Growth, Distribution and Effective Demand: Alternatives to Economic Orthodoxy 0765610094, 9780765610096

Citation preview

G rowth, D istribution, and Effective D emand

GROWTH, DISTRIBUTION, AND EFFECTIVE DEMAND Alternatives to Economic Orthodoxy

GEORGE ARGYROUS MATHEW FORSTATER GARY MONGIOVI, EDITORS

First published 2004 by M.E. Sharpe, Inc. Published 2016 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN 711 Third Avenue, New York, NY 10017, USA

Routledge is an imprint ofthe Taylor & Francis Group, an informa business Copyright© 2004 by Taylor & Francis. All rights reserved. No part of this book may be reprinted or reproduced or utilised 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. Notices No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use of operation of any methods, products, instructions or ideas contained in the material herein. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data Growth, distribution, and effective demand : alternatives to economic orthodoxy : essays in honor of Edward J. Nell/ edited by George Argyrous, Mathew Forstater, and Gary Mongiovi. p. em. Includes bibliographical references and index. ISBN 0-7656-1009-4 (alk. paper) I. Nell, Edward J. 2. Economics. 3. Economic development. 4. Income distribution. 5. Demand (Economic theory) 6. Capitalism. I. Nell, Edward J. II. Argyrous, George, 1963III. Forstater, Mathew, 1961- IV. Mongiovi, Gary. HB 119N455G76 2003 330-dc21

2003042446 ISBN 13:978-0-7656-1009-6 (hbk)

Contents

List of Tables and Figures Preface

ix xi

Part I. Growth, Distribution, and Technical Change 1.

2.

Transformational Growth, Interest Rates, and the Golden Rule Marc Lavoie, Gabriel Rodriguez, and Mario Seccareccia

3

Transformational Growth and the Changing Nature of the Business Cycle Korkut A. Ertiirk

23

3.

A Classical Alternative to the Neoclassical Growth Model Thomas R. Michl and Duncan K. Foley

4.

Wealth in the Post-Keynesian Theory of Growth and Distribution Neri Salvadori

5.

Transformational Growth and the Universalityof Technology Ross Thomson

35

61

81

vi

6.

7.

CONTENTS

Growth, Productivity, and Employment: Consequences of the New Information and Communication Technologies in Germany and the US Harald Hagemann and Stephan Setter Educational Insights from Edward Nell’s Theory of Transformational Growth Thomas E Phillips

98

115

Part II. Money, Employment, and Effective Demand 8.

9.

Labor Market Dynamics within Rival Macroeconomic Frameworks Anwar Shaikh

127

Has the Long-Run Phillips Curve Turned Horizontal? Craig Freedman, G.C. Harcourt, and Peter Kriesler

144

10.

Simulating an Employer of Last Resort Program Raymond Majewski

163

11.

Disabled Workers or Disabled Labor Markets? Causes and Consequences George Argyrous and Megan Neale

181

Interest Rates, Effective Demand, and Financial Fragility: Edward Nell and the Trieste Tradition Louis-Philippe Rochon and Matias Vernengo

203

12.

13.

Sraffa in the City: Exploring the Urban Multiplier Gary A. Dymski

220

14.

Elements of Historical Macroeconomics Thorsten Block

239

Part III. Theory, Method, and the History of Ideas 15.

Thorstein Veblen and the Machine Process Geoffrey M. Hodgson

261

16.

Is There a Classical Theory of Supply and Demand? David Laibman

279

CONTENTS

17.

The Capital Controversy, Stability, and the Income Effect Susan /. Pashkoff

18.

Cumulative Causation a la Lowe: Radical Endogeneity, Methodology, and Human Intervention Mathew Forstater

19.

The Epistemological Status of Economic Propositions Gary Mongiovi

20.

On Some Criticisms of Ricardo’s Discussion of Agricultural Improvements Christian Gehrke and Heinz D. Kurz

Contributors Index

vii

293

309

317

327

347 351

List of Tables and Figures

Tables 1.1 1.2 1.3 1.4 1.5 1.6 1.7 11.1 11.2 11.3 11.4 11.5 11.6 14.1 14.2

ADF Results for Complete Sample Period: United States ADF Test with a Broken Trend: United States ADF Test for Subsamples: United States ADF Results for Complete Sample Period: Canada ADF Test with a Broken Trend: Canada ADF Test for Subsamples: Canada Granger Causality Changes in Employment Levels, 1981/82-1998/99 Labor Force Status of Males and Females, 1971and 2001 Male Labor Force Participation Rates by Age Group, 1971 and 2000 Regression Results Disability Pension Recipients by Age Group Mature Age Allowance Recipients The Changing Nature of the Business Cycle Historical Conditions on Parameters: NBC: 0 < u < l, OBC: N Is “Sticky”

16 16 17 17 18 18 19 184 185 185 194 197 199 248 250

Figures 1.1 1.2

Nell’s Stylized Facts on the Secular Rate of Growth and the Real Rate of Interest Spread (g-i) for US 1947:1-1999:1

13 14

x

LIST OF TABLES AND FIGURES

1.3 2.1 2.2 2.3 3.1 3.2 3.3 3.4 4.1 4.2 4.3 6.1 6.2 6.3 6.4 8.1 8.2 9.1 9.2 10.1 10.2 10.3 10.4 10.5 10.6 10.7 11.1 11.2 11.3 11.4 11.5 16.1 16.2 17.1 17.2 17.3

Spread (g-i) for Canada 1947:4-1996:4 Stable Focus Unstable Focus Bounded Instability The Fossil Production Function The Trajectory of the Distribution Point The Trajectory of the Wage Trajectory of the Growth-Distribution Point with Technical Change Conditions for Steady-State Growth when the Interest Rate Equals the Capitalists’ Rate of Profit Conditions for Steady-State Growth when the Interest Rate is a Function of the Capitalists’ Rate of Profit Conditions for Steady-State Growth when the Interest Rate is a Function of the Overall Rate of Profit Growth Rates in Real GDP, 1992-2000 Capacity Utilization, 1991-2001 Share of ICTs in GDP and Its Growth, 1995-99 GDP and Labor Market in the 1990s: US and Germany Monotonic Convergence in the Neoclassical Labor Market Harmonic Oscillation in the Neoclassical Labor Market Long-Term Unemployment in the US as a Percentage of Total Unemployment (1965-97) The Horizontal Long-Run Phillips Curve Unemployment Rate, 1989-2004 ELR’s Public Service Employment, 1989-2004 Real GDP, 1989-2004 Inflation, Annual Rate Domestic Sales Deflator, 1989-2004 The Net Cost of the ELR, 1989-2004 Federal Deficit, 1989-2004 Net Benefits of the ELR, 1989-2004 Full-time Employment Rates in Australia, 1966-2001 Disability Pension Recipients as Percent of Working-Age Population, 1910-99 Male Disability Support Pension Recipient Rates, 1969-99 The Relationship between Full-Time Employment and DSP Rate for Males, 1969-99 The Relationship between Changes in Employment and Changes in DSP Rate for Males by State, 1971-99 Anomalous Price Adjustment Market Price Adjustment to Discrepancies between a and v Supply, Demand, and Excess Demand Curves Single Technique Two Techniques

14 31 32 33 42 46 49 51 69 72 75 99 107 108 109 130 131 145 153 170 171 172 172 173 174 175 183 188 189 191 195 281 287 297 302 303

Preface

The essays collected here honor the work of Edward J. Nell, who has been pushing the envelope of economic theory for forty years. Nell has been a member of the Graduate Faculty of the New School for Social Research (now New School Uni­ versity) since 1969. His writings and teaching reflect the intellectual values cher­ ished by the German-speaking émigré economists who formed the backbone of the Graduate Faculty at its inauguration in the early 1930s— methodological eclecti­ cism, a critical outlook toward conventional economic thinking, and a powerful commitment to the idea that the ultimate purpose of social science is to assist the achievement of progressive outcomes. Nell’s intellectual breadth is indicated by the range of issues to which he has made important contributions: growth theory, the capital theory debates, monetary economics, macroeconomic theory and policy, drug policy, structural economic change, economic methodology, economic history, analytic philosophy, and the philosophy of knowledge. Edward Nell was bom in Chicago in 1935, graduated from Princeton University in 1957, and then went to Oxford University as a Rhodes Scholar. He taught at Wesleyan University from 1962 to 1967, and at the University of East Anglia until 1969, when he joined the Graduate Faculty; since 1993 he has been Malcolm B. Smith Professor of Economics. Nell’s interest in how economic forces evolve through history dates back to his days at Oxford, and has driven virtually all of his scientific work. He has been a perceptive critic of neoclassical economics, calling into question its methodological foundations (most notably in Rational Eco­ nomic Man, written with Martin Hollis, who passed away while the present col­ lection was in the planning stages) and its internal logic (particularly in his writings on the orthodox treatments of capital and growth). As a constructive critic of his

xii

PREFACE

fellow dissenters from mainstream orthodoxy, Nell has encouraged much useful engagement with difficult but crucial analytical issues. We suspect that all of the contributors to this volume have, on one occasion or another, had the energizing benefit of Edward’s comments on their work. “You’re on to something interest­ ing . . . , ” he often begins, before pointing out an overlooked consideration with important implications. Nell is of course not only a critic: over the years he has had a great deal to say about the structural dynamics of capitalism, and in developing his ideas on this topic he has drawn inspiration from across the spectrum of analytical traditions. Nell’s commitment to intellectual openness is well exemplified by his edited volume Growth, Profits and Property, which was published by Cambridge University Press in 1980. This book was the culmination of a project, initiated in the 1970s, to knit Keynesian, Kaleckian, Sraffian, and Marxian traditions into a unified PostKeynesian account of modem capitalism. A series of summer schools held in Tri­ este during the 1980s continued this project, and Nell was a central figure in fostering constructive dialogue across the various traditions, not least because of his rare ability to discern the complex and subtle ways these different approaches can reinforce one another. He continues this work of critique and synthesis in his ambitious transformational growth project, which has been under way since the late 1980s. The essays in this book develop three interconnected themes that run through modem heterodox economics, and through Nell’s work. The first, and perhaps the most fundamental, of these themes is the idea that economic growth is demanddriven. This premise, which of course derives from Keynes, stands in stark contrast to the conventional view that growth is mainly conditioned by the willingness of economic agents to defer current consumption. A second theme concerns the connection between economic growth and the structural characteristics of a market economy. An important branch of the nonneoclassical literature since the 1950s has sought to establish whether capitalism exhibits certain general “laws of motion” in the pattern of demand, the degree of industrial concentration, the size of the government sector relative to the private sector, and the ways in which financial arrangements evolve and influence produc­ tion, income, and wealth. Again, there is a large and vibrant body of nonmainstream work in this area, stemming from Keynes, Sraffa, Lowe, and Minsky. These positive themes are closely linked to a critical point of view, also asso­ ciated with Nell, that calls into question key elements of orthodox economics. The last set of essays in this collection aims at buttressing the various nonorthodox approaches to growth and distribution by critiquing particular aspects of the con­ ventional theory, by elaborating neglected themes in nonorthodox theory, or by exploring overlooked methodological ideas. The three of us had the privilege of studying with Ed at the New School during a particularly exciting period in the Graduate Faculty’s history. Among the many valuable things we learned from him, perhaps the most useful lesson was that

PREFACE

xiii

scholars ought not to be afraid to take calculated intellectual risks. By example he has shown that progress in economics can occur only if theorists have the courage to abandon the easy road when difficult paths may lead to more promising desti­ nations. Our contributors join with us in expressing our deep respect and warm regard for Edward Nell as teacher, friend, and colleague. With this collection we offer him a small portion of what he has given to our discipline. George Argyrous Mathew Forstater Gary Mongiovi

Part I Growth, Distribution, and Technical Change

1 Transformational Growth, Interest Rates, and the Golden Rule Marc Lavoie, Gabriel Rodríguez, and Mario Seccareccia

Introduction

We first met Edward Nell in the winter of 1980, during his visiting appointment at McGill University.1 Along with our colleague, the late Jacques Henry, we had invited Ed Nell to give a seminar at the University of Ottawa. For the occasion, he had chosen to lecture on “The Simple Theory of Effective Demand” (Nell 1978); the title of his presentation reflected the usefulness of the apparatus he had devel­ oped to illustrate the relevance of the principle of effective demand, as it applied to a modem industrial economy with constant variable costs. Because of its obvious pedagogical usefulness and positive heuristic merits, we have utilized Nell’s ap­ paratus ourselves, both in the classroom and in our published work. Nell made extensive use of it in his subsequent publications, for instance in his Prosperity and Public Spending (1988a) and in his magnum opus, The General Theory o f Transformational Growth (1998). Although of wide interest, Nell’s model of effective demand is not the subject we have chosen to address. When Nell visited Ottawa, we discovered that he was also highly interested in monetary and liquidity issues, and that he was ready to entertain these matters over a drink until the early morning hours! As early as 1967, Nell had published a paper on Wicksell’s monetary circuit in the Journal o f Political Economy. There, he explored several of the questions that were to become the prime focus of the French circuit theorists, and that were highlighted again, first in a little-known working paper (Nell 1986), and then in the highly successful book that he edited with Ghislain Deleplace in 1996, Money in Motion. In his 1967 paper on monetary circulation within a Wicksellian framework, Nell

4

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

took into account the impact of interest payments on aggregate demand and on the closure of the monetary circuit. Ever since 1967, we believe, Nell has been tan­ talized by the relationship between the real and the monetary aspects of the econ­ omy. In “The Simple Theory of Effective Demand” (1978), he dealt with the monetary and credit consequences of a fall in wage costs and prices, arguing along Kaleckian lines that lower prices would tend to reduce effective demand, rather than increase it, because the higher real debt would push some firms and households into bankruptcy. In that same paper, for simplification purposes, Nell assumed that the amount of profits would equal the value of investment. This assumption, which in dynamic terms implies that the rate of growth of output is equal to the rate of return on capital (the so-called Golden Rule), plays a key role in The General Theory o f Transformational Growth. With Nell’s added assumption of capital ar­ bitrage, the Golden Rule implies that there must be a tight relation between the rate of return on financial capital and the rate of growth of capital and output. It is this relation that is the subject of our chapter. First, we shall try to capture the main hypotheses and stylized facts proposed by Nell regarding the relations among the rate of profit, the rate of interest, and the rate of growth of output. In so doing, we shall show how his own views are distinct from those of most Sraffians who have tackled the same issue. In the second section, we shall see whether the stylized facts proposed by Nell can be verified by modem time-series techniques involving the use of American and Canadian observations pertaining to the last half-century. Theoretical Relations among Interest Rates, Profit Rates, and Growth Rates Rates o f Profit and Rates o f Growth In the classical tradition, the normal rate of profit was determined by the existing real wage for a given technology. The rate of interest was given some leeway, but it could be no higher than the normal rate of profit and, in general, it would be much lower, so as to compensate for the “toil and trouble” of engaging in entre­ preneurial activity and investing in real capital. In both classical and Marxist views, in the long period, the rate of accumulation, or the rate of growth of output, g, is determined essentially in accordance with the Cambridge equation, g = rnsc9 where rn is the normal rate of profit and sc is the propensity to save out of capitalist income. Hence, the higher the normal rate of profit or the degree of thriftiness, the higher the rate of accumulation. In the work of Post-Keynesian and Sraffian economists, the Cambridge equation must be read in reverse. For a given degree of thriftiness, a higher rate of accu­ mulation will induce a higher rate of profit. To which rate of profit such a relation applies, however, has been the subject of controversy. For first-generation Post-

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

5

Keynesians, such as Nicholas Kaldor and Joan Robinson, the rate of accumulation determined the normal rate of profit, rn. Nell agrees with this causal relation, which has also been elaborated by other more recent Post-Keynesian writers within an oligopolistic framework: namely Eichner (1976), Wood (1975), and Harcourt and Kenyon (1976). Indeed, this has become an important feature of the Post-Keynesian theory of the firm, whereby firms are assumed to fix their price markups as a function of their long-term investment plans. This implies, as Nell argues, that the secular (or normal) rate of accumulation determines the normal rate of profit. How­ ever, most modern-day Kaleckians and Sraffians interpret the Cambridge equation as saying that the actual rate of accumulation determines the actual rate of profit, r, and not necessarily the normal rate of profit. The cause of this divergence among Post-Keynesians is that Kaleckians and Sraffians consider that the rate of capacity utilization may deviate from its normal or standard value even in the long period. In other words, while entrepreneurs may attempt to achieve a normal rate of capacity utilization by modifying the rate of increase of capacity relative to the expected rate of growth of their sales, the dy­ namics of effective demand may be such that they may not succeed in achieving this goal. As a result, the actual rate of capacity utilization over several periods is not necessarily equal to its normal value, and hence the actual rate of profit may also be different from its normal value. By contrast, Joan Robinson (and perhaps also Nell) assumes that there are underlying forces propelling the actual rate of capacity utilization toward its normal value (and hence that the actual rate of ac­ cumulation is equal to the secular one, as estimated by entrepreneurs). If, however, such forces do not exist (i.e., if the long-period rate of utilization is not necessarily the normal rate of utilization), then what will determine the normal rate of profit? The usual Kaleckian answer would be that such a rate of profit is somehow related to the degree of monopoly of firms. Another possible answer could be that the normal rate of profit depends on the power struggle within the labor market, whereby, for given technological conditions, the ensuing real wage rate determines the normal profit rate. Finally, another view, put forth by Sraffians, postulates that the rate of interest, resulting from monetary conditions (and allowing for entrepreneurial risk), is the variable that regulates the level of the normal rate of profit. If one assumes that there is a single determinant of the normal rate of profit, then clearly only one of the following two views is possible: either the normal rate of profit is determined by the Cambridge equation, by the secular rate of accu­ mulation along Robinsonian lines, or the normal rate of profit is determined in the Sraffian sense by the normal rate of interest (i.e., the rate of interest of the present monetary regime). One would thus understand why authors such as Nell, who are supportive of the former theory, would quite adamantly reject the latter. We feel, however, that there is no need for such rejection. Suppose, along the lines of the Kaleckian theory of increasing risk, that over

6 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

the long run firms can borrow only a certain multiple p of their retained earnings. Assume also that firms must pay a rate of interest i on the funds borrowed, and that they face a dividend payout ratio which is roughly equivalent to this interest cost. Interest and dividend payments will thus be equivalent to iK, where K is the value of capital, with the final financing of new capital projects being done either by retained earnings or by borrowing in the financial markets. Under these circum­ stances, firms will need to obtain a certain level of profits to achieve the desired level of investment, given the financing constraints and the existing rate of interest. With n being the level of profits of the firm, investment / will be equal to / = (k — iK) + p(n ~ iK).

( 1)

Dividing by K to obtain growth rates, we have rn = i + g j( l + P),

(2)

where rn is the normal rate of profit and gs is the secular rate of accumulation (Lavoie 1992: 111). Hence, for a given leverage ratio p, the normal rate of profit depends both on the long-period rate of interest and on the secular rate of accumulation. The value of p, presumably, would depend on conventions ruling in the banking industry regarding creditworthiness. Within this representation, the older Post-Keynesian view (as endorsed by Nell) and the new Sraffian view (as put forth by Garegnani 1979; Pivetti 1985; and Panico 1985) are not incompatible. They are two aspects of the same problem, that of being able to engage in the long-term financing of investment while respecting existing financial conventions. The above shows that it is possible to argue that the long-period rate of interest is one of the determinants of the normal rate of profit, while still admitting that firms would take the secular rate of accumulation into account when setting their price markups. If Nell were merely to argue that there exists a positive relation between the rate of accumulation and the normal rate of profit, the compromise reached in the above equation would be sufficient to reconcile his viewpoint with that of other postclassical authors. Nell, however, goes somewhat farther. Nell maintains that the normal rate of profit determines the long-period rate of interest, and argues that there is a Golden Rule mechanism which concomitantly moves all three main rates: the rate of accumulation, the normal and the realized rates of profit, and the rate of interest. In fact, a large portion of his 1998 book is devoted to a demonstration of this rule, and to its consequences. Now, while at various places Nell recognizes that actual measures of the rate of profit show that it is twice the level of the rate of accumulation, most of the book is devoted to the proof that the rate of interest (as determined by the rate of profit) and the rate of accumulation must move in tandem over the long run.

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

7

Determination o f the Rate o f Interest As is well known, one can distinguish two Post-Keynesian views of the determi­ nation of the rate of interest. On the one hand, there is the horizontalist view, which is mainly associated with the work of Kaldor (1982) and Moore (1988), but also with the circulation approach and the work of several Sraffians. On the other hand, there is the so-called structuralist view, which has been largely associated with Post-Keynesians such as Pollin (1996) and, possibly, Nell. In the horizontalist view, the central bank sets the base rate of interest— the overnight rate— and this short-term rate becomes the standard for all other rates of interest in the monetary system. In the short run, discrepancies between the short­ term and long-term rates of interest can emerge, as agents’ expectations of future short-term rates vary. However, through arbitrage, long-term rates will eventually adapt to the short-term rates set by the central bank. For instance, as borrowers recognize that short-term rates are persistently below long-term rates, they will stop issuing long-term bonds, thus driving up their price and pushing down the rates of return on long-term bonds until they adapt to the short-term central bankadministered rates. In the horizontalist view, as Deleplace and Nell (1996: 36) put it, “the central bank plays a crucial role, because it may fix the short-term rate of interest.. . . Then, the long-term interest rate adjusts to the short-term one.” By contrast, within the structuralist approach, while the central bank may be able to set interest rates exogenously in the very short run, eventually the central bank must give in to market forces and adjust short-term rates under its control to long-term rates that are essentially determined by market factors. This is quite clear in the writings of Pollin (1996), who tries to demonstrate the validity of this view by relying on Granger causality tests which indicate that long-term interest rates determine short-term ones. Nell himself is somewhat more ambiguous, claiming that “the long-term interest rate is relatively autonomous in relation to the short­ term rate, which is fixed by the central bank.” On the other hand, “the long-term rate is determined by arbitrage between equity and bonds, where the prices of shares are understood to be influenced by the expected rate of growth in the real sector. Hence the long rate reflects the rate of growth” (Deleplace and Nell 1996: 36). In a straightforward manner, causality runs from the first to the last of the following variables: rate of accumulation, realized rate of profit, normal rate of profit, normal long-term rate of interest, and, eventually, the short-term rate of interest. The arbitrage between equity and bonds, and between growth rates and interest rates, is outlined in more detail in Nell (1998: 269), where he explains that “in­ vestors can choose between growth through investment in real assets, on the one hand, and growth through compound interest on bonds, on the other.” Under these circumstances, he argues, “arbitrage should tend to bring the real rate of interest into line with the rate of growth.” This view is very similar to that presented by Randall Wray. According to Wray (1991: 18), “in the long run, the interest rate is

8 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

determined by the state of long-term expectations— that is, by expectations of profit on real assets.” Much like Nell, Wray (1991: 16) also links the rate of growth of real assets to the rate of interest: “In the absence of uncertainty, the interest rate would be pushed up to the point where it just equalled the rate of profit if markets were perfectly competitive. Each borrower would issue bonds to purchase assets up to the point where the expected rate of growth of the value of the assets just covered the rate of growth of the liabilities entailed in the bonds issued (which is the interest rate).” From the pure history-of-economic-thought perspective, it seems that the horizontalist approach is consistent with the causality outlined in chapter 17 of The General Theory (1936), where Keynes discards the doctrine of the Wicksellian natural rate of interest. As John Smithin (1994: 156-57) points out, “in the absence of a natural rate of interest, it can also be argued that central bank control over short real rates will ultimately influence the entire structure of interest rates in the economy, including long rates.” Eventually, Smithin adds, “the real economy must adjust to the policy-determined interest rate, rather than vice-versa. This is therefore the precise opposite of the natural rate doctrine.” In contrast, while Nell and ad­ vocates of the structuralist view recognize that high rates of interest will adversely affect investment projects and profitability in the short run, they claim that this causality is reversed over the longer run, and that it is the rate of interest which will have to adjust to the prevailing rates of return on real assets. For this reason, it may be argued that the structuralist view of how rates of interest are determined over the long period is not very different from the ideas defended by Wicksell in his theory of the natural rate of interest. The horizontalist view of the determination of the rate of interest appears to be an intrinsic outgrowth of the Cambridge critique of the neoclassical theory of capital. As explained by Rogers (1989) and Garegnani (1990), the Cambridge cap­ ital controversies extend to monetary theories and to theories of unemployment. The Cambridge critique shows that there can be a multiplicity of possible equilib­ rium rates of interest and rates of unemployment. Central banks, through their ability to set real rates of interest, have the ability to change what entrepreneurs reckon to be the normal rate of profit, and thereby the ability to set sustainable rates of growth and rates of unemployment. Monetary policy is not neutral: it does influence the real economy. This, it seems to us, is the main message of a monetary theory based on a monetary analysis, in contrast to a monetary theory based on real analysis, to adopt a terminology employed by Schumpeter (1954) and Rogers (1989). Sraffa’s (1960: 33) famous sentence, endorsed by Garegnani (1979), to the effect that the money rate of interest is susceptible of determining the normal rate of profit, would reflect this belief that monetary policy is not neutral. Nell (1999: 278) cites Joan Robinson, who believed that it was “excessively fanciful to suppose that the Central Bank could set a long-period interest rate (or that there is such a thing!).” Robinson (1979: 180) had indeed written that Garegnani’s suggestion that the normal rate of profit “could be determined by monetary

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

9

policy seems to be excessively fanciful.” However, it should also be pointed out that in 1952 Robinson put forth a theory of distribution that was not very different from the one more recently proposed by the Sraffians. Robinson (1952: 96) wrote at the time that “the distribution of income between labour and capital depends . . . on the rate of profit, that is, the rate of interest plus the rate of net profit. If the rate of interest were (and always had been) lower or the entrepreneurs were habit­ uated to a lower rate of net profit, the share of labour and the level of real wages would be so much the higher.” Here, no doubt, Robinson means that the normal rate of profit is lower when the perceived rate of interest is lower. This is exactly what is claimed by Pivetti (1985: 87) when he says that “lasting changes in the rate of interest will cause corresponding changes in profit rates, and inverse changes in the real wage.” Surely, Robinson had forgotten what she had written thirty years before in claiming that such a proposition was “excessively fanciful.” Reasons Leading to the Rejection o f the Horizontalist Causality Why does Nell reject so strongly the causality claimed by horizontalists and the Sraffians? There appear to be several reasons. First, Nell believes that a normal rate of profit ought to be determined by some “normal” variable. The normal rate of profit, or the target rate of return, is an element of “benchmark prices” set by industry leaders, prices that must depend essentially on the long-term rate of ac­ cumulation. Like the Joan Robinson of 1979, Nell does not believe that there is such a thing as a normal rate of interest or a “lasting change in the rate of interest” (Nell 1988b: 264). By contrast, he argues that there is a normal rate of growth of demand, and that firms will set benchmark prices to ensure that their retained earnings are sufficient to finance growth (Nell 1999: 270, 284). This argument is rather difficult to swallow. From about 1980 to 2000, it could easily be ascertained that OECD economies lived under a regime of high real rates of interest, with real rates of government (no-risk) bonds averaging 5% to 6%, whereas during the pre1980 era real rates were much lower, around 2%. Similarly, in less developed countries, real rates easily exceeded 10% or 15% in the 1990s, whereas they were much lower before. These monetary regimes can no doubt be identified by firms and their entrepreneurs. Second, even if it can set the nominal rate, Nell claims that the central bank is unable to set the real rate of interest (Nell 1999: 269). However, it would be difficult to support such an assertion. Monetary authorities in Canada, for instance, control the overnight rate of interest literally on a daily basis with perhaps only a tiny margin of error. In the case of short-term interest rates, the relevant inflation rates with which to compute ex ante values are usually monthly or quarterly inflation rates that can be taken as essentially predetermined. As has been argued elsewhere (Seccareccia 1998), given their commitments in favor of price stability, central banks very much have the power to target short-term real rates of interest and set

10 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

them at levels that they believe, for instance, would be consistent with a presumed NAIRU or noninflationary output gap. A third possible objection is raised by Nell when he argues that one way “the monetary authorities set interest rates is by manipulating the money supply; to raise interest rates the money supply will be constricted, and vice versa to lower them. But if prices move directly with interest rates, where these latter are governed by policy, then prices must move inversely to the supply of money! This flatly con­ tradicts virtually all thinking on the role of the Quantity of Money” (1988b: 264; see also Nell 1999: 290n.3). Despite having accepted the view that the money supply is endogenous, the above statement is somewhat problematic. On the one hand, Nell sees that if interest rates are higher, prices relative to unit labor costs ought to be higher, and hence the money supply required to absorb the increased monetary transactions will be rising endogenously. On the other hand, Nell believes that for interest rates to rise, the money supply must diminish. This latter statement is rather perplexing. In the circulation approach, an approach that is now unwit­ tingly endorsed by a good number of central bankers (Henckel et al. 1999), interest rates may be modified by central bank authorities without any change in the supply of money or of high-powered money. In the past, because of their need to borrow from the discount window, fluctuations in the discount rate would be sufficient to bring about a change in the whole array of interest rates. Within present-day central bank arrangements in such countries as Canada and the UK, it is enough for the monetary authorities to announce a change in its base interest rate for the overnight rate on settlement balances to adjust immediately. There is no need for the central bank actively to pursue open market operations to achieve such a result. The over­ night rate will change, and with it changes in the whole vector of interest rates will ensue, following the announcement from the monetary authorities, without any accompanying variation in the money supply. Indeed, this would be entirely con­ sistent with the horizontalist view, as espoused by Moore or Kaldor, whereby the (overnight) rate of interest can be considered a fully exogenous variable. Since the money supply is endogenous, higher interest rates may initially induce higher prices or higher rates of inflation, and hence higher rates of growth of the money supply. At the same time, higher interest rates may also induce a slowdown in inflation and money growth as the nominal variables grow more slowly because of the higher interest rates impacting negatively on aggregate demand and employment. However, not only does Nell reject the causal role that interest rates could play as a determinant of the normal rate of profit; he also claims that the so-called Golden Rule must hold, that is, that an economy’s rate of growth should be equal to the long-term rate of interest (and to the rate of profit). Why is Nell so convinced of the normative necessity of this rule, even though he recognizes that in practice the two rates will move together rather than be exactly equal? There are no doubt several reasons justifying the normative necessity of the Golden Rule, but one stands out in particular. Nell (1998: 261) claims that to preserve a stationary rate of inflation, the rate of increase of the money supply ought to equal the (nominal)

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

11

rate of growth of capital. This, he says, will be achieved when the (nominal) rate of interest set by the central bank is equal to the (nominal) rate of capital accu­ mulation (see also Nell 1998: 282).2 Even if we were to accept such a hybrid Wicksellian monetary growth rule, why does Nell believe the rate of growth of the money supply ought to equal the rate of interest? Here, apparently, he falls prey to an obvious oversimplification. It is as if the profits of banks were exactly equal to the interest payments made by borrowers, while banks were assumed to pay no interest on their liabilities: When the rate of interest equals the rate of growth, the earnings of the banking system enable it to expand its monetary issue— or its credit lines— at the same rate as the economy is growing, without increasing the banking system’s degree of risk. That is the banking system’s earnings will be proportional to the interest rate; hence bank capital will accumulate at the rate of interest, while deposits and loans will grow as the rate output is growing. When the two are equal the expansion of the money supply, to keep pace with the needs of trade, will take place at a constant ratio of bank capital to liabilities. (Nell 1998: 269)

Such a statement is highly equivocal, and hardly provides a justification for the normative Golden Rule. Bank profits are not a function of the level of interest rates. Rather, bank profits depend on the spread between the loan and deposit rates, and on the amount of bank service charges relative to operating costs. In reality, there need be no relationship whatsoever between bank profits and interest rate levels.3 These various theoretical conundrums notwithstanding, Nell’s key hypothesis arising from his espousal of the Golden Rule is that the rate of interest (namely, the long-term rate) must necessarily gravitate around the rate of growth (rate of profit), with the interest rate tracking the latter via capital arbitrage. This hypothesis is sufficiently precise to be subjected to econometric evaluation; the following sec­ tion explores the empirical validity of Nell’s recent restatement of the Golden Rule. The Empirical Relation between Rates of Interest and Rates of Growth A Stable Spread until a Break Occurs An examination of the determinants of the normal rate of profit would be full of embûches, especially owing to the difficulty of empirically defining an economy­ wide rate of profit. We shall therefore confine ourselves to a much simpler empirical exercise, mainly inspired by Nell (1998; 1999). Since he defends a particular in­ terpretation of the Golden Rule that establishes a formal link between the rate of interest and the rate of growth, we shall focus on the empirical relation between the long-term rate of interest and the rate of growth of output in Canada and the United States.

