end of this section. However, so as to be able to indicate the main features of my argument at this juncture, I shall here allow myself the simplifying assumption that demand and price stand in a so-called linear relation to one another. This means that price rises (or reductions) of the same amount—say, 1 mark—would also always lead to a decline (or increase) in the (annual, monthly or weekly) sales by an identical amount—say, 100 units of the commodity in question. If we imagine that the changes in price and sales occur in accordance with the same law even in still smaller gradations, then one could plot the price in each case as the abscissa and the sales quantity belonging to it as the ordinate of a curve, and under the given assumptions this curve would be a straight line. Now let us suppose, to begin with, that at a price that merely covers the monopolist’s running production costs—his overhead costs can obviously be completely ignored in this connection—the sales quantity during the time period under observation is 2,000 units. Then it is clear that the most profitable monopoly price must lie precisely 10 marks above the production cost for 66

ON THE THEORY OF TAX INCIDENCE

each unit of goods. At that price, to be sure, according to our assumptions, the demand has already sunk to 1,000 units, but at this sales quantity the monopolist’s net earnings have risen to 1,000·10=10,000 marks, whereas at the price first mentioned, for example, he of course earns absolutely nothing. At a price of 9 marks (always reckoned above cost) he would only receive 1,100·9=9,900 marks, and at a price of 11 marks, 900·11, i.e. again just 9,900 marks. Now if for example a quantity tax of 2 marks per unit is imposed, the most profitable price will lie at 11 marks (above production cost), for at this price the monopolist’s net profit after taxes are deducted will still be 900·(11-2)=8,100 marks, whereas at the old price it would now come to just 1,000·(10-2)=8,000 marks, and at a price of 12 marks (above cost) it would similarly be just 8,000 marks. If one draws up the geometrical figure suggested above (see Figure 5.1) the monopolist’s gross earnings are represented by a rectangle whose sides express the price level and sales quantity in each case. If one draws a parallel to the sales axis at a distance k from the origin (where k stands for the cost of one unit of the commodity), then the part of the rectangle to the right of this line represents the net earnings. The surface area of this rectangle as is well known is at a maximum when the sides of the triangle in which it is inscribed are halved. The most profitable monopoly price is therefore k+10 marks. Now if a tax of 2 marks per unit is levied, this obviously has the same effect as raising the cost per unit by the same amount, and one has to draw up a similar rectangle in the triangle that is bordered by the dotted vertical line plotted at a distance from the origin of k+2 marks, and so on.

In this case, therefore, the additional charge will come to precisely half the tax rate. But one would be very much deceived if one thought for a moment that the producer and the consumers now each bear a half of the tax amount payable. In fact, the consumers lose precisely 900 marks as a result of the price rise (not counting the loss of consumption utility); the net earnings of the monopolist, however, are evidently curtailed by 1,900 marks, while the total sum flowing into the tax treasury only comes to 1,800 marks. Here, therefore, a sum of money of 1,000 marks, i.e. more than half the tax, goes to waste without any use whatsoever.11 If the monopolist had held firm to the old price he would clearly have lost all of 2,000 marks. By means of the price rise he consequently realizes a relative profit of 100 marks. This profit would, however, have turned out much higher at a higher tax rate. If for example a tax of 10 marks is imposed on each unit of goods—which would obviously swallow up the entire net earnings if the price remained unchanged—it is most profitable to raise the monopoly price by 5 marks. Of course, since this means that sales will be reduced by half, the monopolist’s net earnings before taxes will fall to just 7,500 marks, but at the same time the tax sum payable falls to just 5,000 marks, so that he is left with a net profit after taxes of 2,500 marks. Here, however, the added loss to the (remaining) consumers and to the monopolist 67

INCOME, TAXES AND DUTIES

himself would come to 2,500 marks each and would therefore when combined be equal to the total tax amount. If on the other hand the tax rate falls, then it can readily be recognized that the relative profit we have talked about diminishes not just at a constant but at a quadratic rate. At a tax rate of 1 mark this profit would not as one might suppose come to 50 but rather only to 25 marks. At a tax rate of 20 pfennigs, the monopolist would lose 200 marks if the sales price of his commodity remained unchanged, but he would still lose 199 marks if the price were raised by 10 pfennigs (which in theory would now be the most profitable price for him). Here, therefore, his relative profit would come to just a single mark. He would probably hardly go to the trouble of raising prices in order to achieve this tiny advantage; hence, this method of raising taxes might appear unobjectionable in this case. However, if he does go through with the price rise, then in order to earn him this 1 mark the consumers will have to suffer a pecuniary loss of no less than 99 marks—and in the process the public treasury will none the less lose 2 marks. It is worth emphasizing that this theoretical result is entirely independent of the law governing the connection between price and sales, in other words of the form of the demand curve, since in general, when small changes are involved, each function can be substituted by an expression of the first degree and each curve by a straight line. There doubtless lies in the circumstances described above at least a partial explanation of a fact that people often claim to have observed, namely that a production or consumption tax of a moderate amount has occasionally led to absolutely no change in price. This is supposed to have been the case, for example, when the Bavarian malt duty was raised, especially in the larger cities, according to the investigations of G.Schantz (Schmoller’s Jahrbücher, 1882). Now it may be pointed out that, as Schantz himself emphasizes, precisely in the larger cities of Bavaria the beer breweries possess a monopolistic character. For this reason the possible relative profit which they could 68

ON THE THEORY OF TAX INCIDENCE

realize by raising prices would probably be utterly negligible if the tax rate is low, or, which amounts to the same thing, a small increase is made to an existing tax whose effects have already settled. If, however, under such conditions a price increase really does occur as a result of the tax, the tax must necessarily be regarded as a failure, since the consumers’ sacrifice, as we have seen, then occurs completely in vain.

