Macroeconomic Measurement Versus Macroeconomic Theory (Routledge Frontiers of Political Economy) [1 ed.] 0815353340, 9780815353348

Table of contents :
Cover
Half Title
Series
Title
Copyright
Contents
List of tables
List of graphs
List of figures and diagrams
Preface
1 Introduction
Introduction
1.1 Why this book?
1.2 Conceptual differences extend to the cores of neoclassical macro theory and statistical macro-measurement
1.3 The two very different meanings of ‘micro-founded’
1.4 The differences between macro-models and macro-statistics have a history
1.5 Macroeconomic events outside the framework of neoclassical macroeconomic models as well as macroeconomic statistics
1.6 The changing boundary of our idea of ‘the (macro) economy’ and the immensely measurable nature of monetary transactions
1.7 Hey, national accounts are political accounts
1.8 Overarching integration: the modern flow of funds / national accounts
1.9 An overview of the key differences between DSGE macro-models and macro-measurements
1.10 Mitchell-style business cycle indicators, the accounts and DSGE models
1.11 The conceptual model of the book (1): five interrelated phases of development of a macro-statistical variable
1.12 The conceptual model of the book (2): cases
1.13 The conceptual model of the book (3): meta-formulas
2 Money, prices and pricing
2.1 Introduction
2.2 Transactions
2.3 Prices and pricing
2.4 Money
2.5 The nature prices and pricing in a monetary society
2.6 The nature of the statistical production boundary
2.7 National accounts as an instrument of control
3 Money and how it’s estimated
3.1 Introduction
3.2 Money and its measurement
3.2.1 Monies, manuals and measurement
3.2.2 The flow of funds as an overarching model
3.2.3 The monthly monetary press release of the ECB and the macroeconomic formula of everything
3.2.4 Single accounting concepts of the account of money (1): Friedman and Schwartz
3.2.5 Single accounting concepts of the amount of money (2): Divisia indices
3.2.6 Quadruple accounting aggregates: money and debt
3.3 Money and the models
4 Labor and unemployment
4.1 Introduction: the discussion at the heart of macroeconomics
4.2 The concept of involuntary unemployment
4.3 Measured employment, unemployment and labor flows
4.3.1 Where do the concepts and definitions come from? A bit about the International Labor Organization
4.3.2 The ILO definitions of employment and unemployment and the influence of the US Congress
4.3.3 The ILO ‘periodic table’ of monetary and non-monetary work, unemployment and leisure
4.3.4 The difference between people and jobs
4.4 Neoclassical ideas about labor and the working of the labor market
4.4.1 Concepts and definitions: leisure and the difference between people and hours
4.4.2 Deconstructing the future
4.4.3 Fundamental and non-fundamental ways to make the models consistent with high unemployment
4.4.3 Models which take the statistics at face value
4.5 Voluntary and involuntary declines in hours of labor and unemployment during the Great Depression and beyond
4.5.1 The anomaly
4.5.2 The institutional/statistical explanation of the anomaly
4.5.3 The neoclassical explanation of labor during and after the Great Depression
4.6 Summary
5 Capital (and land)
5.1 Introduction
5.2 The (r)evolutionary nature of capital
5.3 Capital and the statisticians
5.3.1 Concepts and definitions
5.3.2 Measurements
5.3.3 Valuations
5.3.4 Volumes
5.3.5 Natural capital?
5.4 Capital and the neoclassicals
5.4.1 The concept, implicit or otherwise
5.4.2 Back to the way ahead
5.5 A comparison
6 Consumption
6.1 Introduction: the concept
6.2 Consumption in the national accounts: definitions and operationalizations
6.3 Consumption in the DSGE models
6.4 A comparison
7 I stands for gross fixed capital formation
7.1 Introduction
7.2 Concepts: the changing nature and definition of gross fixed capital formation
7.3 Statistical definitions
7.4 ‘Investment’ as a variable in the DSGE models
7.5 Time consistency
7.6 Measurement and our intertemporal understanding of the macroeconomy: the rise and decline of the investment rate of the Western world, ca. 1807–2018
7.6.1 The USA: exceptionalism
7.6.2 France and the UK: catching up versus falling behind
7.6.3 Germany, the Netherlands and the Nordics: archetypes
7.6.4 Italy, Spain
7.7 Overview
8 Unreal production
8.1 Introduction: how real is ‘real’ production
8.2 Single and double deflation
8.3 Putting formulas to the test: superlative index numbers and Dutch agricultural production and prices, 1851–2016
9 Macroeconomic unit labor costs as we measure them are no indicator of competitivity
9.1 Introduction
9.2 NULC, RULC and ULC
Epilogue
Index

Citation preview

Macroeconomic Measurement Versus Macroeconomic Theory

Ideally, scientific theory and scientific measurement should develop in tandem. In recent years this has not been the case in economics. There used to be a time when leading economists, or their students, established or led statistical offices and took care that the measurements were consistent with the theory and vice versa. Not anymore. Macroeconomic theorists and macroeconomic statisticians do not even speak the same language any longer. They do use the same words, such as ‘consumption’, ‘investments’ or ‘unemployment’ but the meanings they attach to these words are often wildly different. This book maps the conceptual and definitional differences between macroeconomic theory and measurement and explores them in some detail while also tracking their intellectual, historical and, in some cases, ideological origins. It also explores possible policy implications. In doing so, the book draws on two separate strands of literature which are seldom used in unison: macro-statistical manuals and theoretical macro-papers. By doing so, the book contributes to the effort to bridge the gap between them without compromising on the idea that a meaningful science of economics should, in the end, be based upon individual people and households and their social, political, juridical and cultural embedding instead of a ‘representative consumer’, or Robinson Crusoe figure. This work is essential reading for students, economists, statisticians, and professionals. Merijn Knibbe was born in Veldhoven, The Netherlands. He studied economics at the Rijksuniversiteit (University of) Groningen. He published on production, income and institutional changes in Dutch agriculture between 1850 and 2015 as well on and a habilitation on developments in Frisian agriculture between 1505 and 1832 and worked at the Centraal Bureau voor de Statistiek in the Netherlands. He has also published on historical flows of feed, food and minerals in the Netherlands as has written many blogs on economic developments and economic theory after the Great Financial Crisis and the extent to which economic metrics and models can be used to map and analyze the nature of this crisis.

Routledge Frontiers of Political Economy

Cognitive Capitalism, Welfare and Labour The Commonfare Hypothesis Andrea Fumagalli, Alfonso Giuliani, Stefano Lucarelli and Carlo Vercellone Political Economy for Human Rights Manuel Couret Branco Alternative Approaches to Economic Theory Complexity, Post Keynesian and Ecological Economics Edited by Victor A. Beker The Dark Side of Nudges Maria Alejandra Caporale Madi Inequality and Governance Andreas P. Kyriacou A New Approach to the Economics of Public Goods Thomas Laudal Marx’s Capital after 150 Years Critique and Alternative to Capitalism Edited by Marcello Musto The Political Economy of Prosperity Successful Societies and Productive Cultures Peter Murphy Macroeconomic Measurement Versus Macroeconomic Theory Merijn Knibbe For more information about this series, please visit: www.routledge.com/books/ series/SE0345

Macroeconomic Measurement Versus Macroeconomic Theory Merijn Knibbe

First published 2020 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 Merijn Knibbe The right of Merijn Knibbe to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. 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. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book has been requested ISBN: 978-0-8153-5334-8 (hbk) ISBN: 978-1-351-13670-9 (ebk) Typeset in Galliard by Apex CoVantage, LLC

Contents

List of tablesix List of graphsx List of figures and diagramsxii Prefacexiii 1 Introduction Introduction 1 1.1  Why this book?  2 1.2 Conceptual differences extend to the cores of neoclassical macro theory and statistical macro-measurement  5 1.3  The two very different meanings of ‘micro-founded’  9 1.4 The differences between macro-models and macro-statistics have a history  10 1.5 Macroeconomic events outside the framework of neoclassical macroeconomic models as well as macroeconomic statistics 12 1.6 The changing boundary of our idea of ‘the (macro) economy’ and the immensely measurable nature of monetary transactions 14 1.7 Hey, national accounts are political accounts  22 1.8 Overarching integration: the modern flow of funds / national accounts  23 1.9 An overview of the key differences between DSGE macromodels and macro-measurements  25 1.10 Mitchell-style business cycle indicators, the accounts and DSGE models  28 1.11 The conceptual model of the book (1): five interrelated phases of development of a macro-statistical variable  30 1.12  The conceptual model of the book (2): cases  31 1.13  The conceptual model of the book (3): meta-formulas  32

1

vi  Contents 2 Money, prices and pricing 2.1 Introduction 38 2.2 Transactions 38 2.3  Prices and pricing  41 2.4 Money 42 2.5  The nature prices and pricing in a monetary society  47 2.6  The nature of the statistical production boundary  59 2.7  National accounts as an instrument of control  68

38

3 Money and how it’s estimated 3.1 Introduction 75 3.2  Money and its measurement  75 3.2.1 Monies, manuals and measurement 75 3.2.2 The flow of funds as an overarching model 77 3.2.3 The monthly monetary press release of the ECB and the macroeconomic formula of everything 84 3.2.4 Single accounting concepts of the account of money (1): Friedman and Schwartz 88 3.2.5 Single accounting concepts of the amount of money (2): Divisia indices 91 3.2.6 Quadruple accounting aggregates: money and debt 94 3.3  Money and the models  97

75

4 Labor and unemployment 4.1 Introduction: the discussion at the heart of macroeconomics 105 4.2  The concept of involuntary unemployment  106 4.3 Measured employment, unemployment and labor flows  112 4.3.1 Where do the concepts and definitions come from? A bit about the International Labor Organization 112 4.3.2 The ILO definitions of employment and unemployment and the influence of the US Congress 113 4.3.3 The ILO ‘periodic table’ of monetary and non-monetary work, unemployment and leisure 116 4.3.4 The difference between people and jobs 118 4.4 Neoclassical ideas about labor and the working of the labor market 121