12 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Nell points out that the rate of interest and the rate of growth must closely track one another in the long run. This tendency has to do essentially with the fact that portfolio holders continually engage in capital arbitrage. Once one allows for the appropriate adjustment for risk/liquidity factors, wealth holders ought to be indif­ ferent in the long run between holding “real” or “financial” assets (Nell 1998: 24346; 1999: 285-87). If any deviations exist between the real rate of interest (i) and the rate of growth (g), there would be a tendency for these two rates to fall back into line because of the market incentives that these deviations generate. For in­ stance, Nell argues that in a pro-growth regime in which the gap g -i becomes inordinately positive, asset holders would be shifting away from financial assets in favor of “real” ones, thereby simultaneously pushing up g, pulling bond prices downward, and raising interest rates. With the tendency of both these rates to move upward, and because of a capacity constraint on the increase of g, an appropriate difference, g-i, will ultimately ensue that would be compatible with zero capital arbitrage. Conversely, in a pro-rentier regime in which the gap between g and i becomes excessively negative, financial holders would be pushing up the price of bonds as they shift away from the real sector, thereby lowering i and pulling down g until a suitable difference between g and i is reestablished. As long as the basic underlying parameters of the regime remain unchanged, a gap between g and i will persist. However, as soon as the regime breaks down due to some change in key parameters, a period of structural turbulence will follow until a new regime is in place and another long swing in economic growth begins: The claim is not that a persistent gap between i and g will last forever; the point is rather that it will not close until there is a change in basic parameters. Hence the argument here is relevant to the analysis of “long swings” in economic behaviour. Thus i < g is consistent with the long postwar boom, from the end of the war to the first oil shock, and i > g with the growing stagnation since then. The gap in each case will tend to close with changes that bring about the turning points—but rather than equilibrium, a new regime of persistent divergence is likely to be established. (Nell 1999: 287)

The key point is that once a specific regime emerges, the difference between i and g will not disappear. Instead, the economy will stabilize at a new Robinsonian state of tranquillity with the two rates moving pari passu with one another until a further shock perturbs this state. The economic history of Western mass production econ­ omies can thus be characterized by long periods of relatively tranquil states dis­ turbed intermittently by shocks which modify the underlying parameters that determine the relative position of the financial vis-à-vis the real sector, and con­ sequently the relation between i and g. Because of their empirical implications, Nell’s capital arbitrage relation and his interpretation of the Golden Rule form a testable hypothesis that can be subjected to econometric evaluation. As shown in figure 1.1, Nell suggests that during the

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

13

Figure 1.1 Nell’s Stylized Facts on the Secular Rate of Growth and the Real Rate of Interest

early postwar period, the gap between i and g followed a stationary path until the latter half of the 1970s, when the Western economies reacted to a series of oil price hikes. The structural break that resulted from the oil price shock pro­ voked a regime shift that during the post-1980 period has been associated with stagnant growth and the reversal of the previous long-term relation between i and g. In addition to studying this gap between i and g, we can also evaluate whether the long-run causal link between i and g postulated by Nell (1998: 279) was in­ deed a “temporally causal” one in the sense that it can actually be supported em­ pirically by means of standard Granger tests, using both US and Canadian observations. We have used quarterly data for real output growth (measured as the firstdifference in the logarithm of constant-dollar GDP) and real interest rates (mea­ sured as the yield on long-term corporate bonds less the growth rate of the Consumer Price Index).4 In the US case, the data used span the period 1947:1 to 1999:1; for Canada, they cover the period from 1947:4 to 1996:4. Moreover, since Nell’s hypothesis is concerned with the gap between i and g, the principal variable used in our empirical analysis is the spread between output growth and the real interest rate series, which will be denoted herewith as the “spread” or “g-i.” The actual evolution of the spread can be observed in figures 1.2 and 1.3 for

14 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Figure 1.2 Spread (g -i) for US 1947:1-1999:1

Figure 1.3 Spread (g -i) for Canada 1947:4-1996:4

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

15

the US and Canada, respectively. Simple graphical inspection would suggest that around 1980-82 a break point in the series occurred in both countries. While recognizing the possible existence of such a structural break that would undoubt­ edly support Nell’s claim of a structural shift in its pattern during that period, we shall first analyze the stationarity of the spread for the complete sample period without considering the possibility of a break point in the trend function. This will then be followed by the application of unit root tests that could account for a single possible break point. Following Perron (1989), the nonrejection of a unit root in a time series can be a consequence of ignoring a potential break point in the trend function. To evaluate this possibility, we have applied unit root tests with a broken trend that have been proposed by Perron and Rodriguez (2003). Though Perron and Rodriguez analyze two models to evaluate structural changes, we have chosen to apply the more general specification that allows a potential break in both the intercept and the slope of the trend relation. Moreover, we have adopted a variety of rules to select the appropriate lag length for the ADF test5 and two separate methods to choose the break point.6 Finally, once the break point has been selected for the respective country time series, we analyze the sta­ tionarity of the spread in each relevant sub-sample period obtained from the pre­ vious step. US Evidence Table 1.1 displays the results of the application of the standard Augmented DickeyFuller (ADF) test to the spread for the complete sample period (an intercept and a time trend were included in the ADF test). Although four lags seem more rea­ sonable, given the frequency of the data, we present evidence for lags of one to four quarters. The numbers in parentheses indicate the level of significance for the rejection of the null hypothesis of a unit root in the time series. Our findings indicate that the spread for the complete period can be considered as a nonstationary process. Results from the application of unit root tests for Nell’s hypothesis of a broken trend in the time series are presented in table 1.2.7 The evidence is overwhelmingly in the direction of a strong rejection of the unit root hypothesis in favor of sta­ tionarity with a broken trend. It is also noteworthy that these results were so re­ gardless of the particular rule used to select the lag length or the break point. Indeed, the break point was calculated to be between 1978:2 and 1981:1, and for this reason, two subsample periods could be identified: one from 1947:2 to 1978:1 and the other from 1981:2 to 1999:2.8 Table 1.3 presents results of the application of the standard ADF test for each subsample with varying lags from one to four quarters. Unlike the complete sample period, evidence from the separate sample periods is highly supportive of Nell’s hypothesis of stationarity for the pre- and postbreak subperiods. In fact, we find a strong rejection of the unit root hypothesis for all subsamples. In other words, the nonstationarity of the spread found initially in the series for the complete sample

16 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Table 1.1 ADF Results for Complete Sample Period: United States

ADF

Lag 1 2 3 4

-3 .2 7 (10%) -3 .2 6 (10%) -3 .1 8 -3 .1 2

Table 1.2 ADF Test with a Broken Trend: United States

M ethod

ADF

Lag

Break point

Infimum t-sig AIC BIC MAIC MBIC

-6 .2 2 -7 .1 6 -7 .1 6 -6 .1 2 -6 .1 2

(1%) (1%) (1%) (1%) (1%)

4 0 0 0 0

1981:1 1981:1 1981:1 1978:2 1978:2

-6 .2 2 -7 .1 4 -7 .1 4 -5 .8 8 -5 .8 8

(1%) (1%) (1%) (1%) (1%)

4 0 0 1 1

1981:1 1981:1 1981:1 1981:1 1981:1

Supremum t-sig AIC BIC MAIC MBIC

period was induced by the presence of a break point. Without the structural break, the spread variable would be trend-reverting. Canadian Findings Our results from the application of the standard ADF test for the complete Canadian sample period are shown in table 1.4. There is some evidence from these obser­ vations in support of stationarity using the ADF test procedure with only one lag. However, this lag length was not sufficient to generate white noise residuals. In­ deed, when we use higher lags, the evidence against a unit root is weak. Conse­ quently, as in the US case above, we can presume that the Canadian spread is an integrated process when using the complete sample period. In much the same way, results from the tests for a unit root with a broken trend

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

17

Table 1.3 ADF Test for Subsamples: United States

Sample period: 1947:2-1978:1 ADF Lag

Sample period: 1981:2-1999:2 Lag ADF

1 2 3 4

1 2 3 4

-4.37 -5.41 -5.50 -5.90

(1%) (1%) (1%) (1%)

-4.00 -4.03 -5.32 -6.00

(5%) (5%) (1%) (1%)

Table 1.4 ADF Results for Complete Sample Period: Canada

Lag 1 2 3 4

ADF -4.38 (1% -3.15 (10%) -3.08 -3.07

are shown in table 1.5. As with the US data, the evidence lends some support to Nell’s contention in favor of the hypothesis of stationarity with a broken trend. In the Canadian case, the break point was detected to be 1980:2 or 1981:2, depending on the method used to estimate the break point. From the above evidence on the break points, we have chosen to analyze two subperiods, from 1947:4 to 1980:1 and from 1981:3 to 1996:4. Table 1.6 exhibits the results of the application of simple ADF tests for each subsample. Our findings for the first subperiod are obvious: the spread is indeed stationary, as was the case for the US. However, for the second subsample, the evidence was not so clear. We reject the unit root hypothesis with four lags (at 5%), but this was not possible for other lags. Since evidence from Canada for this subsample does not support the stationarity of the spread variable, it may be argued that Nell’s hypothesis finds only partial confirmation from Canadian time series. Causality One final issue we wish to explore is Nell’s postulated causality going from the rate of growth (g) to the long-term rate of interest (/). Needless to say, testing “causality” in the Granger sense indicates only the ability of one variable to predict

18 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Table 1.5

ADF Test with a Broken Trend: Canada M ethod

ADF

Lag

Break point

Infimum t-sig AIC BIC MAIC MBIC

-4 .5 7 -4 .5 7 -4 .5 7 -4 .5 7 -4 .5 7

(2.5%) (5.0%) (2.5%) (1.0%) (1.0%)

1 1 1 1 1

1980:2 1980:2 1980:2 1980:2 1980:2

-4 .3 2 -4 .3 2 -4 .3 2 -4 .3 2 -4 .3 2

(2.5%) (5.0%) (2.5%) (1.0%) (2.5%)

1 1 1 1 1

1981:2 1981:2 1981:2 1981:2 1981:2

Supremum t-sig AIC BIC MAIC MBIC

Table 1.6 ADF Test for Subsamples: Canada

Sample period: 1947:4-1980:1 Lag ADF

Sample period: 1981:3-1996:4 Lag ADF

1 2 3 4

1 2 3 4

-6 .2 7 -4 .4 3 -4 .1 7 -4 .1 8

(1%) (1%) (1%) (1%)

-3 .1 2 -3 .3 4 (10%) -3 .2 4 (10%) -3 .5 4 (5%)

the evolution of another variable. From this point of view, causality does not nec­ essarily imply a cause-effect relation. It could, however, offer useful insights on the temporal “causal” link between two variables. Similar to the previous analysis of unit roots, tests of Granger causality require the selection of a lag length. In this case, we used the Akaike (AIC) criterion. Table 1.7 presents the results of the Granger causality test. For the US spread, the AIC rule selected four, eight, and four lags for the first subsample, the second subsample, and the complete sample, respectively. For the complete sample period, the causality runs primarily in one direction, from the real interest rate variable to output growth. However, for the two subsample periods a two-way causality was found. This bidirectional causality would seem to give some support to Nell’s arbitrage mechanism, but of course the causal impact of growth rates on long-term

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

19

Table 1.7 G ranger Causality Samples

Causality

p-value

k

I to g i to g g to i i to g g to I

1.5% 2.5% 6.7% 0.0% 0.0%

4 4 4 8 8

i to g i to g

0.0% 3.1% 8.3%

2 8 1

United States Complete period First period Second period

Canada Complete period First period Second period

____ LJiLS______

interest rates could also arise as a consequence of the reaction function of the Federal Reserve, via the short-term rates set by the latter. For Canada, the AIC rule selected eight, one, and two lags for each sample period, respectively. When all the samples are considered, the overwhelming evi­ dence for Canada suggests that the causality runs primarily in one direction, from the real interest rate variable to output growth. Concluding Remarks The Golden Rule implying a long-term relation between the rate of interest and rate of growth (via the rate of profit through capital arbitrage) has been the subject of significant controversy. While most Post-Keynesians would support the existence of such a long-period relation on theoretical grounds, opinions differ radically re­ garding the postulated causal link among the three key variables: rate of interest, rate of profit, and rate of growth. Nell’s writings have played a pivotal role in this debate. From being primarily a theoretical concern among earlier Post-Keynesian economists, his most recent contributions have shifted the debate to the empirical level. In fact, in his recent work, Nell makes a series of claims, in the form of stylized facts, that we have tried to evaluate empirically with the help of modem techniques of time series analysis. The evidence from about 1950 both for the US and Canada generally supports Nell’s claim that except for a structural break in the behavior of real long-term interest rates and output growth rates centered around 1980, the spread between i and g has been stable during the pre- and post-1980 subperiods. The only possible exception is Canada for the latter subperiod. While the long-term relation hypoth­ esized by Nell and a good number of other post-Keynesians is partially substanti­ ated by the evidence presented, the long-run causal link that he postulated going

20 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

from output growth to real long-term interest rates has found only moderate sup­ port, at least by means of traditional Granger tests. Instead, our findings appear to lend more support to those who would argue that the short-run postulated causal link going from i to g should also be extended to the long-run behavior of such variables. The empirical evidence seems, therefore, to be more consistent with the horizontalist and Sraffian views, which claim that the real interest rate is the ex­ ogenous variable, determined by monetary policy and not by real forces such as the rate of growth of the economy. Notes 1. This anecdote involves only Marc Lavoie and Mario Seccareccia. Gabriel Rodríguez has an indirect link with Edward Nell since one of his teachers in Peru, Javier Iguiñiz, had been a Ph.D. student at the New School for Social Research during the late 1970s. 2. The reader can contrast this rule with a normative rule we prefer, according to which the real rate of interest ought to equal the rate of productivity growth (see Pasinetti 1981). 3. Indeed, an increase in interest rates may have detrimental consequences on the profits of banks, since a large portion of their assets may be less liquid and have a longer-term horizon than their liabilities. 4. The use of yields on long-term corporate bonds for the test is based on the widely held structuralist belief that in the long run, capital arbitrage between the “real” and “finan­ cial” sectors applies especially to long-term corporate securities that can serve as appropriate substitutes for the holding of “real” assets. For the US, we have used Moody’s long-term corporate bond yields averages (twenty years and above); constant-dollar GDP was obtained from the Bureau of Economic Analysis (NIPA) of the US Department of Commerce. For Canada comparable series on corporate bond yields were obtained from the CANSIM da­ tabase (label no. D89860), and the constant-dollar GDP series was obtained from Statistics Canada, National Income and Expenditure Accounts (13-001). Because of a methodological change in the Canadian estimate of GDP after 1996, post-1996 data were excluded from analysis for Canada. The respective CPI inflation rates were calculated from time series obtained from the CANSIM database: label nos. P100000 (Canada) and D 139136 (US). 5. We use the Akaike (AIC), Schwarz (BIC), Modified Akaike (MAIC), Modified Schwarz (MBIC) and sequential t-sig methods to select the lag length (see Ng and Perron 2001; Perron and Rodríguez 2003). For the AIC, BIC, MAIC, and MBIC rules, we use Áanax=int[10*(T/100),/d, and for the sequential t-sig method, /cmax= int[4*(77100), where T is the sample size. The lower bound is always £=0. 6. The first method selects the structural break as the point that yields the minimal value of the ADF statistics. The second method selects the break point such that the absolute value of the t-statistic on the coefficient of the change in slope is maximized. The two methods are labeled infimum and supremum, respectively. Perron and Rodriguez (2003) show that the first method performs better than the latter method. 7. Following the recommendation of Perron and Rodriguez (2003) and because of the sample sizes, finite critical values for 7=200 and for different selection lag rules were used. 8. One caution is needed here. Notice that a break point estimated using infimum and supremum procedures is not consistent for the true value of the break point when the data-

TRANSFORMATIONAL GROWTH AND THE GOLDEN RULE

21

generating process contains a break. In other words, the break data point may be around the estimated break point. We divide the sample using estimated break points, and no obser­ vations are allowed between the estimated break point at the end or beginning of the sub­ samples. Regardless of the procedure, however, our results remain unchanged.

References Deleplace, G., and Nell, E.J., eds. 1996. Money in Motion: The Post Keynesian and the Circulation Approaches. London: Macmillan. Eichner, A.S. 1976. The Megacorp & Oligopoly: Micro Foundations of Macro Dynamics. Cambridge: Cambridge University Press. Garegnani, P. 1979. “Notes on Consumption, Investment and Effective Demand: II.” Cam­ bridge Journal of Economics 3: 69-82. --------- . 1990. “Quantity of Capital.” In The New Palgrave: Capital Theory, ed. J. Eatwell, M. Milgate, and P. Newman. London: Macmillan. Harcourt, G.C., and Kenyon, P. 1976. “Pricing and the Investment Decision.” Kyklos 29: 449-77. Henckel, T., Ize, A., and Konaven, A. 1999. “Central Banking without Central Bank Money.” International Monetary Fund Working Paper WP/99/92. Kaldor, N. 1982. The Scourge of Monetarism. Oxford: Oxford University Press. Keynes, J.M. 1936. The General Theory of Employment, Interest and Money. London: Macmillan. Lavoie, M. 1992. Foundations of Post-Keynesian Economic Analysis. Aldershot, UK: Edward Elgar. Moore, B.J. 1988. Horizontalists and Verticalists: The Macroeconomics of Credit Money. Cambridge: Cambridge University Press. Nell, E.J. 1967. “Wicksell’s Theory of Circulation.” Journal of Political Economy 75: 38694. --------- . 1978. ‘T he Simple Theory of Effective Demand.” Intermountain Economic Review (Fall): 1-33. --------- . 1986. “On Monetary Circulation and the Rate of Exploitation.” Thames Papers in

Political Economy. --------- . 1988a. Prosperity and Public Spending: Transformational Growth and the Role of Government. Boston: Unwin-Hyman. --------- . 1988b. “Does the Rate of Interest Determine the Rate of Profit?” Political Economy, Studies in the Surplus Approach 4: 263-67. --------- . 1992. “Demand, Pricing and Investment.” In Transformational Growth and Effective Demand: Economics after the Capital Critique. London: Macmillan. --------- . 1998. The General Theory of Transformational Growth: Keynes after Sraffa. Cam­ bridge: Cambridge University Press. --------- . 1999. “Wicksell after Sraffa: Capital Arbitrage and Normal Rates of Growth, In­ terest and Profits.” In Value, Distribution and Capital: Essays in Honour of Pierangelo Garegnani, ed. G. Mongiovi and F. Petri. London: Routledge. Ng, S., and Perron, P. 1999. “Lag Length Selection and the Construction of Unit Root Tests with Good Size and Power.” Econometrica 69: 1519-54.

22

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Panico, C. 1985. “Market Forces and the Relation between the Rates of Interest and Profit.” Contributions to Political Economy 4: 37-60. Pasinetti, L.L. 1981. Structural Change and Economic Growth. Cambridge: Cambridge Uni­ versity Press. Perron, P. 1989. “The Great Crash, the Oil Price Shock and the Unit Root Hypothesis.” Econometrica 57: 1361-1401. Perron, P, and Rodriguez, G. 2003. “GLS Detrending, Efficient Unit Root Tests and Struc­ tural Change.” Journal o f Econometrics 115: 1-27. Pivetti, M. 1985. “On the Monetary Explanation of Distribution.” Political Economy, Studies in the Surplus Approach 1 (2): 73-103. Pollin, R. 1996. “Money Supply Endogeneity: What Are the Questions and Why Do They Matter?” In Money in Motion: The Post Keynesian and the Circulation Approaches, ed. G. Deleplace and E.J. Nell. London: Macmillan. Robinson, J. 1952. The Rate o f Interest and Other Essays. London: Macmillan. --------- . 1979. “Garegnani on Effective Demand.” Cambridge Journal of Economics 3: 179— 80. Rogers, C. 1989. Money, Interest and Capital: A Study in the Foundations of Monetary Theory. Cambridge: Cambridge University Press. Schumpeter, J.A. 1954. History o f Economic Analysis. Oxford: Oxford University Press. Seccareccia, M. 1998. “Wicksellian Norm, Central Bank Real Interest Rate Targeting and Macroeconomic Performance.” In The Political Economy of Central Banking, ed. P. Arestis and M.C. Sawyer. Cheltenham, UK: Edward Elgar. Smithin, J. 1994. “Cause and Effects in the Relationship between Budget Deficits and the Rate of Interest.” Économies et Sociétés 28 (1-2): 151-70. Sraffa, P. 1960. Production o f Commodities by Means of Commodities. Cambridge: Cam­ bridge University Press. Wood, A. 1975. A Theory o f Profits. Cambridge: Cambridge University Press. Wray, L.R. 1991. “Boulding’s Balloons: A Contribution to Monetary Theory.” Journal of Economic Issues 25 (1): 1-20. --------- . 1993. “Money, Interest Rates and Monetary Policy: Some More Unpleasant Mon­ etarist Arithmetic?” Journal of Post Keynesian Economics 15: 541-69.

2 Transformational Growth and the Changing Nature of the Business Cycle Korkut A. Erttirk

In recent years, Edward J. Nell has persuasively argued that economic growth should be studied as a process of qualitative transformations in technology, orga­ nization of production, social institutions, and the structure and pattern of con­ sumption. As part of this overall project he has also argued that the nature of the business cycle and macroeconomic adjustment has been transformed as the craftbased economy gave way to the mass-production system at the turn of the twentieth century. In Nell’s view, during this process, the nature of macroeconomic adjust­ ment, and with it the very nature of economic volatility, changed as well. In the empirical literature, much agreement exists that economic volatility has been transformed in the post-World War II era (Altman 1992; Bailey 1978; De Long and Summers 1986). For instance, Sylos-Labini (1991) identifies different empirical trends in what he calls the “classical” business cycle (1800-1913) and the “new” business cycle after World War II; the irregular and violent fluctuations in the interwar period, in his view, constituted yet another era. Nell provides what might be a compelling theoretical explanation for this em­ pirically well established change of pattern in economic volatility. Moreover, his argument refutes the view that the distinctive characteristic of the Keynesian system is sticky money wages. In Nell’s account, sticky money wages are as much an attribute of the craft system as of the mass-production economy, and thus play no unique role in explaining the Keynesian business cycle associated with the massproduction economy. This chapter develops a simple dynamic model to draw an analytical contrast between cyclical behavior in the two types of economy by formalizing Nell’s ar­ gument on the changing nature of the business cycle. The exercise shows that while

24 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

only damped oscillations are possible in the craft system, the mass-production economy is intrinsically unstable, generating explosive oscillations. This tends to support the view that the containment of business cycle instability in the massproduction economy required the emergence of the activist state. The Old and the New Business Cycle According to Nell, differences in economic organization, in the nature of the labor process, and in technology account for why output adjusts to changes in demand differently in the two systems (see Nell 1992, part III; Nell and Phillips 1995). Whereas fixed employment and variable productivity characterize the craft system, the mass-production economy is characterized by fixed productivity and a variable capacity utilization/employment ratio. In a craft economy production is to order, and current costs are largely fixed. Work tends to be skilled and specialized, and, more important, requires close co­ ordination among workers. Each member of the work crew is indispensable and thus has to be present for production to take place. Start-up and shutdown costs are prohibitive, and storage technology is underdeveloped. The product is not stan­ dardized, and few, if any, economies of scale exist. Firms generally lack market power, and a “natural” size of operation limits their size. Long-run growth in de­ mand is accommodated not by the expansion of existing firms but by the formation of new ones subject to the same size constraints. Under these conditions of fixed employment, prices are highly flexible whereas money wages, set by custom, re­ main stable in the short run. Moreover, because employment does not change much in the short run, changes in demand produce little, if any, immediate pressure on wages. In the face of changes in demand, price movements play a stabilizing role. Rising (falling) in inverse proportion to the elasticity of output, price changes in­ duce a higher (lower) supply of output along with rising (falling) marginal cost— since productivity rises at a diminishing rate as a function of the intensity of labor. At the same time, the change in prices affects consumption demand through its impact on real wages. Following an increase in demand, rising prices lower con­ sumption demand by decreasing real wages and thus tend to balance the initial increase in demand. In a similar vein, falling demand is associated with falling productivity (the work crew works more slowly), falling marginal costs, and falling prices. In this case, the higher consumption demand caused by a higher real wage tends to offset the initial fall in demand. In other words, productivity varies with demand while employment remains fixed, and exogenous changes in demand tend to be offset by those induced by the initial change. Thus, consumption and invest­ ment move in opposite directions. In a mass-production economy, the organization of production is completely transformed. Integrated tasks replace activities carried out by teamwork in produc­ tion where current costs become largely variable. Work is no longer specialized,

CHANGING NATURE OF THE BUSINESS CYCLE

25

and thus individual workers become dispensable. Standardized goods are produced to stock. The existence of scale economies implies no natural size of operations for firms. As a rule, oligopolistic markets and advanced storage technology enable firms to build plants with significant amounts of reserve capacity and to administer prices. In response to changes in demand, output, employment, capacity utilization, and energy costs can all be varied pari passu and productivity can be kept constant. Output tends to move with demand because any change in autonomous spending gives rise to induced spending in the same direction. Thus, investment and con­ sumption expenditures rise and fall together. The output adjustment is based on variations in employment governed by the multiplier-accelerator mechanism. Prices no longer play a decisive role in output adjustment. The Model Consider a simple system with two state variables, real output (F) and the general price level (P): Y — Y (Y,P) P = P{Y,P\

(1)

where the dots above variables indicate time rates of change.1 It is assumed that both functions can be continuously differentiated. Without specifying any explicit functional form, the dynamic properties of the system in (1) can be discussed on the basis of the signs of the elements of its Jacobian matrix, which consists of the partial derivatives of the individual equations with respect to the two state variables. Provided that Nell’s arguments on the old and the new business cycles can be summarized as a series of different assumptions about the signs of the four partial derivatives in (1)— Yy, YPj P y, P p— conclusions can be drawn on how the dynamic behavior of the system differs between the two cases. As argued below, in the context of this simple dynamic model the massproduction economy is distinguished from the craft economy on account of the signs of Yy and Pp. Both are positive in the mass-production economy and negative in the craft system, while the other two partials maintain the same sign in both types of economy: YP < 0 and P Y > 0. We now turn to a discussion of the signs of these partial derivatives. Yy: The Impact of the Level of Output on Its Rate of Change By definition, output is equal to the product of labor productivity (/r) and the quantity of labor employed (L): Y =

kL .

(2)

26 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

The rate of change of output is thus given by Y = 7CL + Ln.

(2a)

In the craft economy, equation (2a) is reduced to Y =

kL,

(2b)

since output adjustment is based on variations in productivity where employment basically remains constant, L = 0. The level of productivity in turn has two determinants: the intensity of labor (X) and the scale of output: n = n (X, Y), and thus n = n j i + KyŸ,

(2c)

where nx > 0 and nxx < 0 (productivity rises with intensity of labor at a dimin­ ishing rate), n Y < 0 (there might be diseconomies of scale because of the small scale of production), and kyy < 0 (speed up of production creates additional costs as it becomes harder to maintain work discipline, morale, and instruments of pro­ duction, etc.). It is assumed that nXY = nYX — 0. Rearranging (2c) and substituting it in (2b) yields

(2d)

Treating the change in the intensity of labor as exogenous2 and taking the partial derivative of (2d) with respect to the level of output, it can be seen that Y y = dY — < 0. dY By contrast, in the mass-production economy, output adjustment is based on variations in employment where productivity remains by and large fixed, which implies that (2a) is reduced to Y = nL,

0)

where employment varies with changes in investment demand and other autono­ mous expenditures (A) and the expenditure multiplier (LA). With a given level of productivity, this can be expressed as L = La(A - A),

(3a)

CHANGING NATURE OF THE BUSINESS CYCLE

27

where the expenditure multiplier and the trend value of autonomous expendi­ tures (A) are assumed constant, and the deviation of A from its trend value is given by (A - A) = mY + A(Y).

(3b)

The first argument in (3b) is the accelerator, where the coefficient m is constant, while the second argument reflects the effect of economies of scale (and thus A y > 0). Substituting (3b) in (3a) and utilizing (3) yields

(3c)

Assuming that (1 — LAm n > 0), and because A Y > 0, the partial derivative of (3c) with respect to Y is positive; thus, in the case of a mass-production economy, Yy

> 0. YP\ The Impact of the Price Level on the Rate of Change of Output In the short run, money wages are assumed constant in both types of economies. Thus, changes in the price level affect the level of output through their impact on the real wage in both systems. However, in the craft economy, the real wage affects not only workers’ consumption on the demand side, but also the quantity of output firms desire to produce on the supply side. By contrast, in the mass-production economy, output is solely demand-determined. Given the structure of production, the increase in output does not involve rising marginal cost. Thus, changes in the real wage affect output through their impact on aggregate demand (i.e., on workers’ consumption and firms’ investment expenditures). In order to introduce the price level into the analysis, we multiply both sides in (2) by the money wage rate and divide by the price level. Rearranging,

(4)

where N is nominal and Y the real output. The real wage, X = —, is a function of the price level only, since the money wage is assumed constant, w = w. Because this is the craft economy, we set L = 0 and assume that money wage remains constant. This implies that output varies with the price level, labor productivity, and the real wage

28

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Y = f [ P , 7t(X{P)\ X(P)]

(4a)

with YP < 0, Yn > 0 and Yx > 0. The labor productivity varies with the intensity of labor, which in turn is a function of the price level. Rising prices induce producers to increase the intensity of labor, implying that, K = 7Z{X(P)\

(4b)

and, thus; k = KxXpP

(4c)

where output per unit labor rises as a function of the intensity of labor, nx > 0 and where the intensity of labor varies positively with the price level (XP > 0) at a diminishing rate (XPP< 0). In the case of the real wage, we have X = X(P) and thus, X = XpP where XP < 0 since w = w. Using (4a) the rate of change of output then is given by: Y = [ Y P + YjiyXp +YxXp]P

(4d)

Whether output rises of falls with a given increase in the price level depends on the relative magnitudes of the negative (YP + YxXP) and positive iY jtxXP) terms in (4d). The former reflects the impact of the real wage on the demand side, while the latter term reflects the impact of the price level on the supply side. The former is always negative since XP > 0, while the latter is positive below some maximum level beyond which intensity of labor cannot increase. This also implies that the impact of the price level on the rate of change of output (YP) is likely to be negative. Setting the change in price to unity, (P = 1), YP — Ypp + YnftxXpp + YxXpp

(4e)

It can be recalled that X p p < 0 by assumption while it is reasonable to assume YPP = XPP = 0, and thus it follows that Y P < 0.

CHANGING NATURE OF THE BUSINESS CYCLE

29

In the case of mass production, YP is also negative, provided that a higher (lower) real wage has a positive (negative) effect on autonomous expenditures. Differentiating both sides in (4), but now setting it - 0, we get

(5)

Taking the partial derivative with respect to the price level yields

(5a)

Assuming that the real wage affects employment only through its impact on au­ tonomous expenditures, and not through factor substitution (LP — 0), and that XP — 0, equation (5a) is reduced to

(5b)

From (3a), we have the change in employment given as a function of autono­ mous expenditures and the expenditures multiplier: L = La(A - A).