With direct taxation of net earnings, this consequence would fail to occur. But the question here arises whether one could not go a step further in this regard. Suppose that instead of imposing a tax on monopoly goods the state were to grant the producer in question a bounty of 2 marks for each unit of his commodity consumed (in the home country). Then it would be most profitable to the producer to lower the sales price by 1 mark, by which means the demand would grow to 1,100 units. The monopolist’s net profit would consequently now be 11·1,100=12,100 marks, while to be sure, the subsidy paid by the state would amount to 2,200 marks. But the growth in the monopolist’s net profits caused by the state subsidy could be collected back into the state treasury without infringing on his interests by an appropriate increase in the taxation of his income. This would mean that the state would be left with a loss of just 2,200-2,100=100 marks, at most, but against this loss would be set a savings of 1,000 marks for the previous consumers, as a result of the price reduction. In addition, there would be a certain increase in utility for the new consumers. From the point of view of the common weal, therefore, this procedure would undoubtedly have a great deal to recommend it. To be sure, I do not claim that it is applicable in practice, and indeed, the intended purpose—to make better use of a private monopoly without infringing on the interests of the monopoly owner—can in general be attained in many other ways. What I have said is only intended to draw attention to a circumstance that, as we shall see later, is also of far-reaching significance for other parts of fiscal and economic theory.12

APPENDIX Let the annual demand for monopoly goods be represented by a function F(p) of the unit price p, a function that we assume to be completely continuous. The total gross income will then be expressed by p·F(p). Now if the monopolist’s (annual) overhead costs are denoted by K and the running costs per unit of goods by k, then the net profit is obviously (p-k)·F(p)-K. As is well known, this expression reaches its maximum value when its differential quotient relative to p disappears, i.e. when 69

INCOME, TAXES AND DUTIES

(p-k)F’(p)+F(p)=0 or

This relation, incidentally, is capable of a simple geometrical interpretation, as the expression to the left represents the subtangent of the demand curve at the point in question (if p is plotted as the abscissa and F(p) as the ordinate of the curve). As our formula shows, and as is also self-evident, the monopolist’s overhead costs, which are independent of the volume of sales, have no influence on the formation of the most profitable price, nor will any tax of a fixed amount be able to bring about any price shift whatsoever. If a tax of this kind were to swallow up the entire net profits previous to its imposition, it would make the business as such impossible. Matters are just as simple in the case of a tax that takes away a set fraction of the net earnings. Such a tax, too, as is immediately obvious from our equations, would leave the setting of prices completely unaffected; it would only curtail the monopolist’s net profits by the amount of the tax, no more and no less, and would not affect the consumers. Similarly, a progressive tax rate, which approaches asymptotically a set percentage as net earnings grow, would, as is easily seen, have exactly the same consequences. If, however, the tax is imposed at a fixed rate for each unit of the sales quantity, then this obviously has the same effect as if the running costs were raised by the rate of the tax. If the tax rate is s, then the most profitable price will now occur at

which, as can be proved without difficulty, always signifies a price rise. If, in contrast, the tax is assessed at a set percentage, e.g. of the gross receipts, or, and this comes to the same thing, of the sale price of each unit of goods, then obviously the monopolist’s net profits are now expressed by

and this quantity reaches its maximum value when

70

ON THE THEORY OF TAX INCIDENCE

Evidently, this level of prices will be the same as with the fixed rate quantity tax just discussed, as long as this rate comes to precisely of the running production costs per unit of the goods in question. For example, an ad valorem tax that comes to a tenth of the gross earnings whould lead to the same price level as a quantity tax that came to a ninth of the running costs. An ad valorem tax taking away half the gross earnings would be equivalent in every respect to a quantity tax equal to the running costs, and so on. This quite peculiar relation, too, is evidently completely independent of the form of the demand curve. On the other hand it can easily be shown, using the equation above, that the amount of the ad valorem tax in this case would be greater than that of the quantity tax. For since the expression to the left is always positive, obviously we have

and therefore

and so on. But then, by the same token, if the same amount of tax is to be raised in one way or the other, the price increase must necessarily be smaller in the case of the ad valorem tax than in the case of the quantity tax. The former tax will therefore be more to the consumers’ advantage than the latter—yet it will also be more advantageous to the monopolist himself, since he obviously loses more the higher he has to set his prices as a result of the tax. (Here it is self-evident that Seligman’s claim, mentioned above, is wrong. It would in fact only be correct if the monopolist’s running costs completely fell away.) The general proof of the second of the theses advanced in the text, namely, that the relative profit the monopolist can realize by means of increased prices declines at a quadratic rate relative to the tax rate, if this is sufficiently small, will be easy for any reader with a training in mathematics. Incidentally, as already stated, it is possible here without detracting from the general validity of the proof to substitute the demand function in the vicinity of the monopoly price in question with an expression of the first degree, such as F(p)=a-m·p, which makes the formulation of the proof exceedingly simple. 71