105

Contents vii 4.4.1 Concepts and definitions: leisure and the difference between people and hours 121 4.4.2 Deconstructing the future 125 4.4.3 Fundamental and non-fundamental ways to make the models consistent with high unemployment 125 4.4.3 Models which take the statistics at face value 127 4.5 Voluntary and involuntary declines in hours of labor and unemployment during the Great Depression and beyond  129 4.5.1 The anomaly 129 4.5.2 The institutional/statistical explanation of the anomaly 130 4.5.3 The neoclassical explanation of labor during and after the Great Depression 135 4.6 Summary 140 5 Capital (and land) 5.1 Introduction 145 5.2  The (r)evolutionary nature of capital  145 5.3  Capital and the statisticians  150 5.3.1 Concepts and definitions 150 5.3.2 Measurements 154 5.3.3 Valuations 156 5.3.4 Volumes 159 5.3.5 Natural capital? 162 5.4  Capital and the neoclassicals  164 5.4.1 The concept, implicit or otherwise 164 5.4.2 Back to the way ahead 167 5.5 A comparison 170

145

6 Consumption 6.1  Introduction: the concept  177 6.2 Consumption in the national accounts: definitions and operationalizations  178 6.3  Consumption in the DSGE models  185 6.4 A comparison 194

177

7 I stands for gross fixed capital formation 7.1 Introduction 198 7.2 Concepts: the changing nature and definition of gross fixed capital formation  199 7.3 Statistical definitions 202

198

viii  Contents 7.4  ‘Investment’ as a variable in the DSGE models  205 7.5 Time consistency 211 7.6 Measurement and our intertemporal understanding of the macroeconomy: the rise and decline of the investment rate of the Western world, ca. 1807–2018  212 7.6.1 The USA: exceptionalism 213 7.6.2 France and the UK: catching up versus falling behind 214 7.6.3 Germany, the Netherlands and the Nordics: archetypes 215 7.6.4 Italy, Spain 216 7.7 Overview 220 8 Unreal production 8.1  Introduction: how real is ‘real’ production  226 8.2  Single and double deflation  227 8.3 Putting formulas to the test: superlative index numbers and Dutch agricultural production and prices, 1851–2016 233 9 Macroeconomic unit labor costs as we measure them are no indicator of competitivity 9.1 Introduction 243 9.2  NULC, RULC and ULC  243

226

243

Epilogue251 Index254

Tables

1.1 An overview of the key differences between DSGE models and the flow of funds/national accounts 25 2.1 Number of domestic servants, selected years, 1899–1981 (thousands)63 3.1 An excerpt from the flow of funds, US Federal Reserve Board (use and source table)78 3.2 From whom to whom matrix, long-term loans, 2017 end of third quarter ex. pension assets and liabilities 82 3.3 The ECB definition of the eurozone monetary aggregates M1, M2 and M3 (including deposits which are left out of these aggregates) 84 4.1 Flows of labor, UK, 2001–2018 119 5.1 Fixed capital items listed in the SNA 2010 152 8.1 Use and prices of artificial fertilizer in the Netherlands, 1914–1923239

Graphs

1.1 Credit advanced to Irish resident private-sector enterprises: real estate, land and development activities – outstanding amounts and transactions 6 1.2 Real private consumption, real private investment and real government consumption plus investment, USA: year on year change (%) by quarter, chained 2012 USD, seasonally adjusted annual rate 29 2.1 Other accounts receivable and payable, flows, Ireland 45 2.2 Farm gate milk prices, the Netherlands, before and after globalization of the market 52 2.3 Workers of the world 55 3.1 Distributional accounts, USA, top 1%, next 9%, next 40% and bottom 50%, assets and liabilities (selected assets and liabilities as well as totals), USD millions 83 3.2 From MFI credit to M3 money 87 3.3 Stock and flow of ‘long-term loans’ (mainly mortgages) of Irish households 88 3.4 M3 money and Divisia money year on year growth rates, eurozone, original 12 members 93 4.1 Unemployment and year on year change of earnings (%), UK, 1850–1939 106 4.2 Unemployment, UK and the Netherlands, 1910–2015 109 4.3 Headline and broad unemployment, USA, 1955–2017 110 4.4 Gross job gains and losses, USA, quarterly data, 2Q1992–2Q2018120 4.5 Declines in number of hours worked, USA, 1929–1933 and 1944–1949130 4.6 Unemployment, USA and Germany, 1910–2018 133 5.1 Resident non-financial companies, Ireland, long-term loans (liabilities): total and per lending sector, EUR€ millions 151 5.2 Ownership of fixed assets, households, the Netherlands, 2016 154 5.3 Worldwide vehicle registrations, 1960–2016 167 6.1 Historical development of different components of household consumption, Germany and Greece 180

Graphs xi 6.2 Components of household consumption (% of GDP), Germany and Greece 181 6.3 Worldwide production of cars, regional data 185 7.1 Nominal gross fixed investment rate, Sweden, 1807–2018 199 7.2 Nominal gross fixed investment rates, USA (% of GDP) 214 7.3 Nominal gross fixed investment rate, France and the UK, 1815–2017215 7.4 Nominal gross fixed investment rate, Germany and the Netherlands, 1807–2017 217 7.5 Nominal gross fixed investment rate, Finland, Denmark and Ireland, 1844–2017 217 7.6 Nominal and a ‘real’ gross fixed investment rate Italy and Spain, 1850–2017 218 8.1 Intermediate inputs (% of gross output), Dutch agriculture, 1851–1950, current and fixed prices 230 8.2 A Törnquist and a chained Fisher Price Index, Dutch gross arable output, 1851–2016 235 8.3 Producer prices of arable and livestock products, the Netherlands, 1851–2016 236 8.4 Gross arable and livestock output, Törnquist deflated volume, the Netherlands, 1851–2016 236 8.5 Gross arable and livestock production (% of total gross agricultural production), current and fixed prices, 1851–2016 237 8.6 Prices of different kinds of artificial fertilizer, plus Törnquist and chained Fisher Price Index 238 9.1 Nominal unit labor costs, selected countries in the eurozone, 2000–2018245 9.2 Compensation of employees per hour worked (% of German compensation), selected eurozone countries 245 9.3 Labor share, highest and lowest share per sector, sectoral data, the Netherlands, 2008–2013 246

Figures and diagrams

2.1 4.1 4.2 5.1 6.1

A Dutch stamp from 2017: private money Proposed classifications of people in the labor force framework Examples of search behavior The relation between the national accounts and ‘natural capital’ Sectoral consumption flows according to the SNA

43 116 117 164 181

Preface

A book like this, comparing theoretical and empirical concepts and definitions of (neoclassical) macroeconomic variables with the concepts and definitions of the (macro-)statisticians had to be written. As there isn’t one while considerable differences do exist. Which leads to misunderstandings. To be able to do write such a book, one has to be a ‘Jack of all trades’. I qualify. But as this proverb goes there is a trade-off to this: I’m a master of none. I’m not professionally occupied either with labor statistics or data on consumption, capital or, investment or with constructing ‘real’ variables (although I have constructed and estimated such estimates). For all these subjects there are people better equipped to write about the concepts and definitions, even when the economic history approach I use in this book does add value to the often technocratic analyses of the specialists. Also, I did not publish any DSGE model (and do not intend to do so. A systematic comparison of the statistical and the theoretical concepts is, however, lacking – so somebody had to bite the bullet. As far as I know, this book is the most systematic and in-depth overview available of the differences between the concepts and definitions of theoretical, neoclassical macro-economics and the macro-statistics as embedded in the National Accounts, the Flow of Funds and the Labor Statistics. As knowledge about the differences between the theoretical and statistical concepts is pivotal to the interpretation of the data as well as the models, it fills a void – it is for instance important to know that, generally, the neoclassical macro concept of consumption deals only with non-durable, non-government goods and services. My endeavor will often be wanting, shallow or even plain wrong, but it does show that such differences can and have to be pointed out if we want to move to a better science of economics. It is a call to arms. Not many economists share an interest in comparing basic traits of the statistics to those of neoclassical macro-theory even when many are interested in either the statistics or the theory. Some, however, do. Nathan Tankus provided me, unwittingly but intentionally, via his Twitter account, with invaluable links to articles and authors. The ideas of the economist Means, which are pivotal if one wants to move from a theory based on market prices only to statistics based on market but also on administered and regulated prices, would not be in this book without him. Early attempts of several of the chapters in this book have been read and criticized by Josh Mason and Dyane Coyle, while an early draft of the chapters

xiv  Preface on money has been presented to Daniel Mügge, Professor of Political Arithmetic, and colleagues at the University of Amsterdam. I owe a debt of gratitude to all of them. The ideas of Frits Bos, long ago a fellow student as well as a former colleague at the Centraal Bureau voor de Statistiek, permeate the book. The idea that rigorous quantification of social processes does not exclude a genuine and in depth interest in the fate of real ‘common people’, and even is a necessary part of investigating their lives, which is one of the guiding principles behind this book, is a clear consequence of long term contacts with the Economic History group at the University of Groningen.

1 Introduction

Introduction This book discusses the complex conceptual differences between macroeconomic measurements and neoclassical macroeconomic theory, which this chapter outlines in three parts: • •

First part of Chapter 1. Paragraph 1.1 The ‘why’ of the book, the three basic questions as well as some sources and methods of investigation. Second part of Chapter 1. Paragraphs 1.1–1.10. Theoretical backgrounds to the framing, discussion and interpretation of the results of the investigation. • 1.2 Conceptual differences extend to the cores of neoclassical macro theory and statistical macro-measurement; • 1.3 The two very different meanings of ‘micro-founded’; • 1.4 The differences between macro-models and macro-statistics have a history; • 1.5 Macroeconomic events outside the framework of macro-models as well as macro-statistics; • 1.6 The changing boundary of our idea of ‘the (macro) economy’ and the immensely measurable nature of monetary transactions; • 1.7 Hey, national accounts are political accounts; • 1.8 Overarching integration: the modern flow of funds (FOF) / national accounts; • 1.9 An overview of the key differences between DSGE macro-models and macro-measurements; • 1.10 Mitchell-style business cycle indicators, the accounts and DSGE models.

•

Third part of Chapter 1. Paragraphs 1.11–1.13 Elements of the structure of the story. • • •

1.11 The conceptual model of the book (1) – five interrelated phases of development of a macro-statistical variable; 1.12 The conceptual model of the book (2) – cases; 1.13 The conceptual model of the book (3) – meta-formulas.