(3a)

We now drop the accelerator and ignore economies of scale in (3b) since neither is a function of the price level. Allowing instead the possibility that autonomous expenditures might deviate from their trend value as a function of the real wage,4 we here write the rate of change in employment as L = LaB(MP)),

(5c)

where B = A — A, and it is assumed that a higher real wage raises autonomous expenditures (Bx > 0). It follows that

L„ =

W

r

Substituting (5c) and (5d) in (5b), we get

0, XP < 0 and LA > 0. Py. The Impact o f the Level o f Output on the Rate o f Change o f the Price Level

Here, we will assume that prices tend to vary as a function of the output gap in both systems (i.e., P Y > 0). This assumption, and the notion of a natural level of output it implies, is relatively straightforward in the case of a craft economy. How­ ever, the rate of change in the price level can also be shown to vary with the level of output in excess of some “normal” level of output in the case of a massproduction economy with administered prices and stable markups over prime cost, provided that inflationary expectations are a part of the wage bargain (Blanchard 1999, chs. 6, 7). P p\ The Impact o f the Price Level on Its Rate o f Change

Markup pricing implies that price changes can be cumulative. Thus, we will assume that the sign of this partial derivative is positive in the mass-production economy, and negative in the craft system. The Analysis of the Dynamic Behavior of the Equation System The Jacobian matrix for the equation system in (1) is given by

The sum of the two characteristic roots is equal to, trJE = Yy + Pp, and their product is given by IJE\ = YyPP - P yYp. Given the signs of the partial derivatives for the craft system, it can readily be seen that the Jacobian matrix is always positive, and the trace is always negative. Thus, a saddlepoint, unstable node, and unstable focus are ruled out. The likely outcome is a stable focus, as shown by constructing the phase diagram of the system in (1) and drawing the trajectories of motion.4 Recurrent oscillations require ongoing external shocks, since only damped oscillations are possible in this system. Using the implicit function rule, we can ascertain the slopes of the Y = 0 and P = 0 isoclines:

CHANGING NATURE OF THE BUSINESS CYCLE

31

Figure 2.1 Stable Focus*I

Given our assumptions for the mass-production economy, the determinant of the Jacobian is still positive, but the trace changes its sign and is now positive, implying an unstable focus.5 This again can be seen by constructing the phase diagram of system (1) under the set of assumptions for the mass-production economy. Thus, the dynamic behavior implied by the assumptions made to characterize the mass-production economy is characterized by explosive oscillations. This begs the question of how stability is attained. The explanation Nell suggests, along with many others before him, is that the institutional framework of the post-World War II era made countercyclical intervention by governments possible. Instability was thereby kept within bounds because when output deviated sufficiently from its trend value, countercyclical forces came into play. In the context of our model, this implies that for intermediate values of Y, the sign of the partial derivative Yy is still positive, but for values that are sufficiently large or small, the sign becomes negative. This makes the Y = 0 isocline nonlinear, with its slope negative at first, then positive, and then negative again. The direction of the streamlines in the phase diagram of the revised system below show that on the assumption that countercyclical policies are effective, the unstable focus in figure 2.2 becomes a limit cycle. If we denote the compact set D of R as the set {(7, P)0 < Y ^ 7,,0 ^ P ^

32

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Figure 2.2 Unstable Focus

P,}, then any positive semi orbit in R starting outside D will eventually enter D. It can be seen that at any point below (above) the Y = 0 isocline dyldt is positive (negative), and thus movement in Y will be from left (right) to right (left). Likewise, at any point to the right (left) of the P = 0 isocline, P is positive (negative), and thus movement is from down (up) to up (down). Since the singular point of the system is unstable, and in the bounded region D no singular point other than E exists, and it is impossible for a trajectory within the region to exit, then by the Poincare-Bendixson theorem there exists in D at least one attracting closed orbit (see figure 2.3). Conclusion Our exercise has shown that the stylized ideas that have been advanced in drawing a contrast between a craft-based economy and a mass-production system do indeed imply a fundamental difference in the nature of the business cycle in these two systems. While only damped oscillations seem possible in the craft-based economy, in a mass-production economy the system is intrinsically unstable and oscillations that emerge are explosive. Thus, it is plausible that under the conditions of the new

CHANGING NATURE OF THE BUSINESS CYCLE

33

Figure 2.3 Bounded Instability*I

business cycle, instability could be contained only by the emergence of the activist state. By contrast, the fact that the business cycle was not inherently unstable during the craft-based system could have meant that the system could go on functioning without the active intervention of the state. Notes I would like to thank Gary Mongiovi for his helpful comments. I alone am responsible for the remaining errors. 1. The two state variables can also be thought of as deviations from trend values. 2. Below, the intensity of labor is expressed as a function of the price level. Here, however, because prices are assumed to be constant, the intensity of labor is treated as exogenous. 3. In other words, this amounts to assuming that changes in workers’ consumption ex­ penditures are lumped together with autonomous expenditures. 4. The sufficient condition requires that the discriminant of the characteristic equation be negative: (trJE)2 < A\JE\. 5. The sufficient condition is again a negative value for the discriminant of the charac­ teristic equation. If, however, PY = 0, then the fixed point becomes an unstable node.

34 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

References Altman, M. 1992. “Business Cycle Volatility in Developed Economies, 1870-1986: Revi­ sions and Conclusions.” Eastern Economic Journal 18: 259-75. Bailey, M.N. 1978. “Stabilization Policy and Private Economic Behavior.” Brookings Papers on Economic Activity 1: 11-50. Blanchard, O. 1999. Macroeconomics. 2nd ed. Upper Saddle River, NJ: Prentice-Hall. De Long, J.B., and Summers, L.H. 1986. “The Changing Cyclical Variability of Economic Activity in the US.” In Studies in Business Cycles, voi. 25, ed. R. Gordon. Chicago: University of Chicago Press-NBER. Nell, E.J. 1992. Transformational Growth and Effective Demand. New York: New York University Press. Nell, E.J., and Phillips, T.F. 1995. “Transformational Growth and the Business Cycle.” East­ ern Economic Journal 21: 125-46. Sylos-Labini, P. 1991. “The Changing Character of the So-Called Business Cycle.” Atlantic Economic Journal 19: 1-14.

3 A Classical Alternative to the Neoclassical Growth Model Thomas R. Michl and Duncan K. Foley

This chapter lays out a modem version of the classical growth model and compares it with the standard neoclassical model.1Since it is now clear that population growth in many advanced industrial countries has reached a plateau (see Foley 2000), and demographers project that economic development may push the boundaries of this demographic transition to large areas of the world, our classical model includes the transition from a labor-surplus economy to a labor-constrained economy, with par­ ticular attention to the complications introduced by biased technical change along nonsteady-state paths. The chapter is organized as follows. The first section details the classical model component by component, and includes an alternative interpretation of the aggre­ gate production function based on the hypothesis that biased technical change pre­ dominates. The second section elaborates the dynamics of the classical model without technical change. This section introduces the important idea that classical growth theory, unlike its neoclassical counterpart, does not assume that full em­ ployment represents the natural initial state of a growing capitalist economy, which raises the question of the transition from labor-surplus to labor-constrained growth. The third section elaborates the model with technical change of a capital-using, labor-saving variety that captures the idea of secularly increasing mechanization and automation. A major preoccupation of this section is the analysis of the nonsteady-state paths induced by biased technical change. The fourth section draws out some implications for the critique of neoclassical growth theory, which seems altogether fitting in a volume celebrating Ed Nell, an economist who has contrib­ uted seminally to this effort (e.g., see Nell 1967, 1970, 1992). The final section

36 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

offers some observations on the practical and theoretical significance of the classical alternative to the neoclassical growth model. A Standard Classical Model Models of growth, no matter their orientation, are cut from similar cloth: a theory of the labor market, a theory of the product market, a theory of asset markets, a theory of production and technical change, and a theory of consumption and saving. Different schools of thought make different choices about how best to close their growth models, and this provides a useful way to arrange any comparison of al­ ternative models or theories.2 Our standard classical model (like the standard neo­ classical model) is a com model, in which one homogeneous good is produced by means of accumulated stocks of itself (called capital goods) and undifferentiated labor. Any technical change is of the disembodied variety, and the capital goods are putty-putty, meaning that it is possible to retrofit the existing capital goods to new engineering specifications without cost. We make no claim that the results generalize to more sophisticated models with heterogeneous capital goods. Labor The classical vision does not naturally and exclusively regard distribution as de­ termined in the markets for labor and capital. The original classical economists believed that the real wage was determined by historical and institutional forces. By taking the distribution of income to be exogenous, they were free to explore the factors that determine growth and capital accumulation. The key to this for­ mulation was the view that the labor force is endogenously constituted to meet the needs of capital accumulation, through demographic laws in the case of Ricardo and Malthus, or through historical processes in the case of Marx. The first of these approaches asserts that fertility and mortality rates adjust so that workers are bio­ logically regenerated to satisfy the demands of industry. In short, labor supply adjusts to labor demand at some conventional real wage. The latter asserts that historical processes such as the Enclosure movement continuously regenerate an actual surplus of labor, the active and latent reserve armies of labor. We will stay closer to the latter interpretation. The notion of a conventional wage offers little appeal in the presence of persis­ tently increasing labor productivity. In its place, we assume that historical and institutional forces set a floor under the wage, w, and that this floor adjusts contin­ uously with every advance in labor productivity, jc, so that the conventional wage share remains constant. This floor operates under conditions of labor surplus or endogenous labor supply. On the other hand, when the accumulation of capital has exhausted the surplus labor available, we will assume that the wage adjusts to equate the supply of labor with the demand for labor on a continuous basis.3 If this adjustment mechanism fails and capital overaccumulates, so that the demand for labor exceeds its supply, we will assume that the wage will rise until it absorbs

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

37

the entire gross product per worker. Formally, suppressing time subscripts, using It to represent the profit share associated with the wage floor, ND to represent labor demand and IVs labor supply, we have

( 1)

This formulation draws a sharp distinction between a labor-surplus growth regime and a labor-constrained growth regime. We need some notation to distinguish be­ tween the regimes; let w° represent the wage floor and w ] the market-clearing wage. We characterize the market-clearing wage below, using the model of capital accumulation. We assume that the supply of labor grows exponentially at rate n. To save on notation, we normalize the initial labor force to be unity, so that Ns = ent.

(2)

The demand for labor depends on the level and technique of production, to which we now turn. Production and the Product Market At any point in time, the technique of production is described by a Leontief pro­ duction function. We discuss extensively the nature of the technology— meaning the whole set of techniques available at a point in time— but for now assume that the appropriate technique has been selected. Let output consist of a homogeneous, all-purpose good, represented by X, produced by means of the accumulated stock of capital, X, and labor, N. We represent the productivity of capital under conditions of full utilization of capacity by p —XIK. The production function is X = min (pK, xN). Under the assumption that many identical capitalist firms compete vigorously within the same product market, each will receive the same rate of return on capital, and the price will consist of unit labor costs plus unit profits at the average rate. In terms of the parameters of the production function, it is convenient to express the trade-off between wages and profits in terms of the wage rate-profit rate sched­ ule or wage-profit schedule:

(3)

38

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

where v represents the gross rate of profit, equal to up. The rate of profit is also equal to r + 5, where S represents the rate of depreciation of capital goods. We assume throughout this chapter that the rate of depreciation is constant through time and across different techniques of production. We will refer to v as the rate of profit, and r as the net rate of profit. We adopt the convention that r° represents the net rate of profit associated with the wage floor, w°, and that r 1 represents the net rate of profit associated with the labor market clearing wage, w1 (which is defined below). The net rate of profit will be —8 when labor demand exceeds labor supply. The wage-profit schedule provides a useful visual summary of the structure of the capitalist economy. Its intercepts at x and p and its slope, which is equal to —jfc, the capital-labor ratio, compactly characterize the technique of production. Because much policy attention has been focused on the national saving rate, we want to identify the level of consumption per worker, c. The consumption-growth schedule shows the combinations of consumption per worker and rates of capital accumulation gK, supported by the technique of production:

(4)

We adopt the convention that gz represents i / z, the growth rate of a variable, and that z represents its time derivative. We also suppress time subscripts unless they are needed for clarity. Equations (3) and (4) are common to any model of economic growth, and because they have exactly the same mathematical form, we refer to them collectively as the growth-distribution schedule. Assuming full utilization of capacity, the demand for labor is a function of the level of capital,

(5)

The demand for labor thus depends on the technique selected and the volume of capital that has accumulated. Consumption and Saving The classical tradition distinguishes itself by an appreciation of the class structure of saving and consumption. In a classical spirit, we assume that workers treat their wage as a subsistence wage (with “subsistence” defined historically rather than absolutely), which implies that as a class, workers contribute zero net social saving. The assumption that workers as a class live a hand-to-mouth existence frees us to pursue the classical theme that capitalist saving is the driving force of economic

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

39

growth, but it should not be viewed as anything more than a simplifying assump­ tion.4 For example, it does not necessarily imply that individual worker households do not save, since their contribution may be canceled out at the social level by the dissaving of retired workers. We simplify by assuming that these cancellations are complete. The treatment of investment, on the other hand, is more problematic. Since the object of analysis is the long-run trajectory of a capitalist economy, we have chosen not to distinguish between enterprise and thrift. But it has not escaped our notice that this leaves the model vulnerable to the complaint registered by Nell (1985) against a similar model of Marglin, namely, that it relies to an unacceptable ex­ tent on Say’s Law. As we argue in Foley and Michl (1999: ch. 10), neoKeynesian models, which incorporate a separate investment function and accom­ modate equilibria having unutilized capital stock, are more convincingly interpreted as representations of the short run and are therefore more useful in analyzing highfrequency fluctuations in capitalist economies than long-run growth in potential output. Let us assume that the representative capitalist displays altruism toward his or her descendants, so that the consumption problem of the whole dynasty is to max­ imize utility over an infinite horizon, discounted at the pure rate of time preference, /?, which plays the role of a thrift factor. The logarithmic utility function simplifies the solution considerably because the wealth and substitution effects of a change in the profit rate wash out. We will assume perfect foresight of the path of future profit rates on the part of capitalists. The capitalist dynasty’s consumption problem can be written

This problem can be solved through control theory.5 The optimal stream of con­ sumption is a constant fraction, /J, of the instantaneous wealth of the dynasty, K. This translates into an equation for capital accumulation that we refer to as the Cambridge equation: (6 )

A great advantage of this utility function is that it uncouples accumulation from expectations of the profit rate. A change in the profit rate leaves the propensity to consume out of wealth, CIK, constant and equal to /J. Thus equation (6) is true at each instant no matter what path has been taken or is expected to be taken by the profit rate. This greatly simplifies the transition between a labor-surplus regime and

40

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

a labor-constrained regime, or in general any dynamic trajectory involving changes in the rate of profit.6 Technical Change and the Fossil Production Function We model technical change as purely exogenous,7 and consider two alternative cases. First, following common practice, we will take technical change to be purely labor-augmenting (Harrod-neutral). Second, we consider labor-saving, capital-using technical change, which we will call Marx-biased technical change. Technical change is modeled as exponential change in the coefficients of the most advanced or best-practice technique. We will consider exogenous technical change as a process that alters the bestpractice technique (using a tilde to represent best-practice technical coefficients) as follows: x

=

x0e*

P = P f* k = k0ea~x)t.

(7)

Marx-biased technical change is the hypothesis that y > 0 and x < 0. Harrodneutral technical change is the hypothesis that y > 0 and x ~ 0. For notational clarity we use gx, gk, and gp to refer to the actual growth rates of output per worker, capital per worker, and the output-capital ratio, respectively. These can differ from y, y — x> and X when the best-practice techniques are adopted with a lag. The hypothesis that technical change has some kind of bias can hardly be called extravagant; a leading neoclassical explanation (e.g., Berman et al. 1998) for growing wage inequality, for example, depends upon such a conjecture. But this hypothesis has not been fully explored in the context of models of labor and capital accumulation. Some evidence that a pattern of rising labor productivity, rising capital intensity, and declining capital productivity has attended the eco­ nomic development of capitalist societies, at least during significant historical ep­ isodes, is presented in Foley and Marquetti (1997) and Foley and Michl (1999: ch. 2). Marx-biased technical change provides an alternative way to understand the production function. At any point in time, we assume for now that capitalistmanagers will adopt the best-practice technique, continually updating their choice of technique as innovations with higher labor productivity (but lower capital pro­ ductivity) become available. Thus, the best-practice technique will be evolving through time, leaving behind a sequence of discarded techniques. Insofar as these “fossil” techniques remain part of the engineering culture and have not yet been

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL 41

forgotten, they can be described by an aggregate production function we call the fossil production function. Rather than resting on the arbitrary assumption of in­ variable technical proportions, our classical model is based on a well-specified theory of technology that explains the aggregate production function as a repre­ sentation of the technological history of the economy.8 To derive the fossil production function, let us assume constant Marx-biased technical change prevails. Equations (7) represent the parametric form of the fossil production function. Assume that technical changes are adopted without delay, so that gx = y and gk = y — Then we have gx = y/(y - x)gk. Integrating from some initial year to the present gives us

It is immediately obvious that the rectangular form of the fossil production function for past techniques is the familiar Cobb-Douglas equation. A complete description of the technology at some arbitrary time, when Kt units of capital have been ac­ cumulated, is x = Ak° for x < m in [pt(KtIN), Jcj,

(8)

where A represents a consolidated constant of integration that can be definitized from the initial conditions. We can visualize the fossil production function in logarithmic form in figure 3.1. The sequence of past techniques forms the segment with upward slope a . The most recent best-practice technique is the most capital-intensive technique yet in­ vented, and it sets the limit on labor productivity. It is possible to increase the amount of capital per worker using the most recent technique, but giving the same number of workers more machines just adds unusable capacity and has no effect on their output, as represented by the flat segment. On the other hand, switching to an old technique by reassembling the existing stock of putty-capital makes it technically possible to employ more workers. It is possible to choose a position down the fossil production function, but it is not possible to choose a technique more mechanized than the best-practice technique.

42

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Figure 3.1 The Fossil Production Function

The fossil production function has a Cobb-Douglas segment for all past techniques start­ ing from some initial position. The best-practice technique (£„ xt) is the most mechanized technique yet invented.

Dynamic Properties without Technical Change It makes sense to explore the simplest case, with no technical change, before we tackle the complications of the fossil production function. Equations (l)-(6) define a fully specified classical model of growth, determining the endogenous variables, w, v, c, gK, N°, and Ns. The rate of capital accumulation determines the rate of growth of output. The model’s character depends on whether or not the labor supply constraint is binding.

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL 43

Labor-Surplus Economy The model begins in a labor-surplus regime as long as K0 < k, making the initial capital stock insufficient to fully employ the initial labor force (normalized to one labor unit). The wage is determined by the wage floor. The wage floor then deter­ mines the profit rate, through the growth-distribution schedule, and the profit rate determines the rate of capital accumulation through the Cambridge equation. gK = r° - p. It should not escape our notice that this is an endogenous growth model, as long as the labor constraints are not binding. A decrease in the capitalists’ rate of time preference, for example, elevates the rate of capital accumulation. The rates of growth of output and labor demand follow, as workers are drawn into waged em­ ployment from the reserves of labor to accommodate the increase in accumulation. If gK < n, of course, this regime will go on forever. On the other hand, if accumulation eventually depletes the labor reserves, the model transforms into a labor-constrained regime. From Labor-Surplus to Labor-Constrained Economy The transition from a labor-surplus to a labor-constrained regime in principle starts affecting the model at the moment when capitalists anticipate the exhaustion of labor reserves. But as we have already noticed, with the log utility function this realization will not change anything, and the Cambridge equation remains valid because the impending decline in the rate of profit will not affect the current pro­ pensity to consume out of wealth. Thus, at the moment of transition, T, the rate of accumulation abruptly declines from r° — /} to n. The moment of transition occurs when the capital stock employs all workers, or K r = ken'\ so T = (In K0 — In k)/(n — r° + /J). Once the system passes into the labor-constrained regime, the wage rises so as to equate the rate of growth of labor demand to the natural rate of growth of the labor force. The Cambridge equation changes roles, and now determines the rate of profit that sustains full employment, or r] = n + /?. This abrupt decline in the rate of profit appears as a movement along the growthdistribution schedule. The wage that clears the labor market can be obtained from the growth-distribution schedule: w' = Jt - (r] 4- S)k. If we wanted to make life interesting, we could assume a more general capitalist utility function, but we will leave that for the separate treatment it deserves.9

44

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Dynamics with Exogenous Technical Change The classical model with exogenous technical change is a dynamical model con­ sisting of equations ( 1)—(6) together with equation (7). Its character depends crit­ ically on the value of x • Harrod-Neutral Technical Change Introducing Harrod-neutral technical change (x = 0) into this model adds an ele­ ment of realism without much complexity. If we rescale the per-capita variables, and express them in effective labor units, the results of the previous section go through. In particular, the wage floor and market-clearing wage in effective labor units can be determined in exactly the same way. Measured in actual labor units, the wage floor and the market-clearing wage rise at the rate of labor-saving tech­ nical change. The moment of transition occurs when the capital stock employs all effective workers, or KT = k0e(r+n)T, so T = (In K0 - In k0)/(y + n — r° + fi). Marx-Biased Technical Change Accommodating the hypothesis of Marx-biased technical change (x < 0) requires that we abandon the steady state. Again, it is useful to distinguish between the labor-surplus and labor-constrained regimes. In either case, capitalists have the op­ tion of declining to adopt the new techniques as they become available. Capitalists will adopt only viable technical changes that are expected to increase their private rate of profit. It can easily be demonstrated that the viability condition is

n < a .10 We will call a the viability parameter to acknowledge its role in the selection of new techniques. Note that the viability parameter can also be interpreted as the output elasticity of capital in the fossil production function, but only over the set of past techniques. The output elasticity of capital using the best-practice technique is zero. There is substantial empirical evidence (Foley and Michl 1999: ch. 7) that a exceeds the profit share by a large margin, which presents the problem for neo­ classical theorists that the wage exceeds the “marginal product of labor,” in con­ tradiction to their theory of distribution. The best-practice technique is evolving and techniques may be adopted with a time delay, creating a pocket of unchosen techniques. Thus, the rate of capital deepening is not restricted to the rates of technical change, and follows

(9)

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL 45

The second option presents itself when techniques are adopted with a delay, in which case there will be a segment of Cobb-Douglas technology available. In this case, the usual marginal productivity equations apply. In particular, w = (1 — a)x so that gw = gx = agk. To anticipate, during its labor-constrained phase the econ­ omy will make a transition to the second option when n = a. Labor-Surplus Economy

If the viability condition fails to be satisfied (n > a) at the profit share established by equation (1), then the new techniques will be rejected because the wage is so low that they are not economical (recall that they are capital-using). The current technique will remain in use as long as the system enjoys a labor surplus, and the model reduces to one without technical change in its labor-surplus phase. We will assume that (n < a) so this does not happen.11 If by some fluke the viability condition is satisfied as an equality (n = a), the new techniques will be adopted. The path of the distribution point, (v, w), will then lie along the locus of switch points formed by the intersections of growthdistribution schedules adjacent in time. To see this, define the distribution point at a switch point to be (cr, co). Then for techniques (k , x) and (k\ x') adjacent in time, a) — x — ok (x) = x f — o k ’

where x ’ = x + dx = x + yxdt k' = k + dk = k H- (y — x)kdt.

It follows that (o = (1 - a)x

a = ap. We will call this locus the envelope of switch points. It corresponds to the factor price frontier in neoclassical theory, although the envelope exists only for past techniques, whereas the neoclassical construction is defined over an unlimited spec­ trum of preexisting techniques. The envelope of switch points is the dual, in wageprofit space, of the upward-sloping segment of the fossil production function. Its rectangular form can be obtained from the parametric equations by using either equation (7) or equation (8):

46

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Figure 3.2 The Trajectory of the Distribution Point

The trajectory of the distribution point (v, w) lies to the northwest of the envelope of switch points in the labor-surplus economy with exogenous, Marx-biased technical change. The separation depends on the difference between the profit share, 7r, and the viability parameter, a . The envelope of switch points is the dual of the Cobb-Douglas segment of the fossil production function.

where B = In (1 — a) + In x0 + aJ{ 1 — a )(ln a + In p0) and the restrictions from equation (8) that w > (1 - a) x and v < ap apply. If the viability condition holds as an inequality (jt < a), the new techniques will be adopted but the trajectory of the growth-distribution point will always lie to the northwest of the envelope of switch points. This kind of trajectory is illus­ trated in figure 3.2, which exploits the fact that both the envelope of switch points and the trajectory of the labor-surplus economy with a constant profit share are log-linear functions having slope - a l{ 1 - a). The distance between them (ex­ pressed in level form rather than logs) is the norm of the vector difference (v, w)

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL 47

—(ap, (1 — a)jt), which depends on the difference a — m, since v — Ytp and w = (1 — 7t)x, Because Marx-biased technical change reduces the output-capital ratio, the rate of profit declines over time in the labor-surplus economy. Marx called this process the law of the tendency for the rate of profit to fall, and attributed deep historical significance to it.12 As the rate of profit declines, it brings down the rate of capital accumulation through the Cambridge equation. The capital stock evolves according to the law of motion:13 K = (,7ip0e* - (S + p))K. Growth thus decelerates under the labor-surplus regime even before the reserves of labor are exhausted. As the model is formulated, there is really no meaningful floor under this process, which is probably why Marx regarded this as a potential Achil­ les heel of capitalism. Once the rate of profit falls below the capitalist discount factor, the decumulation of capital begins. We will assume that this point is never reached. Indeed, the transition to the labor-constrained economy is contingent on maintaining a sufficiently positive rate of accumulation. It is worth remarking that rising wages, guaranteed through the wage-floor as­ sumption, maintain the technical dynamism which has these profound effects on profitability. If the wage were constant (as with a strict subsistence wage), the profit share would rise to the level of the viability parameter, at which point technical dynamism would cease because labor costs are too unimportant to justify further mechanization. The labor-surplus regime ends if and when the capital stock absorbs the available labor supply, or at the transition point 7, when Kn = k0e{n + 7 “ X)T{. This transition point depends on the initial conditions and parameters in the obvious ways. We will assume it exists. From Labor-Surplus to Labor-Constrained Economy As before, in the transition to the labor-constrained system the Cambridge equation changes roles. Under the labor-surplus regime, it determines the rate of accumu­ lation. Under the labor-constrained regime, the rate of accumulation must adjust to the natural rate of labor force growth through changes in distribution, so that n = 8 n = 8 k ~ 8k = r' ~ P ~ (Y ~ X)- The profit share continuously satisfies n = Po]e~*‘(n + S + P + (y - *)) as long as it remains below the viability parameter, a, where gk = y - x* Before impact with the labor constraint, the rate of accumulation must be high enough to ensure that the labor constraint is met, or gK > n + y~X- The path of K meets the labor constraint from below. The growth rate must generally be reduced then at impact by an increase in wages above the wage floor sufficient to reduce the profit share to satisfy 7in = pf,1 (n + S + P + (y — x ))• Thus, at 7,,

48

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

the model moves abruptly along the prevailing growth-distribution schedule to a higher wage and lower profit rate, much as it did in the absence of technical change. From this point on, however, the profit share must increase by gn = ~ x to compensate for the continuing decline in the output-capital ratio and maintain the net rate of profit at its equilibrium value, which is r 1 = n + /? + (y - x)- The rate of capital accumulation that just maintains full employment when all technical changes are adopted is gk = n + y~X- The profit share will stop increasing when it reaches the viability parameter and the system undergoes a transition to the next phase in its development. The increasing profit share requires wages to grow more slowly than labor pro­ ductivity. In fact, the wage share, after first rising above the old floor, will even­ tually decline below the conventional wage share that had been operative during the years of labor surplus since the profit share rises to a while it began as It < a. The growth rate of the wage can be found by logarithmic derivation of w = (1 — n)x to be

It can easily be verified that gw < 0, so that the wage is decelerating, although it is clear that gw itself remains positive. The behavior of the wage in this model is unlike its behavior without technical change (or with neutral technical change). There, with the onset of labor constraints, workers enjoyed a permanent increase in the wage and the wage share. Here, main­ taining full employment requires that the wage share eventually fall below the conventional wage share that had been operative during the years of labor surplus, and that the wage fall relative to the wage floor, as illustrated in figure 3.3. It is not self-evident how this change might be accomplished during a period when the labor reserves have been exhausted. It is interesting to contemplate briefly an alternative scenario in which the wage floor is maintained. Under this treatment, after rising when the labor constraint is reached, the wage returns to the path described by the rising wage floor. In figure 3.3, this occurs where the bold line intersects the thin line representing the wage floor. The profit share at this point is still below the viability parameter, so the technical changes remain viable. Thus, the wage rises proportionately to labor pro­ ductivity, and the profit share remains It. As in the labor-surplus regime, biased technical change and declining capital productivity drive down the rate of profit. This deterioration in the conditions of accumulation results in ever deepening eco­ nomic stagnation. Unemployment, which had been wiped out while the profit share was temporarily low, reemerges on the other side of the transition to the labor-

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL 49

Figure 3.3 The Trajectory of the Wage

The trajectory of the wage is shown by the bold line. The thin line shows the rising wage floor established by the conventional wage share, or w = (1 - n)x. The transition to the labor-constrained economy occurs at 7,. The economy meets the envelope of switch points at 72, and achieves its steady state at 7*.

constrained regime and grows over time. This would certainly be a source of social conflict, and is perhaps close to the kind of crisis of overaccumulation that Marx had in mind. Workers’ aspirations about the minimum share of output to which they are entitled are inconsistent with the maintenance of a rate of accumulation sufficient to sustain full employment under conditions of technological dynamism. We return to the original scenario in which the wage floor fails to reassert itself. This scenario turns out to be useful in establishing the precise relationship between

50

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

our classical model and the neoclassical growth model. As we will now show, the two models converge once the profit share has risen sufficiently. Joining the Envelope o f Switch Points When the profit share has risen to the viability parameter (n = a), the best-practice technique parts company with the technique that will be chosen. Call this transition point r 2, where T2 = (In (p0(n + S + ft + y — x)/a))/x- New techniques invented after T2 will be adopted with a time lag (if they are adopted at all). The model has now joined the envelope of switch points, and it behaves exactly as would a model with a well-behaved neoclassical production function. The profit share, determined by the usual marginal productivity equations, will remain equal to the viability parameter, n = a. The capital-labor ratio evolves according to the familiar fun­ damental equation, which can be written as follows:14 k = ocAka — (n + S + fS)k. It is clear that the rate of capital deepening, gk = gK - n9decelerates as profitability declines and capital accumulation slows. We see that gk < y — *, so that the separation between actual and best-practice techniques is verified. We are now ensconced in the world described by the neoclassical growth the­ orists. The dynamics of the economy at this point are identical to the transitional dynamics of a Solow-type growth model. It should come as no surprise, then, that this process of capital deepening comes to an end when the system achieves the steady-state technology. Call this transition 7*. The steady-state technology will be * n + p + 8 p* = ------ ------- . a

x*

=

Am-a) p*at(a-\)

fa *

=

^ 1 /(1 -« ) p * l / ( a - l )

The steady-state profit rate will be v* = ap*, and the wage, w* = (1—a)x*. Presumably, new, more mechanized best-practice techniques continue to be in­ vented, but like old radio programs drifting out into space unheard, they never make it into production. We can visualize this whole drama in figure 3.4, which shows the path of the distribution point, (v, w), starting in the labor-surplus regime at t = 0 and contin­ uing through the two portions of the labor-constrained regime marked off by 7„ 72, and T*.

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

51

Figure 3.4 Trajectory of the Growth-Distribution Point with Technical Change

With exogenous technical change the trajectory of the growth-distribution point, (v, w), in log form begins in the labor-surplus economy and ends at the steady state. The transition to the labor-constrained economy occurs at 7,. The economy joins the envelope of switch points at 72, and achieves its steady state at 7*.

Neoclassical and Hybrid Growth Models Having elaborated a classical growth model, we are well situated to compare it in a critical spirit with the neoclassical growth model. Three features that sharply distinguish the Solow-Swan model from our classical model are the proportional saving assumption, the neoclassical production function, and the assumption that the wage adjusts so that the labor market clears continuously. The proportional saving assumption treats households as identical units that save

52 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

a constant fraction s = 1 — c of their total wage and capital income. The accu­ mulation of capital follows gK = sp - 6.