INCOME, TAXES AND DUTIES

III TAX INCIDENCE UNDER CONDITIONS OF FREE COMPETITION. SIMPLEST HYPOTHESIS. INTEREST AND WAGES As indicated above, in a theoretical discussion of tax incidence the distinction between producers and consumers can be completely ignored if the entire economic life of a country is understood as a single production process, the result of which, after an exchange of products has taken place, is represented in a certain real income for each person. In viewing the economy in this way, we disregard the occurrence of monopoly incomes as well as any entrepreneur’s profits that would denote something other than recompense for the labour of management. Instead, we assume that in economic terms the entire economic capital of the country is a single, undifferentiated mass, all parts of which are equally capable of employment in one type of production or another. The more precise definition, or rather the necessary limitation of the concept of capital will become evident of itself from the following observations. In order in general to give the matter as simple a form as possible and thereby to allow the essential point that we should like to expose to emerge clearly, we shall in the first instance permit ourselves the assumption that there exists in an economic union, e.g. in a country, which is equipped with a certain quantity of capital, labour and land, only a single branch of production, i.e. only a single object of consumption (W) is produced, while everything else that is consumed in the country is acquired from outside by means of direct exchange, and to be precise under conditions of exchange that can be regarded as determined by the state of the world market. This means that the capital of the country consists essentially partly in stocks of the finished commodity (W), partly in equipment and raw materials for its production. One could even think of the income of each individual in the first instance as drawn directly in (W). Clearly, a tax affecting domestic production under such conditions could give no cause for any changes in the price of the commodity (W) or of other consumer goods. Rather, its effect will be seen exclusively in the altered income of the various producers, the capitalists, the labourers and the landowners. For the time being, however, we shall ignore the productive powers of land by assuming them as free, though in the next section we shall drop this restriction. Judging the influence of any tax on the level of interest and wages, that is, on the remuneration of the various classes of workers, would be a relatively easy task under such conditions, provided only that it was known how and by what causes those two economic quantities in general are determined. But until recently one could have searched through the whole literature on economics without receiving an answer to this quite elementary question of economic theory that was at all satisfactory. 72

ON THE THEORY OF TAX INCIDENCE

We can probably leave out of account here the answer that the older economists gave to our question, referring to the tendency of the population to multiply, namely that in the long run the labourer draws just precisely what is necessary to maintain his own life and that of his family, while the capitalist retains the entire remaining output for his own account. Even if it is still only too true, unfortunately, that a large part of the world’s working population lives close to the subsistence level or even below it, this can none the less not be considered a completely general phenomenon, still less an economically necessary fact—and incidentally, the latter was never claimed by Ricardo, for instance. There is also good reason for ignoring the well-known variant of this doctrine presented by Karl Marx. He claims, for instance, that the employer buys the commodity labour at its cost value, i.e. the food necessary for the maintenance of one working-class family, but sells it (in the form of finished products) at its consumption value, and that the difference is surplus value, and so on. Further, this depressed price of labour is arrived at not by means of natural population growth, but rather by creating a kind of artificial surplus in the working-class population, an ‘industrial reserve army’. The facts confirm neither of these views. In our day, there are countries where for several decades the wages even of unskilled labour have almost without exception stood far above the subsistence level. As to Malthus’s law of population in particular, its significance lies in the fact that it teaches us about the nature of the various ‘checks’ that have the function of counteracting the physiological tendency of the population to increase, and that it shows the relative necessity of these checks, but not in the eternal inadequacy of these checks having been proved or ever asserted by Malthus. All that really remains to be discussed, therefore, is the wage fund theory, which attempted to make the level of wages dependent on the ratio between the size of the available capital intended for the payment of wages and the number of workers. This argument was, of course, accompanied by the conclusion that the level of interest, conversely, tends to rise when capital grows in relative terms (or the working population decreases in relative terms), and tends to decline in the opposite case—and this incidentally is very probable in itself. With reference to the theory of tax shifting, it would follow from the opinions of the older economists that a tax, no matter how it was imposed, could always successfully be shifted from workers to employers, but never from employers to workers. According to the wage fund theory, in contrast, it would only be possible to shift a tax if either the size of the capital or the number of workers (i.e. the supply of labour) changed as a result of the tax, but given such a change, the tax would then be shifted irrespective of how the tax was imposed. This theory has remained the dominant one, although the wage fund theory itself, as is well known, is no longer as highly regarded as it used to be. 73

INCOME, TAXES AND DUTIES

These days, the wage fund theory is considered a false doctrine, yet I do not think I am overstating the case when I claim that most of those who have criticized it have not quite hit on the point where it is really in error, which is also why nobody has yet succeeded in putting a more perfect theory in its place. It is hardly worth polemicizing against authors who join Macleod, Karl Marx or Henry George in simply denying that the workers are maintained by the capital. One can only be amazed that sensible men have let themselves be misled by such superficial scholarship. Karl Marx even goes so far as to claim that it is rather the workers themselves who advance the capitalists the amount of their wages, since after all in common practice wages are often not paid out until a week or several weeks after the work in question has been done.13 Of course this is merely splitting hairs. It is true that the workers do advance the wage during this week or these weeks, and in doing so they themselves can be regarded as capitalists with respect to this sum. But this falls far short of what is required, for of course in the broader economic perspective the wages of labour have to be advanced up to the moment when the finished consumer goods, the goods of the first order (Menger), can be exchanged for other consumer goods, and this period of time on average amounts to not weeks but years. If the workers were able to support themselves for this entire period of time, then in fact no other capital would be necessary. It can of course occur in exceptional cases that the wages of labour are not paid out until after the completion and sale of the product; but if this is the case then the capitalist as such can make no claim to any part in the product of labour either. In his review of my work Über Wert, Kapital und Rente (Schmoller’s Jahrbuch, vol. XIX, 1, p. 336) Professor Lexis observes that to be sure the worker ‘can for the most part only turn [his labour] to account by participating in the production of future goods’. But this, he writes, is ‘not the decisive point, for even if, like for example the baker’s apprentice, he produces a good to be consumed on the day of its production, he is none the less forced, because he is unable to exploit his labour independently,…to sell it for a more or less adequate means of subsistence while renouncing all claims to the product of his labour’. But how does Lexis know that the worker in this case has to renounce his claims to a part of the product of his labour? For in general, the value of his work cannot be independently ascertained, but can rather only be assessed in combination with the other labour employed in production, and this must for the most part be paid for in advance. That the capitalists as such—ignoring monopoly incomes and the like—on average draw more than the interest on the aggregate wages advanced directly or indirectly by them, does not seem to me obvious in itself, but rather self-contradictory.