2  Introduction

1.1  Why this book? What is macroeconomics all about? After 2008, I re-educated myself as a macroeconomist, and this became a pressing question, one that ideally has two answers. An empirical one and a theoretical one, which are supposed to be two sides of the same coin. But it turned out that in macroeconomics this is not always (and has not always been) the case. Until 2008 the theoretical side – as embedded in the neoclassical dynamic stochastic general equilibrium (DSGE) models – had dominated the field for several decades. But the worldwide economic downturn showed that theoretical macroeconomics provided insufficient answers to the events of the downturn as stated in terms of the national accounts, the flow of funds and the labor data. The measurement of macro-economic data had greatly evolved but not in unison with the development of theory (see Card (2011) about this for an in depth overview of ‘Labor’ and ‘Unemployment’). Two kind of ‘currencies’ had developed and the exchange rate between these two currencies was and is not easily tractable (Andrle, Brůha and Solmaz 2017). It turned out that even when the same words are used, the theoreticians crafting the models and the statisticians estimating the real world are not always talking about the same thing. Consumption as it’s measured is not ‘consumption’ as it’s used in the models, capital is not ‘capital’ and ‘unemployment’ is not ‘unemployment’. Also, different words are used to denote, at first sight, comparable concepts. Quite some neoclassical theorists among whom quite some Noble laureates for instance use the word ‘leisure’ to denote wat statisticians call ‘unemployment’. This linguistic confusion is not new. In 1944 Haavelmo noted, pondering the relation between economic theory and the then-burgeoning craft of measuring the macroeconomy, ‘the confusion . . . caused by the use of the same names for quantities that are actually different’ (Haavelmo 1944, p. 7). Another example: according to the majority of models, all government expenditure, called ‘government consumption’, is wasteful. By definition. It uses resources but adds nothing of value. This means that, according to the large majority of DSGE models, government expenditure on education or health care does not add to prosperity. According to the statisticians, though, ‘government consumption’ does add value which is the very reason to call it ‘consumption’. Clearly, two remarkably incongruous concepts of consumption are used. Ceci n’est pas une science? Yes, there are DSGE models that do accept the possibility that government expenditure might actually produce useful public goods or services (Stähler and Thomas 2011; Iwata 2012). But the core central bank models used to inform policy (Bokan et al. 2016) and, crucially, models used in education (Sims 2015) do not follow this road. These models argue that government expenditure is wasteful, while statisticians hold an opposite view. So, it turns out that we have to answer two questions, not one. The first is about theory: ‘what are neoclassical DSGE macro-models all about?’ The second is about measurement: ‘what are macro-statistics all about?’ Answering these questions is not always easy, especially when it comes to theory. As indicated, not all DSGE models use the same conceptualization of variables. Furthermore, many

Introduction 3 neoclassical macroeconomists are less than precise when it comes to d ­ efining variables. I’m not the first one to notice this. According to Thorstein Veblen, this fuzziness is a defining feature of what he called neoclassical economics. He stated about the tendency to translate everything into an unobservable internal discounting of pleasure and pain by atomistic individuals: It is not simply that the hedonistic interpretation of modern economic phenomena is inadequate or misleading; if the phenomena are subjected to the hedonistic interpretation in the theoretical analysis they disappear from the theory; and if they would bear the interpretation in fact they would disappear in fact. (Veblen 1909, p. 175) It was one of the reasons for him to coin the phrase ‘neoclassical’ for this kind of economic discourse; this book can be understood as an endeavor to check if this Veblen quote still has truth to it, over one hundred years later. Anyway, this lack of precision and the often implicit nature of assumptions makes it quite complicated to investigate the concepts of the variables used in the neoclassical models. The all-important concept of consumption which is at the heart of neoclassical macro has to be gleaned from occasional remarks in papers, footnotes and by reading between the lines. Macro-statisticians, to the contrary, take great care to define their variables. When it comes to consumption, Eurostat states: Actual individual consumption . . . refers to all goods and services actually consumed by households. It encompasses consumer goods and services purchased directly by households, as well as services provided by non-profit institutions and the government for individual consumption (e.g., health and education services). In international comparisons, the term is usually preferred over the narrower concept of household consumption, because the latter is influenced by the extent to which non-profit institutions and general government act as service providers. (Eurostat 2018c) Impressive tomes are written detailing and operationalizing these definitions to enable consistent measurement and even to enable measurement at all, while modelers conveniently often leave the government and also non-profits out of their concept of consumption, most of the times not being explicit about this. Such differences in culture – occasional footnotes versus impressive tomes, explicit and detailed statements versus implicit use – make it difficult to compare the concepts and definitions of the theoretical models with those of the statisticians. Sometimes, concepts can even only be understood by spotting omissions. Non-profits (churches, unions, amateur sports clubs, charities like the UK National Trust which with 2.551 square kilometer is the largest private landowner in the UK) are an explicit sector in the statistics. But I’ve never seen them mentioned in the models, not even between the lines. They are seemingly left

4  Introduction out of the neoclassical production and consumption boundary. Another example: most DSGE models not only exclude government production of public goods and services from their consumption concept but also exclude durable consumption goods, like cars or the proverbial use value yielding chair of Alfred Marshall (Marshall 1920). I only understood this when I ran into a DSGE model which explicitly included them and mentioned that including durables was the exception to the neoclassical macro rule. Another difference: statisticians by definition look at the recent or, sometimes, distant past. The models, choice oriented as they are, look at the near but also at the distant future. The main actor of the models, the ‘representative consumer’, makes, within the in the long run binding confinement that the economy always returns to a benign equilibrium, choices which designate this future (in a subsequent chapter I’ll discuss HANK models, models which have more than one representative consumer). The reason consumer durables are often not included in the models is related to this: existing consumer goods restrain the representative consumers in their choices (having a chair influences the discussion to buy a chair) which complicates modelling this future and enabling the ‘representative consumer’ to choose freely. Statisticians however have to include purchases of consumer durables as they are part of total monetary expenditure which means that you can’t leave them out with running into inconsistencies (why is paying for the services of a taxicab considered ‘consumption’ and buying a car not?), which points to another difference: the choice theoretical background of the models requires that the representative consumer is not constrained by other people. The measurements, to the contrary, are fundamentally about such constraints: transactions like buying a car are always with other people or organizations. The models with one representative consumer rules out such transactions by definition. You do not have monetary transactions with yourself. Summarizing: the differences between the models and the measurements are large and fundamental. This book sets out to flesh out and discuss some of these differences. Which means that a third question has to be added: ‘what are the main conceptual and definitional differences between the DSGE macro-models and the macro-statistics?’ Subsequent chapters will set out to answer the three questions as I’m not aware that any book doing this in a systematic fashion exists. The rest of this chapter will provide some background information and explain the multilayered structure of the book as well as the basic methodology used to answer the questions. Some additional remarks: in the first subquestion I introduced the word ‘neoclassical’. The reason to use this phrase is simple but pressing: the DSGE models are a conscious effort to embed the ideas of neoclassical economics about the nature of capital, labor and, especially, the homo economicus in a macroeconomic framework. There are other macro models, these will not be investigated. The attentive reader might have noticed that I did not use the phrase ‘gross domestic product (GDP)’. This variable is a keystone of the national accounts – but a keystone can only be a keystone in relation to a building. When I relate to measurements it is to the building, not the keystone.

Introduction 5

1.2 Conceptual differences extend to the cores of neoclassical macro theory and statistical macro-measurement Differences between neoclassical macro-theory on one side and macro-measurement on the other are manifold. This is not just the case when it comes to mere definitions. The rift extends to the foundation the statistics and the models are built upon, to how concepts are related as well as to the basic approach of how science should be done. Statisticians go to great length to define their variables as precisely as possible, engage in discussions about these concepts and are eager to adapt time series to definitional changes. These ideas are written down in specialized articles and elaborate handbooks. Unemployment is a case in point. It consists not only of long and short-term unemployment but also, among other things, of involuntary part time work and of people available for work but who, for whatever reason, are not seeking work. The manuals are clear on this. But: what is the binding element of all these statistical variables? The same question can be posed for the models. It took me some time to realize that Edward C. Prescott, one of the winners of ‘The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel’ (TSRPiESiMoAN), when using the word ‘leisure’, actually writes about what statisticians define and measure as ‘unemployment’ (Prescott 2016). He is far from the only TSRPiESiMoAN winner who does this even when no neoclassical manuals seem to exist which explain this specific use of language. Why do they do this? But there is more to this. What is the foundation of their belief that calling unemployment ‘leisure’ is warranted? Such questions lead us to the core of theory and measurement. A smart non-economist observer blessed with undue perseverance and starting to read neoclassical macroeconomic theory and trying to answer such questions about unemployment might, after reading tons of often bland and boring journal articles and central bank working papers, answer the question ‘what are Neoclassical DSGE macro-models all about’ with a single word: Utility! Utility is the non-monetary psychological variable which is central to the ‘representative consumer’, the single person also called ‘social planner’ which in the workhorse neoclassical macro-models stands for society and whose preferences about today and tomorrow permeate the models from beginning to end. Or better: from top to bottom. True, in more advanced models there may be more persons or even banks and entrepreneurs – but these maximize ‘utility’ too. It is the relentless maximization of utility present and utility yet to come which makes the model-word go round! As work does not yield utility but, to the contrary, disutility being unemployed is better than being employed! Which means that unemployment can be called ‘leisure’. This ‘utility’ of the ‘representative consume’, who stands for society, is somehow related to the variable with the same name in the microeconomic models of the homo economicus, a variable which pervades economic theory and which has done