(10)

This equation can be justified within neoclassical theory (Barro and Sala-i-Martin 1995, ch. 2) as the solution to the dynamic optimization problem of altruistic house­ holds with Cobb-Douglas technology and utility function in the form u = In c. Thus, a fundamental distinction is between the neoclassical model’s typical as­ sumption of one representative household and the classical model’s assumption of a class structure to saving. This distinction reveals an underlying difference in the vision of these two schools of economic thought. The classical vision rejects the neoclassical school’s strict methodological individualism, and takes account of distinct social and his­ torical structures larger than the family or household that frame economic behavior. Thus, the two-class saving function emerges as a representative social relationship in which agents appear as the personification of characteristic social relationships that are identified by economic theory in close consultation with historical obser­ vation. It is all the more important to identify this fundamental difference in vision because modem neoclassical economists no longer regard the two-class assumption as anathema, and are willing to adopt it when it suits their needs (Smetters 1999). The interpretation consistent with the neoclassical vision is that in the long run, accumulation is controlled by the most patient agents, while the classical attitude is that the agents in the position to control accumulation exhibit behavioral char­ acteristics appropriate for the reproduction of the dominant system of social relationships. With Neoclassical Technology The neoclassical production function is generally assumed to be continuously dif­ ferentiable and to display diminishing marginal products of capital and labor with constant returns to scale. The obvious candidate in any comparison with the clas­ sical model is the Cobb-Douglas function, x = Aka.

(11)

The Cobb-Douglas function satisfies the Inada (1963, 120-21) conditions, which include x \k ) > 0, x ”(k) < 0, lim*_,0jc'(A:) = °°, and limb ^x'ik) = 0. The assumption of full employment works in conjunction with the neoclassical production function and the assumption of perfect competition, which selects the technique whose mar­ ginal product of labor equals the wage. With the Cobb-Douglas function, of course, this means the profit share will remain equal to a at all times, so that the wage

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

53

satisfies w = (1 — a)ka. For any capital stock and labor supply, it will always be possible to clear the labor market by

( 12)

Thus, equations (10)-(12), in addition to equations (2), (3), and (4), define the familiar Solow-Swan growth model. The dynamic path is provided by the solution to the fundamental equation gk = sAka 1 - 8 -

h,

(13)

given an initial capital stock, K0. The transitional dynamics depend upon whether the initial capital stock (per worker) exceeds or falls short of its steady-state value. The steady-state capital-labor and output-capital ratios are

If the initial capital stock falls short of its steady-state value, the transitional dy­ namics of this model compare to those of the classical model with exogenous biased technical change during the phase from T2 to 7* when the economy has joined its envelope of switch points. With a Fossil Production Function We can gain some insight into the importance of the neoclassical production func­ tion for the Solow-Swan model by replacing it with exogenous Marx-biased tech­ nical change. This hybrid model consists of equations (10), (12), and (7) in addition to equations (2), (3), and (4). The fossil production function imposes two restrictions on the techniques avail­ able that violate the Inada condition of strict positivity of marginal products. First, the technique cannot be (nontrivially) more capital-intensive than the most ad­ vanced technique yet invented. Second, the technique cannot be (nontrivially) less capital-intensive than the oldest technique that remains in the collective memory of engineers and entrepreneurs. We began the classical model at an initial position where these two margins were the same, and then allowed biased technical change to operate so that eventually the fossil production function would contain a CobbDouglas segment between these two extremes. Approaching the neoclassical model in the same way, let us ask whether a full employment equilibrium exists under the proportional saving assumption. Recall

54 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

that we normalized the initial labor supply to unity. Thus, the condition for supplydemand equilibrium with respect to the level of employment is simply K0 = k0. Clearly, unless we assume some previous accumulation of technical knowledge (so that k can be chosen from an unlimited number of techniques), we must impose the assumption of full employment on the initial conditions. Even granting an initial full employment position, satisfying the dynamic con­ dition for full employment that gN = n, Harrod’s existence problem (Harrod 1939) requires further restrictions. In the long run, the growth rate, gN = gK = sp - 5, will match the natural rate, n, only if some right-hand-side variable is allowed to change. The classical resolution associated with Kaldor (1956) and Pasinetti (1974) is to argue that the saving propensity can change, through changes in profitability. The neoclassical resolution, associated with Solow (1956) and Swan (1956), is to argue that the technique of production can change in response to movements in factor prices. The steady state technique satisfies p* = (n + S)/s. Will this work with the fossil production function? If p0 < p*, the neoclassical resolution fails to eliminate an excess supply of labor. The excess supply of labor results in wages that decline to zero. Even though profits absorb all the national income, and the profit rate achieves its maximum value (p), the system is incapable of achieving sufficient growth to maintain full employment. The Solovian mechanism, reducing the capital-labor ratio to soak up the excess supply of labor, is prohibited by the limited spectrum of techniques. Ifpo > p*, the neoclassical resolution achieves only limited success. This case presents the option of modulating the rate of growth of capital intensity by ac­ cepting new techniques with a delay. The general condition for full employment is that n — gN = gK — gk. The rate of capital deepening, gk9 will be y —x if techniques are accepted without a lag, but it can be a free variable if techniques are accepted with a time delay. Thus, the slowest growth of labor demand possible is sp0 — S

+ (r - *)• Therefore, if only p* < p0 < (n + 8 + (y — x)Vs>the economy can maintain continuous full employment through changes in the wage and the technique of production, which adjust to satisfy the familiar fundamental equation of the SolowSwan growth model, equation (13). The transitional dynamics of the Solow-Swan model with the fossil production function are similar to the phase between T2 and 7* in the classical model with exogenous technical change described above. However, if p0 > (n + 8 + (y — *))/,$, the neoclassical resolution fails to eliminate the excess demand for labor. In this case, the excess demand for labor would raise the wage to its limiting value. Even though wages absorb the entire national income, the rate of capital accumulation is undiminished. The Solovian mechanism, speeding up the rate of capital deepening, is unavailable in this case because the most capital-intensive technique available is already in use. Doubling the wage of a nineteenth-century machinist would not have caused firms to intro­ duce numerical control technology, because it had not yet been invented. The lesson here would seem to be that it takes more than variable proportions

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

55

to resolve Harrod’s existence problem. It is the lack of restrictions on the available spectrum of techniques which ensures that the demand for labor, both at its initial level and in its growth, matches the supply of labor. This was all relatively obvious to Solow (1956: 75-76), but the classical theory of the production function invites us to reconsider just how special the neoclassical production function is. With Class-Structured Saving Finally, consider a hybrid model similar to Pasinetti (1974), which consists of equations (5)-(7) plus the full employment assumption, equation (12). This is the classical model with exogenous technical change plus the assumption that the wage continuously clears the labor market, if possible. Again, we ask when this is pos­ sible. In the previous case, we found difficulties when the initial output-capital ratio was either too high (excess demand for labor) or too low (excess supply of labor). The Cambridge equation in the classical model creates a direct link between profitability and capital accumulation that is lacking in the traditional Solow-Swan model. The full employment condition is n = gN = np — 8 — j3 — gk9 where gk is determined by equation (9). If the initial output-capital ratio is too low, or p0 < n + S + /?, the distributional mechanism cannot rescue the full employment equilibrium because the required profit rate that would generate the natural rate of growth cannot be found on the growth-distribution schedule. Even if workers could “live on air,” the rate of profit would be insufficient to sustain full employment. On the other hand, the distributional mechanism rules out the possibility that the initial output-capital ratio is too high. An excess demand for labor raises the wage share and reduces the profit rate until the rate of accumulation drops to the level required by the natural rate of population growth and technical change. There are three possibilities. First, if p0 > (n + S + /? + y - *)/a, the system will begin with a profit share below the viability parameter, or tZq < a. Thus, biased technical changes will be adopted without delay, and ; r = ( n + 5 + / J + y — xVP will be maintained continuously. The profit share rises to compensate for the decline in capital pro­ ductivity. In this position, the model behaves as the classical model with exogenous technical change when it first encounters the labor constraint, from Tx to T2. Second, if (n + 5 + / J + y - * )/a < p0 < (n + 8 + /J)/a, the profit share will equal the viability parameter, or n0 = a. Thus, biased technical changes will be adopted with a delay to satisfy the fundamental equation gk = n + 8 + f l ­ ap. In this position, the model behaves as the classical model when it has joined the envelope of switch points, from T2 to 71*. In both these cases, the dynamics of the model are transitional to a steady-state equilbrium that is achieved when the output-capital ratio reaches p* = (n + 8 + p)/a. There is a third possibility in this model that does not arise with a neoclassical aggregate production function. If the initial output-capital ratio lies in the interval

56 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

(n + S + /}) < p0 < (n + S + p)/ct, the model immediately achieves a “neutral” steady-state equilibrium. The profit share that maintains full employment growth will be ;r* = (n + S + /J)/p0, and since this is greater than the viability parameter, tv* > a, technical changes will be rejected. Concluding Comments The classical model presented here draws a sharp distinction between a laborsurplus and a labor-constrained economy. It suggests that the transition between the two is likely to be dramatic, typically involving sharp declines in the profit share. A labor-constrained economy with the kind of Marx-biased technical change that generates capital deepening characteristic of significant parts of the historical record would require continuous increases in the profit share. The classical model poses the question of whether this transition has occurred, even in advanced cap­ italist countries, and provides some hints about how one might provide an answer based on the time profile of the profit share and its relationship to the viability parameter. It also invites us to probe into the sources of bias in technical change. The classical research agenda for the analysis of economic policy and other issues encompasses both the labor-constrained and labor-surplus cases. Think of the debates about Social Security in the US, for instance, where demographic con­ straints are usually projected to slow capital accumulation, and it is clear how much the classical approach broadens the discussion. The trustees of the Social Security Administration publish three projections for the next seventy-five years. Most policy discussion focuses on the intermediate projection that output will grow by less than 1.7% per year over this horizon, ignoring the more optimistic projection of 2.3% annual growth. By comparison, output growth averaged 3.3% per year over the decade from 1990 to 2000 (Social Security Administration 2001: table V.B.8). The trustees foresee that the growth rate of the labor force will decline from 1.2% per year from 1990 to 2000 to 0.3% by 2030 in the intermediate forecast and 0.5% by 2030 in the optimistic forecast. Meanwhile, they foresee that productivity growth will drop from 1.7% per year from 1990 to 2000 to 1.5% per year in the inter­ mediate scenario and increase to 1.8% per year in the optimistic scenario (Social Security Administration 2001: tables V.B.l and V.B.2). Thus, most of the projected slowdown in output growth is attributable to slower employment growth. It is hard to exaggerate the significance of these different projections. Under the intermediate projection the Social Security trust fund will be exhausted by 2038, necessitating a policy response. But under the optimistic scenario it will remain intact for the next seventy-five years and beyond. The classical model raises the question of whether these projections reflect labor constraints or capital constraints on growth. On the one hand, if labor constraints are currently binding or if a transition to the labor-constrained economy looms ahead, we should begin to see signs of the impending stagnation of the economy in prominent economic aggregates. Notably, it is hard to discern any in the New

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

57

Economy of the 1990s. On the other hand, in the labor-surplus economy the stag­ nation implied by either the intermediate or the optimistic projection can only be ascribed to changes in the model’s primitives: the productivity of capital, the profit share, or the capitalist saving propensity. In the absence of some good reason to think that the profitability of capital or capitalist propensity to invest will deterio­ rate, an analysis premised on the labor-surplus model suggests that growth at his­ torical rates will prevail, as long as reserves of labor remain available from such sources as increased labor force participation or immigration. From this perspective, even the optimistic scenario of the trustees looks unduly dour. The key point is that the classical dogma does not cultivate the habitual view that predetermined growth of the labor supply necessarily represents an ineluctable constraint on capital accumulation. This catholic perspective creates opportunities for economic research and policy analysis. Neoclassical economists, by contrast, regard the full employment of labor and other resources to be a central dogma of economic theory (conditional on constraints imposed by informational issues raised by New Keynesian theory), presumably because they envision the allocation of scarce resources to be the paramount economic problem. Their ideological hegem­ ony in public debates may help account for the widespread uncritical acceptance of the pessimism expressed in the trustees’ actuarial projections. The neoclassical production function provides structural support for the central neoclassical tenet of scarcity. And while, as has been known at least since Solow et al. (1966), the production function is not the defining feature of neoclassical growth theory, it certainly helps paper over a weakness of the classless represen­ tative agent model, which is its separation of the profitability of capital from the accumulation of capital. The classical two-class theory of accumulation reduces the range of initial parameters for which the existence of a full employment path cannot be guaranteed. Ultimately, though, the quality of intellectual debate would improve measurably with some recognition that the issue which separates neoclassical from classical theory is less one of logic than one of vision. The fossil production function is intended to provide an intuitive interpretation of the empirical patterns in the macroeconomic aggregates across countries and across time. A recent preoccupation among neoclassical economists has been rec­ onciling their theory with quantitative observations which suggest that the output elasticity of capital is about twice as large as the observed profit share.15 Of course, the classical model presented here does not consider this an anomaly since equality of the profit share and viability parameter is merely contingent. The main resolution to this puzzle among neoclassical economists has been to redefine “capital” to include not just accumulated output but also accumulated intellectual and produc­ tive skills or “human capital.” It is not yet clear that this resolution will carry the day even among neoclassical economists, but even if it does, the classical approach raises a question of whether the neoclassical production function suffers from some deeper problems that cannot be solved merely by broadening the definition of capital until it fits the empirical observations.

58

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Notes 1. We use the term “classical” in the original sense deployed by Marx, to refer to the early and seminal political economists, rather than in the later sense deployed by Keynes; but we include Marx within this broad taxonomy. Economists in the classical tradition, such as Joan Robinson, Nicholas Kaldor, and Luigi Pasinetti, have also been influenced by Keynes and Kalecki. The current chapter elaborates on Foley and Michl (1999). 2. This insight seems to have been first applied to macroeconomic theories by Sen (1963), a tradition that was continued effectively by Marglin (1984). 3. The assumption that the wage clears the labor market in this labor-constrained regime will strike some economists as lacking in classical spirit. It could be replaced by Goodwin’s (1967) assumption that the growth of wages depends upon the unemployment rate to allow for a labor-constrained regime with persistent unemployment. 4. See Michl and Foley (2003) for a classical model that nests workers who save for life-cycle reasons with capitalists who save for bequest reasons. The traditional approach among classical economists has been to adopt the Cambridge saving equation, gK = spr, where sp represents the saving propensity of capitalists or property owners. As a first ap­ proximation, there can be no objection to this approach, but we will find that a carefully specified solution to the representative capitalist’s intertemporal consumption problem offers some insight. 5. Define the Hamiltonian H = e~pt InC + p(rK — C) where p is the costate variable. Then from the first-order condition, Hc = 0, and Hp = K, and HK — —p we find that gc = gK = r - C/K = r - j3, so that C/K = p. 6. The proportional saving assumption loses some of its appeal during such a process since it is hard to defend the presumption of a constant saving rate with a declining net rate of profit. With our version of the Cambridge equation, the rate of capitalist saving out of profit income is a direct function of the net rate of profit, or 1 — /?/r. 7. For an elaboration of a classical model in the context of endogenous technical change and scarce natural resources, see Foley (2003). 8. There is a similarity here with Cobb and Douglas’s interpretation (Cobb and Douglas 1928) since they estimated the production function from historical data without any allow­ ance for technical change. The difference is that they envisioned movement along a preex­ isting function. For elaboration of the fossil production function with disembodied and embodied technical change, see Michl (1999) and Michl (2002), respectively. 9. If the utility function has a constant elasticity of intertemporal substitution, an an­ ticipated decline in the rate of profit would affect the current propensity to consume out of wealth. For example, if the elasticity is less than unity, as some empirical evidence suggests (see Barro and Sala-i-Martin 1995: ch. 2), it will lead to lower consumption. Capitalists will start reducing their consumption right away (consumption smoothing), so that the rate of accumulation will rise and the point of transition at T will be brought forward in time. An elasticity greater than unity reverses all this, and postpones the transition. 10. Call 9 = np the rate of profit with best-practice technique, and v = (1 — w/x)p the rate of profit using the existing technique. Then d9 = nxpdt and dv = (—p/x)dw or, recalling that dw = ywdt, dv = yp(n — 1)dt. Then d9 > dv => n < y/(y — x) and the statement is proved. 11. But note that starting with n > a, the profit share will fall to n = a when and if

CLASSICAL ALTERNATIVE TO THE NEOCLASSICAL MODEL

59

the economy reaches the labor constraint. At this point, the path looks much like the path between T2 and 71* described below because from t = 0 to when the labor constraint binds, a backlog of unchosen techniques has accumulated. 12. For a modem discussion of Marx’s law, see Duménil and Lévy (1994). 13. The solution to this first-order differential equation is

K = C exp

pQe*r — (Ô + ($)t.

C is a constant of integration that can be obtained from initial conditions. 14. The solution to this Bernoulli equation is

where À = n + Ô + ¡5 and 6 = M 1 — a)(T2 - t). 15. For example, see Mankiw et al. (1992). We have ignored an even more basic issue of the role of technical change and the Solow decomposition in reconciling the neoclassical model to the empirical record. For discussion and critique of the concept of total factor productivity from the perspective of the fossil production function, see Foley and Michl (1999: 164-65) and Michl (1999: 203-04). In a word, the fossil production function renders total factor productivity nugatory.

References Barro, R.J., and Sala-i-Martin, X. 1995. Economic Growth. New York: McGraw-Hill. Berman, E., Bound, J., and Machin, S. 1998. “Implications of Skill-Biased Technological Change: International Evidence.” Quarterly Journal of Economics 113: 1245-79. Cobb, C.W., and Douglas, P.H. 1928. “A Theory of Production.” American Economic Review 18: 139-65. Duménil, G., and Lévy, D. 1994. The Economics of the Profit Rate. Aldershot, UK: Edward Elgar. Foley, D.K. 2000. “Stabilization of Human Population through Increasing Returns.” Eco­ nomics Letters 68: 309-17. --------- . 2003. “Endogenous Technical Change with Externalities in a Classical Growth Model.” Journal o f Economic Behavior and Organization. 52: 1-24. Foley, D.K., and Marquetti, A. 1997. “Economic Growth from a Classical Perspective.” In Proceedings: International Colloquium on Money; Growth, Distribution and Struc­ tural Change: Contemporaneous Analysis, ed. J. Texeira. Brasilia: University of Brasilia, Department of Economics. Foley, D.K., and Michl, T.R. 1999. Growth and Distribution. Cambridge, MA: Harvard University Press. Goodwin, R.M. 1967. “A Growth Cycle.” In Socialism, Capitalism and Growth, ed. C.H. Feinstein. Cambridge: Cambridge University Press. Harrod, R. 1939. “An Essay in Dynamic Theory.” Economic Journal 49: 14-33. Inada, K.I. 1963. “On a Two-Sector Model of Economic Growth: Comments and a Gener­ alization.” Review of Economic Studies 30: 119-27.

60 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Kaldor, N. 1956. “Alternative Theories of Distribution.” Review of Economic Studies 23: 83-100. Mankiw, N.G., Romer, D., and Weil, D.N. 1992. “A Contribution to the Empirics of Eco­ nomic Growth.” Quarterly Journal of Economics 107: 407-37. Marglin, S.A. 1984. Growth, Distribution, and Prices. Cambridge, MA: Harvard University Press. Michl, T.R. 1999. “Biased Technical Change and the Aggregate Production Function.” In­ ternational Review of Applied Economics 13: 193-206. --------- . 2002. “The Fossil Production Function in a Vintage Model.” Australian Economic Papers 41: 53-68. Michl, T.R., and Foley, D.K. 2003. “Social Security in a Classical Growth Model.” Cam­ bridge Journal of Economics 27: forthcoming. Nell, E.J. 1967. “Theories of Growth and Theories of Value.” Economic Development and Cultural Change 16: 15-26. -------- . 1970. “A Note on the Cambridge Controversies in Capital Theory.” Journal of Economic Literature 8: 1245-79. --------- . 1985. “Jean-Baptiste Marglin: A Comment on ‘Growth, Distribution and Inflation.’ ” Cambridge Journal of Economics 9: 173-78. -------- . 1992. Transformational Growth and Effective Demand: Economics after the Capital Critique. New York: New York University Press. Pasinetti, L.L. 1974. Growth and Income Distribution: Essays in Economic Theory. Cam­ bridge: Cambridge University Press. Sen, A. 1963. “Neo-classical and Neo-Keynesian Theories of Distribution.” Economic Rec­ ord 39: 53-64. Smetters, K. 1999. “Ricardian Equivalence: Long-Run Leviathan.” Journal of Public Eco­ nomics 73: 395-421. Social Security Administration. 2001. Annual Report of the Board of Trustees. Washington, DC: US Government Printing Office. Solow, R.M. 1956. “A Contribution to the Theory of Economic Growth.” Quarterly Journal of Economics 70: 65-94. Solow, R.M., Tobin, J., von Weizacker, C., and Yaari, M. 1966. “Neoclassical Growth with Fixed Factor Proportions.” Review of Economic Studies 33: 79-115. Swan, T.W. 1956. “Economic Growth and Capital Accumulation.” Economic Record 32: 334-61.

4 Wealth in the Post-Keynesian Theory of Growth and Distribution Neri Salvadori

Introduction The Post-Keynesian theory of growth and distribution was first proposed by Kaldor (1955-56). Kaldor called his new theory “Keynesian,” even though, he stressed, Keynes had never developed it himself. The idea that the long-term rate of accu­ mulation determines the distribution of income can, in fact, be traced back to the “widow’s cruse” parable in Keynes’s Treatise on Money: If entrepreneurs choose to spend a portion of their profit on consumption. .. » the effect is to increase the profit on the sale of liquid consumption goods by an amount exactly equal to the amount of profits which have been thus expended.. . . Thus, however much of their profits entrepreneurs spend on consumption, the increment of wealth belonging to entrepreneurs remains the same as before. Thus profits, as a source of capital increment for entrepreneurs, are a widow’s cruse which remains undepleted however much of them may be devoted to riotous living. (Keynes, CW V,

125) In the present chapter, wealth is introduced as a determinant of saving within the Post-Keynesian theory of growth and distribution. There is a rich literature on this theory branching out in several directions, including Edward Nell’s paper “On Long-Run Equilibrium in Class Society” (Nell 1989). However, while the problem of wealth and wealth formation is inescapable, to the best of my knowledge it has never been investigated systematically or in sufficient detail. To introduce wealth in such a framework would be a pointless exercise if we had reason to presume

62

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

that it wouldn’t change much. Yet this is not so: introducing wealth can be expected to make a real difference in the argument. This has actually been shown in another context, when wealth was introduced into the original Keynesian theory of em­ ployment and gave rise to the real balance effect.1 The introduction of wealth as a determinant of saving allows also a further generalization. Fazi and Salvadori (1985; see also Salvadori 1991) generalized the threefold savings ratio model introduced by Chiang (1973) by introducing a model where workers’ saving function is nonlinear. On the contrary, capitalists’ saving was supposed to be linear since it was assumed to be a function of their profits only (and it needs to be homogeneous of degree one). The introduction of wealth as a determinant of saving allows also that the capitalists’ saving function need not be linear.2 The section on “Preliminaries” states the model and the basic assumptions. “Fur­ ther Preliminaries” introduces some formal results that will be used throughout the next section. A further section studies the case in which the rate of interest equals the rate of profit. The results obtained are similar to those of the Pasinetti model: a two-class economy may or may not exist, but if it exists the rate of profit is determined by the growth rate and the saving habits of capitalists only. The next two sections study two models in which the rate of interest paid by capitalists to workers is lower than the industrial profit rate; first, the interest rate is modeled as a function of the capitalists’ own rate of profit, and then as a function of the overall rate of profit, which is also the industrial profit rate. We then move on to consider a special model in which, as in the Kaldor model, the rate of profit is determined independently of the rate of interest, provided that the latter is smaller than the former. The final section contains some concluding remarks. Preliminaries There are two social classes, workers and capitalists. Workers’ earnings comprise wages (W) and profits (Pw) as interest on loans to capitalists. Capitalists earn only profits (Pc). Workers’ and capitalists’ savings (Sw, Sc, respectively) are defined by the functions ( 1)

(2)

where Kw is workers’ capital loaned to the capitalists and Kc is the capitalists’ own capital (Kw + Kc = K). Moreover, since money, financial assets, and land are not contemplated in this chapter, capital is the only form in which wealth exists, and therefore Kw and Kc are also the amounts of wealth owned by workers and capi­ talists, respectively. For functions (1) and (2) to hold irrespective of the numeraire chosen, in which all value magnitudes are expressed, F(-) and G(-) must be ho­ mogeneous of degree 1. It will also be assumed that

WEALTH IN THE POST-KEYNESIAN THEORY

63

(3) (4) (5)

If the derivatives mentioned in inequalities (4) are constants (which is neces­ sarily the case when wealth does not matter in determining capitalists’ saving), it is quite natural to add, among the assumptions, the inequalities

(6)

But if they are not constants, then if inequalities (6) are added among the assump­ tions, they are assumed to hold for any (Pw9 W, K J and any (Pc, Kc) in which the functions F(-) and G(-) are defined. This is quite a strong assumption, and I avoid it in this chapter. Furthermore, steady-state growth is assumed. Then workers’ and capitalists’ cap­ itals grow at the same rate (n); that is, the following constraints hold: (7) (8 )

The Post-Keynesian theory of growth and distribution also contains a techno­ logical relationship in addition to the equilibrium conditions (7) and (8). Let v E V(r, n \

0 < n, r < R

be such a technological relationship (a correspondence), where v is the capitaloutput ratio, and R is the technologically feasible maximum rate of profit. Corre­ spondence V(r, n) has the following properties: rv < 1 if v E V(r, n) and 0 < n, r < R Rv = 1 if v E V(R, n) and 0 < « < / ? . Let us briefly show how this technological relationship is built up. Let us assume that there are m commodities. For each commodity i there is at least one process (a, ef, /) that is able to produce it: the m-vector a is the material input vector, the i-th unit m-vector e, is the output vector, and the scalar / is the labor input. A collection of m processes, each producing a different commodity, is called a tech­ nique and is described by the pair (A, I), where A is the material input matrix and 1 is the labor input vector. The identity matrix I is the output matrix. If technique (A, 1) holds, commodities are consumed in proportion to vector d > 0, the growth

64 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

rate equals n ^ 0, and one unit of labor is employed, then the intensity vector x and the consumption per unit of labor c must be such that xT = cdT + (l+ /i)x TA xTl = 1. If technique (A, 1) holds, the rate of profit equals r ^ 0, and the numeraire consists of the consumption basket d, then the price vector p and the wage rate w must be such that p = (l+ r)A p + w\ dTp = 1. Hence, at the growth rate n and at the rate of profit r, the capital-output ratio relative to technique (A, 1) is

It is known (see, e.g., Kurz and Salvadori 1995: ch. 5) that for each rate of profit r such that 0 < r < R, there is a cost-minimizing technique and, as a consequence, for each r and each n such that 0 < n < r < R, there is a capitaloutput ratio v. If more than one cost-minimizing technique exists for a given r, then they share the same price vector, but in this case we have a range of v’s because cost-minimizing techniques may be combined. Further Preliminaries Let i = (P J K J be the rate of interest paid to workers on their loans to the capi­ talists, let n = (PJK() be the capitalists’ rate of profit, and let r = (Pc + Pw)l K be the overall rate of profit, which equals the rate of profit on industrial investments. If Kc > 0, a straightforward transformation of equation (8) is all that is needed to determine n since Pc = nKc\ G(nKc, Kc) = nKc. And since G(-) is homogeneous of degree 1, G( k , 1) = w, which implicitly determines the relationship between n and n: K = 0, equation (7) determines the (W /K J ratio as a function of the interest rate i and the growth rate n. In fact, from equation (7) we obtain

66 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

which implicitly defines ( WIKw) as a function of i and n. Let

(10)

be this function. By differentiating the identity F(i, f(i, n), 1) = n totally with respect to i and n, we obtain, respectively,

and from Euler’s Theorem,

Therefore

Thus, from inequalities (4), we obtain

WEALTH IN THE POST-KEYNESIAN THEORY

from which it is immediately obtained that

67

and, more precisely,

(i d

In the special case in which F(P„,, W, Kw) := spwP„ + s„,W (0 < sw s spw < 1)

The following inequalities will be used in the diagrammatic expositions used in this chapter: (12) where a(ri) is such that f(a(n), n) = 0. Since

whatever is a , a(n) is implicitly defined by the equation

Since

is an increasing function of

if and only if (13)

which is certainly the case since inequalities (5) hold and if and only if

. Similarly,

(14) which is certainly the case since inequalities (5) hold and A Rate of Interest Equal to the Capitalists9 Rate of Profit In this section it is assumed that the rate of interest i paid to workers on their loans to the capitalists is equal to the capitalists’ rate of profit n and that therefore both

68

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

equal the overall rate of profit r. If Kc > 0, equation (9) is all that is needed to determine the rate of profit: (15) The result that the rate of profit is determined by the growth rate and capitalists’ saving habits is referred to in the literature as the “Pasinetti process” or the “Pasinetti theorem” or even the “Pasinetti paradox.” It is a direct consequence of the assumptions that the rate of interest is equal to the rate of profit and that Kc > 0. l f K c > 0, equation (7) merely serves the purpose of determining the capital shares (KJK) and (KJK) via the (W /KJ ratio. In fact, from equation (10) and the fact that i = r we obtain

where v is the capital-output ratio. Hence, (KJK) < 1 if and only if (16) Furthermore, if / = r = n and Kc = 0, then equation (8) is always satisfied, and Kw = K since capitalists have disappeared. Therefore equation (7) determines a relation between v and r: (17) These results are illustrated in figure 4.1, where the overall rate of profit r is measured on the horizontal axis and the output-capital ratio (1/v) is measured on the vertical axis. The 45° line OD cuts the first quadrant in two parts: only above O D are wages positive (W > 0); along OD wages vanish (W = 0). Curve AD represents equation (17): it is nonincreasing because of inequalities (11) and cuts the 45° line OD at D = (a(n), a(n)), where a(ri) satisfies inequalities (12). Capi­ talists’ capital is positive only below curve A D because of inequality (16). Line BC represents equation (15). Steady-state growth is feasible only along either segment A D or segment BC. Taking into consideration the technological relationship between v and r, v E V(r, n), a long-run equilibrium exists whenever such a relationship cuts curve AD or segment BC. If it meets BC at C, then only capitalists earn an income. If it cuts AD (point B included), then there is a one-class, long-run equilibrium in which capitalists’ capital equals zero. A two-class, long-run equilibrium is possible only

WEALTH IN THE POST-KEYNESIAN THEORY

Figure 4.1 Conditions for Steady-State Growth when the Interest Rate Equals the Capitalists’ Rate of Profit

69

70 GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

if the technological relationship v E V(r, n) cuts the segment BC excluding the extreme points B and C. Hence a two-class economy exists at the growth rate n if and only if

For the sake of completeness, a condition for the existence of an odd number of equilibrium solutions—hence at least one— for a one-class (workers) economy should be given:

which can be written as

where F(0, l/v0, 1) is saving per unit of capital if the whole income is given to the workers as wages, and F(/?, 0, 1) is saving per unit of capital if the whole income is given to the workers as profits. The model investigated in this section exhibits several of the properties of Pasinetti’s 1962 model: a two-class economy may or may not exist, and this depends on the growth rate, workers’ and capitalists’ saving habits, and technological data; if there is a two-class economy, then the rate of profit is determined by the growth rate and capitalists’ saving habits only; but if there is a one-class economy, then the profit rate is determined by the growth rate, workers’ saving habits, and tech­ nological data. The Rate of Interest as a Function of Capitalists’ Rate of Profit In this section it is assumed that the rate of interest i paid to workers on their loans to the capitalists is a given function of the capitalists’ rate of profit n, whose derivative is smaller than 1. The assumption of a rate of interest lower than the capitalists’ rate of profit has been made by a number of authors (e.g., Laing 1969; Balestra and Baranzini 1971; Moore 1974; Pasinetti 1974, 1983; Gupta 1976; Fazi and Salvadori 1981, 1985; Salvadori 1991). If capitalists exist (Kc > 0), then n is determined by equation (9). If capitalists do not exist (Kc — 0), then n is both irrelevant and undetermined. If the rate of interest is a function of n, then it can be considered given once n is determined. Clearly, i < n.