The other objections to the wage fund theory mostly come down to the view that it is not certain in advance how large a part of his property the capitalist wants to employ in production. That may be true, but it surely hardly constitutes a sufficient reason to reject the theory in question. After all, at 74

ON THE THEORY OF TAX INCIDENCE

any given moment the wealth employed in production has a specific magnitude, and as the level of annual savings, or any possible increase in the unproductive consumption of capital, is mostly small relative to the size of the total stock of capital, this stock can even be considered virtually constant over any period of time that is not too long. The real error in the wage fund theory probably lies instead in the fact that the wage fund itself only represents a part of the entire productive capital, and indeed, a part that for the present is completely undetermined. It has usually been identified more or less with the ‘circulating’ portion of the capital, as distinct from the portion ‘invested in more or less fixed ways’—though, incidentally, neither of these concepts is very specific. Now if it is asked what determines the division of capital into ‘fixed’ and ‘circulating’ parts, then in general the answer given is that the state of the technology is decisive. This answer, however, does not contain the whole truth; rather, alongside the purely technological element an element of economic practice can also be clearly distinguished. It is a well-known fact that the introduction of certain technical improvements, machines and the like, is often only profitable when wages have already reached a certain level, whereas on the other hand if the level of wages is low it is more profitable to the entrepreneurs to use predominantly manual labour in spite of its technical inferiority. We will return in a moment to the explanation for this; if we assume it to be an established fact, then the dilemma results that on the one hand, given a certain total capital, the level of wages is supposed to depend on the distribution of this capital into fixed and circulating portions, but on the other hand, this distribution (all other things being assumed equal) is supposed to depend on the level of wages in each case. Now in itself, this is not in the least surprising. In other areas of economics, too, similar phenomena occur—e.g. in a market economy supply and demand are dependent on the price, but the price in turn is dependent on supply and demand—but here at least it emerges that we are dealing with two mutually dependent quantities. In order to be able to determine them, we then obviously have to know not just one but two relations connecting them. Of these two the wage fund theory in fact delivers one; but its main mistake is to have regarded this one relation as sufficient. In order to become more closely acquainted with the second necessary relation, one needs to know above all why it is that increased prices for manual labour regularly lead to the introduction of mechanical equipment. This occurs not just indirectly by spurring on the spirit of invention, but also directly, in that inventions already made only become economically viable at a higher level of wages. This phenomenon requires explanation all the more since at first glance it even appears rather paradoxical. For in fact the machines themselves are products of labour; if therefore labour becomes more expensive, then it might 75

INCOME, TAXES AND DUTIES

at least appear that the machines must become correspondingly more expensive, too. However, in general, as Ricardo has shown, this will not be the case.14 For the price of the machines, says Ricardo, includes not only wages but also profit, namely above all the interest on the capital by means of which the labourers employed in producing the machines were maintained. Now if wages rise over the whole line of production, and if in consequence—all other things being assumed equal—the rate of profit sinks, then this means that the price of the machines becomes cheaper relative to manual labour. Ricardo even claims that the price of the machines should not go up at all, though this is not shown by his argument. Incidentally, according to his well-known principle, wages could only go up as a result of a rise in the prices of foodstuffs, though this is of no relevance here.

If one pursues somewhat more precisely this line of thought, which Ricardo has merely explained by a numerical example, it becomes clear that the heart of the matter lies in the fact that the entire process of production is lengthened by means of the introduction of the machines in question. In Ricardo’s example this process is lengthened from one year to two, which means on the one hand that the same original quantity of labour produces a greater quantity of finished products, but on the other hand that the capital employed in the maintenance of the labourers is invested for a longer time, say, for two years instead of one, and must therefore ultimately be compensated with two-year instead of just one-year interest. This then necessarily leads, as we shall soon show, to the longer-term capital investment (the introduction of the machine) only becoming economical at a quite precise level of wages. It is reflections of precisely this kind that have enabled Böhm-Bawerk— though to be sure the relevant passage in Ricardo probably never occurred to him at the time—to succeed, in the first instance for stationary conditions, in setting up a real, consistent theory of the reasons determining interest and wages. The only earlier attempt known to me to find out by rational enquiry in what way interest and wages are related (disregarding the misguided speculations of Thünen on the natural wages of labour) occurs in an article in vol. I of the Danish Nationaløkonomisk Tidsskrift (Copenhagen, 1873). The essay, which, significantly, was written not by a professional economist but rather by two mathematicians, F.Bing and Julius Petersen, bears the title, ‘Bestemmelse af den rationelle Arbejdsløn samt nogle Bemærkninger om Økonomiens Methode’ [Determination of the rational wages of labour, with some observations on the methodology of economics]. A few objections can perhaps be made to the details of this study, but its fundamental thought is closely related to that of Böhm-Bawerk, and if developed further would doubtless have led to a theory very similar to his. However, I cannot find any trace of the essay having been paid any attention by Danish economists. In a note, the editors of the journal reserve to themselves the right to respond to the article, but as far as I know, such a response has never appeared. 76