6  Introduction so for quite some time (Marshall 1920), even when no mechanism for adding individual utility of individual people is provided.1 But ‘social’ utility is why neoclassical DSGE economists equate unemployment with leisure. Optimizing utility, and not money, makes the world go round in the models. But ‘utility’ is not at the heart or even the fringe of economic measurement. Far from it. An equally smart non-economist observer blessed with an even larger amount of perseverance would answer the question ‘what are macro-statistics all about?’, after reading tons of always bland and boring manuals about the NA and the FOF and after investigating screen after screen of data and scores of press releases, also with one word. But a quite different one: Transactions! Whatever the statisticians measure, it’s not ‘utility’. They do not even try. They measure transaction based monetary flows of income, production, labor, all kinds of expenditure and also credit in a granular way. Hey, you really should take a look at the data about domestic credit to the Irish construction sector before and after 2008, information which is routinely available in the datasets of the Central Bank of Ireland!   They also measure stocks of financial assets (including different kinds of money) and they measure sectoral debts and the monetary value of fixed assets. And they also directly measure one non-monetary variable: paid labor. And unemployment. And they measure prices and interest rates and, neat, the flows are interconnected. A change in one flow is by accounting necessity connected to offsetting changes in other flows. They do this by bottom up aggregating

120,000 100,000 80,000 60,000 40,000 20,000 0 –20,000

Outstanding amounts (stock)

Transacons (flow)

Graph 1.1 Credit advanced to Irish resident private-sector enterprises: real estate, land and development activities – outstanding amounts and transactions Source: Central Bank of Ireland, financial statistics Table A.14

Introduction 7 individual transactions by individuals, companies and government entities and even charities and churches. But they do not measure ‘utility’. They do not even try. Transactions, not optimizing utility, show how the world is going round. That’s also why statisticians call unemployment ‘unemployment’: people without work want to enter an employment transaction but aren’t able to find a counterpart meaning that they are ‘un-employed’. So it’s looking forward against looking backward, utility against transactions, bottom up against top down, granular data against stylized sectors and explicit optimizing against the evolution of society. Other models, more consistent with the statistics do exist (one example: Godley and Lavoie 2007). In this book I will however restrict my attention to the neoclassical ‘DSGE’ macromodels. The very design of these models complicates comparing statistical with theoretical concepts and definitions – as the theoretical ones are equated with psychological variables while a rather limited production and consumption boundary is used. Remember the absence of consumer durables, nonprofits, public goods and services and indeed even, as we will see, money and financial markets. As Willem Buiter stated about pre crisis DSGE models, precisely one century after Veblen: ‘Both the New Classical and New Keynesian complete markets macroeconomic theories not only did not allow questions about insolvency and illiquidity to be answered. They did not allow such questions to be asked’ (Buiter 2009). Consistent with Veblen’s prophecy, money and transactions disappeared from theory and fact. As present day DSGE models state: they are based on ‘sound’ analysis of ‘deep’ and ‘fundamental’ psychological relations which escape the laborious and gritty work of actual measurement and day to day action and hence the discussion about the fact one wants to measure and how this should be done. According to Arjo Klamer, this emphasis on deep, abstract and eternal truth below the factual surface is typical for the modernist view of life (Klamer 2001). Unfortunately, it also means one has to dig hard and deep and to read between the lines to find out what the modelers are really writing about. Even then, as there are no official manuals and formal definitions one DSGE model might use a concept in a slightly different way than another. Digging hard and deep however does yield results. Returning to the concept of consumption in the models, neoclassical authors state the next things: ‘Dynamic stochastic general equilibrium (DSGE) models typically (as do most models) treat government spending as wasteful; such spending does not contribute to enhancing private sector utility or productivity’ (Russel Kincaid (2008). Or to quote Daragh, Jacquinot and Lozej (2014, pp. 4, 5): ‘Neoclassical economic theory typically assumes that government expenditure is wasteful’ and ‘Regarding consumption goods, we maintain the assumption from many models that government consumption expenditure is wasteful’. Such sentences have to be taken literally. It is about all government spending, including spending on roads, education or healthcare. Mind that statisticians do not define government consumption as intermediate use by the government but as the production of public goods and services. But what if for instance as happened in the UK, the

8  Introduction postal services are privatized? A DSGE model about the situation after privatization is, taken the model literally, about another economy than a DSGE model about the situation before privatization. The boundaries do not change – but the area within does. The citations above are clear. But they are also hard to find. The models are silent on this – even when the citations above do seem to be consistent with the verdict of Veblen. There is more to this. The neglect of government production is remarkable. In the larger body of economic thought, expenditure on public goods and services – either individual consumed ones or public ones – is considered to be a part of total consumption (Samuelson (1954). The entire fifth ‘book’ of The Wealth of Nations (Smith (1784, first edition: 1776)) is dedicated to this. And analysis of government expenditure is of course a viable branch of the tree of economics. It’s not as if economists turn a blind eye to the government. The problem of the models is however already indicated in the last sentences of the famous 1954 Samuelson three page article about government expenditure: To explore further the problem raised by public expenditure would take us into the mathematical domain of ‘sociology’ of ‘welfare politics’. . . . Political economy can be regarded as one special sector of this general domain, and it may turn out to be pure luck that within the general domain there happened to be a subsector with the ‘simple’ properties of traditional economics. (Samuelson 1954, p. 389) Traditional aka, in the definition of Samuelson, political economy means, in this context: the neoclassical analysis of markets. Somehow, it’s not easy to use this apparatus to analyze non market transactions. But even then some DSGE models do accept the idea that government expenditure can be a boon to society (Stähler and Thomas 2011; Iwata 2012). A comparable situation exists for consumer goods. Often durable goods, like cars, are excluded for reasons of convenience: leaving them out circumvents the problem of how ‘use-value’ of durable goods after the purchase adds to utility.2 Most models exclude them. But some don’t leave this out (example: Monacelli 2009). It can be modelled, which means that defining government expenditure as wasteful by definition and omitting cars and chairs out of the models is a conscious choice which defines the core of the models. It’s all about utility. And activities yielding utility in most models are quite restricted, however this is often not well described in the models. Statisticians take the opposite approach. Take, again, consumption. When statisticians make international and historical comparisons, the concept of ‘Actual Individual Consumption’ (AIC) is used. This is an aggregate of household purchases of private durable and non-durable goods and services plus government expenditure on individual consumed public goods and services like education and health care but excluding collective public goods like roads and law and order, which are measured separately (Eurostat 2018b). There are good reasons to do this. Importantly statisticians are more or less forced to include ‘government consumption’ like education into their concept of consumption as there are non-trivial differences

Introduction 9 between countries as well as large historical changes within countries when it comes to the boundaries between individually consumed ‘private’ and ‘public’ goods and services. Healthcare is a case in point. The UK National Health Service (NHS) is public but some of the services provided by the NHS are private in the USA. Taking the inner conceptual consistency of DSGE models serious – which I do – means that the models consider expenditures on the NHS ‘wasteful’ while they consider comparable expenditures in the USA as ‘adding to utility’. The statisticians however look at transactions: as long as the services are part of the ‘money economy’ and based on transactions, like wages paid to nurses, they are within the production boundary, it does not matter if they are provided by the government, companies or even households themselves or non-profits, like churches (or ‘NPISH’, non-profits institutions serving households, in the terminology of the national accounts). Privatizing the postal service does not matter for the measurement of AIC. Monetary transactional activities are the core of macro-measurement – a net cast wide. If we want a consistent body of macro theory as well as macro measurement, the DSGE models will also have to cast their net wide, much wider than at the present. Neither the core variables – utility and transactions – or the production and consumption boundary of the respectively the statistics are consistent with each other.

1.3  The two very different meanings of ‘micro-founded’ The differences between statisticians and theoreticians stretch into semantics. Both use the phrase micro-founded. It’s the core idea holding DSGE models together. But modelers give it another meaning than the statisticians. The theoreticians assume that the neoclassical microeconomic utility formula which defines their homo economicus also applies to society as a whole. There is, according to them, one neoclassical ‘microeconomic’ utility curve for the entire macro-society, which means that the macroeconomic ‘representative consumer’ (their model of society) behaves in the same way as the microeconomic homo economicus. The relation between social and individual utility is not clear – the aggregation problems mentioned by Veblen (1909), Marshall (1920) and Arrow (1950) still persist today. But a utility curve for entire societies shaped like a utility curve for the homo economicus is still assumed. All of society acts as one giant homo economicus. Hence the phrase ‘micro-founded models’, which is shorthand for ‘neoclassical macroeconomic models which assume that a society behaves like one neoclassical individual’. In the case of a number of representative consumers with either different preferences, different power or different constraints on their actions this still holds: all of them behave like the micro economic homo economicus. Hence the phrase ‘micro founded’. Statisticians also use the phrase ‘micro-founded’. But with a totally different meaning. For them it means that their aggregate data are, in the end, based upon measurement and aggregation of myriads of monetary micro-transactions. Only one example: the title of an OECD paper by Coli and Tartamella (2015): ‘The Role of Micro Data in National Accounts. Towards Micro-founded Accounts for the Household Sector’. Instead of describing the influence of top down decisions on future states of the world it’s about bottom up

10  Introduction aggregation. The theoreticians and statisticians do not always seem to be aware of this difference – the words imprinted on their respective coins sound the same but are part of different languages and people do not seem to be aware of the existence of different languages. Also a transactions based measurement system like the national accounts (NA) or the flow of funds (FOF) by definition implicates that the macro-statistics are fundamentally not about a single ‘representative consumer’ but about relations between multiple people and companies and the government and banks and non-profits. The idea of a single representative consumer is opposite to the foundations of the NA/FOF. The accounts map and aggregate the monetary flows engendered by an incredible complicated maze of persons and organizations which all have monetary transactional relations which each other, in the case of debt and credit and in fact many production and consumption contracts even of a long-term nature. It is the opposite of the ‘representative consumer’ which, in the models, does not trade but who does have ‘trade-offs’ between the present and the future. Good things are happening. Especially since 2008, the models have been extended with more consumers, banks, foreign countries and even class: institutional detail is added and transactions are becoming important. Also, academic economists like Thomas Piketty and Moritz Schularick are extending the use and scope of existing statistics, also by adding their new own long term estimates of economic variables to the existing body of knowledge. And they are not the only ones making long-term and very long-term data are made available (Reinhart and Rogoff 2009; Dimsdale, Hills and Thomas 2010). But the rift is still deep and wide. Micro-founded is, by many an economist, still understood as a phrase which harkens back to theoretical neoclassical micro-economics. And by statisticians as a method of measurement and aggregation.