WEALTH IN THE POST-KEYNESIAN THEORY

71

From equation (10) we obtain

(18)

By definition the overall rate of profit is a weighted average of i and n,

that is,

(19)

which has the remarkable property of being linear in r. Moreover, from equation (18) we get that (KJK) = 1 if and only if (20) It is worth noting that (i < (pin))3, (i) (ii) The above analysis is presented graphically in figure 4.2, where the curve AD and the segment BC of figure 4.1 are also drawn. The line EC is defined by equation (19). Segment EC cuts the 45° line at point C, where r = 0 implies g(r, n) > f(r, n) (since f(r, n) is decreasing in r), (because of inequalities (11)), (because of inequalities (11)).

74

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

Furthermore,4 (i) p(0, n) = f(0, n)\ (ii) p( 0, then we obtain from the equilibrium equations (7) and (8) that

whereas if Kc = 0, equation (8) is not a constraint and the rate of profit is deter­ mined by equation (7) with Kw = K. In order to prove that the reverse is not true in general, let us provide an example in which the above statement (1) holds, but function F(-) does not have the form (25). Let

where sc is a constant (0
0; it is homogeneous of degree zero (as a consequence F(Pw, W, Kw) is homogeneous of degree 1); and inequalities (3) are satisfied provided that y > 1 and 8 is close enough to zero. Similarly, it is recognized that function F ^ , W, K J cannot be written in the form (25) unless 5 = 0 , which is not the case. However, from equation (7) we obtain

78

GROWTH, DISTRIBUTION, AND TECHNICAL CHANGE

which, inserted into equation (19) (or into equation (17) for i = r, or into equation (23) for i = h(r)) gives

which is independent of i and, as a consequence, statement (1) above holds. State­ ment (2) also holds. In fact, if i = r = n and Kc > 0, then W = 0. Finally, I show that if wealth does not matter in determining savings, that is, if (27) then if statement (1) above holds, the saving function F(-) has the form (25). First of all, by exploring the section “Further Preliminaries,” we obtain that if identities (27) hold, then both functions 0, and b, = kb > 0.

(5a)

130 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

Figure 8.1 Monotonie Convergence in the Neoclassical Labor Market*1

This simple linear first-order system is monotonically stable around w* = a,/&,, which from equation (3a) implies v* = a — bw* = a - b (a jb x) = a - b[k(a 1)fkb] = 1 (full employment).4 Figure 8.1 illustrates the adjustment process from an initial state of excess demand. But when we instead use the second wage adjustment function, equation (4b), we get quite a different picture. Differentiating equation (4b) and substituting it into equation (3a) gives w" = a, — bxw, where a v b x are as defined previously

(5b)

Equation (5b) also has an equilibrium at w* = a jb x and v* = 1 (full employment). This particular dynamic equation is known as a harmonic oscillator (Hirsch and Smale 1974: 15), and it has the property that the actual levels of w and v oscillate endlessly around their equilibrium values with possibly substantial fluctuations. Figure 8.2 illustrates this second adjustment process. As we can see, the mere existence of a full-employment equilibrium does not imply that the system will come to rest at this point. It may instead over- and undershoot it endlessly.5 The foregoing brings out two critical features of the static neoclassical labor market story. First of all, the assumption that the real wage responds solely to the

LABOR MARKET DYNAMICS

131

Figure 8.2 Harmonic Oscillation in the Neoclassical Labor Market

excess demand for labor, as in equations (4a)-(4b), implies that the real wage is represented solely as a market-clearing price, not a socially determined variable.6 Second, in this formulation the equilibrium real wage is independent of social forces. It is determined solely by the technology (through the marginal productivity of labor, which determines labor demand) and by exogenously given household preferences about work and leisure (which determine the supply of labor). It is of course true that the interventions of unions and of the welfare state may push the real wage above its putative equilibrium level, thereby giving rise to unemployment. But these would be disequilibrium phenomena. The equilibrium real wage and employment levels are purely psychotechnical. The equilibrium level of employ­ ment in turn determines a particular level of output, and hence productivity of labor, via the aggregate production function (Godley and Shaikh 2002: 426-28). It follows that the wage share, the ratio of the real wage to productivity, is determined entirely by technical and psychological structures. There is no room for unions and the state within this story, except of course to prevent equilibration.7 This is most obvious in the ubiquitous Cobb-Douglas production function, in which the equi­ librium wage share is equal to the labor elasticity parameter of the production function.8 Neoclassical growth dynamics extends this story to allow for population growth

132 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

and technical change. The labor market is assumed to be in equilibrium at all times, but now the real wage, productivity, and the capital-labor ratio all grow in response to population growth and technical change. These latter factors now also influence the equilibrium levels of the real wage and wage share, but once again they are determined independently of any direct struggle over wages. In the case of an aggregate Cobb-Douglas production function undergoing neutral technical change, the wage share continues to be directly determined by the function’s labor param­ eter, which is independent of social forces (although it may change as technology changes). Nonetheless, we have seen that the existence of a stable equilibrium does not imply that the wages, productivity, the capital-labor ratio, and even the wage share are actually at their equilibrium values. Quite independently of any social forces, the internal dynamics of the adjustment process may lead them to fluctuate endlessly around their equilibrium values. Thus even within the internal logic o f the neoclassical representation o f the labor market, we cannot thereby take ob­ served values o f variables to be the same as equilibrium values.9 Labor Market Dynamics within Standard Keynesian Macroeconomics Within the standard Keynesian model, the variable v stands for the employment rate (the ratio of actual employment to available labor), and 1 -v represents the unemployment rate. The basic argument is best approached by combining the ex­ pressions in equations (1) and (2) as v = LIN = YKyN) < 1.

(2a)

In the static case, productivity y and labor supply N are given, so the employment rate varies solely with output Y. This in turn is said to be directly determined by demand Z, which in the simplest case is a multiple of autonomous demand A = / + G = investment + government spending. Y = Z [short-run equilibrium] Z = A/ a — (/ + G)/(7 [multiplier] v = YIN — A /{a yN ),

(6) (7) (8)

where < 7 = 5 + ¿(1—5) = the private propensity to save + the tax rate = the “leakage rate.”10 Within this framework, fiscal policy (G, t) plays a central role, for if autonomous investment is insufficient to generate something close to fullemployment share (v 1), then some combination of a higher G or lower t is called for. What of the distribution of income? Keynesian theory usually insists that wage bargains are made in money terms, and that prices are set as fixed markups on unit costs. Both wages and prices are often taken to be “sticky” in the short run, by

LABOR MARKET DYNAMICS

133

which it is generally meant that they do not immediately respond to unemployment. More important, it has been argued that fixed markups imply that prices rise in the same proportion as money wages (Sawyer 1985: 117-18; Asimakopulos 1991: 29).n This would imply that even if money wages were to respond to unemploy­ ment (at least at some point), real wages would nonetheless remain unchanged. However, if this were so, then real wages, and hence the wage share (in the present static case), would also be utterly impervious to social and institutional pressures. But the logic of the Keynesian argument does not actually imply that real wages are impervious to unemployment. Indeed, Keynes himself conceded that persistent unemployment would erode not only money wages but also real wages (Bhattacharyea 1987: 276-79). The debate about the “stickiness” (nonlinearity) of the realwage response to unemployment is not equivalent to a debate about the direction of the response. The next two equations show why. Recall that W = the money wage and P = the price level, so that the real wage w = W/P . Equation (9) says that the money wage rises when the level of employment is above some threshold, and falls in the opposite case. W /W = f(v v0),

(9)

where v0 is some threshold level of employment. The parameters that determine the level and steepness of this function may then be taken to represent the strength of social pressures on the money wage. And equation (10) shows that if prices are set as fixed markups on costs, they will change less than money wages because some part of costs is independent of money wages.12 P IP =

k(WVW),

(10)

where k < 1 is the share of wage costs in wages plus fixed costs. This makes it evident that even when workers bargain in terms of money wages and firms set prices by fixed markups, if money wages respond at some point to (un)employment, then so will real wages. The response may be slow and socially painful, as Keynes argued, but it will be inevitable. Since the real wage w — W/P , we can write w'/w = F (v—v0),

(11)

where F(v—v0) = (1 —x)*f(v—v0), and K < 1, as previously noted. Equation (11) is really a real-wage Phillips curve. The question now arises: What impact might a change in real wages have on the Keynesian story about employment? Note that in the static case, productivity (y) is given, so the wage share u = xv/y moves with the real wage, and the profit share (1—u) and profit rate move inversely to it. Then there are two possible channels discussed in the Keynesian literature, both of which lead to the same conclusion. The first of these is the familiar Kaldor-Pasinetti

134

MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

linkage between the private savings rate and the division between wage and profit share. Let real total savings S = savings out of wages + savings out of profits = swwL + sJJ, where sw = the propensity to save out of wages, sK = the propensity to save out of profits, and 77 = total real profits. Since total output Y = wL + 77, we can write s = S/Y = snu + sn( \ - u ) = the private savings rate,

(12)

where u = the wage share = wLIY = w/y, and (1-w) = the profit share = THY = (Y — wL)/Y. If the propensity to save out of wages (sw) is lower than that out of profits ( s j, a fall in the wage share u will shift the division of income in favor of profits, thereby raising the average private savings rate s. To the extent that tax rates are also higher for profit income, the average tax rate t also will move in the same direction. The “leakage rate” share a will therefore rise, and as is evident from equations (6)-(8), this will lower demand, output, and the employment rate, other things being equal.13 But other things will not remain equal, because lowered output implies lowered capacity utilization, which then operates through the second channel to undermine investment, which in turn further lowers output and employ­ ment. Both channels therefore affect the employment rate in the same direction: a drop in the employment rate that is sufficient to lower real wages will spark further drops in employment, and so on.14 A rise in employment would obviously have the opposite effect. This is where the “stickiness” of nominal wages becomes crucial, because it translates into real wage stickiness. Insofar as the distribution of income does not respond, the static system remains stable. But if employment (unemployment) changes are strong enough to trigger nominal wage changes, then the static Keynes­ ian model is knife-edge unstable— toward depression on one side, toward infla­ tionary full-employment on the other. The problem may be put another way. The Keynesian model implicitly relies on the presence of some unstated automatic mechanism that stabilizes the distri­ bution of income. Such a mechanism would have to be substantially independent of social and institutional forces, because if it were not, then the change in the wage share would trigger knife-edge instability. In this way we once again arrive at the conclusion that the static Keynesian model, like its neoclassical counterpart, implies that the wage share is independent of social forces. The last step is to consider the growth dynamics of the Keynesian model. Al­ lowing for changes in variables over time, we can write equation (8) as

v(o = Am oym m

m

where autonomous demand A(t) = 7(t) + G(t) = investment + government spend­ ing; share a — s 4- t ( l —s) where s and t are savings and tax rates, respectively;

LABOR MARKET DYNAMICS

135

and y(t) is the productivity of labor. Population growth and technical change will persistently raise N(t) and y(t), which will tend to erode employment. But even if autonomous demand A(t) is growing, there is no particular reason why the growth in its two autonomous components should precisely offset the growth in population and productivity. The general imbalance between the two sets of growth rates will then make the employment rate v persistently rise or fall. In the absence of some feedback between v and the other variables, the Keynesian growth model is unstable. We have seen that a changing employment ratio is likely to change money wages, at least at some point. If fixed markup pricing were indeed to lead to equiproportional changes in prices, then real wages would be unaffected and em­ ployment would be unstable. Unfortunately, a flexible real wage turns out to make matters even worse. Even in the dynamic case with technical change, a fall in the employment ratio will lower the wage share. As previously, this would raise the leakage rate share G and exacerbate the problem of a falling employment rate v. This is the Keynesian paradox of thrift once again, this time in a growth context. Under the standard Keynesian assumptions of exogenous technical change and au­ tonomous investment, the problem of labor market instability therefore seems intractable. This is the point at which the difference between the Keynesian and Harrodian frameworks becomes decisive. In both cases, when the unemployment rate is above some critical level, the real wage falls. This leads to a rise in the average savings rate. In the Keynesian case, a rise in the savings rate reduces the level of output by reducing the multiplier, and hence further worsens the employment situation. In the Harrodian case, the very same rise in the savings rate raises the long-term (warranted) rate of growth, which improves employment. Thus, whereas the de­ pendence of the savings rate on the distribution of income destabilizes the em­ ployment rate in the Keynesian model, it stabilizes it in the Harrodian one. As we shall see, the crucial difference in the two results stems from a critical difference in their analysis of investment. Labor Market Dynamics within the Harrodian Tradition The difference between Keynesian and Harrodian treatments of effective demand is best understood by considering their common starting point: the simple multiplier relation (i.e., with balanced budgets and balanced foreign trade). Yt = IJs.

(13)

Keynesian economics portrays investment (7t) as “autonomous” in the short run, in the sense that it is independent of current outcomes. From this point of view, investment is the proximate “cause” of output, via the multiplier. Harrod’s point is that this conception of investment contains a fundamental inconsistency. The mul­

136 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

tiplier effect of investment, he notes, is only half of the story. The very purpose of investment is to expand capacity, and this requires not only the anticipation of demand but an evaluation of the utilization of existing capacity. For investment to be self-consistent, the two aspects must mesh. It follows that the investment path is endogenous, not exogenous as the Keynesians would have it. These considerations led Harrod to derive the self-consistent path of investment, which he calls the “warranted path.” If Yc = capacity output, then R = YJK = capacity-capital ratio, which Harrod takes to be constant over time (Harrod-neutral technical change). Dividing both sides of the multiplier relation in equation (13) by K, and noting that UK = K /K = gK = the rate of growth of capital and vc = YIYC = the capacity utilization rate (not to be confused with the wage share w), we get vc = gA sR X

(13a)

Equation (13a) is merely another way of expressing the multiplier relation, and it tells us that in short-run equilibrium the actual rate of capacity utilization will depend on how close the rate of growth of capital is to sR. Alternately, it tells us that only when gK = sR will the capacity created by investment match the demand induced by investment spending. Only then will capacity be fully utilized so that vc = l . 15 Thus the “warranted” rate of capital accumulation is given by gw K = sR-

(14)

It is at this point that the labor market enters into the picture.16 From equation (la) the employment rate v = K/(kN). Taking rates of change, defining gK = K/K, gk = k'/k, and gn — N'/N, and noting that equation (14) implies that gK = sR along the warranted path, we get the fundamental Harrodian employment dynamic:17 v’/v = gK - (gk + g„) = SR - (gt + g j .

(15)

This tells us that the warranted rate of employment (unemployment) will be chang­ ing continuously whenever the warranted rate (sR) is not equal to what Harrod calls the “natural rate” (gk + gn). But, as he points out, if s is exogenously given by savings habits, gn isgiven by populationcharacteristics, and R and gkare given by technical change, there is no mechanism toclose any gaps andhence prevent v from rising or falling to its limits. It would appear, then, that the employment rate is inherently unstable. It is here that the dependence of the average saving rate on the distribution of income, which played a destabilizing role in the static Keynesian model, now plays a stabilizing role in the Harrodian employment dynamic. We saw in equation (12) that the average savings rate is a negative function of the wage share u: s — s(u) such that s' < 0. From equation (15), whenever the warranted rate of growth sR is

LABOR MARKET DYNAMICS

137

less than the natural rate of growth (gk + gn), then v'/v < 0 and the employment rate v will start to fall (unemployment will rise). Therefore the wage share will also start to fall,18 which will in turn make the average savings rate s rise. Since the natural rate of growth (gk + gn) is given, the rise in the savings rate will reduce the initial gap between the warranted and natural rates. This process will continue until the gap is closed and the employment rate is stabilized. The preceding Harrodian dynamic has a very powerful implication, namely, that there is only one wage share that will stabilize the employment (unemployment) rate (i.e., make v'/v = 0). Since the savings rate s(u) is a monotonic function of the wage share, and R, gk, and gn are all exogenously given, there is only one wage share that will suffice. Moreover, what was implicit in the Keynesian argument now becomes explicit: the requisite wage share is completely independent o f worker strength, because it is completely determined by savings propensities, technology, and population growth. Labor Market Dynamics within the Marx-Goodwin Model The Harrodian analysis of the labor market relies on the notion that a fall in the employment rate v will undermine the wage share u. But it is the direction of response that is central to that discussion. The actual path, and its implications, are not addressed. We do not know, for instance, whether or not the adjustment process leads the economy to full employment. Nor do we know whether we end up at the long-run employment rate and corresponding wage share, or merely oscillate around them as in figure 8.2. It was Goodwin’s contribution to take up the latter two issues in his elegant formalization of Marx’s notion of a reserve army of labor (endogenous rate of unemployment).19 He accomplishes this by combining the real-wage Phillips curve that is implicit in the Keynesian argument (equation (11)), the explicit KaldorPasinetti dependence of the savings rate on the wage share (equation (12)), and the employment dynamics implicit in the Harrodian argument (equation (15)). These three equations, which are reproduced below, constitute the basic structure of the Marx-Goodwin model. w'lw = F(v—v0), where v0 = some threshold rate of employment s = S/Y = swu + sn{1 - u) = s(w) = the private savings rate v’/v = gK - (gt + gn)

(11) (12) (15)

Goodwin directly adopts three central assumptions of the Harrodian formulation: that the economy is on the warranted path, so that the actual rate of accumulation gK equals the warranted rate of growth sR; that the natural rate is constant (gk + gn) because the rates of technical change and population growth are constant; and that the capacity-capital ratio R is constant over time (Harrod-neutral technical change). Since output is equal to capacity along the warranted path, R = Y/K =

138 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

ylk, which in turn implies that gy = y'/y = gk = k'/k, both of which are also constant. This allows us to transform the real-wage reaction function in equation (11) into a wage-share reaction function, since u = w/y implies u'/u = w'/w — gy. Finally, Goodwin’s original formulation contains three specific simplifications, which al­ though they are not essential for the general results, we retain in order to reproduce Goodwin’s original equation system. These are that the wage reaction function is linear, that workers do not save (sw = 0), and that capitalists save everything (s„ = 1). Goodwin’s nonlinear dynamical system is therefore given by u'/u = h (v -v 0) ~ gy v7v = (1 " u)R - (gy + gn).

(16) (17)

This 2X2 nonlinear differential equation system is known as the Lotka-Volterra “predator-prey” system. In the first equation, the parameter v0 is the threshold rate of employment that triggers real-wage increases, and the parameter h is the sensitivity of the wage share to disequilibrium in the labor market. Both of these may be interpreted as aspects of labor strength. Note that v0 < 1 implies that workers are strong enough to begin raising real wages even while there is some unemployment. Therefore a lower v0 constitutes greater worker strength, as does a higher h. Goodwin’s model has four properties that are relevant to our present discussion. First, as in the modified neoclassical wage adjustment function of equations (4b)(5b) and figure 8.2, the Goodwin model yields a perpetual oscillation around its equilibrium points.20 Second, as in Harrod, the equilibrium wage share is com­ pletely independent of “class struggle.” This follows from the Harrodian employ­ ment dynamic in equation (17), since v'/v = 0 implies a particular wage share w* = 1 — (gy + gn)/R in which neither of the labor strength parameters (v0, h) appears. Third, equilibrium in the labor market will generally yield some persistent rate of unemployment, since u'/u = 0 implies v* = v0 + (gyJh\ and this can be less than 1 (but not above it because v = 1 represents actual full employment). Finally, while labor strength does not affect the equilibrium wage share w*, it does affect the equilibrium employment rate v*. Unfortunately, the effects of greater labor strength are unambiguously negative: a rise in labor strength (a fall in v0 and/or rise in h) will lead to higher equilibrium unemployment. Given that Goodwin’s model is an attempt to formalize Marx’s arguments about labor market dynamics, it is particularly striking that it leads to the conclusion that “class struggle” over wages would not only be completely ineffective in changing the rate of surplus value, but would also harm employment conditions. It should be noted that these conclusions do not arise from the simplifying assumptions of Goodwin’s original model, but are rather implicit in both Keynesian and Harrodian formulations also.

LABOR MARKET DYNAMICS

139

Summary and Conclusions This chapter has attempted to analyze the manner in which alternative macroeco­ nomic frameworks portray the dynamics of the labor market. Two types of dynam­ ics have been of interest, both of which depend upon the mutual interactions between the wage share and the employment rate. In disequilibrium dynamics, the issue is the manner in which these variables respond to imbalances in the labor market, while in growth dynamics the issue is their response to technical change and growth in labor supply. We examined the basic neoclassical, Keynesian, Harrodian, and Marx-Goodwin models, since each embodies a particular approach to macroeconomics. Dynamics require explicit analysis of stability of various equilibria. But even the existence of a particular stable equilibrium need not imply that the economy will be at or even near that point. The analysis of the neoclassical model demon­ strates that if real wages respond to the current excess demand for labor, then the labor market converges to a particular wage at full employment (figure 8.1). But if real wages respond to the cumulative excess demand for labor, then the labor market would exhibit endless and possibly large fluctuations in real wages and excess labor demand, around but not at, the equilibrium real wage and full em­ ployment (figure 8.2). This second type of response is reminiscent of Goodwin’s elegant representation of Marx’s argument about the reserve army of labor, except that in his model the center of gravity is a persistent level of unemployment, not full employment. In any case, this type of disequilibrium dynamic reminds us that we should be careful to distinguish between equilibrating paths and equilibrium points. At an empirical level, this cautions us not to confuse observed variables with their putative equilibrium levels. In the case of growth dynamics, a second type of finding emerges. It turns out that in each of the four macroeconomic approaches, the paradigmatic case is one in which the organizational or institutional strength of labor has no influence what­ soever on the path of real wages and on the level of the wage share. In all of the approaches, it is technical factors and labor supply growth that determine the stan­ dard of living of workers. The degree of labor strength in the struggle over wages has no effect at all. In the neoclassical case, this is instanced by the ubiquitous Cobb-Douglas production function, in which the labor elasticity parameter directly determines the wage share. Hence the profit-wage ratio is determined entirely by production conditions. In the standard Keynesian case, the corresponding outcome arises from markup pricing, in which changes in money wages are said to cause equiproportional price changes. This not only leaves the real wage unchanged, but also implies that it is unchangeable. In the Harrodian framework, unemployment affects the wage share, which in turn affects the warranted rate of growth via the dependence of the savings rate on the wage share, a la Kaldor and Pasinetti. This feedback loop leads the system to stabilize around full employment in the long

140 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

term. But it also implies that the wage share is completely determined by the rates of technical change and population growth, completely independent of labor strength. Finally, even in Goodwin’s classic formalization of Marx’s theory of the reserve army of labor, “class struggle” over wages has no effect whatsoever on the rate of surplus value. Indeed, greater labor strength would only serve to increase the long-run equilibrium rate of unemployment. This is a particularly unkind cut for a Marxian model. Two critical questions are raised by the general theoretical finding that wage shares are independent of labor strength. First of all, is it at all empirically plau­ sible? The stability of wage shares is a well-known “stylized fact.” But so are differences between wage shares across nations and across levels of development. Are these differences reducible to those arising solely from technical factors and conditions of labor supply? Alternately, if social forces do indeed influence the wage share, how might such a mechanism operate? The key expression to consider is equation (15), in which the rate of change of the employment ratio depends solely on two critical variables: the rate of accumulation gK = s(u)R and the rate of mechanization gk, assuming that the rate of growth of the labor supply gn is exogenous. v’/v = gK - (gt + gn) = s(u)R ~ (gk + gn)

(15)

We saw that if the output-capital ratio R and the mechanization rate gk are exog­ enously given, then there is only one wage share, u = w*, consistent with a stable employment rate (i.e., with v’/v = 0). But this conclusion would not be altered if R and gk, and indeed even gn, were to also depend on the wage share.21 What is needed, therefore, is some other mode of feedback between the employment rate and one of these variables. A particularly simple one is to suppose that the rate of mechanization depends not only on the wage share (i.e., indirectly on the employ­ ment rate through its effect on the relative cost of labor) but also directly on the employment rate (i.e., directly on the relative availability of labor). Rowthom (1984: 203-5) notes that this is precisely the argument in Marx.22 Then gk = f( m,v), and v’/v = gK - (gk + gn) = s(u)R - [g*(w,v)+ gni

(15a)

The results of this apparently minor extension are dramatic. Suppose we consider the extreme case in which the wage share is now entirely determined by “class struggle,” so that u = w0. Then if v'/v > 0 initially, the employment rate v will rise, which will raise the mechanization rate gk(u0, v), thereby bringing the em­ ployment rate back into balance. It follows that the same result would also obtain if we assume that the wage share depends on both “class struggle” and the em­ ployment rate. Thus the preceding simple modification completely reverses the general theoretical conclusion that the wage share is independent of labor

LABOR MARKET DYNAMICS

141

strength, for now there is plenty of room for the influence of the relative strength of labor. Notes 1. Strictly speaking, we should also distinguish between virtual and actual magnitudes of Y, W, P, etc. But this leads into the issues of expectation formation and adjustment, which are secondary to our present concerns. 2. Such a linear function can come about as the actual or approximate ratio of nonlinear labor demand and supply functions. 3. The assumption that a > 1 ensures that the lowest possible wage, w = 0, corresponds to a positive excess demand for labor. This way, as w rises, v falls, so that full employment (v = 1) corresponds to some positive level of w. 4. We can rewrite equation (5a) in the form w' = 6,(w* — w), in which case it is clear that if w > w*, w' < 0 and w declines steadily until w = w*. Conversely, if w < w*, w' > 0, and w rises steadily until w = w*. 5. We could of course combine the two adjustment processes in equations (4a)-(4b), in which case the system will exhibit oscillatory convergence. Adding random shocks to this process will then result in perpetual erratic oscillations around full employment and a corresponding real wage. 6. This is a direct consequence of the Walrasian assumption that each potential worker expects to be able to sell as much labor as he or she would like. The influence of (expected) demand is therefore eliminated from the start. 7. We could of course create some room for social determination by allowing the house­ hold preference structure to respond to politics and institutions. But this would take us outside the standard framework of this school. 8. The Cobb-Douglas production function is of the form Y = AKpL l~p. This can also be written in per-unit-of-labor form as YIL = y = Akp, where k = K/L = the capital-labor ratio. The marginal product of labor MPL is the partial derivative of Y with respect to L, and through perfect competition this is set equal to the real wage w: MPL = (1—j3)AKPL~^ = (1 —j3)Akp = (1—P)y = w. Thus the wage share u = w/y = (1 —/?), where (1—j3) is a technological parameter representing the partial elasticity of output with respect to labor. 9. There is, in addition, a separate question of whether the neoclassical growth model would indeed be stable in the face of real-wage adjustment processes such as that in equa­ tions (4a)-(4b). 10. In the standard derivation, Z = C + / + G, where here consumption C = c(Y — 7), taxes T = tY , and I and G are exogenous in the short run. The assumption of short-run equilibrium Y = Z then implies that Y = c(Y — T) + / + G, so that (1 - c)(Y — T) + T = i(y - tY) + tY = [s + t( 1 - s)]Y = I + G, where s = 1 - c = the private savings rate and t = the tax rate. Since both s and t are leakages from expenditures, share a = [s + t( l—5)] may be termed the “leakage rate.” 11. In a pure circulating capital model, if all prices are constructed from fixed markups on costs, then all costs can be resolved directly or indirectly into wage costs. It follows that if markups are held constant, prices will change in the same proportion as money wages. 12. If we define prices as fixed markups on unit costs, then P = (1 +ju)*(a0 + Pm +

142 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

Wl), where ¡a = the fixed markup on unit costs, a0 = the autonomous component of unit costs (such as fixed costs and costs of imports), m = materials used per unit output, and / = labor used per unit output. This gives us the expression P [l —(1 + jn)m] = (1 4- jj)(a0 + Wl), and differentiating this yields P '[ l- ( 1 + fl)m] = W( 1 + ju)/. Dividing the latter relation by the one preceding it and simplifying gives us PIP = k(W/W), where k = (Wl)/ (aQ+ Wl) = the share of wage costs in nonmaterial costs. 13. This is a version of the Keynesian paradox of thrift, in which a higher savings rate lowers the level of employment (Foley and Michl 1999: 185-86, 189). 14. One might add that the rise in potential profitability consequent on a fall in real wages might stimulate investment, and hence counteract the other effects. Keynesian eco­ nomics recognizes that investment depends on both the marginal efficiency of investment (the potential rate of return on new investment) and the rate of interest (the opportunity cost of new investment). But it tends to require both being determined elsewhere in the system, and hence ignores this potential stabilizing reaction (Rogers 1989: 260-61; Panico 1988: 181-90). 15. Capacity represents economic capacity, not engineering capacity. Thus capacity is fully utilized when it is at the most profitable point of utilization, which includes the optimal amount of reserve capacity needed to meet the demands of business and fend off competitors. A firm has excess capacity when its utilization is below this point, and has a deficiency of capacity when it is above this point. Either instance will provoke a response in investment plans. 16. A separate issue has to do with the apparent instability of the Harrodian warranted path. This path is in fact quite stable (Shaikh 1989, 1991). We will not pursue that question here. 17. Since Y, K, and Yc all grow at the same rate along the warranted path, gK = gY. And since Y = Yc along the warranted path, R = YJK = Y/K = y/k. Then the assumed constancy of R (Harrod-neutral technical change) implies that gk = gy. With these substitutions, equa­ tion (13) can be written in the more familiar Harrodian form v'/v = gY — (gy + gn) = sR - (gy + £„)'

18. If the real-wage Phillips curve of equation (11) is expressed in linearized form, w'l w = h(v — v0) = —hv0 + hv. This is the form used by Goodwin (1967), and it implies that the rate of change of the wage share u = w/y is given by u'lu = Wlw — y'/y = ~(hv0 + gy) + hv. Thus the wage share will rise once the employment rate has exceeded the threshold (v0 + gjh). 19. Solow (1990: 35-36) justly observes that the Goodwin model is a “beautiful paper” which “does its business clearly and forcefully.” 20. This oscillation is of a somewhat different character, though, since this equilibrium point of the Goodwin model is a quasi-stable center. 21. If the latter relations were nonlinear, it might be true that there would be more than one wage share which might work. But even so, none of these would be dependent on labor strength, for the same reasons as previously. 22. Rowthom (1984: 204) points to Marx’s “often expressed and often cited view that capital can always overcome labour shortages by adapting its rhythm of work and methods of production . . . [thus] shortages of labour. . . can eventually be overcome by reorganizing methods of production or mechanizing or redesigning the work process . . . given time, cap­ ital can adapt itself to whatever supplies of labour are available.”

LABOR MARKET DYNAMICS

143

References Asimakopulos, A. 1991. Keynes's General Theory and Accumulation. Cambridge: Cam­ bridge University Press. Bhattacharyea, A. 1987. “Keynes and the Long-Period Theory of Employment: A Note.” Cambridge Journal o f Economics 11: 275-84. Foley, D.K., and Michl, T.R. 1999. Growth and Distribution. Cambridge, MA: Harvard University Press. Godley, W., and Shaikh, A. 2002. “An Important Inconsistency at the Heart of the Standard Macroeconomic Model.” Journal o f Post Keynesian Economics 24: 423-41. Goodwin, R.M. 1967. “A Growth Cycle.” In Socialism, Capitalism and Economic Growth: Essays Presented to Maurice Dobb, ed. C.H. Feinstein. Cambridge: Cambridge Uni­ versity Press. Hirsch, M., and Smale, S. 1974. Differential Equations, Dynamical Systems and Linear Algebra. New York: Academic Press. Panico, C. 1988. Interest and Profit in the Theories o f Value and Distribution. New York: St. Martin’s Press. Rogers, C. 1989. Money, Interest, and Capital: A Study in the Foundations of Monetary Theory. Cambridge: Cambridge University Press. Rowthom, B. 1984. Capitalism, Conflict, and Inflation: Essays in Political Economy. Lon­ don: Lawrence & Wishart. Sawyer, M.C. 1985. The Economics o f Michal Kalecki. Armonk, NY: M.E. Sharpe. Shaikh, A. 1989. “Accumulation, Finance, and Effective Demand in Marx, Keynes, and Kalecki.” In Financial Dynamics and Business Cycles: New Prospects, ed. W. Semmler. Armonk, NY: M.E. Sharpe. --------- . 1991. “Wandering Around the Warranted Path: Dynamic Nonlinear Solutions to the Harrodian Knife-Edge.” In Kaldor and Mainstream Economics: Confrontation or Convergence, ed. E.J. Nell and W. Semmler. London: Macmillan. Solow, R.M. 1990. “Goodwin’s Growth Cycle: Reminiscence and Rumination.” In Nonlinear and Multisectoral Macrodynamics: Essays in Honour of Richard Goodwin, ed. K. Vellupillai. New York: New York University Press.