ON THE THEORY OF TAX INCIDENCE

In my work Über Wert, Kapital und Rente, published in 1893,15 I have endeavoured to give Böhm-Bawerk’s theory a strict mathematical form and to supplement it in certain respects. For a fuller understanding of the following elaborations, perhaps I may refer the reader both to that work and to the fundamental work of Böhm-Bawerk. However, since the fundamental idea of Böhm-Bawerk’s theory has not yet found general acceptance and has recently been energetically attacked by leading authorities, I think I have good reason to present and defend it briefly once again. This will also give me the opportunity to improve my earlier presentation in a few particulars. In place of the conventional but hardly tenable division of capital into fixed and circulating capital, following the procedure of Jevons and BöhmBawerk the entire stock of capital should be conceived of as mobile, i.e. as circulating more or less slowly. In virtual terms it is regarded as it were as a single sum of finished consumer goods, which was originally intended for the maintenance of the workers but has for the most part already been exchanged for labour and then continues to exist in the form of previously performed labour (tools, raw materials, preliminary products, etc.) until it is again replaced by the proceeds of the sale of the final products. However, when looking at things in this way, in my opinion it is better to exclude extremely durable capital items, such as buildings, roads and the like, from the concept of capital. Under stationary conditions, which as the simplest hypothesis we want to make the basis of our argument here, only very small additions are made to the stock of such objects, whereas the other capital items such as tools, machines, raw materials, etc., have to be renewed from time to time. In the first instance, therefore, those durable items can be placed alongside land as ‘rent-generating goods’, whose connection with the problem will not be considered until later. The time that passes between the investment of an individual piece of capital in the purchase (payment) of labour and the restitution of this capital by means of the sale of finished consumer goods (goods of the first order) is called the circulation or investment time of the small portion of capital in question. Now if, as is usual, labour is employed in the production and sale of a commodity, and in consequence capital has to be advanced at various points in time, then we say that the capital sum involved has an average investment time. If, for example, a week of labour is applied today, and another week next year, and so on, until the finished commodity is sold after four years, years, and if no other then on average the labour pays off after expenses had to be incurred in the production of the commodity, the price it is sold for minus the four weeks’ wages, which thus constitute the entire invested capital in this case, will represent the interest on that capital over 2 years (of course, we are only using simple interest in our calculations here). 77

INCOME, TAXES AND DUTIES

If on the other hand a piece of work is performed today that only pays gradually over a longer period of time (such as, for example, the manufacture of a machine that takes several years to pay its costs), here too it will always be possible to calculate the average investment time of the capital advanced. If, for example, a capital investment k brings in receipts of p each year for four years, and then nothing more, the present value of the four revenues (calculating with simple interest) is obviously

where z signifies the rate of interest that is still to be worked out, and this expression must exactly equal k. If z can here be assumed to be sufficiently small, then the expression can ultimately be simplified to: p(4-10z)=k, whence we derive

But now, if t denotes the average investment time of the capital, we have

i.e. after eliminating z,

Since 4p is necessarily greater than k, the investment time of the capital will accordingly be somewhat longer than the average length of the interval that passes between the expenditure of the capital and its full reimbursement,* for of course in our case this interval is also

In a similar way, all production processes, which are often quite complicated, can always be reduced to a specific investment time of the portion of capital involved, and thus there ultimately emerges a specific average investment time for the total capital of the business involved. Of course in this calculation one always has to imagine the various stages of the same production process as belonging to a single business. In other words, for the purposes of our study, the capital remains invested even if the capital item in question (e.g. a machine) passes from the possession of one entrepreneur to another. 78

ON THE THEORY OF TAX INCIDENCE

I have felt it advisable here to revive the older expression ‘investment time’ used by Jevons in place of the concept introduced by Böhm-Bawerk, ‘production period’, i.e. the time that passes between the beginning and end of the production process. Of course the two concepts are closely related. If, for example, the same number of workers is employed with no variations for the whole of the production period and their wages are paid out gradually, then, as can easily be discovered, the average investment time of the capital is precisely half the production period. If, however, as is usually the case, the quantities of labour employed are distributed quite unevenly over the production period, then it is not so easy to find out the ratio between the investment time and the production period. In general, the concept of the ‘production period’ is somewhat unclear, and sometimes cannot even be precisely defined. But ultimately in dealing with the questions to be discussed here, what matters is not the length of the production period but rather only the length of the investment time of the capital, and this is always easy or at least easier to grasp and to define. There therefore appears to be every reason for using only the latter concept when exact relations are at issue.

Now the main point of Böhm-Bawerk’s theory consists in the assumption that it will always be possible to raise the average productivity of the labour employed by interpolating where appropriate one piece or another of preparatory labour—which ultimately is equivalent to a lengthening of the average investment time of the capital applied. He means this, of course, not just in the sense that more labour yields more products, but rather in the broader sense that a given quantity of labour spread as it were over a longer period of time will prove more productive than the same quantity of labour employed in a shorter production process of the same kind or still more in (capital-free) immediate production. This would in fact be a perfect counterpart to the phenomenon known by the name of division of labour. If several workers together carry out the production of a certain kind of commodity while dividing up the labour as far as possible, each of them will be able to deliver more products using the same amount of labour than if he worked under conditions where the division of labour were less developed, or still more, if he worked on his own. I have used the word ‘assumption’ above, as I completely agree with Lexis when he says that ‘the thesis that there is a functional connection between the length of the production period and the productivity of labour…has to be proved’, and proved ‘in concreto and separately for all different kinds of goods production’.16 Admittedly this proof will require both comprehensive and difficult statistical enquiries for differing periods and countries. However, it seems to me that Böhm-Bawerk has demonstrated the inherent probability of the thesis not just ‘by means of a few general observations’, but rather very thoroughly. In his review of my work Über Wert, Kapital und Rente, mentioned above, Lexis believes himself able to prove that, on the contrary, Böhm-Bawerk’s thesis stands in contradiction to the facts. However, his argument shows in my opinion that he has probably failed to arrive at a quite clear idea of what is really involved. 79