1.4 The differences between macro-models and macro-statistics have a history The dichotomy between neoclassical theory and measurement is not a recent occurrence. The transactions central to the statistics are part of the same ‘money economy’ which was emphasized by the theory and measurements of Thorstein Veblen’s best student and one of the dominating economists of the first half of the 20th century, Wesley Mitchell. In 1916 he stated: ‘Among recent tendencies in economic theory none seems to me more promising than the tendency to make the use of money the central feature of economic analysis’ (Mitchell 1916, p. 140). He pitted this against the psychology and utility oriented approaches of John Stuart Mill and especially William Stanley Jevons, in the event approvingly quoting a 1915 article of a young Cambridge professor by the name of J.M. Keynes. He is however quite positive about Alfred Marshall as, in line with the criticisms of Veblen and contrary to earlier utility oriented economists, Marshall emphasized real prices actually paid, which means: the transactions central to macroeconomic measurement (in this book, the 8th edition of his Principles of Economics from 1920 will be cited, Mitchell cites the 6th edition of 1912).

Introduction 11 Mitchell would, during the rest of his long career, put monetary transactions where his mouth was. He published the first real national income accounts of the USA, for the 1909–1919 period (King et al 1921) and the much more detailed companion book (King et al 1922), stressing the importance of these data for especially an analysis of the distribution of income. Later, he guided economists like Simon Kuznets and Morris Copeland and helped to conceptualize and organize their work on NA and FOF. Next to this, he was crucially involved with statistical business cycle analysis while he also stood at the cradle of the Friedman and Schwartz book about the monetary history of the United States (Friedman and Schwartz 1963). The hypothesis that no other 20th century economist contributed more to estimating the ‘money economy’ and translating the criticisms of Veblen of utility and neoclassical economics into measurable concepts can’t easily be refuted, which all serves to underscore the idea that the dichotomy between neoclassical macro-utility on one hand and macro-estimates of the monetary economy and the aggregate value of different kinds of transactions was already visible during the first decades of the 20th century. And while aggregate transactions were already measured in detail in King et al (1922), the problems with aggregating utility were also already clear to the mind of perceptive economists. To quote Alfred Marshall: the task of adding together the total utilities of all commodities, so as to obtain the aggregate of the total utility of all wealth, is beyond the range of any but the most elaborate mathematical formulae . . . if the task be theoretically feasible, the result would be encumbered by so many hypotheses as to be practically useless. (Marshall 1920, p. 521) This problem of measuring utility is still not solved: ‘utility’ in the models is not aggregated from the bottom up but introduced in the models as a top down unmeasured variable. A relation between individual utility of real persons and utility of the representative consumer representing society is not specified or mentioned. This circumvents but does not solve the aggregation problem already mentioned by Alfred Marshall and more thoroughly analyzed by Kenneth Arrow, an analysis which led him to coin the impossibility theorem (Arrow 1950). Nowadays, these criticisms still stand (Morreau 2014). Unless everybody is exactly equal it is not possible to aggregate individual preferences into consistent social preferences. The statisticians to the contrary do not use utility but ‘monetary transactions’ as their core variable. As every profit and loss account shows, there is a clear and identifiable relation between individual and aggregated transactions and, within the confines of a company or a person or an economic sector or even a country, between different kinds of expenditure. Macro-measurement is as such almost the antithesis of the models, while utility defies measurement and aggregation, monetary transactions are immensely measurable and enable aggregation in a big way. In the meanwhile the concept of micro-utility has regressed to the purely psychological variable already criticized by Thorstein Veblen and has not become a more advanced version of the ‘use-value’ related variable defined by

12  Introduction Alfred Marshall and dear to modern marketeers. It still defies measurement.3 While the NA are by now measured as a matter of routine, on a quarterly basis and the world over. In a sense and when it comes to macroeconomics, economic science has come full circle since the criticisms of Veblen. Albeit, when it comes to the models, on a more sophisticated level of mathematical description. And when it comes to measurement, on a much higher theoretical as well as practical level of statistics and measurement. The differences between models and measurement also extend to the social realm. Modelers often work at universities, are pressured to ‘make a name’ and do not have the same publication avenues as the often anonymous statisticians. ‘Heroes’, to borrow a phrase from the analysis of corporate culture, are people who published a lot in prestigious journals and who get cited a lot. But who are our statistical heroes? I’ve lauded Wesley Mitchell, an economist-statistician who in his age was very well known. But while lesser contemporaries like John Bates Clark, Alfred Marshall and Irving Fisher are still well known he is all but forgotten. John Bates Clark, one of the neoclassical economists starting to tinker with the concept of a classless, genderless and race less society modelled with the concept of the representative consumer even has a very prestigious price named after him. A Wesley Mitchell prize of economic measurement is lacking. . . . It’s not just that there is no such price. It’s also that economic statisticians do not seem to feel the need for it. Their culture is not aimed at making a name or becoming a scientific hero. But it’s aimed at measuring the economy in an anonymous way, while theoreticians literally need to make a name to keep their job and are part of a culture which glorifies individual honor bestowed upon them by society. The highest honor, TSRPiESiMoAN, is not awarded to institutes. Statistical work for various reasons often requires anonymity while main results are often anonymously published by institutions. As stated: differences are huge – even when TSRPiESiMoAN winners like Jan Tinbergen, Milton Friedman, Simon Kuznets and George Stigler started out as economist-statisticians – the last three of these at the national Bureau of Economic Research (the total number of prize winners associated with the NBER is 27). But the social and cultural world of the statisticians is not the same as that of the theoreticians.

1.5 Macroeconomic events outside the framework of neoclassical macroeconomic models as well as macroeconomic statistics Also, even when a variable like unemployment is covered by the statistics but not (at least not in a way consistent with the definitions and measurements of the statisticians, as we will see) by the models, both do not seem to bother to directly measure financial bubbles, even though these can be and are, to an extent, inferred from the quarterly NA data and even when upswings and downswings were the explicit subject of investigation of Mitchell (1913, 1927); Burns and Mitchel (1946) and Tinbergen (1939) and even when Friedman and Schwartz (1963) made an explicit distinction between what now are called financial crises

Introduction 13 and ‘normal’ upswings and downswings of the business cycle. The statisticians are large and by satisfied with measuring data on a yearly and not on a cyclical time scale. Also, modelers left many if not most of the causal variables identified by Tinbergen (1939) as well as the monetary variables identified by Friedman and Schwartz (1963) out of their models. Figure 1.1 shows credit to the construction sector in Ireland. As is well known, there was a construction bubble (and bust) in Ireland between 2000 and 2009. Which is perfectly visible in the data on credit (mind that the flows are quarterly data). These data are part of the integrated NA/FOF data – they are routinely available on the Eurostat website, among other places. But even when the event is measured the statistics do not label it ‘a bubble’ in any formal way. The NA data enable the identification of bubbles – but do not formally measure them. They focus, taking the year as their fundamental time unit, on expenditure, income and production and the way these flows are financed. The flow of funds also take lending and borrowing into account, as well as stocks of credit and debt, looking at banking but also at receivables and payables. As the FOF uses the same (sub-)sectors and terminology as the national accounts this means that the systems can be and are integrated. A major feat of economic measurement. To be more precise, in the modern integrated NA/ FOF accounts ‘income’ is not equal to expenditure but {income plus net credit} is equal to {expenditure on goods and services as well as expenditure on new and existing financial assets}, which enables the measurement of bubbles. But ‘bubbles’ are not part of this conceptual apparatus as the actual measurements are based upon years and quarters meaning that multi-year events are outside of the conceptual scope of formal measurement. In paragraph 1.10 we will see that cycle instead of calendar oriented statistics do exist, however. For quite some time bubbles also where outside of the conceptual set up of the DSGE models (Buiter 2009). Banks – one of the well-defined sectors of the national accounts – were for decades left out of the models. During the last years, theoreticians as well as institutional inclined economists have tried to extend their models with banks to enable their models to show bubbles. Claudio Borio coined and operationalized the idea of the financial cycle (Borio 2012). In this book the model of Bokan et al. (2016), an elaboration of part of the core DSGE model used by the European Central Bank, will often be used as a kind of default DSGE model. This model enables bubbles to exist. Contrary to Borio, Bokan et al. however do not use the measurements of time series data of the integrated accounts, even if these do show bubbles. The rift still exists. Another exemption: economic class. Economic class is not based on income and education but on ownership. Some people have to work for a living, others can live from the ownership and capital and land. Bokan et al. (2016) re-introduce a class they call ‘entrepreneurs’ which literally owns and rents out all the capital into their model. But they do not even bother to check if such a class exists and how it has to be conceptualized, defined and measured while this can be done. Fessler and Schürz (2017), who also tackle the question of ‘capitalists’ in the modern economy, go to great lengths to investigate if such a class exists, how it can be measured and how large it is. Their results are based on more or less the same classical, ownership related

14  Introduction economic definition of class which, among many other 19th-century economists, was adopted by Marx: Intergenerational wealth transfers are a main driver of class location. Our class typology can serve as an excellent proxy for the position of a household in the wealth distribution. We discuss why this class typology has many potential advantages with regard to the measurement and analysis of wealth. Class is key in order to understand wealth inequality. (Fessler and Schürz 2017, summary) Bubbles and busts and ‘class’ are, as yet, not an integrated part of the accounts or the models, let alone of an integrated science of economics were measurements and theory tally, even when Wesley Mitchell pioneered the estimation as well as the dating of business cycles while ‘class’ as an economic category was prevalent in the writing of many 19th-century economists. At present, monthly measurements of producer of consumer sentiment or industrial production are used to measure cycles while the work of Bokan et all. and Fessler and Schürz show that the class gap can be tackled, too. But class and the swings of the business cycle data are not yet well tied to either the national accounts or the models. Next to mending a rift there is a road to travel, too.