9 Has the Long-Run Phillips Curve Turned Horizontal? Craig Freedman, G.C. Harcourt, and Peter Kriesler

Any statistical relationship will break down as soon as it is relied upon for policy purposes. —Charles Goodhart, articulating Goodhart’s Law1 We all look for patterns to make sense out of life. Economists are no different in this respect from anyone else. It is the location of that search rather than the process itself that distinguishes them from other sifters of data. The danger is that the need to order that environment may lead to creating and preserving tools well past their usefulness. Even worse, we may come to believe that our own constructs, these very same tools, represent some inherent and invariant natural relationship. By doing so, we blind ourselves to the ever changing pattern of economic relationships. Starting in the 1970s, most developed countries experienced a rising trend of inflation. After each economic downturn, inflation remained higher than at its pre­ vious trough. In contrast, the 1980s saw an ever rising trend of unemployment in those same countries. After each upturn, unemployment was higher than at the previous economic peak (OECD data, 1970-90). Moreover, the long-term portion of this rising unemployment was itself increasing (figure 9.1). It is interesting that in the US, where monetary policy over the relevant period was generally less stringent than in corresponding European countries, the growth of the long-term unemployed was far less dramatic. At its height, long-term unemployment in Italy accounted for nearly 70 percent of total unemployment (The Economist, December 2, 1995: 85). This should arouse the interest of any economist. Judging from the literature that is rapidly accumulating on this topic, many economists have been intrigued

HAS LONG-RUN PHILLIPS CURVE TURNED HORIZONTAL?

145

Figure 9.1 Long-Term Unemployment in the US as a Percentage of Total Unemploy­ ment (1965-97)

by these two disparate trends. A shift this fundamental must reflect significant changes in some underlying essential causal factor. If we examine the two decades from 1970 to 1990, we see a clearly discernible shift in macroeconomic policy. Monetary authorities in OECD countries moved from targeting unemployment through expansionary fiscal policy to constricting accelerating inflation by means of restrictive monetary policy. It is our thesis that the observed trend alterations in both unemployment and inflation are connected to the concomitant policy switch. We can look to the cause of the former in the latter decision. Macroeconomic policy generates and changes underlying economic relationships in ways that we only selectively understand. The profession seems to relearn the validity of this propo­ sition at periodic points of theoretical disarray. Keynesian economics came to grief when its theory, ballasted by the inflationunemployment trade-off articulated in the Samuelson-Solow (1960) Phillips curve model, proved to be floated more by the desires of its fabricators than by any empirical sea of data. The “menu choice” that economists promised to policymakers proved to be ephemeral. The optimism underlying the model was superseded by the pessimism represented by its long-run relative. Our claim is that the expectations-augmented Phillips curve, the standard model in the economics pro­ fession since the 1980s, has in turn become obsolete for the very same reason that the simple Phillips curve was vanquished. The parallels here are intriguing. In both cases the generators and defenders of these models insisted that they had specified an inherent natural relationship. A heuristic model was elevated to an undeserved

146 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

level of generality. In a sense, belief in the generality of the model destroyed its applicability.2 The trade-off between unemployment and inflation depended on assuming re­ versibility as a defining characteristic of the Phillips curve.3 This is what Friedman (1968) and Phelps (1967) successfully undermined. They failed to realize that the model they devised to replace the garden variety Phillips curve is itself irreversible. The trade-off implicit in the long-run Phillips curve makes any given stable inflation rate obtainable if policymakers are willing to pay the opportunity cost in the coin of rising unemployment. In other words, having argued that a naive belief in the efficacy of the original Phillips curve had pushed policymakers into implementing policy which propelled the economy up the long-run Phillips curve, they then ar­ gued that the procedure was reversible, and that “coming down” was symmetric to “going up,” involving the reverse trade-off. This is exactly the trade-off that we claim does not exist. The long-run Phillips curve had a specific unemployment policy as its midwife. Changing policy to target inflation generates irreversibilities. We enter the parallel universe of the long-run horizontal Phillips curve and a sig­ nificantly new framework for analysis.4 Irreversibility characterizes all inflationary models, including the one we pro­ pose. Traditional macro policies that aim to target unemployment and/or inflation, must necessarily undercut themselves; none are economically sustainable. Although we may accept that Keynes successfully identified the problem behind any resigned acceptance of the business cycle,5 nevertheless it is also true that he failed to supply the solution. Subsequent policy attempts to resolve underlying economic problems through macro policy have in turn come to grief.6 The profession’s brief intoxi­ cation with the Phillips curve best demonstrates this unfortunate tendency. A Search for Order Where None Existed We might then reason as follows: the near chaos is only apparent, there must be some order here . . . let us try dating the points and observe the sequence in which they occur—when we do this we observe the most remarkable phenomenon-----Now our original chaos has resolved into truly remarkable order. . . one is left, at the end of all this, with a feeling of the possibility of a truly scientific study of human behaviour in the economic sphere, and with an attitude of optimism about the long-term development of such a science. — Robert Lipsey, quoted in Leeson 1994: 19

Phillips’s original work (1954, 1958) was not met with anything approaching uni­ versal acclaim. Aspects of the statistical foundation of his work were questioned by Knowles and Winsten (1959), Routh (1959), Kaldor (1959), Reynolds (1960),

HAS LONG-RUN PHILLIPS CURVE TURNED HORIZONTAL?

147

Pechman (1960), Nourse (1960), Schultze (1959), Conrad (1959), Dicks-Mireaux and Dow (1959), Ozanne (1959), and Turner (1959). As Leeson (1994: 1) notes: Phillips had uncovered an interesting nineteenth-century relationship, but after 1913 his curve did not describe the data at all well. The inter-war period was clearly a disaster for the curve; the post-war fit was the product of Phillips’ innocence with respect to data analysis— he had geared, without adjustment, two unemployment se­ ries which were measured in quite different ways. The econometric evidence in the post-1962 period was also very mixed.

Adoption of the curve seems most of all to have been policy driven, an experience that was to recur with its successor. The empirical or theoretical arguments for adopting this particular piece of apparatus were, as mentioned, less than compel­ ling. It filled a perceived need for a tool that could provide a clear-cut menu for decision makers. As always, precision provides the politician with the illusion of control. Faith in the Phillips curve spread so rapidly in influential comers of Wash­ ington and London exactly because it offered politicians what their preexisting predilections would have wished. Like everyone else, they are most impressed when they hear echoes of their own desires dressed up in technical attire.7 The trade-off implied by the curve reduced economic policy dilemmas to a simple question: Inflation or unemployment? Even those who were able to stop their ears to the siren call of political influence, of advising the mighty if not the wise, found the curve not without theoretical appeal. It quite nicely plugged a gap in the generally accepted IS-LM framework by allowing price changes to be endogenously incorporated. This eliminated the need to treat prices exogenously, which had added an unwanted ad hoc element to the theory.8 It also undermined the danger of a Keynesian free lunch by assuring the profession that there was an opportunity cost to be paid for employment growth; to the ears of the assembled ranks of economists, it had the quiet but sweet ring associated with competitive markets. Perhaps the Phillips curve gained strength and was so heatedly defended because its ambiguity lent itself to multiple uses. In the marketplace of ideas it met the existing demands of the profession. The underlying basis for this supposed correlation was left largely unexplained. Although the given data were also consistent with a simple business cycle rela­ tionship, it conveniently became transformed into a stable menu of choice. To have such a menu implied reversibility. The collapse of the Phillips curve became an event waiting to happen. The introduction of a natural rate of unemployment (the long-run Phillips curve) banished any thoughts of a simple Keynesian trade-off. Market forces once again triumphed. A natural rate allows economists to abdicate any role in reduc­ ing unemployment.9

148 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

Since as shown by Osberg (1988), all unemployment suddenly became frictional (or voluntary), it followed that any existing level of unemployment, no matter how high it may be, is full employment according to this definition! However, to avoid some­ what the tautological implications of their theory, neoclassical economists postulated the existence of a positive long-term or “natural” rate of unemployment that would be compatible with an equilibrium state in the economy in which there are no labourmarket pressures for wages and prices to rise. Except for short-run deviations due to unexpected shocks, the tendency would thus be towards this equilibrium long-term level of voluntary unemployment. From this it follows that the only way of perma­ nently reducing this “natural” rate of unemployment is to remove all institutional barriers and “imperfections” in the labour market that render wages inflexible down­ wards, such as unemployment insurance benefits, welfare payments, trade union ac­ tivity and minimum wage laws. (Seccareccia 1991: 51)

The natural rate shock troops faced a choice following the overthrow of the simple Phillips curve standard. They could have eschewed causality, considering their model to be merely heuristic or operational.10 Equating the natural rate with the nonaccelerating inflation rate of unemployment (NAIRU) avoids any commitment on the part of the theorist. There in fact seemed to be no compelling empirical reason to tie it into a general equilibrium model of market-clearing real wages, except for some given a priori theoretical predilections. By eschewing this chance, they would have let an opportunity slip out of their grasp. Thus the idea of a natural rate as the market-clearing rate of employment is an attempt to provide a particular microfoundation underpinning to the relationship in order to make it more than simply operational. Unfortunately, this left their position vulnerable once the natural rate appeared to slip its mooring. It does seem a bit queer, if not unnatural, that the equilibrium rate should slyly trail the actual rate. It is hardly compatible with our notion of what constitutes an equilibrium. Nonetheless, the current debate, centering on hysteresis, tries to grap­ ple with this observation.11 The dynamic relation between short-term and total unemployment is in fact a complex dynamic relation, where the level of short-term unemployment depends both on changes in and the level of unemployment. An increase in the flow into unemployment initially sharply increases the fraction of short-term unemployment, but may eventu­ ally be associated with a decrease in this fraction as total unemployment rises. Even taking account of these complications, the general result remains that if the long-term unemployed exert little or no pressure on wages, an increase in long-term unemploy­ ment increases equilibrium unemployment for some time. Like the insider model, this implies that short sequences of shocks will have little effect on equilibrium unem­ ployment, while long sequences will increase equilibrium unemployment for some time. (Blanchard and Summers 1987: 294)

The attempts by proponents of the long-run Phillips curve to shore up the model’s preeminent position by resorting to structural and other institutional defenses is

HAS LONG-RUN PHILLIPS CURVE TURNED HORIZONTAL?

149

reminiscent of the rearguard battles fought by the retreating Keynesians facing the onslaught launched against their own model.12 The similarity between these two situations seems to have escaped the notice of all of the protagonists. The parallels are quite distinct. The Keynesian Phillips curve promised a trade­ off between unemployment and inflation, a matter of simply setting the policy control dials. The promise, as already stated, was based on the belief in the re­ versibility implicit in the model. The expectations-augmented Phillips curve also assumes a type of reversibility. A short-term willingness to pay in the currency of rising unemployment guarantees that inflation rates are the result of policy-setting. As it turns out, the observed hysteresis, or shifting of the natural rate, flows from the model’s lack of reversibility. Just as stagflation represented the collapse of the Phillips curve standard in the profession (though vigorously defended by the ad hoc argument of curve shifting), the higher unemployment rates of the 1980s, associated with lower inflation, represent the collapse of the vertical Phillips curve, notwithstanding the equally ad hoc defense of hysteresis. There is nothing natural about the natural rate. It is merely the artifact of policyinduced shocks, in this case, the postwar push to target unemployment. Prior to this, especially in the period before World War I, a trendless business cycle pre­ vailed. The simple Keynesian Phillips curve was only an artifact of the business cycle in the pre-World War I era, with both price and unemployment sequentially rising and falling. The postwar decision to interfere with the business cycle created the long-run vertical Phillips curve. Demand management reduced the cost of job loss. The resulting upward trend of wages would ordinarily have created a profit squeeze dampening investment and ending prosperity. Ever increasing debt inflation perpetuated the boom. It is thus no coincidence that sustained, accelerating inflation was not a feature of the prewar business cycle. Keynes, but more particularly his self-appointed apostles, convinced policymakers that blind obedience to the un­ folding of business cycles was slavery and, even more important, unnecessary. The blind spot of many current theorists is not to see that the issues now sur­ rounding the natural rate arose as policymakers changed their focus from unem­ ployment to inflation.13 Targeting inflation requires the discipline of the business cycle. The mechanism for doing so is a series of sharp and extended recessions. Rising unemployment dampens wage claims. Almost as a side effect, this cycle of induced recessions yields rising levels of the long-term unemployed. Subsequent economic growth fails to spur employment sufficiently. Firms move toward staffing patterns that reflect the discipline of the business cycle, even if the cycle itself has been deliberately manufactured and guided. The era of large numbers of protected, permanent employees is the natural child of unemployment targeting. In demanddominated economies, interrupted production flows are prohibitively costly. In con­ trast, business cycles bring increased price competition, a greater need for flexibility, and a smaller level of core employees. Given the acknowledged shift in target policy, we would expect to see a cor­ responding shift in the economic model needed to comprehend this clear break

150 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

with the past. We in fact conclude that the long-run Phillips curve has turned horizontal, indicating a given economy’s natural rate of inflation or nonaccelerating unemployment rate of inflation (NAURI).14 Through the Looking Glass: Living with a Long-Run Horizontal Phillips Curve The Red Queen shook her head. “You may call it ‘nonsense’ if you like,” she said, “but I’ve heard nonsense, compared with which that would be as sensible as a dictionary!” —Carroll 1871: 27

For a selected group of OECD countries starting in the late 1970s or early 1980s, the policy switch to inflation fighting led to the deliberate creation of recessions. Government spending and tax cuts removed automatic stabilizers. A lower social wage increased the cost of job loss. A rising cost of job loss kept wage inflation in check. While targeting unemployment had created a vertical Phillips curve, the imposition of an artificial business cycle, as a way of reducing inflation, trans­ formed this hypothetical long-run relation from vertical to horizontal. As a consequence of this policy shift, new layers of unemployed have to be sequentially created in order to reduce inflationary pressure, since it is presumably a rising level of short-term unemployment that best keeps inflation falling by keep­ ing wage demands in check (Layard and Nickell 1986). This implies that not only the level, but also the rate of change, of unemployment affects inflation by en­ couraging wage restraint (see, e.g., Romer 1996). Employed workers worry not only about the probability of finding another job if fired, but also the probability of losing their current job. Without the fear of being laid off, workers could adjust to different sustained levels of unemployment, given that labor markets are not intrinsically regulated via a price auction. During recessions, quantity adjustments in the labor market mean that laid-off workers will boost the ranks of the short-run unemployed (Akerlof et al. 1966). Even with no subsequent increase in long-term unemployment, due to offsetting job creation, the visible increase in layoffs should temper wage demands. In the last half of the 1990s, large-scale job creation operated simultaneously with sig­ nificant numbers of layoffs as major corporations continued to restructure.15 Wage demands, despite tightening labor markets, remained moderate. Wage restraint is not simply the result of creating fewer new jobs or vacancies but of increasing the number of people losing their jobs. The newly unemployed compete against others in the same situation. The long­ term unemployed are not equal competitors for job vacancies. Employers consider the very fact that they have failed to find work to be a signal of their inherent inferiority or to reflect a decay of human capital.16 In other words, effective down­ ward pressure on wages does not come from the total pool of the unemployed, but

HAS LONG-RUN PHILLIPS CURVE TURNED HORIZONTAL?

151

from that proportion of the pool which is not part of the long-term unemployed.17 A rising short-term unemployment rate will be more effective in reducing wage pressure than a simple high unemployment rate, since it both increases the prob­ ability of losing one’s job and reduces the probability of finding another. However, since not all of those in this residual short-term pool are likely to find employment, the pool of long-term unemployed must also subsequently increase. Governments wishing to keep pressure on prices have no other option than to generate increased short-run unemployment by tightening monetary policy. An unintended conse­ quence of such measures must be an ever deepening pool of the long-term unemployed. This is essentially a stock/flow problem. From the current stock of short-term unemployed, a proportion will reenter employment, while the residual will join the ranks of the long-term unemployed. To keep downward pressure on the inflation rate, the stock of short-term unemployed workers must steadily increase (see Blanchflower and Oswald 1995). The slow rate of job creation, which makes this possible, means that an increasing proportion of the short-term unemployed become long-term unemployed. Since correspondingly fewer of the long-term unemployed rejoin the work force, their stock should rise. The rate at which it rises will depend on the determination of the monetary and fiscal authorities to bring down inflation. Such a policy has lingering effects. The absolute and relative numbers of long-term unemployed increase. Once the authorities ease policy enough to allow for eco­ nomic growth, unemployment will drop at a slower rate than in previous periods (see Koretz 1995). Just as full employment does not mean 0% unemployment, so price stability need not mean 0% inflation. Price stability means minimizing the cost of any price effects. Pushing inflation below what we term the economy’s natural rate requires ongoing, costly policy shocks.18 These shocks keep inflation low by causing short­ term unemployment to rise continually. The unemployment rate is stabilized at the natural rate of inflation. This natural rate is consistent with any nonaccelerating rate of unemployment, short-term as well as total. Maintaining inflation rates below the natural rate leads to increasing both short- and long-term unemployment. Our general thesis, then, is that if short-term unemployment remains constant, inflation will be stable at a level given by its previous history. This stability will be associated with what Blanchard and Summers call “fragile equilibria”: A physical analogy is useful here. Consider a ball on a hilly surface. If the surface is bowl-shaped, there will be a single uniquely and sharply determined equilibrium surface— at the bottom of the bow l.. . . If the surface contains two pronounced val­ leys, or is extremely flat or contains many mild depressions, the ball’s position will depend sensitively on just how the ball is shocked. We use the term “fragile equilib­ rium” to refer to situations of this type— where outcomes are sensitive to shocks and may be history dependent. (Blanchard and Summers 1988: 184)

152 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

Reducing inflation below any fragile equilibrium combination of unemployment and inflation requires accelerating the level of short-term unemployment. A sub­ sequent fragile equilibrium at the natural rate of inflation will consist of the equi­ librium level of short-term unemployment added to a higher stock of long-term unemployment. An increasing proportion of the unemployed must necessarily be long-term.19 Therefore we would expect a perceived hysteresis to develop when calculating the natural rate of unemployment given a policy shift to inflation tar­ geting. The shifting natural rate of unemployment represents succeeding fragile equilibria along the long-run horizontal Phillips curve. Each observed natural rate of unemployment, under these circumstances, is dependent on the existing natural rate of inflation. Like the natural rate of unemployment, the corresponding natural rate of infla­ tion is characterized by short-run downward stickiness. There is an increasingly weaker correlation between rising unemployment and the rate of change of money wages at very high levels of unemployment. As evidenced in the Great Depression, workers will resist reductions in money wages despite high unemployment. Graphically, the long-run horizontal Phillips curve specifies a family of shortrun curves (see figure 9.2). Each short-run curve is associated with an underlying level of long-term unemployment.20 Pushing inflation below its natural rate will initially increase short-term unemployment, which permits the lower level of infla­ tion. Over time, the increased stock of short-term unemployed will flow into a higher proportion of long-term unemployed. We then move to a new short-run Phillips curve that is to the right of our starting point. An unwavering determination to reduce inflation below its natural rate will push the economy out to succeeding short-run Phillips curves that lie ever more rightward along the horizontal longrun curve. Inflation will continue to return to its natural rate as employment re­ sponds to the lifting of monetary constraints. The cost of attempting to reduce inflation below its natural rate will be accel­ erating increases in the level of unemployment. The move from one short-run Phil­ lips curve to the next along the horizontal long-run curve parallels the upward movement along the traditional vertical long-run Phillips curve. It is important to note that the natural rate of inflation is not associated with a single stable level of unemployment. Rather, because it depends on historical circumstances, unemploy­ ment can stabilize at any rate. The natural rate of inflation is taken by us to be an operational construct, though this is only our point of departure. Institutional fac­ tors, such as the previously mentioned reduction in the number of permanent em­ ployees, can cause both the nature of jobs and the level of competition to change. Institutional rigidities should also influence the observed results.21 These factors are intensified by the heterogeneous nature of firms and the wage contracts that distinguish them.22 Or, taking the lead from Irving Fisher (1926), we could see this rate as providing a necessary amount of flexibility to the economic system. A lower rate of inflation fails to keep the system buoyant because firms are too tightly price constrained, working under an inadequate margin for error.

HAS LONG-RUN PHILLIPS CURVE TURNED HORIZONTAL?

153

Figure 9.2 The Horizontal Long-Run Phillips Curve

Any rate of inflation incorporates some prices that are increasing faster than the inflation rate, encouraging expansion, while others, lagging behind, act as a brake on growth. A critical ratio between the two yields stable employment growth. Newly created jobs balance job losses, thus stabilizing the short-term unemploy­ ment rate. In other words, at this level the rate of change of short-term unemploy­ ment is zero. This weighted average defines the natural rate of inflation. We might point out that any construct of this sort is speculative. Whatever combination of causal links applies, the operational model is itself unaffected. The essential feature of the model is a basic nonsymmetry in the responsiveness of inflation and unemployment to policy shocks. If the target is the unemployment

154 MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

rate, we are in a world of the vertical long-run Phillips curve. Expansionary policy can only induce a short-run trade-off between unemployment and inflation. In the long run there is no trade-off, and the expansionary policy is associated with ever increasing rates of inflation. Inflation is stable only if unemployment is at the NAIRU. Any attempts to reduce unemployment below that level will increase in­ flation. We believe that a symmetric argument can be applied to the analysis of contractionary policy aimed at reducing inflation. By targeting inflation, we move to the world of the horizontal Phillips curve. Contractionary policy may induce short-run trade-offs, but in the long run there is no trade-off. Contractionary policy is associated with an ever increasing rate of unemployment. Unemployment will stabilize only if inflation is at its natural rate. Finally, if government abandons the attempt to target either inflation or unemployment by macroeconomic policy, then an apparent trade-off will arise. Periods characterized by economic booms will be associated with low unemployment and a potential for an eventual rising level of wages, and vice versa, without any causal link established. The entire relationship is then driven purely by the business cycle. This argument can be described by the following equation:*1

( 1)

where

a, b , and Ô are constants (a> 0, b1);

Du is a dummy variable, 0 if policy targets unemployment and 1 otherwise; Dp is a dummy variable, 0 if policy targets inflation and 1 otherwise; u is the unemployment rate; and p is the inflation rate.

In the situation before the government used macroeconomic policy to target either inflation or unemployment, both Du and Dp were equal to 1, so equation (1) became23 1 + au — kp = 0. Note the analogy to the Keynesian multiplier suggested by equation (5): Z can be considered a “final demand” vector; to generate it, a larger amount of goods, X, is required (since 1 > [/ — b'a0 — 54]). I discuss this analogy further in the next section. The key feature of equation (5) is that the surplus arises in production: the “economy” is itself productive.7 The problem of equilibrium relative prices is now

230

MONEY, EMPLOYMENT, AND EFFECTIVE DEMAND

more difficult than before, even in the no-region case, because some subset of agents must “earn” the surplus, even though the source of this surplus is simply production per se.8 In effect, a rule for distribution of claims on the surplus product is required. Under capitalism, the surplus accrues to the owners of the means of production; those who own their labor power must sell it to capital owners, for them to use in production in exchange for a wage. The wage rate w need not equal a pure subsistence level as above. Here, the wage rate is given by w = b'P + w0.

(6)

Note that w is socially determined, not technically determined; it depends on the wage bargain struck between capital and labor, and this depends on the relative strengths of the two sides. Whatever portion of the surplus isn’t ceded to labor accrues to the owners as a rate of profit, r, on the capital they have advanced. For simplicity, r is assumed to be uniform across sectors. Assuming also that profit is earned on both wages and capital goods advanced, then equilibrium prices are given by P — a0 w (l+ r) + A(S + rT)P. Solving for P gives P = [/ - A{8 + rl)]-' [floH0 >0 » 0 > 0 (labor intensity) > 0 (Kaldor effect) > > 0 ,< b 6 =0 > > 0 ,> b 5 =0

d

>0 >0 >0 >0 » 0 =0 >0 =0 0, 0 =0 >0

as a fraction (w) of potential output. Equation (8) is a reformulation of the produc­ tivity definition, and (9) defines the profit share. Already at this stage the model clearly distinguishes itself from the typical marginalist model in two main respects. First, investment is the variable driving the system instead of being assumed to be savings-constrained. The firm’s decision to accumulate, determined by profit and activity levels, is the crucial variable driving long-term growth and the business cycle in capitalist societies.9 Second, adjustment to changes in investment spending takes place via income distribution and/or output adjustment instead of substitution processes in response to relative price shifts. In a typical neoclassical setting, for example, higher real wages are assumed to set in motion the substitution of capital for labor, thus in­ creasing investment and reducing labor demand. In the heterodox model both the cost and the effective demand implications of higher wages play a role. On the one hand, higher wages will increase unit cost and lower profits, implying a negative impact on investment. On the other hand, higher wages stimulate consumption demand, and therefore investment via the accelerator term in (3). In general, the wage increase is more likely to have a stimulating (depressing) impact the higher (lower) the accelerator effect and the greater (smaller) the difference in savings propensities. Consequently, economies can be profit-led or wage-led, depending on the values of the coefficients. The value, variation, and explanation of the crucial coefficients are ultimately a historical and empirical question. In addition to investment-driven savings and income distribution effects, there are two additional distinguishing model characteristics. Income distribution is not determined by marginal factor productivities but rather by power relations and social contracts. As discussed earlier, this follows directly from the adoption of classes or income groups instead of optimizing individuals as the units of analysis. In the Regulation Theory model, for example, during the postwar period real wages

ELEMENTS OF HISTORICAL MACROECONOMICS

251

are set by a social (Fordist) contract as a share of increasing productivity. Finally, the model is dynamic in nature since it makes productivity growth endogenous and expresses wage- and price-setting in rate of change form. The Transformational Growth model contrasts the craft economy of the nine­ teenth century with the postwar mass-production economy (table 14.1). The first relevant feature of a craft economy is that actual output fluctuates around a “nor­ mal” level. Instead of potential output, the variable 0). Employment will therefore become flexible in (8), representing the familiar Keynesian employment multiplier. In terms of our model, it means that the coefficient of utilization in the consumption equation (b2) will be positive in the New Business Cycle. In addition, the process of setting wages and prices and their interaction have undergone a profound transformation. In equation (6) the excess demand element drops out and is replaced by a markup rule. Benchmark prices are set as a strategic decision by firms in accordance with cost and market share considerations. The actual prices will vary somewhat with changing variable cost (here represented by the wage share) but will otherwise be stable even if demand fluctuates. This is partly due to the variability of output and employment, but also reflects the shift in the nature of competition, from a concern with prices to a race for new tech­ nology (Nell 1993).