INCOME, TAXES AND DUTIES

Lexis ‘would like to maintain that since the beginnings of civilization, technological progress has had a twofold tendency, which has been marked by increasing success: on the one hand, certainly, it has tended to reduce the number of labourers supported by an equal capital in different economic enterprises, but on the other hand, at the same time it has tended to shorten the production period’. In a way, this claim already involves a contradiction in terms, for at least if the number of workers supported by an equal amount of capital has been reduced more than the average level of wages has increased, then according to the concept of capital used above, this is evidence that a lengthening of the average production period has taken place. This follows easily if one thinks of the existing capital broken down into labour. But let us hear how Lexis continues. ‘Consider the amount of time that has been saved in the course of our century, at every stage of production and the movements of trade, by means of new methods of transportation.’ Well now, just how certain is this? To be sure, the transportation of a given quantity of commodities over a given distance demands far less time today than it used to. But on the other hand, we must consider that the portion of total production entrusted to any means of transportation at all at the beginning of the century was utterly negligible. Most products, by far, were consumed at the place of production itself or in its immediate vicinity. These days, however, the populations of entire countries are supplied with their most important consumer goods from the most distant parts of the earth, and it is almost an exceptional case when any commodity comes to be consumed at the place it is produced. If one takes this into account—and the point is, here one really has to pay due attention to everything—and if one considers further that the labour employed in the manufacturing of ships, railroad equipment, etc. (not to mention the construction of the railroads themselves) has to be paid for several years in advance, before the capital involved is reimbursed, then there can hardly be any doubt that modern means of transportation have not in themselves led to a shortening, but rather to a very substantial lengthening of the entire production process. Lexis continues: ‘Making a sharpened stone axe with a hole in it for the shaft must have cost neolithic man many times as much time as making an iron axe requires today, even if all preparatory labour is included in the calculation’, since ‘given modern mass production only a very small part [of this labour] could be attributed to a single axe.’ Now of course this is obvious. But here Lexis is evidently mixing up two very different concepts—the length of the production period (or of the investment time of the capital), and the working time needed to make one unit of the commodity. The latter has unquestionably been reduced, otherwise of course no technical progress would have been achieved. With the same total expenditure of labour—say, x days of labour—that a stone axe required at that time, it may be possible to make ten, twenty or more iron axes today. But those x days in neolithic times—as far as can be guessed from the workmanship—were applied in an unbroken sequence, whereas today for the most part they span a period of several years, so that on average only a few minutes of each calendar day in this period are used in the production of precisely those ten or twenty axes. This is the real difference, and this is also the reason why present-day production can usually only be carried out by capitalistic methods. Something similar probably applies in the other examples Lexis adduces. But here, finally, it must be pointed out that Böhm-Bawerk has never denied, indeed, he has explicitly emphasized, that there are undoubtedly some kinds of technical progress that in every respect have caused a shortening of the previously customary production process. However, in doing this they did not demand any new accumulation of capital; on the contrary, at least in the long run they set free capital—which, incidentally, always directly benefits the workers. 80

ON THE THEORY OF TAX INCIDENCE

But if such inventions had really made up the majority of cases of technical progress, the inevitable consequence would have been that the total capital fund relative to the number of workers (or actually to the total sum of the workers’ annual wages) would gradually have declined. But Lexis himself admits—if I have understood him at all correctly—that the contrary is the case, and by this admission, as was observed above, he has invalidated his own argument from the very outset.

Now if, however, this general postulate of Böhm-Bawerk’s theory is conceded, then it might at first glance be supposed that the capitalist would always strive to progressively lengthen the investment time of his capital, at least for as long as the interest that might be lost during the transitional period can be ignored. However, this will not be the case at all; at any given level of wages there is rather one investment time that is the most profitable, as the following example will show.17 For the sake of simplicity we assume that finished products are only completed at long intervals, so that the capital invested in the business all becomes free at the same time. In reality this is generally not the case, instead production occurs in stages, but this makes no essential difference here. Now with a three-year investment time, for example, the capital advanced, i.e. the sum of the wages paid out in instalments might be reimbursed by selling the finished goods with a mark-up of 25 per cent; the annual profit on capital is consequently 25 : 3=8 per cent. Let us suppose that with a four-year investment of capital it is possible to reimburse the wage sum advanced with a mark-up of 32 per cent, which is equivalent to an annual interest rate of just 8 per cent. Consequently, the three-year investment is the more profitable one for the capitalist. But if wages rise by one-tenth, then in the first case the mark-up will only constitute approximately 14 per cent (15 :110) of the present wage sum, which corresponds to an annual interest rate of 4 per cent. With the fouryear investment, on the other hand, the remaining profit margin would still be 20 per cent (22:110) of the wage sum, and the annual interest on the capital would therefore be 5 per cent. That means that the four-year period is now the more profitable. The length of the most profitable investment time will consequently be greater when wages shift upwards. This result has general validity, as long as we can assume that though every lengthening of the investment of capital raises the productivity of labour, the increase in productivity is not directly proportional to the increase in time, but instead the returns increase on a declining scale, which seems entirely consistent with reality. The most profitable length of the investment of capital is therefore in fact determined by the level of wages. However, it is equally true that the latter quantity is dependent on the average length of capital investment, and this we see as follows. Any capital whatever—always assuming the productive powers of land to be free—can only find productive employment in one way, namely by hiring labour. But 81