1.6 The changing boundary of our idea of ‘the (macro) economy’ and the immensely measurable nature of monetary transactions One can pose the question if the rift between macro-theory and measurement has always been as large as it is today. As shown, the rift has been present for quite some time but it can be argued that it has not always been as large as it is today. To show this, this paragraph contains a short introduction into historical aspects the development of national accounts and, somewhat later, the flow of funds, mainly focusing on the ‘production boundary’ or the question what is and what isn’t covered by theories and measurements. And how this differs from neoclassical economics. The focus of this investigation will be the government: should government production be covered by theory and measurement or shouldn’t this be the case? This won’t cover the entire production boundary but it will serve to show that boundaries changes, albeit in in the case of macroeconomics in an opposite way when it comes to the measurements and the models. We’ve seen that the NA/FOF include government expenditure in ‘final demand’ as it adds a boon to society. As stated, most of the models take the radical libertarian position that government expenditure is a waste, by definition. Recall the statement of Samuelson about the realm of what he considered to be ‘traditional economics’ – government expenditure was simply not covered by ‘political economy’ as he defined it, as this kind of political economy was only occupied with markets. Is it right to use this particular definition of ‘political economy’ and, more importantly, can we call this kind of exclusive market

Introduction 15 oriented thinking ‘macroeconomics’? Did ‘political economists’ indeed purge the government from their thinking and is defining the macro-economy as solely the realm of consumer choice of private goods sound economic procedure? There are reasons to doubt this. The government has for a long time been an important element within the production boundary as defined or used by generations of economists. Mitra-Kahn (2011) argues that it was Keynes who introduced the government into the accounts while earlier economists left the government out of their production boundary. He discusses Adam Smith, who in 1776 published The Wealth of Nations (Smith 1784, first edition 1776). According to MitraKahn (who also discusses earlier developments) this book had a large influence on our definition of the macro-economy. It is about the transactional monetary economy. Smith labelled many paid services (but not all) ‘unproductive’, an idea which, according to Mitra-Kahn, influenced government statistics (which, to be sure, at the time were a far cry from the modern national accounts) but somehow Mitra-Kahn is silent about Smith’s ideas about the government and seems to assume that Smith excluded the government from the production boundary. The boundary proposed by Smith was, according to Mitra-Kahn, around 1900 successfully challenged by Alfred Marshall who included more services but not yet the government in his idea of productive work and whose students, when they started to rise through the ranks of the UK bureaucracy, managed to change the statistics accordingly. The concepts of Marshall were at the end of the 1930s brushed aside by John Maynard Keynes, who (therewith extending the work of Colin Clark), introduced the production of the state into the national accounts to enable an analysis of non-inflationary wartime spending (Keynes and Rothbarth 1939; Keynes 1940). To be able to do this he also established the present day Office for National Statistics (ONS) in the UK, which had to measure the variables defined by Keynes while he also managed, during his trips to the USA, to get his ideas accepted in the US government offices, therewith sidelining Simon Kuznets and his more welfare oriented approach to national accounting. As a consequence, national accounting was set on a track which was meant to enable an analysis of a war economy and a central role for the government which, despite some tinkering at the margins (Bos 2003, 2013; Eurostat 2013), basically endures to this day. A system which, according to Mitra-Kahn, is not fit for an apt analysis of welfare oriented policies. The ideas of Mitra-Kahn are highly interesting. But they are UK-centered. And they need elaboration. The main conceptual contribution of Keynes to national accounting was not the introduction of the government into the accounts. The role of the state had long been stressed by economists like Adam Smith. And not just by him. It has been included in national accounts and preceding estimates of the national economy from the very beginning. In 1608 Simon Stevin, a Flemish/Dutch polymath who among many other things also invented or at least perfected decimal notation, improved windmill design and developed an ultra-fast ‘wind powered beach carriage’, published ‘Vorstelicke Bouckhouding’ (‘Accounting for Princes’) (Stevin 1608). This book (actually: chapter of a larger book) was inspired by an earlier French example. It was about keeping accounts

16  Introduction for the estate of the prince, not about accounting for the state or the nation. But such approaches to the wealth of the sovereign instead of the commonwealth would wane. A few decades later William Petty, a polymath who among other things developed a device for ‘double writing’ and who charted Ireland, wrote his Political Arithmetick (published in 1690 but written around 1676). This book argued and estimated among other things ‘That the Power and Wealth of England, hath increased above this forty years’. It was not about the estate of the prince. It was about the state of the nation. He wrote about employment, expenditure and wealth and made some comparisons with other states like the ‘Hollanders’ and the French. According to Mitra-Kahn (2011) this was not the first or last individual endeavor to define and estimate national prosperity. But it serves to illustrate the growing focus on the nation or, as Petty also called it, the ‘Commonwealth’. It also paid special attention to the productive role of the government, as the title of the second chapter alleges: ‘That some kind of Taxes, and Publick Levies, may rather increase than diminish the Common-Wealth’. This last point was not lost on Adam Smith when he published his Wealth of Nations about one century later. Not mentioned by Mitra-Kahn (and many other economists) the entire fifth and last part of the Wealth of Nations is devoted to productive government expenditure and the increasing role of the state. Clearly, ideas about productive government actions have a long pedigree when it comes to ‘political economics’ – the use of Samuelson of this term is clearly idiosyncratic even when, in a rhetorical sense, it enabled him to showcase his analysis. Anyway, Stevin as well as Petty as well as Smith focused on the money economy. This would change. Like the emphasis on the role of the government. Marshall is accredited by Mitra-Kahn with refocusing the attention of economists on markets and individuals and including many, albeit not all, services but not the state in the productive economy (Mitra-Kahn 2011). Reading Marshall’s Principles of Economics (8th edition, Marshall (1920)) shows that services are indeed included and a focus on individuals and markets is prevalent. As such his thinking was heavily influenced by economists focusing on ‘utility’ like Jevons and Böhm-Bawerk. According to him, the economy was in the end not about producing furniture, but about the utility a chair yielded even when real money prices did play a role on markets. For Marshall it however did not matter if the chair was produced by the market or the government. He admits the existence of government factories as well beneficial government policies related to health while he notes the relatively fast growth of the number of government employees. But despite this and contrary to Adam Smith (and while writing at the very height of British colonial power) he does not devote any special attention to the government as such, which is a mayor deviation from Petty and Smith. Also, national income or, as he prefers to call it, ‘the national dividend’ does play quite a role in his book. But he was behind the times when it comes to this subject. The extensive list of literature on national accounting in Bowley (1942) shows that in 1920, when the 8th edition of Economics was published, there already was quite an extensive literature about this ‘national dividend’. The idea propagated by Mitra-Kahn (2011) that it were mainly the ideas of Smith and Alfred Marshall

Introduction 17 which influenced the measurement of the wealth and income of the nation is clearly incorrect – scores of economists tried to estimate and define it and made large progress. Somehow Marshall missed this and did not use this literature to define the ‘national dividend’ in any precise way, though his ideas seem to be related to the flow of income (see also Bowley 1920). Much more attention is devoted by him to aspects like utility and consumer surplus – fundamentally nonmonetary variables – than to the boundaries of the money economy. It is well possible – my knowledge of the history UK statistics is not extensive enough to be able to corroborate this – that his students indeed changed the nature of UK statistics by putting more emphasis on services. But did they, as the work of MitraKahn implicates, also like Marshall more or less forget about the government as a distinct category of the national economy? If so, they were clearly behind their times. To show this we have to cross the ocean. Around the same time as Marshall published his 8th edition, an economist of the same stature but working in the USA was already making his mark: Wesley Mitchell. He became the head of the newly established National Bureau of Economic Research (NBER). As an economist, he did not put emphasis on utility but on the ‘money economy’ and its measurement, being well aware that the ‘money economy’ was not just about earning money but also about the challenging task of wisely spending money – for instance by housewives (Mitchell 1912). Comparing this with the bland statements of Marshall about the ‘utility’ of furniture, Mitchell was decades ahead. In Mitchell’s thinking, there was conscious life and dedicated effort, and not just furniture, at the other side of the monetary production boundary. Anyway, one of Mitchell’s most remarkable results was an early NBER study, ‘Income in the United States: Its Amount and Distribution 1909–1919’ (King et al 1921). This book, which estimates value added in the different sectors of the USA economy, discusses the production boundary in a much more specific way than Smith or Marshall. In its little read detailed companion volume (King et al 1922) we can find the next phrase in the chapter abut the government, written by King (King 1922, p. 201): In dealing with the product of government, the same criterion is used that has been applied in the industrial fields previously studied; namely what book or money income do individuals, as such, derive therefrom. Evidently, governmental units expend great amounts for wages and salaries, but they pay no dividends. Large sums are, however, disbursed in interest, mostly to private individuals, but to no inconsiderable extent to banks. Clearly, the government was included in the first major national accounts study of what already was the major economy in the world, based upon a purely monetary criterion: government employees are paid a wage. Including the government in the ‘national dividend’ was – contrary to the remarks of Mitra-Kahn – not an invention of Keynes. More important: actual measurement forced economists to be much more precise and clear about what they were talking about than Smith and Marshall ever were forced to be, which eventually led to concepts like

18  Introduction ‘actual individual consumption’, which includes the consumption of individually consumed goods and services provided by the government, like large chunks of education. Excluded are the depletion of stocks of natural assets as a negative and also excluded household work (King et al. 1921) but this omission is explicitly mentioned and a very rough estimate of the importance of household work is provided. Inclusion of the depletion of stocks of natural resources as a negative still has to be incorporated in the accounts. The influence of people like Mitchell but also Bowley and Colin Clark (who estimated integrated accounts for the UK (Clark 1937) can also be witnessed in the work of Keynes himself. The Economic Consequences of the Peace (Keynes 1919) still used a large amount of not well integrated data to prove Keynes’s point about the state of an (admittedly fragmented) national economy. Eighteen years later, in the autumn of 1939 when he published the articles which would become his ‘How to pay for the war’ in the London Times, he used national accounts data based upon the work of Colin Clark but (guided by Keynes) extended and updated by Rothbarth (Keynes and Rothbarth 1939). There are several interesting parts to this epoch making piece. First, it showed the power of national accounting as a tool for policy. Second, it contained income estimates per income group, a mayor advancement and a deliberate goal. As Keynes (1940) stated in the acknowledgments: ‘I have been assisted throughout on the statistical side by hlr. E. Rothbarth of the Statistical Department, Cambridge University, who is responsible, in particular, for the estimated division of total income between the income groups’. Third, it rightly envisioned government expenditure as expenditure with the same main goals as private expenditure: consumption and investment. This enabled Keynes to provide a much more systematic analysis of the economy than he could provide in ‘The economic consequences of the peace’: his work was finally up to the standard of King et al. (1921). He was well aware of the distributional effects of high wartime inflation: the poor would suffer most. Eager to prevent this he was able to state, with some quantitative precision, how much disposable income of different groups of consumers had to be restrained to lower consumption expenditure to be able to free resources for the war effort while keeping the purchasing power of the lower income group afloat at least to an extent. Importantly, he used the three basic national account approaches (income, expenditure and production) in combination. Nowadays, such data are a staple of GDP press releases in the shape of the ‘supply and use tables’ (Eurostat 2018a) though it has to be stated that the standard national accounts do not contain data on different income groups – Keynes’ main national income innovation has been discarded! See also Tily (2009). There is a nasty twist to this. Patnak (2018) argues that, at least inspired if not guided by John Maynard Keynes, during World War II the British government pursued a policy of ‘profit inflation’ in India, increasing profits and decreasing the purchasing power of the poor to free resources for the war effort – a policy which led to the Bengal famine of 1943–1944 which claimed the live at least three million people (though Keynes might not be ‘credited’ with rejecting the free grain offered to the British government for use in Bengal by the USA and Canadian governments).