ELEMENTS OF HISTORICAL MACROECONOMICS

253

The key advantage of mass-production technology is its ability to adjust variable cost, the wage bill in particular, to changes in the demand for the product. The markup may change cyclically, not because of supply and demand interactions but because of strategic considerations in an oligopolistic market setting. This possi­ bility is reflected in the addition of the capacity utilization level in (6). Kalecki, for example, developed the idea that the markup varies countercyclical^ (c6 A

(6)

If we now check Ricardo’s second numerical example against the background of this formulation, we see that it violates condition (5), because

CRITICISMS OF RICARDO’S DISCUSSION

333

Hence Ricardo’s numerical example is not strictly correct. It is perhaps best inter­ preted as a crude approximation to a correct example. Such numerical approxi­ mations were quite common at the time when Ricardo wrote. There are several ways to rectify the example. We may, for example, start from the initial set of capitals employed on the different qualities of land (50, 60, 70, 80) and any one of the new figures for the capitals after the improvement. Let us start from marginal land, land 4, and assume with Ricardo that the improvement allows one to save five units of capital expressed in com, so that the new capital amounts to 75. The capitals employed on the lands 1, 2, and 3 after the improvement satisfying the condition of unchanged com rents would then have to be 46.875 instead of 45; 56.250 instead of 55; and 65.625 instead of 65. The differences between the cor­ responding elements of the two sets of numerical values are not very large, so that Ricardo’s example may be considered a first approximation to a fully correct il­ lustration. (The corresponding savings of capital would respectively be 3.125 on land 1; 3.750 on land 2; and 4.375 on land 3.) Obviously, with these new values for the post-improvement capitals, condition (5) would be met. (Had we taken as our starting point Ricardo’s values of the capital on any one of the intramarginal qualities of land instead of those on the marginal land, then we would, of course, have got different values for the post­ improvement capitals on the remaining lands and correspondingly different absolute amounts of capital savings.) Notice that while this change performs the task for which it was designed, we are not yet able to say anything definite about the levels of the rate of profit before and after the improvement, except— as expressed by inequality (6)— that the latter will be larger than the former. Thus, the case contemplated would, for example, be compatible with ra = 74 and rp = 79. In fact, it would be compatible with any pair of (nonnegative) values of the two profit rates satisfying the equation

We are also not yet able to say anything definite about the numerical values of the com rents obtained on the intramarginal lands. To fill this lacuna, we have to provide one further element of information. For example, we may fix the rate of profit in the pre-improvement situation. In many of his numerical examples Ricardo, for simplicity, assumed a rate of profit of 10%. If we set r“ = 0.1, we can calculate the corresponding output on the marginal land (which, by construction, equals the output on the intramarginal lands) and obtain that X = 88. All the other pre- and post-improvement magnitudes can then be

334 THEORY, METHOD, AND THE HISTORY OF IDEAS

ascertained. We shall come back to this closure in our discussion of Harry Johnson’s 1948 note. Alternatively, we may fix the output obtained on each and every kind of land. Let us assume X = 90 quarters of com. This is tantamount to assuming that the levels of the rate of profit before and after the improvement are

(These values of the rate of profit are, of course, also compatible with the conditions on the intramarginal lands.) We can now calculate the com rents obtained on the intramarginal qualities of land by subtracting from com output the com capitals appropriately discounted forward. We obtain for land 1: com rent = land 2: com rent = land 3: com rent =

90 90 90

- (1 - (1 - (1

+ r«)50= 90+ ra)60= 90+ ra)70= 90-

(1+ (1+ (1+

^46.875 r^56.250 ^65.625

The arithmetical imprecision involved inRicardo’s numerical example should now be obvious. In order to satisfy condition (3), which requires that with regard to each quality of land the pre- and post-improvement com rents are equal, the savings of capital cannot be of equal absolute size, but must be of equal relative size— relative to the original amount of capital employed on each quality of land culti­ vated. Does Ricardo’s imprecision thwart the general thrust of his argument? The answer is no. The flaw can easily be remedied. The correctness of Ricardo’s ar­ gument does not stand or fall with the correctness of the numerical illustration he provides: the special case of purely capital-saving and com rents-preserving im­ provement he has in mind can be given a precise analytical expression and a precise numerical illustration. (For a different numerical illustration, see Gehrke et al. 2003). Caiman’s Objections to Ricardo’s Analysis of Agricultural Improvements Edwin Cannan (1917: 258-61) appears to have been the first author to point out that Ricardo had committed an “error” in his second arithmetical example (see also O’Brien 1975: 128-29). Cannan suggested the following corrected version of Ri­ cardo’s example: If the number of quarters of com produced by each of the “four portions of capital” be x , then the original com rent will be of the 80, nil, of the 70, — 80

of the 60,

= 33.75 = 22.5 = 11.25

CRITICISMS OF RICARDO’S DISCUSSION

20

30

335

3

—x, and of the 50, — x, in all ~x, and if a quarter is worth £4, money rent will be

oU 80 4 £3x. After the improvement, the com which regulates the price can be produced with 5 — “less capital, which is the same thing as less labour,” and consequently the price 80 3 of com falls from £4 to £3-, the com rent rises from 4

3 4 and the money rent remains £ 3 - X -x = 3jc, exactly the same as before. 4 5 (1917: 259-60)

He added: “If equal absolute amounts are taken away from the ‘four portions of capital,’ com rent will always rise, and money rent will always remain the same” (1917: 260). The example, Cannan concluded, was erroneous and thus unfit to convey “Ricardo’s doctrine”— which he took to be that improvements of the second type must always reduce money rents and leave com rents the same.6 As was shown in the preceding section, Ricardo’s numerical example is indeed not com rent-preserving. However, there are elements in Cannan’s interpretation of Ricardo’s theory of agricultural improvements that are dubious or difficult to sus­ tain. Cannan’s exposition of Ricardo’s “error” is followed by the remark: Curiously enough, Ricardo himself, in the Chapter on Taxes on Raw Produce, recog­ nises, the converse case, namely, that the addition of an equal absolute amount to each of the four portions will diminish the com rent and leave the money rent unal­ tered. (Cannan 1917: 260)

This remark is misleading. For what Ricardo in fact demonstrated in the chapter “Taxes on Raw Produce” is that when a tax is imposed on raw produce, then the net com produce (net of tax) that is obtained from equal “portions of capital” (which are expressed in terms of com) will be reduced by equal percentages. The “portions of capital” in terms o f com are left unchanged by the imposition of the tax, while the corresponding “portions of capital” in terms o f money are raised by equal absolute amounts and by equal percentages (because, by assumption, the pretax amounts of those “portions of capital” are all equal). There is thus noth­ ing “curious” about the fact that Ricardo should have given “the right answer” to “his problem” in chapter 9: the problem that he investigated there is completely

336 THEORY, METHOD, AND THE HISTORY OF IDEAS

different from (and is certainly not “the converse case” of) the one contemplated in chapter 2. Cannan regarded Ricardo’s “error” in his second arithmetical example as merely a particularly striking example of his misconceived “attempt to show that improve­ ments must temporarily lower rent” (1917: 260). With regard to Ricardo’s first type of improvements, Cannan observed: “In the arithmetical example which Ricardo gives to illustrate his doctrine it happens to be true . . . but this is only so because he supposes in the example—what is not very likely to occur—that the improve­ ment adds ‘an equal augmentation,’ that is, an equal absolute amount, to the produce of each of the successive qualities of land or portions of capital employed” (1917: 254). Cannan then tried to ridicule Ricardo’s arithmetical example by producing a number of counterexamples, in which the post-improvement com rent rises rather than falls, because the cost differentials are augmented. He then summarized the results of these exercises as follows: To make Ricardo’s doctrine true of com rent, we must suppose what we have no grounds for believing, and what seems prima facie improbable, that improvements always add an equal absolute amount to the produce of each of the successive “layers” of capital, or at any rate that they never add a larger absolute amount to the produce of the more productive layers, than to that of the less productive layers. (1917: 256)

However, according to Cannan, it is unclear whether “Ricardo’s doctrine” was meant to apply to com rents or to money rents. He therefore also investigated the requirements that have to be met in order to make it true of money rather than com rents: To make Ricardo’s doctrine true of money rent, we must suppose that improvements always add an equal percentage to the produce of each of the successive layers of capital, or, at any rate, that they never add a greater percentage to the produce of the more productive layers than to that of the less productive. (1917: 258)

According to Cannan, there is some evidence that the “doctrine” was meant to apply to money rents because Ricardo “boldly asserts that improvements ‘probably’ do add equal percentages to the different layers” (1917: 258). Cannan’s reference is to a passage in chapter 32, “Mr. Malthus’s Opinions on Rent,” of the Principles, where Ricardo had maintained: Nothing can raise rent, but a demand for new land of an inferior quality, or some cause which shall occasion an alteration in the relative fertility of the land already under cultivation. Improvements in agriculture, and in the division of labour, are common to all land; they increase the absolute quantity of raw produce obtained from each, but probably do not much disturb the relative proportions which before existed between them. (Works I: 412-13; emphasis added)

CRITICISMS OF RICARDO’S DISCUSSION

337

For Cannan the italicized statement contains the “bold assertion” that improve­ ments are likely to cause equiproportional augmentations in com outputs on each quality of land (thus making “Ricardo’s doctrine” true of money rent). However, what is in fact asserted is that improvements “probably do not much disturb” the relative proportions which before existed between the different qualities of land— not that they will remain strictly the same. What Ricardo seems to have had in mind is that the existing net output differentials (in terms of com) between the different qualities of land, and hence the com rents obtained from each, are “not much disturbed.” He was at any rate clear about the fact that improvements which cause equiproportional increases in com outputs on each quality of land will in­ crease the cost differentials, and hence the com rents obtained from each. This is evident from a passage in his Notes on Malthus,1 where he wrote: “Mr. Malthus may say that improvements on the land, if they increase the produce on all land in equal proportions, will increase the difference, in com produce between equal capitals employed on the land” (Works II: 134-35). Cannan raised a further objection against Ricardo’s theory of agricultural im­ provements: When [Ricardo] says the two kinds of improvements do not affect rent equally, he apparently means that improvements of the first class lower it more than those of the second class, because they lower both money rent and com rent. (Cannan 1917: 258— 59; emphasis added)

Note that the idea that Ricardo’s distinction obtains its significance from the fact that rent is lowered “more” by improvements of the first class than by those of the second class is Cannan’s, not Ricardo’s. It is therefore rather disingenuous of Can­ nan to criticize this statement— which is his own interpretation of Ricardo’s mean­ ing— with the (correct) observation that “an improvement which affected money rent only might of course lower money rent more than an improvement which affected both money and com rent, but Ricardo does not think of this” (1917: 259n). Finally, there is Cannan’s rather odd claim that Ricardo had neglected altogether to consider those alternative cases of agricultural improvements in which neither com nor money rents remain unaffected: As he professes to be dealing with improvements in general, and yet does not think it necessary to consider the case of improvements which cannot be effected “without disturbing the difference between the productive powers of the successive portions of capital,” we must suppose that it did not occur to him that there was such a case. (1917: 256)

This accusation cannot be sustained. Ricardo clearly had made no claim that the two cases contemplated exhausted all conceivable cases of agricultural improve­

338 THEORY, METHOD, AND THE HISTORY OF IDEAS

ments. He had, on the contrary, concluded his analysis of the effects of agricultural improvements on rent, in chapter 2 of the Principles, with the following statement: Without multiplying instances, I hope enough has been said to show, that whatever diminishes the inequality in the produce obtained from successive portions of capital employed on the same or on new land, tends to lower rent; and that whatever increases that inequality, necessarily produces an opposite effect, and tends to raise it. (Works I: 83; emphasis added)

Johnson’s Note of 1948 and His Claims Concerning Sraffa’s Position Harry G. Johnson was at Cambridge as a student in 1945-46 and as an Assistant Lecturer and then Lecturer from January 1949 to March 1956 (see Johnson and Johnson 1978: 87). In 1948 he published a short note titled “An Error in Ricardo’s Exposition of His Theory of Rent” in the Quarterly Journal o f Economics (Johnson 1948). In this note Johnson demonstrated that the improvement contemplated in Ricardo’s second numerical example will raise com rents and leave money rents unchanged. Apparently, Cannan’s criticism had escaped Johnson’s attention. His account of Ricardo’s “error” is in substance the same as Cannan’s, although his exposition is somewhat different and, indeed, rather more confusing than Cannan’s. The interest of Johnson’s note derives from contentions that were later put for­ ward by its author about the immediate reception of his note in Cambridge. In an article published in 1976 he wrote: On arrival in Cambridge [in early 1949] I found Maurice Dobb and Piero Sraffa much excited by my note, which they were sure was wrong; Sraffa even proposed to publish a criticism of my note, thereby breaking nineteen (I think) years of silence. They were then both Fellows of Trinity College, with rooms not too far apart. Nevertheless, I was shown a stack of notes that had passed between them, notes replete with such phrases as “Johnson’s fallacy is___ ” I would have been oveijoyed to have smoked Sraffa out of his long silence, especially as I felt sure that my mathematics were correct. But unfortunately, or fortunately, Sraffa eventually looked up Edwin Cannan’s history of thought and found the error already noted there, whereupon he dropped the subject. (Johnson 1976: 86)8

Johnson’s contentions are reiterated by Denis O’Brien in an article on “Edwin Cannan as an Interpreter of the Classical Economists”: Cannan was able to show, by the employment of some neat algebra, that Ricardo’s conclusion that improvements would lower rents was simply mistaken___ The cri­ tique is an object lesson in the detailed careful analysis of an original author, but the lessons of Cannan’s achievements go wider. For Sraffa and Dobb in Cambridge were apparently completely ignorant of this. When Harry Johnson . . . produced an identical

CRITICISMS OF RICARDO’S DISCUSSION

339

proof (a quite extraordinary intellectual coincidence in itself), the reaction of Sraffa and Dobb was one of incredulity. Sraffa was indeed on the point of publishing a reply when he found out that Cannan had pointed out the mistake 56 years previously. The two most important figures in the generation of a major branch of Ricardo interpre­ tation were thus ignorant, as Johnson notes, of Cannan’s devastating criticism. (O’Brien 1998: 68)

Johnson’s story, if true, would indeed be quite remarkable. However, it seems not very credible prima facie. Edwin Cannan was undoubtedly one of the leading au­ thorities on the classical economists in the first part of the twentieth century, and it would therefore be rather surprising if the editor of Ricardo’s Works had been entirely unaware of his contributions. Moreover, it is well known that Sraffa and Cannan were in contact with one another during the 1920s and 1930s.9 The facts that we have been able to establish are the following. Piero Sraffa’s unpublished papers, which are deposited in the Wren Library at Trinity College, Cambridge,10 contain no notes (by either Sraffa or Dobb) on Harry G. Johnson’s article of 1948.11 Nor are any such notes to be found in Maurice Dobb’s unpub­ lished papers.12 In Sraffa’s Cambridge Pocket Diaries for the academic years 194849 (see Sraffa’s papers, catalog item E 21) and 1949-50 (see E 22), there are no entries for appointments with Harry Johnson.13 There are numerous entries for appointments with Maurice H. Dobb— which is hardly surprising in view of the fact that Dobb had agreed to collaborate with Sraffa in the preparation of the Royal Economic Society edition of Ricardo’s Works in 1948. In the archival material there is thus no evidence for discussions between Sraffa and Dobb about Johnson’s note in the late 1940s (which does not, of course, prove that no such discussions ever took place). However, in Sraffa’s unpublished papers there are various notes on “improve­ ments in agriculture” (see, in particular, the folders Dl/34: 1-8, Dl/37: 2-17, D l/ 42, Dl/44: 36-52, and D3/11/5). Most of these notes are written in Italian, and they are often headed “Miglioramenti” or “Miglioramenti Ricardiani.” All the notes are undated. The catalog of the Sraffa Papers, which was prepared by the archivist at the Wren Library, Jonathan Smith, contains the following statement: “Items D l/ 2-D1/51 were bundled together and marked ‘Notes up to 1927’ by Sraffa. In most cases this is our only evident terminus ante quern.” However, it seems possible to be more precise with regard to the dating of the documents under consideration. Circumstantial evidence suggests that Sraffa presumably wrote the notes that are now collected in folder Dl/37 in, or around, 1924. The evidence in support of this conjecture is the following. First, the folder contains an envelope that bears the date “1924” (see document Dl/37: 1). Second, the notes in folder Dl/37 are written in a notebook from an Italian manufacturer; the same type of notebook was used by Sraffa for excerpts and working notes that he wrote in preparation of his articles of 1925 and 1926 (see folder Dl/41). Third, Sraffa’s notes on “improvements in agriculture” are closely related to— and may indeed at some point have been con­

340 THEORY, METHOD, AND THE HISTORY OF IDEAS

ceived by him as part of—his critique of Marshall’s partial equilibrium analysis. It seems very likely, therefore, that Sraffa had worked on “Ricardian improvements” in the same period in which he was concerned with the preparation of the two articles of 1925 and 1926 that contained his critique of Marshallian supply schedules.14 Some of the notes on “Ricardian improvements” consist merely of bibliograph­ ical references and short citations. Others are (unfinished) manuscript notes, run­ ning over some ten to fifteen pages. The works cited by Sraffa include Ricardo’s, J.S. Mill’s, and Marshall’s Principles; Cannan’s History,; Pigou’s Economics o f Welfare; and Edgeworth’s articles “Appreciations of Mathematical Theories” (1907) and “The Pure Theory of Monopoly” (1897).15 To provide a detailed account of the contents of Sraffa’s manuscript notes is beyond the scope of this chapter. Here we must content ourselves with showing that Sraffa was familiar with Cannan’s critique of Ricardo’s analysis of improvements in agriculture, and that he had in­ deed intensively worked on this problem very early on in his career. As is evident from a heavily annotated copy in his library, Sraffa had carefully studied Cannan’s HistoryJ 6 He excerpted, inter alia, the following statements: “Cannan, Theories, p. 331, ‘No general rule can be laid down with regard to the immediate effect of improvements. It will vary with the nature of the improvement and the circumstances of the country and soil to which they are applied’; pp. 32728 ‘the strangest part of Ricardo’s theory with regard to the effects of agricultural improvements,’ that is the distinction between the two classes; p. 327 ‘ipse dixit of a retired stock broker,’ ” to which he added the comment: “R. with regard to im­ provements is only interested in the argument ad absurdum, and not in the dem­ onstration that improvements are impediments to protectionism, as Cannan says” (Dl/34: 3; emphasis in the original). Initially, Sraffa appears to have been skeptical as to the correctness of Cannan’s objection. He quoted a passage from Bagehot (1911: 208): A well authenticated tradition says that he [Ricardo] was most apt and ready in the minutest numerical calculations. This might be gathered from his works and, indeed, any one must be thus apt and ready who thrives on the Stock Exchange___

Sraffa then added (in Italian): But is it not then highly improbable that R. should have committed the error detected by Cannan? In particular if one takes into consideration that the inverse calculation that must be carried out in order to obtain the exact rent is of the same type as the one which is employed in arbitrage (but did R. occupy himself in trading with bills?). (Dl/34: 2; our translation)

However, scrutiny showed that the objection was correct. A document headed (in Italian) “Malthusian hypotheses: productivity curve” (Dl/37: 2 (1-8)) includes

CRITICISMS OF RICARDO’S DISCUSSION

341

some quotations from Cannan’s History on “Ricardo’s error (second class)”; see in particular document Dl/37: 2 (8). Sraffa again tackled the problem of agricultural improvements after he had been entrusted with the editorship of Ricardo’s Works. Folder D3/11/5 contains a large set of notes, presumably written in 1930 or 1931, on the relationship between Ricardo’s Essay on Profits and the first six chapters of his Principles. Among them is a note in which he completed Ricardo’s second example in terms of a given rate of profit of 10% in the pre-improvement situation, which is equivalent to assuming a com output of eighty-eight units on the marginal land (and, by construction, on all intramarginal lands). He then calculated the rate of profit that obtains after the improvement on the marginal land, which equals ,3/75 or 17!/3%. Applying this rate to Ricardo’s figures of the post-improvement capitals employed on the intramar­ ginal lands, one can calculate the corresponding rents. (Sraffa’s own calculation is very rough.) Sraffa admitted that there is an “error” in Ricardo’s numerical example (see D3/11/5: 53).17 We may now ask two questions. First, why did Sraffa not put in a reference to Cannan’s criticism in chapter 2 of his edition of the Principles? Second, why did he not respond to Johnson’s contentions? Since we are not aware of any written statements by Sraffa regarding these issues, we cannot offer more than a reasoned speculation. First to Cannan. While Sraffa accepted Cannan’s objection to Ricardo’s second numerical illustration, he apparently considered the latter a legitimate approxima­ tion. The “error” was “much smaller than supposed by Cannan” and could easily be eliminated by a trivial correction of the example. In Sraffa’s judgment Cannan’s overall criticism of Ricardo was out of proportion: a sound argument was attacked via an imprecise numerical illustration. There was no need to draw attention to Cannan’s unbalanced criticism. Now to Johnson. First, as admitted by Johnson himself (see Johnson and Johnson 1978: 94), it was Sraffa who had pointed out to him that he was guilty of subjective originality. What is mysterious in Johnson’s account is, first, his reference to having been shown “a stack of notes.” We found no evidence in support of this contention, and it is indeed not clear why a whole “stack of notes” would be needed to make the simple point Sraffa had made a long time ago, in response to Cannan’s criticism, that Ricardo’s numerical example, while not strictly correct, could be envisaged as an approximation. Second, it is not clear why, according to Johnson, Sraffa should only “eventually” have looked up Cannan’s book. Sraffa had studied that book with great care in all its details, particularly the passages devoted to Ricardo. He had written long critical comments on it, paying special attention to Cannan’s exag­ gerated objections to Ricardo’s discussion of agricultural improvements. And now, by the late 1940s, he should have forgotten all this? This is hardly credible. So is it possible that Johnson misinterpreted what was going on? Were the “stacks of notes” he says he was shown perhaps not notes about his 1948 paper but rather some of Sraffa’s earlier papers on this issue? Did the remark, which Johnson reports

342 THEORY, METHOD, AND THE HISTORY OF IDEAS

Sraffa had made, about an intention on Sraffa’s part to publish something on the issue, perhaps refer to Sraffa’s own manuscripts drafted several years earlier rather than to Johnson’s note? And did Sraffa draw Johnson’s attention to Cannan’s book “eventually” or immediately upon confrontation with Johnson’s note? Johnson wrote his account of the story more than two decades after it happened. Is it possible that his memory played a trick on him? After all, the truly surprising fact is that Johnson, and his supervisor Schumpeter (as well as the referees of the Quarterly Journal o f Economics, in which Johnson’s note appeared), were appar­ ently completely ignorant of Cannan’s criticism— an accusation that, as we have seen, cannot be levelled at Sraffa. Why, then, should Sraffa publicly contradict the account given by Johnson? He was not the kind of person who would waste his time on what was all too obvious. C o n c lu s io n

This chapter has scrutinized Ricardo’s discussion of agricultural improvements in the chapter “On Rent” of the Principles and the critical comments it engendered. It was argued that while Ricardo’s numerical illustration of the first kind of im­ provement contemplated— that is, what, for short, was dubbed “land saving”— is correct, his illustration of the “capital alias labor saving” improvement exhibits an arithmetical imprecision: instead of assuming equal absolute savings with regard to each of the capitals employed on the different qualities of land cultivated, Ri­ cardo should have assumed equal relative savings. The slip can easily be remedied. It follows that Ricardo’s argument does not stand or fall with the correctness of the numerical example he provides. We then turned to interpreters and critics of Ricardo’s discussion of agricultural improvements, including Edwin Cannan, Harry Johnson, and Mark Blaug. We ar­ gued that whereas Cannan’s objection to Ricardo’s numerical illustration of im­ provements of the second kind is correct, his overall criticism of Ricardo’s analysis of agricultural improvements is vastly exaggerated. In particular, we do not share Cannan’s view that Ricardo’s attempt to show that improvements are likely tem­ porarily to lower money rents ends “in complete and hopeless failure” (Cannan 1917: 259-60). In addition we pointed out difficulties in some other interpretations of Ricardo’s analysis. The chapter concludes by putting forward evidence that Sraffa was already familiar with Cannan’s criticism in the 1920s. Therefore, it is a mystery what is the factual foundation of Johnson’s reminiscences, which O’Brien interpreted as evidencing that Sraffa had been “completely ignorant” of it in the late 1940s. N o te s

Earlier versions of this chapter were presented in a seminar at the University of Graz in May 2000 and at the Annual Meeting of the History of Economics Society in Vancouver in

CRITICISMS OF RICARDO’S DISCUSSION

343

July 2000. We should like to thank the participants, and especially Christian Klamler, for useful discussions. We should also like to express our gratitude to Mark Blaug, Denis O’Brien, Pierangelo Garegnani, Geoff Harcourt, Neri Salvadori, and Ian Steedman for valu­ able comments on a previous version. 1. This is a companion paper to Gehrke, Kurz, and Salvadori (2003); in the sequel we shall come back to the relationship between the two papers. 2. As we shall see, Ricardo defines the two forms of improvements in terms of given output levels and a given real wage rate, and thus must not be interpreted in conventional neoclassical terms. 3. Ricardo considered gold (money) to be an invariable measure of value that is always produced by means of the same quantity of (direct and indirect) labor per ounce. A reduction in the amount of labor needed to produce a bushel of com on marginal land therefore implied a proportional reduction in the money price of com. 4. Since land of quality 3, which is the marginal land in the post-improvement situation, will not be fully cultivated in order to match effective demand, it will not be scarce, and therefore we can be sure that the rent obtained on it is nil (assuming implicitly, with Ricardo, that the reservation price of the use of land is nil). John Stuart Mill (1871), in his numerical illustration of land-saving improvements, inadvertently constructed an example in which the post-improvement marginal land is fully employed. In this case competition among landlords need not suffice to make the rent on this quality of land vanish: it is possible that the marginal land yields its proprietors a positive rent (although this is not necessarily the case). Ricardo, by mere luck or by ingenuity, avoided the trap into which Mill fell. 5. For a critical discussion of Ricardo’s different definitions of rent in his analysis of agricultural improvements, see Gehrke et al. (2003). 6. Cannan’s criticism of Ricardo’s second example was essentially adopted by Mark Blaug (1997: 113). However, Blaug, unwittingly it seems, put forward a very special form of Cannan’s objection. Blaug contended that “at the margin, rents are zero, so 80 units of capital must receive 80 quarters of wheat” (1997: 113). This implies, of course, that 80 quarters of wheat must also be obtained from all the other “portions of capital,” that is, Cannan’s variable x is given a particular value. More important, the specific numerical value of x chosen by Blaug implies that the rate of profit is zero. (See also the comment on Blaug’s particular specification in the next section.) 7. It must be noted that Ricardo’s Notes on Malthus were published for the first time only in 1928. They were thus not available when Cannan worked on the first edition of his History. 8. Johnson’s paper was reprinted (with small changes) in Patinkin and Leith (1977) and in Johnson and Johnson (1978). In the preface of the latter volume it is stated that the essays presented do not “claim to b e . . . contributions to scholarship and documentation in the history of economic thought” (Johnson and Johnson 1978: x). 9. Sraffa attended Cannan’s lectures on the theory of value and distribution when he was a research student at the London School of Economics in the summer of 1921 and, again, in the academic year 1921-22 (see Potier 1987: 8-9). We also know from Sraffa’s unpublished papers that in the early 1930s he had an extensive correspondence, as well as several meetings, with Cannan. After Sraffa had been entrusted with the editorship of Ri­ cardo’s Works in February 1930, he repeatedly approached Cannan for advice on specific points, which was most willingly given.

344 THEORY, METHOD, AND THE HISTORY OF IDEAS

10. We should like to thank Pierangelo Garegnani, literary executor of Sraffa’s papers and correspondence, for granting us permission to quote from them. References to the papers follow the catalog prepared by Jonathan Smith, archivist at the Wren library. 11. An offprint of Johnson’s article, with the handwritten remark “Compliments H.G.J.” is in Sraffa’s library (catalog no. 5143). There are no comments or annotations. 12. Maurice Dobb’s papers are also deposited in the Wren Library at Trinity College in Cambridge. Brian H. Pollitt, who cataloged Dobb’s papers and still holds some additional material, including Dobb’s diaries, informed us in a personal communication of 25 February 2000 that “I have no record of any discussion concerning Johnson’s 1948 note, but Dobb records meeting with him at 7.45 p.m. on 12 January 1949, obviously before dinner at Trinity. The entry is in his ‘Cambridge Pocket Diary’ of 1948-49.” 13. The name “Johnson” appears twice in Sraffa’s diary for 1949-50: on 29 November 1949: “Audit Dinner (1 guest) (Johnson) black tie”; and on 7 March 1950: “8.30 Kahn’s Seminar: Johnson” (E 22). 14. That the notes in question were indeed written in the 1920s, rather than in the late 1940s in reaction to Johnson’s note, is further corroborated by the following circumstantial evidence: (1) all the literature quoted by Sraffa is pre-1925; (2) there are no references to the manuscripts of his own edition of Ricardo’s Works; (3) there is no reference to Johnson’s 1948 article. 15. Sraffa’s citations are from the Italian version of Edgeworth’s article (see Edgeworth 1897); there are no references to the (abridged and revised) English version of Edgeworth’s paper, which was first published in 1925 in the edition of Edgeworth’s papers prepared for the Royal Economic Society. 16. The two copies of Cannan’s History in Sraffa’s library (catalog nos. 1893 and 7929) are not annotated. However, what apparently was his working copy of Cannan’s History in the archivist’s room of the Marshall Library is heavily annotated throughout, including the section on agricultural improvements. 17. Clearly, the magnitude of the error involved depends positively on the numerical specification of Cannan’s x (or, equivalently, on the pre-improvement rate of profit ra), where, obviously, jc > 80 (ra ^ 0). For example, with * = 80, as assumed by Blaug, the pre­ improvement rate of profit would be nil, the post-improvement profit rate would equal 1/15, and the post-improvement com rents would equal 32, 211/3, and 10% units of com on lands of quality 1, 2, and 3, respectively. Implicitly, this setting amounts to minimizing the numerical value of Ricardo’s error. R e fe r e n c e s

Bagehot, W. 1911. Economic Studies. London: H.S. King. Blaug, M. 1997. Economic Theory in Retrospect, 5th ed. Cambridge: Cambridge University Press. Cannan, E. 1917. A History o f the Theories o f Production and Distribution in English Political Economy from 1776-1848, 3d ed. [1st ed. 1893]. London: P.S. King. Re­ printed 1967. Edgeworth, F.Y. 1897. “La Teoria Pura del Monopolio.” Giomale degli Economisti 13-31 (part I), 307-20 (Part II), 405-14 (Part III). Translated as “The Pure Theory of

CRITICISMS OF RICARDO’S DISCUSSION

345

Monopoly.” In Papers Relating to Political Economy, voi. 1. London: Macmillan, 1925. --------- . 1907. “Appreciations of Mathematical Theories (First Instalment).” Economic Jour­ nal 15: 221-31. Gehrke, C., Kurz, H.D., and Salvadori, N. 2003. “Ricardo on Agricultural Improvements: A Note.” Scottish Journal of Political Economy. Johnson, H.G. 1948. “An Error in Ricardo’s Exposition of His Theory of Rent.” Quarterly Journal o f Economics 2: 792-93. --------- . 1974. “Cambridge in the 1950s. Memoirs of an Economist.” In E.S. Johnson and H.G. Johnson, The Shadow o f Keynes. Oxford: Blackwell, 1978. --------- . 1976. “How Good Was Keynes’ Cambridge?” In E.S. Johnson and H.G. Johnson, The Shadow o f Keynes. Oxford: Blackwell, 1978. --------- . 1977. “Cambridge as an Academic Environment in the Early 1930s: A Reconstruc­ tion from the Late 1940s.” In Keynes, Cambridge and the General Theory, ed. D. Patinkin and J.C. Leith. London: Macmillan. Johnson, E.S., and Johnson, H.G. 1978. The Shadow o f Keynes. Oxford: Blackwell. Marshall, A. 1920. Principles o f Economics, 8th ed. [1st ed. 1890]. London: Macmillan. Marshall, A., and Marshall, M.P. 1879. The Economics of Industry. London: Macmillan. Mill, J.S. 1871. Principles o f Political Economy, 7th ed. [1st ed. 1848]. Fairfield, NJ: A.M. Kelley, 1987. O’Brien, D. 1975. The Classical Economists. Oxford: Oxford University Press. --------- . 1998. “Cannan, Edwin, as an Interpreter of the Classical Economists.” In The Elgar Companion to Classical Economics, voi. 1, ed. H.D. Kurz and N. Salvadori. Chel­ tenham, UK: Edward Elgar. Potier, J.-P. 1987. Un Economiste Non-Conformiste, Piero Sraffa (1898-1983). Lyon: Presses Universitaires de Lyon. Ricardo, D. 1951-73. The Works and Correspondence o f David Ricardo, 11 vols. ed. P. Sraffa in collaboration with M.H. Dobb. Cambridge: Cambridge University Press. Cited as Works, followed by volume number. Samuelson, P.A. 1977. “Correcting the Ricardo Error Spotted in Harry Johnson’s Maiden Paper.” Quarterly Journal o f Economics 91: 519-30.