INCOME, TAXES AND DUTIES

until it is reimbursed by production, it can only ever pay off labour once. Consequently, if K is the capital of a business or a country and if the average investment time of this capital comes to t years, then K:t denotes precisely the annual wage fund and, when divided by the average rate of wages, shows unambiguously the number of workers employed. This statement is axiomatically clear and it constitutes, as Böhm-Bawerk has emphasized, the true meaning of the much-discussed wage fund theory. Now the number of workers determined in this way must obviously in the long run tally with the number of workers really present. If not, then either the working capital itself would be excessive and wages would necessarily have to rise by competition between the capitalists, or on the other hand not all workers would be able to find employment at the rate of wages concerned, which would have to lead before long to reduced wages. Consequently, when equilibrium has been reached, the relation K:t=A·l or K=A·l·t, where A denotes the number of workers and l the rate of wages, must necessarily be satisfied. The level of wages and the length of the investment time therefore determine one another reciprocally, and if the law were known by which the productivity of labour grows with the length of the investment time of the capital, then it would be possible to determine not only these quantities, but also the level of capital profit with mathematical precision, i.e. to express all these things in terms of the quantities K and A presupposed here as knowns.18 Given this situation, if K grows while A remains constant, or conversely A grows while K remains constant, then in theory after some fluctuation to and fro a new position of equilibrium will eventuate. Thus, a relative expansion of the national capital will always lead to a rise in wages, even if this rise will never be proportional to the expansion of the capital. Conversely, a relative contraction of the capital (e.g. as a result of an excessively rapid increase in the population) will lead to a reduction in wages, though this reduction, too, will not occur at the same rate as the contraction of the capital itself. In the process, the rate of interest always changes in the opposite direction to the size of the national capital (it should be observed that here we are always dealing with essentially stationary conditions, so that new inventions and the like can be disregarded). In fact, in the most extreme cases the rate of interest may even decrease or increase at a greater rate than the total capital itself increases or decreases respectively. Now if, as I believe, the law of the relative compensation of the two production factors, labour and capital, has been correctly laid out in its fundamentals, under the assumption of stationary conditions, by the above 82

ON THE THEORY OF TAX INCIDENCE

theory, then from this follows as a simple corollary the decision as to the incidence of a tax on production. In the first instance, it is clear that a tax on the income of the capitalists or of the workers would not have the least influence on the relations between the interest on capital and the wages of labour, if these, as assumed, are determined under free competition on both sides. The highest possible interest will always be the highest possible, even when the state takes away a certain part of it in the form of taxes. But a tax on wages would be equally incapable of causing any change in the most profitable length of the investment of capital. In so far, therefore, as these taxes do not exercise any influence on the formation of capital or on population trends (or, to put it better, the supply of labour), with which we do not want to concern ourselves at present, wages will in all probability remain at the same level as just stated. But production arises out of the combined effect of both these factors; it is the result of labour maintained by capital. If the tax affects production, e.g. by taking a certain part from the value of the product sum or, which in our case is the same thing, from the product sum itself, then both factors of production will necessarily be directly or indirectly affected by it. The entrepreneur cannot completely recoup himself by reducing wages, the lending capitalist cannot demand the same interest as previously. Otherwise, obviously, part of the available capital would become surplus, precisely as a result of the reduced wages, and this free capital would strive to find new employment, which would lead to a renewed rise in wages. However, the workers will be equally incapable of maintaining wages at their previous level, as Smith and Ricardo thought. To be sure, it might seem that there would be nothing to prevent them in so far as the capital is assumed unchanged. But under such conditions the previous investment time of the capital would no longer prove the most profitable; rather, in order to realize the greatest profit on capital now possible the investment time of the capital would have to be lengthened, which would mean that wages would have to decline.19 This emerges with full generality from the mathematical treatment of the problem that we shall undertake in the appendix; but the result can also easily be checked using the numerical example given earlier. If the production figures there—125.5 for the three-year capital investment and 132 for the four-year capital investment (the aggregate of the wages paid out, i.e. the capital in both cases, being set at 100)—are reduced by the tax by, for example, a tenth each, i.e. to 113 and 118.8 respectively, then it becomes clear that the annual interest rate in the first case would now come to 13:3=4 per cent, but in the second case 18.8:4=4.7 per cent. Consequently, the four-year investment will now be the more profitable, instead of the three-year investment, as previously. Evidently, the effect of the tax is exactly the same as if the productivity of labour had declined in proportion to the tax. One could therefore alternatively imagine that in the beginning everything—including the size of the capital 83

INCOME, TAXES AND DUTIES

and the level of wages—had declined at the same rate, while both the investment time of the capital and the interest rate on the reduced capital remained completely unchanged. Now if the capital returns to its previous higher level, then in line with our previous discussion the wages of labour must rise, but not in proportion to the expansion of the capital, i.e. in this case not to such an extent as to attain the level they originally had before the tax; the interest rate on the other hand will fall, but not so much as to oblige the capitalist to bear the entire burden of the tax. Therefore undoubtedly not only capital but also labour is affected by the tax—the relative effect depends on the particular circumstances of production, that is on the law according to which the productivity of labour will increase as a result of any lengthening of the investment of capital. If the scale of added returns mentioned above is a rapidly declining one when the capital investment period is lengthened just slightly compared with the previous investment time, then the tax will chiefly affect the capitalist and spare the workers, relatively speaking. But if, on the contrary, each additional lengthening of the investment time yields virtually the same added returns, then the capitalist will be able to shift most of the tax to the workers in the form of a wage reduction. Let me emphasize that here we are always dealing with the effect of free competition on both sides. If the employers combine against the workers or the workers against the employers, then completely different conditions can arise. Among other things, in the first case a lengthening of the capital investment (expansion of the fixed capital) even in excess of the economically optimal period would become a terrible tool in the hands of the employers for suppressing the wages of labour and so procuring greater profits for themselves (cf. Über Wert, Kapital und Rente, pp. 104f).