Introduction 19 Mitra-Kahn is right to point out the flabbergasting feat that, while not a government employee, Keynes managed to establish an entirely new government office to measure the data he needed, and to stress his influence on the development of the national accounts in the USA in the same period. His catch that Keynes himself admitted that it was only in 1939 that he, according to his own testimony, really figured out why the models of Colin Clark were not fit for his purpose of designing a non-inflationary war effort – they left out ‘supply and use’ by the government but also, not mentioned by Mitra-Kahn and probably more important, the distribution of income – should be in the textbooks (Mitra-Kahn 2011, p. 211). But Mitra-Kahn misunderstands why Keynes shed his last Marshallian feathers. He assumes, that Keynes did not want to show an estimate of private sector GDP which declined because of the war effort and hence opted for a definition of GDP including the rapidly growing war effort of the government – which would show an increase of GDP. This really is of the mark. The entire goal of Keynes was to estimate how much private consumption and investment could and had to shrink. Part of his calculations implied that, as the government produces output and uses resources, too, a reduction of private expenditure of 10 percent won’t lead to a reduction of total expenditure with 10 percent as private expenditure is less than 100 percent of total expenditure. His analysis was no doubt influenced by his memories of the inflation of 1917–1920, which must have taught him that any model used to analyze the situation had to be coherent and consistent and hence to cover aggregate expenditure and total final demand instead of only private expenditure of it or the amount of money going around. And the logic of the model required him to estimate the total tax base in the UK, including government employees, and total investments, including investments of the government. Mitra-Kahn is also not right that the more welfare oriented GDP definitions of Simon Kuznets were ‘sidelined’ by Keynes. Hirschman mentions that Kuznets himself had a decisive influence on the planning of the US war effort with an analysis which, tough production and not income oriented, was highly reminiscent of ‘How to pay for the war’ (Hirschman 2016, pp. 82–83) and basically a kind of input-output analysis. And tough Kuznets (a student of Mitchell) indeed tried to engineer a more welfare oriented concept of GDP the production oriented approach was not a new concept. Actually, the first chapter of Kuznets National Income and Its Composition, 1919–1938, Volume I (assisted by Lillian Epstein and Elizabeth Jenks) is called ‘concepts, classifications and procedures’ and contains a whole taxonomy of different kinds of concepts of ‘national income’ based upon basically the same accounts (see also Bos 2003, pp. 9–16, 2013) for more extensive analysis of such points) while Kuznets (1955) includes military equipment in its definition of fixed capital. The idea that Smith and Marshall influenced statistics is right. But it is not right to state that their ideas decisively influenced the accounts. The development of the accounts knew an own, transaction and measurement based logic while there surely was an international and not just a UK community of economists and statisticians developing national accounting. The very fact of measuring ‘the economy’ forced economists to grapple with concepts and definitions in a way they had never done before. This development

20  Introduction gained momentum when measuring national accounts became institutionalized. Hirschman (2016) states that, before the 20th century, no country published national accounts data even when economists were tinkering with the concepts. This would, however, soon change. A non Anglo-Saxon example: between 1939 and 1941 Jan Derksen published as an employee of the Dutch Centraal Bureau voor de Statistiek (CBS) and no doubt inspired by Jan Tinbergen, a whole string of national account studies for the Netherlands which tackled conceptual and measurement questions (CBS 1939; Derksen 1940, 1941). After the war he would go on to become the head of the national accounts division of the United Nations, coordinating the international discussion about what should be measured and what shouldn’t. The already mentioned literature in Bowley (1942) is arranged by country, which abundantly shows the international character of developments in national accounting and the many persons influencing this development. The point was that ‘the economy’ was defined not just by a limited number of well-known Anglo-Saxon economists but also by an international group of economist-statisticians who defined but also discovered the transactional macroeconomy. It were these statisticians who added rigor, coherence and precision to the concepts and definitions and Keynes could clarify his concepts only after grappling with these statistics. It has to be noted that these statisticians and economists often were no students of neoclassical economists like Marshall but of institutional economists like Veblen and Mitchell (Kuznets got a training as an institutional economists at the Institute of Commerce in Kharkiv, Ukraine). And the most prominent (neo)classical educated economist playing a decisive role in establishing the (measurement) of the national, Keynes himself, put ample emphasis on the time and trouble it took him to get rid of old ideas. As a consequence of such actions and consistent with ideas of those like Samuelson (1954), which were very widely accepted, separate accounts for the government were established. Also, concepts like government consumption (i.e. consumption of individual households of government produced goods or services like education) and AIC were developed. It was a clean sweep. The government produces value, however conceptualized and measured and was included in the production and consumption boundary which were increasingly measured by specialized institutes. Even neoclassical models of the government became more explicit and rigorous. But for one exception. The DSGE models for whatever reason and without discussion purged the idea of a productive government from the models and stuck to the ideas and methods of what Samuelson called ‘traditional economics’. The government was excluded or at least the idea that the government was any good was denounced. Even if several authors showed that government production can be included as a positive. Excluding the government was, and is, a conscious choice. So, the macroeconomic production and consumption boundary changed. On the side of measurement it became much clearer and precise. On the side of theory the government was also was increasingly included in the neoclassical models. Until the rise of DSGE models after around 1970. A major regression. At this point it is necessary is necessary to dwell on the nature of the logic of the national accounts which, unlike the logic of the periodic table, evolves over

Introduction 21 time as economies evolve over time. These accounts are, despite a multitude of imputed posts, deeply rooted in the monetary nature of our economy. This monetary nature of the economy does not only consists of monetary prices but also and by accounting necessity of the social nature of transactions and flows of income and spending, which enables aggregation. This leads to the possibility to classify and measure household income, consumer credit provided by banks, investment expenditure of manufacturing or agricultural production. Monetary transactions – the ‘money economy’ of Wesley Mitchell – are by their very nature eminently and immensely measurable and show, when aptly classified, relations between sectors. The whole point of an individual price is its public and sticky nature. In a market the seller and the buyer both need to know the fixed money price, quality (including delivery conditions and transport) and quantity of a transactions (or at least the procedure how these are decided) ex-ante. In the case of taxes, which are more one-sided than market transactions, the way taxes are set also has to be known in advance. Any transaction enters into at least two sets of sometimes informal but more often formal accounts; the account of the seller and the account of the buyer and, in case of income taxes and VAT, also into the accounts of the government. Three sets of accounts measuring one transaction! In the case of companies, the value of sales is of course decisive for the ability to pay contractual factor incomes and purchased inputs. Bos (2003) shows that there are in fact eight accounts which, on the aggregate level, have to match. And although delineating sub-sectors like ‘transport’, ‘construction’, ‘education’ or ‘agriculture’ is not always as easy as boundaries shift over time it is possible to do this. The same holds a fortiori for the sectors – households, companies, non-profits, financial institutions and the government – and economic categories (wages, profits, interest, rent). These change over time but surely stay recognizable in the short, medium and often also in the long term. Economic actors are interconnected by the flows and stocks of incomes and sales denominated in the unit of account: like wages, profits, interest, sales, credit, debts. The ‘empirical discipline’ of measuring this economy forced economists to think more deeply about this economy than they had ever done before – the issue of actual individual consumption which raises its head once international or historical comparisons of household consumption are made being only one example. Also, all the actors had to be included in the model. Leaving out criminal organizations would yield, by accounting necessity, a black hole in the system. This transaction based logic of the measurements has been pivotal to the development of the national accounts and even more so to the flow of funds and has its implication for scientific economic theory. As Haavelmo stated: The practical conclusion of the discussion above is advice that economists hardly ever fail to give but that few actually follow . . . that one should study very carefully the actual series considered and the circumstances under which they were produced before identifying them with the variables of a particular theoretical model. (Haavelmo 1944, p. 7)

22  Introduction In national accounting, the economists actually did this, which led to the modern NA/FOF data which do not only allow an analysis of the Irish bubble but which the basis for physical as well as monetary input output tables which enable economists to analyze how final demand is related to sectoral production of, say, CO2. Or to measure the labor share of income. Sound measurement: a mayor feat of the science of economics, enabled by the social and immensely measurable nature of monetary transactions.