Contributors

G e o r g e A r g y r o u s

D u n c a n

U niversity o f New South Wales AUSTRALIA

New School University U N ITED STATES

T h o r s t e n

B lo c k M

M aastricht Econom ic Research Institute on Innovation & Technology (M ERIT) T H E N ETH ERLA N D S G a r y

K . F o le y

D y m s k i

U niversity o f California, R iverside U N ITED STATES

a th e w

F o r s ta t e r

University o f M isso u riK ansas City U N ITED STATES

C r a ig

F r e e d m a n

M acquarie University AUSTRALIA

K o r k u t A . E r tiir k

C h r is tia n

University o f Utah U N ITED STATES

University o f G raz AUSTRIA

G e h r k e

348 CONTRIBUTORS

H a r a ld

H a g e m a n n

T h o m a s

R . M

ic h l

U niversität H ohenheim G ERM A N Y

Colgate University UN ITED STATES

G .C . H a r c o u r t

G a r y

Jesus College, Cam bridge U N ITED K IN GDOM

St. John’s University UNITED STATES

G e o ffr e y

M

M

. H o d g s o n

University o f Hertfordshire U N ITED K IN GDOM P e te r

e g a n

T h o m a s

F . P h illip s

Sir Sandford Flem ing College CANADA

o n d

M

L o u is -P h ilip p e

a je w s k i

New York City Council U N ITED STATES

R o c h o n

Kalam azoo UN ITED STATES

G a b r ie l

L a v o ie

U niversity o f O ttaw a CANADA R a y m

London UN ITED KIN GDOM

L a ib m a n

T he G raduate C enter o f the City University o f New York U N ITED STATES a r c

I . P a s h k o f f

D . K u r z

U niversity o f G raz AUSTRIA

M

N e a le

University o f New South W ales AUSTRALIA

S u s a n

D a v id

o n g io v i

K r ie s le r

U niversity o f New South Wales AUSTRALIA H e in z

M

R o d r ig u e z

University o f Ottawa CANADA

N e r i S a lv a d o r i

Université di Pisa ITALY

CONTRIBUTORS

M

a r io

S e c c a r e c c ia

R o s s

T h o m s o n

University o f O ttaw a CANADA

U niversity o f Vermont U N ITED STATES

S t e p h a n

M

S e ite r

U niversität H ohenheim G ERM A N Y A n w a r

S h a ik h

New School University U N ITED STATES

a tía s

V e r n e n g o

U niversity o f U tah U N ITED STATES

349

Index

absentee ownership, 233, 235-36, 273 access to information, 118, 122. See also ICT accumulation, rate of, 5, 6, 7, 47 “Aggregation in Leontief Matrices and the Labor Theory of Value” (Seton), 279 Aglietta, M., 245-46 agricultural improvements: alternative cases, 337-38; capital-saving, 33031; land-saving, 329-30; and pro­ ductivity, 95n.3, 327, 332-33; and rent, 338; Sraffa’s notes on, 33940 Akaike criterion, 18-19 allocation of resources, 57 “Appreciations of Mathematical Theories” (Edgeworth), 340 arbitrariness, problem of, 266-67 Argyrous, George, 181-201 Arkwright, Richard, 84 Arrow, Kenneth, 102, 293, 306n.3 Arrow-Debreu model, 307n.l3 Augmented Dickey-Fuller (ADF) test, 1517 Australia, 182-87, 192-93, 198 Ayres, Clarence, 275-76

Bagehot, W , 340 balanced trade across borders, 228-29, 231 Baldwin, Matthias, 92 Balogh, T., 157n.l2 banks: creditworthiness of, 213; failure of, 212; profits of, 11, 20n.3; role of, 205; speculation by, 213 Barnes, T.J., 224-25, 236n.3 Bartik, T.J., 169 Basel Accord, 213 Baumol, W.J., 224 behavior patterns, 12, 324-25 Bernent, William, 91 Berglund, Per, 178 Berle, A.A., 243 Bhaduri, A., 211, 254n.5 Blanchard, O.J., 148, 151 Blaug, Mark, 327-28, 342, 343, 343n.6 Block, F , 216n.l3 Block, Thorsten, 239-54 Bloomfield, A., 209 Bôhm-Bawerk, E. von, 224, 306n.2 Boulton, Matthew, 88, 90 Bowles, S., 254-55n.7 Boyer, R., 244-45, 254-55n.7 Bretton Woods era, 204, 208-10, 211-12, 215-16n.l0, 216n.l5

352

INDEX

business cycles: effect on economic policy, 154, 155-56; effect on male em­ ployment, 183-84; models for, 23, 247-53, 276n.2; Old and New, 2425, 247, 251-52; and the process of growth, 244; stability of, 23-24 Cambridge capital controversies, 220-22, 293-94, 301-2, 306-7n.l0 Cambridge Criticism, 279 Cambridge equation, 4-5, 39-40, 43, 47, 247, 254-55n.5 Campbell, Donald, 265 Canada, 10, 16-17 Cannan, Edwin, 327-28, 334-38, 341, 342, 343n.9 capacity utilization, 5, 41, 136 capital: accumulation of, 42, 54; deepen­ ing of, 50, 54, 56; endowment of, 301; flows of, 209; formation of, 307n.ll; human, 57; reversing deep­ ening of, 300-301, 303-5; reversing of, 293-94; scarcity of, 216n.l4 Capital and Growth (Hicks), 306n.5 capitalism, and conduct, 267-68 Capitalism, Socialism and Democracy (Schumpeter), 91 capitalist accumulation, 249 capitalist development, 81-82 capitalist dynasty’s consumption problem, 39, 52 capital-labor ratio, 101, 105 capital markets, 210 Capital (Marx), 285 capital movements, 210-11 capital-output ratio, 64, 105-6 capital productivity, decline of, 40 Carlyle, Thomas, 157n.9 Carr, E., 240 Carroll, Lewis, 150 causality: generally, 266, 274; horizontalist, 9-11; macro, 242-43; pro­ cesses of, 241, 265-66; reciprocal, 314-15; relations in, 264-65; universal, 264, 268 central bank: policies of, 211; power of, 9 10, 216n.l6; role of, 7, 204, 209 Chandler, Lester, 81 change: discontinuous, 92; rate of, 25-27, 30; structural, 254n.4; technologi­ cal, 123, 223, 275. See also technical change

Chiang, A.C., 62 Chick, V., 215n.l0 Chifley, Ben, 187 circuit theory of money, 211-13 classical model of growth, 36-40 class struggle, 127-28, 140-41, 225, 231, 286-87 Cobb-Douglas technology, 41, 44-45, 5253, 127, 131, 139 communication technologies, 9 4 ,95n.7,9899, 120. See also ICT competition, 94n.l, 101 Comprehensive Employment and Training Act (CETA), 166 computers, 102-3, 104, 107-8, 112 Conrad, A.H., 146-47 consumption: and demand, 250; and dis­ tribution of income, 164-65, 178; and income, 301; and savings, 3840, 58n.9, 252; workers’, 237n.5, 251 consumption-growth schedule, 38 Corliss, George, 88, 90-91 Corliss engines, 88-89 corporate bonds, 20n.4 Courts, J., 207 craft-based skills, 116, 117, 120-21, 223, 266. See also economies, craft system credit, theory of, 206. See also debt Cripps, F., 207 crisis, 225-26, 236-37n.4 Crisis Decades, 184 cross-industry transfers, 87-89 cultural evolution, 263-64 cumulative analysis, 311, 314-15 Darwin, Charles, influence of, 265-66, 269 Daugert, Stanley, 262 David, P.A., 105 Davidson, Paul, 221-22, 309 debt: burden of, 149, 214; and money, 2067; private, 112; public, 210-11; stability of, 208 deductive logic, and economics, 321-22 Deleplace, Ghislain, 3 demand, 234; changes in, 24, 111-12; growth of, 9, 306n.8; management of, 149; strength of, 288-89 demand-side effects, 103, 110 Dennett, Daniel, 266

INDEX

deregulation of financial markets, 213 Dicks-Mireaux, Louise A., 146-47 digital divide, 112 disability: determining, 185-87, 190-92; and job loss, 196; and males, 187— 88; medicalization of, 190-92; trends in support programs, 187— 88, 195-96 Disability Support Pension (Australia), 192-93, 193 discount rate, 10 disequilibrium processes, 226. See also equilibrium Dobb, Maurice, 94, 291n.4, 338-39, 344n.l2 Dow, J.C.R., 147 Dymski, Gary A., 220-36 dynamic process analysis, 311-12 dynamic stability, 293-94 Eatwell, John, 309 economic geographers, 224-25, 23637n.4 economic historians, 240 economic men, 240 economics: dynamics of, 311-12; history of, 239-40, 253-54; instrumentals method, 324-25; motivations and, 320; postulates of, 321-24; scien­ tific observations in, 323-24; testability of, 320-21; theories of, 246, 319-20, 323

Economics of Imperfect Competition, The (Sraffa), 280

Economics o f Welfare (Pigou), 340 economic systems, modem, 313 economies: craft system, 23-24, 26, 30-31, 95n.5, 208 {See also craft-based skills); demand deficient, 182; mass production, 23-24, 24-25, 2627, 30, 31-32; models of, 25-30; one-class, 70; organization of, 24, 272-74; spatial structures of, 223, 227, 228-29, 231-32, 237n.l0; surpluses in, 222, 232-33, 237n.7, 243; two-class, 62, 76; urban, 227, 230-32; zero-surplus, 226 economies of scale, 27, 82 Edgeworth, Francis Y., 340 education: and the economy, 116-17; late twentieth century, 119-21; learning facilitators in, 122-23; mass pro­

353

duction model, 119, 121; Oxford tutorial model, 118, 120-21, 122; post-secondary, 119; role of, 123; rural schoolhouse model, 117-19 “Edwin Caiman as an Interpreter of the Classical Economists” (O’Brien), 338-39 effective demand: and interest, 209; and prices, 284, 343n.4; theory of, 135— 37, 203, 249; usefulness of, 3, 207 Eichner, Alfred S., 5 Einstein, Albert, 270 ELR (employer of last resort programs): benefits of, 173-75, 196, 199-200; components of, 165; cost of, 170, 173; effects of, 163-64, 169-70; and unemployment, 182. See also unemployment Emmett, Ross, 325, 326n.3 employer of last resort (ELR) programs. See ELR employment: full, 79n.l, 129-30, 151; full­ time, 182, 188-90; and growth, 98. See also unemployment envelope of switch points, 45-47, 50, 53 equation system, dynamic behavior of, 3 032 equilibrium: backward-looking, 236n.l; competitive, 295; condition for, 296; full employment, 53-54, 12930; income effect, 306n.5; long-run, 68-70, 294-96; partial, 280, 286; and production, 177, 300301; stability of, 151-52, 295-96, 298-301; steady-state, 55-56, 23233; two-class, 70-73 “Error in Ricardo’s Exposition of His Theory of Rent, An” (Johnson), 338 Ertiirk, Korkut A., 23-33 Essay on Profits (Ricardo), 341 Euler’s Theorem, 65-66 Euromarket, 210 European Monetary System, 217n.l8 euthanasia of the rentier, 209-10, 216n.l4. See also revenge of the rentiers Evans, Oliver, 88, 92 evolutionary epistemology, 265 FairModel-US, 164-65, 175-76 Fazi, E., 62, 76 federal deficit, 169

354

INDEX

Federal Reserve Board: and interest rates, 18-19, 176, 179; role of, 210; and unemployment, 167, 168. See also central bank Financial Cycle, New, 210-11 Financial Cycle, Old, 208-10 financial instability, 204-5, 211-13, 214 financial markets, deregulation of, 213 Fiorito, Luca, 276 Fisher, Irving, 152 Foley, Duncan K., 35-57, 40 Fonseca, Goncalo, 178 Forstater, Mathew, 309-15, 325 fossil production function, 40-41, 44, 5355, 57 Francis, James, 91 Freedman, Craig, 144-56 Freitas, Fabio, 214-15 Friedman, Milton, 146, 155, 156n.4 Fulton, Robert, 88, 90

64, 211-12; Solow-Swan model, 51, 53, 54; stability of, 102, 135 Gupta, K.L., 76

General Theory o f Transformational Growth: Keynes after S raffa, The

Hagemann, Harald, 98-112, 325 Hall, R., 207 Hamilton, Alexander, 83 Harcourt, Geoff C., 5, 144-56 harmonic oscillators, 130 Harris D., 237n.l0 Harrod, R., 54-55 Harrodian employment dynamic, 135-37 Heilbroner, Robert, 309, 310, 325-26n.l Heimann, Eduard, 318 Henry, Jacques, 3 Hicks, Sir John, 117, 293-94, 306n.6 Hicksian conditions, 296-301, 305-6n.l history: and economic theory, 243-47, 312— 13; importance of, 245-46, 311 History (Cannan), 340-41 Hitch, C., 207 Hobsbawm, E., 181, 184 Hobson, John, 274, 276n.2 Hodgson, Geoffrey M., 243, 261-76 Hollis, Martin, 268, 317 horizontalism, 7, 8, 9-11, 204-5 Howe, Elias, 92 Hoxie, Robert, 272 Hume, David, 264 Hurwicz, L., 293, 306n.3 Hutchison, Terence, 317, 318-21 hysteresis, 157n.ll

(Nell), 3, 4, 81, 100, 116, 181, 310 Germany, 107-9 g-i spread, 13, 15, 19, 204, 208 globalization, 204 Glyn A., 208 Godley, Wynne, 207 gold, 208, 343n.3 Golden Rule, 4, 6, 10-13, 19-20, 204 Goodhart, Charles, 144 Goodwin, R.M., 58n.2, 127 Gordon, David, 309 Graduate Faculty of Political and Social Science, 310, 317-18 Granger causality tests, 7, 13, 19-20 gravitation, problem of, 232-33, 234, 235 Graziani, A., 204 Great Depression, causes of, 216n.l3 growth: and distribution, 43; and interest, 12, 17-19; and investment, 102; and labor, 35; rate of, 19, 47-48,

ICT (information and communication tech­ nologies): employment effects of, 102, 103-5, 109-10; and the global economy, 111; importance of, 98-101, 107 i-g relation. See g-i spread Iguiniz, Javier, 20n.l Imperial Germany (Veblen), 263-64 Inada conditions, 52-53, 53 income: and consumption, 299; distribu­ tion of, 136, 164-65, 250, 293-94 income effects, 293-94, 295, 296, 301, 306n.2, 307n.l4 industrial composition, 183 Industrial Revolution, 94, 95n.5, 262, 270, 271 inflation: controlling, 144-45, 153-54, 163; and interest, 10-11; irreversibility of, 146; and jobs, 167, 170; natural

Garegnani, Pierangelo: and the income ef­ fect, 306n.2; influence of, 309, 343, 344n.l0; on monetary policy, 8-9; on supply and demand, 207, 303 Gehrke, Christian, 327-42 “General Ludd’s Triumph” (song), 271 General Theory o f Employment, Interest and Money; The (Keynes), 8, 203, 216n.l6, 241, 243, 246

INDEX

rate of, 152-53; rates of, 10, 158— 59n.21 information and communication technolo­ gies. See ICT Information Economy, 182 input-output analysis, 221, 223, 224, 226, 236n.2 instrumental analysis, 313 interest rates: determination of, 7-9; and economic growth, 106, 178, 210, 214, 217n.l7; effects of, 175-76, 211-12; and the Golden Rule, 19; normal, 9; overnight rate, 10; paid to workers, 64; and profit, 67-68, 70-74, 76-77, 291-92n.8; real, 20n2, 206; relationship to growth, 11—15; and risk, 5-6; short-term, 215n.6; use in theories, 4 Internet, 120. See also communication technologies; computers intraindustry innovation, 83-87 Invalid and Old-Age Pension Act of 1908 (Australia), 186 invention and growth, 95n.3 investment: deficiency of, 243; effects of, 111, 135-36, 250, 252; and sav­ ings, 235-36 James, William, 265 Jefferson, Thomas, 88 jobless growth, 98-99 jobs: availability of, 188, 193, 198, 199200; full-time, 183-84, 197-98, 199; part-time, 183-84 jobs guarantee programs, 199-200. See also ELR Johnson, Alvin, 317 Johnson, Harry G., 327-28, 333-34, 33842 Joint committee on Social Security (Aus­ tralia), 186-87 Kahn, Richard, 205, 280 Kaldor, Nicholas, 311; on accumulation, 5; on economic growth, 100, 111-12; horizontalism of, 7; influences on, 58n.l, 61, 94n.l, 146-47; on money, 13, 157-58n.l3, 204, 205; Radcliffe memorandum, 215n.2; on returns to scale, 103; on savings, 54 Kaldor model of savings, 62, 76-78

355

Kalecki, M., 253 Kant, Immanuel, 264-65, 266-67, 268, 269 Kaufmann, Felix, 317, 318, 319, 321-24 Kenyon, P., 5 Keynes, John Maynard, 8, 58n.l, 241-42, 245-46, 309 Keynes, John Maynard, works of: General

Theory of Employment, Interest and Money, The, 8, 203, 216n.l6, 241, 243, 246; Treatise on Money, 61; Treatise on Probability, A, 242 Keynesian employment theory, 62 Kindleberger, C., 216n.l3 Kipling, Rudyard, 155 Klamler, Christian, 343 Knight, Frank, 268, 317, 318-21, 326n.3 Knight-Hutchison debate, 318-21 knowledge, diffusion of, 93, 95n.6 Knowles, K., 146-47 Kriesler, Peter, 144-56 Kurdas, Chidem, 178 Kurz, Heinz D., 64, 327^12 labor: classical view of, 36-37; constraint of, 43^44, 56; cost of, 109, 251; demand for, 38, 54, 127-28, 12930, 139, 157n.l2; intensity of, 26, 28-29, 33, 98; productivity of, 40, 48, 134-35; reserve army of, 137, 139, 140; strength, 138, 139; sup­ ply of, 36-37, 43, 45-50, 54, 57, 142n.22 labor markets, 127-28, 156n.4, 182-85, 194-95, 196-97. See also employ­ ment; unemployment labor theory of value, 224, 236n.3 Laibman, David, 279-91 Landes, D., 251 Lavoie, Marc, 3-20, 20n.l, 204, 214 law of motion, 47 Lawson, Tony, 276 leakage rate, 132, 134, 141n.l0 learning by doing, 102, 108 Lectures on Political Economy (Wicksell), 306n.2, 307n.l2 Leeson, L., 146-47, 156n.3 Leontief, Wassily, 220, 221 Leontief model, 37, 223 leverage ratio, 6 linkages, directional, 233-34 Lipsey, Robert, 146

356

INDEX

Lotka-Volterra predator-prey system, 138 Lowe, Adolph: on analysis, 313-14, 318; and economic methods, 317, 32425; relationship with Nell, 310—11; theories of, 310-15; and unem­ ployment, 200 Lowell, Francis, 84-85 Lucas, Robert, 168-69 Lucas critique, 156n.2, 169 Luddites, 271 Maastricht Treaty, 109 machine process, 262-64, 268, 270-71, 273, 276n.6. See also change, technological machinists, 93, 95n.6

Maintaining Full Employment: Public Ser­ vice Employment and Economic Stabilization (Nell, Majewski), 169 Majewski, Raymond, 163-78, 254 Malthus,Thomas, 36 Maneschi, A., 76 market demand, 286, 299-300 markets, analysis of, 243, 296, 297-98 markups, 12, 141-4n.ll Marquetti, A., 40 Marshall, Alfred, 157n.9, 280, 291-92n.8, 327 Marshall Plan, 216n.l2 Marx, Karl: analysis by, 327; on capital­ ism, 82-83, 87, 94n.l; and class struggle, 48-49, 85, 127-28, 32425; critiques of, 220-21; on his­ tory, 36, 254n.2; importance of, 58n.l, 81, 309; on profit, 47; on supply and demand, 285, 287 Mason, William, 85, 92 mass customization, 121-22, 123 mass production system, 181-82, 223, 252 matter-of-fact knowledge, 263, 269, 274 Mature Age Allowance (Australia), 198 Maudslay, Henry, 90 McCloskey, Diedre, 239, 254 McGahey, R., 169 McKay, Gordon, 92 Mead, George Herbert, 265 Means, G., 243 mechanization: effects of, 274-75; and habits of thought, 262-63, 271-72; rate of, 127-28, 271-72 Medicare, 173. See also Social Security Mehlum, Halvor, 254

Methodenlehre der Sozialwissenschaften (Kaufmann), 319 Metzler, L., 293 Michl, Thomas R., 35-57, 40 Milberg, Will, 254, 325-26n.l Mill, John Stuart, 340, 343n.4 Minsky, Hyman, 204-5, 208, 214, 221-22, 254n.5 models: limitations of, 168-69; uses for, 246-47, 247 Modigliani, F., 76 Mohnen, P., 236n.2 monetary circuit, 205-8, 212 monetary policy: and central banks, 25, 159n.24; effect on interest rates, 7 9; failure of, 147, 150, 153-56; and unemployment, 155; use of, 145, 241, 253 money: price of, 286; supply of, 10-11, 175-76, 206; theories of, 203-4, 205-8, 215-16n.l0, 215n.7 Money in Motion (Deleplace, Nell), 3 Mongiovi, Gary, 33, 276, 291, 317-25 monopolies, 103-4 Moody, Paul, 85 Moore, B.J., 7, 204, 205 Moore’s Law, 103 Muckl, W.J., 76 multiplier mechanism, 242 Mumford, Lewis, 271 Myrdal, G., 311 NAIRU (nonaccelerating inflation rate of unemployment), 148, 154, 155 natural interest, 206 natural rights of property, 269, 272 NAURI (nonaccelerating unemployment rate of inflation), 149-50, 155 Neale, Megan, 181-201 Nell, Edward J.: arguments of, 261; on business behavior, 81, 181, 214; on class society, 61; on economic the­ ory, 23, 100, 221, 243, 244, 317; and education, 115-19, 123, 279, 309-10; and ELRs, 163, 165; on interest rates, 11-12; issues ad­ dressed by, 3-4; on knowledge, 268; on money supply, 11; relationship with Lowe, 310-11; on Sraffa, 221-23, 235; and Transfor­ mational Growth, 82, 164-65, 175— 76; and unemployment, 200

INDEX

Nell, Edward J., works of: General The­

ory o f Transformational Growth: Keynes after Sraffa, The, 3, 4, 81, 100, 116, 181, 310; Maintaining Full Employment: Public Service Employment and Economic Stabili­ zation, 169; Money in Motion (Deleplace, Nell), 3; “On Long-Run Equilibrium in Class Society,” 61 Nell, Marcella, 115 Nell’s hypothesis, tests of, 15-16 New Economy, 56-57, 111-12 Newstart Incapacitated (Australia), 198 Newton, Isaac, 270 nonaccelerating inflation rate of unemploy­ ment (NAIRU), 148, 154, 155 nonaccelerating unemployment rate of in­ flation (NAURI), 149-50, 155 Notes on Malthus (Ricardo), 337, 343n.7 Nourse, E.J., 146-47 Nurkse, Ragnar, 209 O’Brien, Denis, 338-39, 342, 343 oil price shock, 13 Okun’s Law, 106, 177 Oliner, S.D., 104 “On Long-Run Equilibrium in Class Soci­ ety” (Nell), 61 ontological commitments, 264-69 output adjustments, 26-27 overaccumulation, crisis of, 48-49 own-substitution effect, 302, 306n.6, 307n.l4 Ozanne, R., 146-47 Parguez, Alain, 204, 214 Park, Cheol-soo, 178 Pashkoff, Susan I., 293-305 Pasinetti, Luigi L., 54, 55, 58n.l, 76 Pasinetti model, 62, 68, 70 patents, 84, 85, 86, 88 Path o f Economic Growth, The (Lowe), 310 Pechman, J.A., 146-47 Peirce, Charles Sanders, 265 permanent control, 322 Perron, P , 15, 20-21n.8 Petri, Fabio, 305 Phelps, E.S., 146 Phillips, Thomas F., 115-23, 146-47 Phillips curve: long-run, 155, 158n.l4; Samuelson-Solow, 145

357

philosophers, influence of, 264-65, 26869, 317, 318 Pieper, Ute, 254 Pivetti, M., 9, 204, 215n.9, 216n.l3 Poincare-Bendixson theorem, 32 Political Economy and Capitalism (Dobb), 291n4 Pollin, Bob, 221 Pollitt, Brian H., 344n.l2 Ponzi finance, 212-13, 214. See also shell games Popper, Karl, 265 population growth, 35, 131-32, 135, 139— 40; and unemployment, 128, 137 Post-Keynesian theory, 61 prices: dynamics of, 151, 288-89; general level of, 25; market, 285-86; natu­ ral, 283-84, 291n.3 Principles (Marshall), 340 Principles of Scientific Management (Tay­ lor), 273-74 Principles (Ricardo), 283-84, 327, 328, 336-37, 341, 342 production function, neoclassical, 52-53

Production o f Commodities by Means of Commodities: Prelude to a Cri­ tique of Economic Theory (Sraffa), 220-21 production with a surplus, 229-30 productivity growth: in the business cycle, 99-100, 105-6; and demand, 94n.l; and interest rates, 20n.2; and labor, 36-37, 102-3 profit, rate of, 334; determinants of, 1112, 64, 68, 207-8; and the Golden Rule, 19; and interest, 7, 9, 67-68, 70-74, 206, 210, 291-92n.8; law of tendency to fall, 47; and sav­ ings, 62, 255n.8; and wages, 21213, 230 profit share, 48 propositions, analytic and synthetic, 319—

20 Prosperity and Public Spending (Nell), 3 Quarterly Journal of Economics, 342 railroads, 91-92, 93-94 real balance effect, 62 real output, 25 reflux principle, 206-7 regulation school, 244-45, 250-51

358

INDEX

religion, outbursts of, 271-72 rents: agricultural, 328-29; com, 330-31, 332-33, 334, 336, 337-38; defini­ tion of, 331; effects on, 328; and monopolies, 237n.7; relationships between, 338 “Report on Manufactures” (Hamilton), 83 retirement age, 198 returns to scale, 99, 102-3 revenge of the rentiers, 211. See also eu­ thanasia of the rentier Reynolds, L., 146-47 Rhetoric o f Economics, The (McCloskey), 239-40 Ricardo, David, 36, 110, 283-84, 32742 Ricardo’s doctrine, 336-37 Rigby, D.L., 236-37n.4 Roberts, Richard, 85, 90 Robinson, Joan: on accumulation, 5; on demand, 280; influence of, 309; in­ fluences on, 58n.l; and interest, 9; on monetary policy, 8-9, 205 Rochon, Louis-Philippe, 203-14, 204 Rodriguez, Gabriel, 3-20, 15, 20-2ln.8, 20n.l Rogers, C., 8 Romer, P.M., 100, 101, 111-12 Rosenberg, Nathan, 89 Routh, G., 146-47 Ruffini, Emiliano, 305 Rutherford, Malcolm, 272, 276 Ryle, Gilbert, 117 Salvadori, Neri, 61-79, 343 Samuelson, P.A., 76, 306n.3, 306n.7, 32728 savings: by capitalists, 8, 58n.6, 62-63, 68, 76-77; class-structured, 52, 5556, 57; and growth, 74; and invest­ ment, 215-16n.l0; private, 133-34; proportional assumption of, 51-52; rate of, 135, 136-37, 139; three­ fold ratio of, 62; and wealth, 6162, 79n.2; workers’, 38-39, 62-63, 77, 78-79, 138 Sawyer, M., 157n.l2 Say’s law, 39 Scarf, H., 293 Schefold, Bertram, 214-15 Schreyer, Markus, 112 Schultze, C, 146-47

Schumpeter, Joseph A., 81, 91-92, 267, 309, 342 Schütz, Alfred, 318 scientific decision, 322-23 Seccareccia, Mario, 3-20, 20n.l, 148, 214 Seiter, Stephan, 98-112 Sellers, William, 91 Serrano, Franklin, 214-15 Seton, Sir Francis, 279 Seton, Sir Francis, works of: “Aggregation in Leontief Matrices and the Labor Theory of Value,” 279; “Supply and Demand,” 279; “The Transfor­ mation Problem,” 279 sewing machines, 92, 93-94 Shaikh, Anwar, 127^11, 309 shell games, 198-99. See also Ponzi finance Sichel, D.E., 104

Significance and Basic Postulates of Eco­ nomic Theory, The (Hutchison), 318-19, 319 Simple Theory of Effective Demand, The (lecture, Nell), 3, 4 Slater, Samuel, 84 Smale, S., 306n.3 Smith, Adam, 283, 287, 311, 327 Smith, Jonathan, 344n.l0 Smithin, John, 8, 211, 214 social order: and causality, 267; con­ straints of, 215n.6, 250, 324-25, 325-26n.l; and innovation, 122; relationships in, 243, 291n.6 Social Security, 56, 173, 191-92 Solovian mechanism, 54, 59n.l5 Solow, Robert M., 54, 57, 104, 240-41 Solowian growth model, 101 Solow Paradox, 99, 105, 111, 112n.4 Sombart, W., 240, 241 Sraffa, Piero: importance of, 309; relations with Cannan, 343n.9; on supply and demand, 207; theories of, 215n.3, 220-36, 280, 290-91, 33839; unpublished papers of, 339-41, 344n.l0 Sraffa, Piero, works of: Economics of Im­ perfect Competition, The, 280; Pro­

duction of Commodities by Means o f Commodities: Prelude to a Cri­ tique of Economic Theory, 220-21, 280

INDEX

stability, problems of, 293-94, 298, 304-5 stagflation, 149, 187 standardization, impacts of, 262-63, 26970 state, role of, 116-17, 254n.5 steamboats, 88, 90 steam engines, 87-89, 92-94; how they work, 270-71 Steedman, Ian, 291n.3, 309, 343 Steindl, J., 211 Stephenson, George and Robert, 92 substitution, principle of, 294-95 substitution effect, 296 Summers, Lawrence, 148, 151 supply, 287, 289-90 supply and demand analysis: balance of, 294; historical aspect of, 290; im­ portance of theory, 291n.l; limits of, 279-80, 285, 324-25; and natu­ ral prices, 284; use of, 205, 215n.l, 221, 222 “Supply and Demand” (Seton), 279 supply curves, 281-83, 286, 287-88 Swan, T.W., 54 Sylos-Labini, R, 23, 244 symmetry, assumption of, 280 taxes: effects of, 168, 253; in labor market dynamics, 132, 134, 167, 173; on produce, 335 Taylor, A., 157n.l2 Taylor, Frederick Winslow, 273-74 technical change: and economic growth, 102; effect on employment, 135, 137; Harrod-neutral, 40, 44; labor saving, 40-42, 98-99; Marxbiased, 40-41, 45-46, 53, 56 technological convergences, 89-94 technological determinism, 261-62, 26364 technological dynamism, 222-23 technological regimes, 181-82 technology, 23; changes in, 35, 40-42, 4 4 45, 81-82, 91-92, 93, 100-101, 244-45; cultural effects of, 26364; diffusion of, 82, 83-85, 86-88, 95n4; importance of, 261; of pro­ duction, 37; social problems with, 271, 272-74; steady-state, 50; uni­ versality of, 82-83, 90, 93-94 telegraph, importance of, 94, 95n.7 Ten Raa, T , 236n.2

359

textile manufacturing, mechanization of, 83-87, 90, 92-94 Thatcher, Margaret, 217n.l8 Theory of Business Enterprise, The (Veblen), 262 “The Pure Theory of Monopoly” (Edgeworth), 340 thermodynamics, 87-88 ‘T he Transformation Problem” (Seton), 279 Thirwall, A., 157n.l2 Thompson, Edward, 271 Thomson, Ross, 81-94 thrift, paradox of, 142n.l3 time, 234, 244 Tobin, A., 158-59n.21, 158n.l4 Tooke, Thomas, 205 Transformational Growth: and education, 119-20; and employment, 182; his­ torical basis of, 310—11; methodology of, 100, 168-69, 204; role of education in, 116; and tech­ nical change, 95n.3, 244; and types of economies, 251; usefulness of, 82, 164-65 transition-to-work incentives, 192-93. See also ELR; unemployment transportation costs, 225 Treatise on Money (Keynes), 61 Treatise on Probability; A (Keynes), 242 Trevithick, Richard, 92 Trieste debates, 309 Trieste economic view, 203 Turner, H.A., 146-47 unemployment, 48-49; and the business cycle, 154, 200; and disabilities, 186-88; effect of mass production on, 181-82, 193; effect on wages, 133-34, 134, 135, 139; and ELR rates, 167; female, 197; and growth, 102, 136-37; and inflation, 104, 146, 147, 149, 158n.l8, 159n.2; and interventions, 131; long-term, 148, 149, 150-51, 152, 155, 158nn.l5-18, 165, 182; male, 184, 193, 194-95; policies on, 6, 155, 156-57n.5, 197; pool of, 158n.l5; rate of, 128, 144-45, 14748, 150, 153-54, 157n.l0; short­ term, 148, 151, 152, 153; unions effect of unions on, 253

360 INDEX

unions, 131, 215n.6, 253, 272 universal machines, 87, 93 universal premises, 268 University in Exile, 310, 317 urban modeling, 223-25, 229-30, 232-34, 235-36 US, interest and growth in, 15-16 Ussher, Leanne, 178 utilization period, 110

Value and Capital (Hicks), 300-301, 307n.ll Veblen, Thorstein, 81, 261-76, 311 Veblenian dichotomy, 262 Vemengo, Matias, 203-14 Vienna Circle, 321 Viner, Jacob, 158-59n.21 Volcker, Paul, 210 Vordoom’s Law, 106, 177 wage-floor assumption, 47, 49-50, 158n.l4 wage-profit schedule, 37-38 wages: behavior of, 247; and consump­ tion, 178; effects of, 48, 290; and mechanization, 140, 272; mini­ mum, 179; money, 163-64, 252; and productivity, 177; rate of, 48, 150, 230; setting of, 51, 253; stick­ iness of, 23, 132-33, 134; and un­ employment, 58n.3, 135, 150-51, 152, 242 wages, real: in agriculture, 328-29, 33233; determining, 36-37, 235, 282; equilibrium in, 27-28, 131; and in­ terest, 210; and investment, 142n.l4; and labor supply, 29, 12729, 130-31, 138, 177, 250; and output levels, 343n.2; and produc­ tivity growth, 101; theory of, 207-8

Walras, Léon, 301, 307n.ll-12 Walras’s Law, 297, 299-300 warranted path, 17, 136, 137-38, 139, 142n.l6 Washburn, Ichabod, 91 Watt, James, 87-88, 90 wealth, 182, 233, 328 wealth effect, 106-7 Wealth o f Nations (Smith), 283 Weaver, 188 Webber, M.J., 236-37n.4 Weber, Max, 268 welfare, 188, 190-91, 192 welfare state, 131, 197-98 welfare system, 182-83, 184-85, 187, 200-

201 welfare-to-work, 184-85, 198-99. See also ELR; unemployment Wicksell, Knut, 301, 306n.2, 307n.l2, 327 Wicksell’s monetary circuit, 3-4 widow’s cruse parable, 61 Wilkinson, David, 84, 90 Wilkinson, John, 90 Winnett, Adrian, 305, 306n.5 Winsten, J.R., 146-47 Wolff, E.N., 224 Wood, A., 5 workers: diffusion of technology by, 84, 90-91, 93; skills of, 86, 106, 16364; standard of living, 254-55n.5; virtual, 111 workfare, 192-93. See also welfare Works (Ricardo), 338-39, 341, 343n.9 Worldly Philosophers, The (Heilbroner), 310 Wray, Randall, 7-8 Young, Allyn, 311