Further details can hardly be elucidated in any other way than by means of a mathematical analysis. Before we undertake this, however, attention may still be drawn to a peculiarity of the tax on production. As a consequence of the longer investment time that in theory, as we have seen, must always occur with this tax, the productivity of labour and consequently also the volume of total production will have to increase by a greater or lesser amount. Consequently a part of the tax will be compensated by the expansion in production, and therefore evaded, in the sense proposed by Hock and Wagner. However, this will not happen in such a way as to diminish the sum flowing into the treasury, but rather so that this sum increases somewhat—a result of the theory that at the very least does not stand in contradiction to certain observed facts. It is tempting to perceive in this a gain for the national economy, which would give the (general) production tax a certain advantage over the direct income tax, which according to what has been said above would lead to no changes in terms of the national economy. But this conclusion would be somewhat premature. 84

ON THE THEORY OF TAX INCIDENCE

For our argument, if looked at more closely, rests on the assumption that the entrepreneur is able to adapt his production a good while in advance to the new conditions arising from an anticipated tax. Otherwise he himself would have to bear the entire amount of the tax without any reduction during the first years after the imposition of the tax. But if he makes the adaptation in advance, then in fact by their diminished wages the labourers pay the part of the tax that falls on them for several years (namely, for as many years as the average investment time of the capital) before the sum of taxes involved is collected by the treasury. If they had done this by paying direct taxes then obviously the state itself would have become stronger in capital terms by this amount, the capital of the country would have been expanded, and the raising of total production we have talked about would perhaps also have been able to occur under these circumstances. In the following section we shall proceed to incorporate in the scope of our reflections the third factor of production, the productive powers of the land (together with the other sources of rent), which have so far been left out of account. First, however, I shall attempt in the appendix to endow the theory set forth above with the exact mathematical form that is its due. APPENDIX In my work Über Wert, Kapital und Rente (pp. 95ff) I have endeavoured to give an account of Böhm-Bawerk’s theory of interest both in the form of algebraic formulae and by means of a simple geometrical interpretation. I shall repeat the latter interpretation briefly here, but with the emendation that in place of the length of the production period the average investment time of the capital is used. So let us regard the latter quantity, which we shall denote t, as the abscissa of a curve whose ordinate represents the annual production, p, of finished consumer goods corresponding to the investment time, calculated for one of the workers taking part in the entire production process involved. Then, obviously, the trajectory of this curve will always have to rise, but concavely with respect to the axis of the abscissae. Of course, in fact no practical limit can be stated beyond which the productivity of labour might not possibly rise, given enormous investment of capital. But it is equally impossible for the (annual) added returns to grow in proportion to the investment time; this would lead to absolutely absurd consequences. The curve will therefore necessarily become flatter and flatter as its distance from the axis of the abscissae increases. Since strictly speaking no production is possible without capital, we allow the curve to begin at the intersection of the co-ordinate axes, though this is not essential. Now if the level of wages, l, is regarded as given for the time being, then we arrive at the most profitable investment time by drawing a tangent to the curve from a point on the axis of ordinates at a distance of l from the origin and locating the abscissa of the point at which the tangent meets the curve (cf. Figure 5.2). For whatever the 85

INCOME, TAXES AND DUTIES

Figure 5.2

investment time is assumed to be, the straight line connecting the corresponding point on the curve with the point mentioned on the axis of ordinates will cut off a piece on the left-hand side of the axis of abscissae, and the size of this piece is expressed by

where z signifies the annual interest rate. The size obviously becomes a minimum, and consequently z a maximum, when the connecting line is a tangent to the curve. But if equilibrium is to be reached between the supply and demand of labour, then, as explained in the text, the equation K:t=A·l must be satisfied, or, which is the same thing, it must be true that

The rectangle bordered by the sides t and l must therefore have the constant area K:A. Now by these conditions both t and l and therefore also z are fixed unambiguously, as soon as K and A are known, together with the equation of the curve. Now the effect of a tax that takes away a specific portion, e.g. a tenth, of the gross returns will obviously be the same as if the productivity of labour 86

ON THE THEORY OF TAX INCIDENCE

had been reduced by a tenth, i.e. as if the curve mentioned above had been replaced by another whose ordinates had a value of only of the corresponding ordinates of the original curve.20 The same procedure therefore now has to be applied to this second curve, and it can easily be shown—e.g. by using the fact that the tangents of both curves, whose points of contact lie on the same ordinate, always intersect one another on the axis of abscissae—that as a consequence of the tax the length of the investment time must be increased, while the level of wages, in contrast, must be diminished, and finally the reciprocal value of the interest rate must be increased, and the interest rate itself must therefore be reduced. All further points now depend on the form of the original curve in the vicinity of the equilibrium point, in the following way. Let us denote by v and -w respectively the two differential expressions and, of which the first, evidently, must always be positive, though