1.7 Hey, national accounts are political accounts Many of the choices made by modelers and statisticians are influenced by values. Most obviously: the national accounts are called ‘national’ for a reason. The nation is preponderant even when recent developments in financial technology and international value chains start to make this choice more burdensome. In this process, social strive played and plays a role. A recent example of the issues at stake: Cravino, Lan and Levchenko (2018) show, using the consumption basket used to estimate consumer price inflation and looking at different incomes, that consumption of middle incomes deviates from the average in a way that makes them more vulnerable for increases or decreases in interest rates than other consumers. In such a situation, not using the consumer price index but for instance a broader index also comprising interest will affect different income groups in a different way. Stapleford (2009) analyses the battles between, among other actors, the government, unions and companies about the concept of the ‘consumer price index / cost of living index’ in the USA. He relates the very existence of the consumer price index to the wish to shift class struggle about the purchasing power of wages from companies and organizations to the more ‘civilized’ realms of statistical offices and official indexation clauses, which is not surprising. The accounts are used for the management of national economies. Keynes (1940) is the primeval example. Which means that they are permeated from their design to their publication by political considerations. This is not the same thing as stating that they are just a dreamed up description of the ideal state. The boundaries between sectors like construction and agriculture have not just been drawn for political and ideological reasons but also because of practical as well as technological and economic reasons. The same holds for the separation between the sector households and the business sector or, to an extent, for a variable like Gross Domestic Product (GDP) (Office of National Statistics 2018) or the consumer price index. Delineating ‘households’ instead of churches or dormitories as a sector is a choice. But households can be delineated: they are not just a statistical artifact. The latest version of the national accounts emphasizes ‘ownership’ however more than previous versions, which, as we will see, can lead to fundamentally different estimates of ‘the economy’. Choices have to be made. The data can only be understood against the background of these choices. And the choices will have a political element. But: other choices could have been made. In the 15th century, there would have been sound economic reasons to delineate ‘The Church’ as a separate sector, while nowadays central banks, which in the 15th

Introduction 23 century did not exist and which would not come into existence for centuries to come, are the only companies with and own sub-sectoral delineation – economies evolve. The accounts have to evolve with them. The fact that ‘national’ accounts become increasingly troublesome because of international flows of income and ownership shows that even the system of nations evolves. The Irish accounts have in fact already be adapted to the unusual large income and ownership flows into and out of this island economy. But even that is in the end a political decision, even if made by the statisticians.

1.8 Overarching integration: the modern flow of funds/national accounts The flow of funds are less well known than the national accounts which means that a short introduction might help the reader. They are not designed to estimate the ‘non-financial’ flows like income, production and expenditure but to estimate flows of lending and borrowing in combination as well as to estimate stocks of financial liabilities and assets. They are of relatively recent origin. In 1944 the NBER (headed by Wesley Mitchell) assigned Morris Copeland with the task to develop, together with the Fed (the central bank of the USA), a financial corollary to the rapidly evolving national accounts: About the circuit flow of payments and its relation to national income and output, our knowledge is exceedingly vague. We do know, however, that the flow of payments does not adjust itself automatically to the flow of goods men are able to produce and need to consume. Indeed, several theorists have argued that cyclical fluctuations in business activity are due primarily to recurring changes in the relative size of these two flows. The findings this investigation promise should put us in a far better position to diagnose our recurrent chills and fevers, and to: seek remedies. (Mitchell 1945, pp. 61–62) Copeland successfully accomplished the task, and only about a decade after Mitchell’s statement was made central banks all over the world started to estimate monetary statistics with the FOF as the overarching framework. The agenda set out by Mitchell still is the agenda of the flow of funds, as shown by a recent quote of the Office for National Statistics (ONS) in the UK which, even when less eloquent, conveys the same message as the Mitchell quote above: An understanding of the economic performance of the UK is especially important for effective policymaking and improving welfare. The non-financial accounts have long been extensively monitored as a health check for the economy, but they do not fully capture the build-up of financial risk. For instance, changes to the underlying resilience of the UK’s source of funding can impact the economy in a way that is not obvious from studying fluctuations in income

24  Introduction or output. . . . We have partnered with the Bank of England to address this, by enhancing the coverage, quality and granularity of financial accounts statistics for the UK. (Office of National Statistics 2018) An example of how these accounts are used can be found in the monthly monetary press releases of the ECB. These show the creation of money as a function of sectoral growth of credit: how much money is lent by the banks, how much is borrowed by households, non-financial companies, the government of financial companies as well as how much of this money ends up ‘abroad’. The credit data shown in Figure 1.1 are part of these statistics on the national level: the money borrowed by Irish construction companies from money creating banks or, as they are called in the accounts, ‘Monetary Financial Institutions’ (MFIs) adds to money growth in the Eurozone. This is however not just about the flow of credit to Irish construction companies and the ‘GDP economy’. An example are data on balance sheet of banks. Mortgages have become by far the largest asset on the balance sheet of the consolidated banking sector and quite often also by far the largest debt on the balance sheet of households. As sales of existing houses do not add to the flow of new goods and services (aside from fees for real estate brokers and the like) and are hence not included in the national accounts, even when they are central to vulnerabilities and might readily lead to economic ‘chills and fevers’, as also argued by the ONS (2018). Interestingly, estimating the flow of funds forces the statistician to use a quite broad concept of money. Many transactions, especially between businesses, are initially financed by payables/receivables, not by commercial credit or direct payments. These payables/receivables are promises to pay with legal tender or deposit money and are a legal way to buy something: a change in ownership takes place. They are a means of exchange. They are also an asset, as they are a classical item on company balance sheets and hence a store of value. The flow of funds by necessity treats these promises to pay as money, which shows that the FOF are not just about money as we pay it but also about other money-like items like payables and receivables or long-term savings and a host of other financial instruments. Another example: when entities like shadow banks are not included in the FOF, ‘black holes’ will appear. In fact, the whole apparatus to measure credit bubbles is available, while the statements of Mitchell in 1944 as well as the more recent statements of the ONS indicate that measuring such bubbles is a prime goal of the FOF. Unfortunately, there still are no official bubble measurements (see however Borio 2012). The modern national accounts have at present incorporated large parts of the flow of funds, which has quite some consequences for economics 101, whose textbooks state that {Income = Expenditure}; modern national accounts to the contrary state that; {Income plus net credit = Expenditure of goods and services plus net acquisition of real and financial assets}.

Introduction 25 These assets include changes in cash and deposits which clearly shows that if people save more money and channel this to savings accounts expenditure on goods and services has to fall, unless net debt increases. Analyses using this framework are for instance Keen (2016) and Ryan-Collins, Werner and Castle (2017). But it has to be stressed that it is included in the basic framework of the modern national accounts and also wealth inequality statistics as these, recently, have been added to the flow of funds by the US Federal Reserve (Batty et al. 2019). This all leads to the next general comparison of the models and the measurements.

1.9 An overview of the key differences between DSGE macro-models and macro-measurements Summarizing the discussion, the main differences can be stated as follows: Table 1.1 An overview of the key differences between DSGE models and the flow of funds/national accounts  

National Accounts/Flow of Funds

DSGE models

Basic model

The circular flow of various monetary streams of incomes, expenditures and productions, powered by myriads of monetary transactions made by millions of households and businesses as well as by the government. All monetary production of new goods and services, including non-market government production and production by ‘NPISH’ (churches, unions, sports clubs etc.) and including money yielding criminal activities. There are some imputations, however, the most important being one for the assumed value of rent of owner occupied houses. Another, ‘FISIM’, is not discussed here

One, two or three representative households which optimize social utility by making a choice between labor, consumption and investments now and in the future. Market production minus production of banks minus public goods and services minus durable consumer goods. Non money-creating banks are increasingly incorporated into the models. Consumption is taken to be the psychological value of purchasing of goods and services. If convenient, it is defined as the use of goods. Production of NPISH is neglected. Some models incorporate government production, some incorporate banks, some incorporate consumer durables. I do not know models which do all of this

Production/ consumption/ income boundary

(Continued)

Table 1.1 (Continued)  

National Accounts/Flow of Funds

DSGE models

Are variables well defined?

The national accounts and the flow of funds data and labor statistics have internationally recognized official compendia which extensively and intensively conceptualize and define the variables.

Relation to welfare or prosperity

The model is expenditure oriented and has no direct relation to individual prosperity. The ‘volume’ of total production can be calculated (real GDP) and is often taken to be a metric of the level and growth of prosperity, partly for its own sake and partly because it is often closely related to (un) employment. The composition of production and consumption is also measured but not used to indicate prosperity. Partly classical (the definition of capital including non-produced capital), partly (old)-Keynesian and old-institutional. Examples are the emphasis on total monetary expenditure, subsectoral divisions; the possibility of involuntary unemployment, the inclusion of NPISH, the treatment of the government and the pervasive role of lending and credit. Some imputations however have a clear neoclassical character. No. Profits/losses and changes in income related to involuntary unemployment as well as changes in stocks and the current account are crucial balancing items in the accounts. Heterogeneous and historical. Qualities and quantities and relative prices change over time which leads to a changing sectoral structure of the economy.

There seem to be no formal compendia which describe and define the variables in the DSGE models in a rigorous, sound, nontrivial way. Sometimes different concepts of variables are used in different models without making this explicit. The sum of present and (discounted) future ‘Social utility’ is taken to be the metric of prosperity and society is assumed to optimize this, given constraints. No clear definition of utility is given, no independent estimates of utility and the discount rate are provided.

Relation to economic ‘schools’

Market clearing required?

Nature of the goods and services

Neoclassical in a restricted version.

Short run Pareto efficient market clearing assumed. Unexpected shocks can wreak havoc but it is assumed that society will adapt. Homogenous and static. Intertemporal relative prices and quantities are set; in a sense the rational expectations about probabilities of future events influence todays structure of prices and productions.

 

National Accounts/Flow of Funds

DSGE models

Basic coordination principles

Markets, the governments, NPISH and household transactions, future transactions only when based upon explicit or implicit legal contracts. No ex ante market clearing required. Prices contain a rent element and can change the distribution of wealth and income in a non Pareto efficient way. Detailed sectoral and subsectoral subdivisions including of financial companies, government production and NPISH

Market transactions including expected future transactions, ex-ante pareto efficient market clearing assumed (in the absence of ‘shocks’).

Structure of production

Basic actors Basic method of estimation

Linkages to other models

Nature of money

Households, firms, government, external sector, financial institutions Aggregation of micro-data, continuous source criticism. Care is taken to make historically and internationally consistent estimates. Especially new products and changing relative prices make this complicated. Labor market accounts, flow of funds, input-output models, environmental accounts (like relation of CO2 production to the structure of production and final demand). Credit originates monies and money like assets. Credit (including trade credits) are originated via transactions between often private agents; credit and lending enables ex-post accounting identities to be ‘true’, even without market clearing.

No or limited sub-sectoral subdivision, sectoral division excludes MFI’s but includes central bank and/or non-monetary financial institutions, sometimes implicitly included in the sector households. Households, central bank, companies, sometimes banks. Use of often detrended cherry picked macrodata to calibrate main variables. These are however not measured in a consistent way. Detrended national account variables are used to calibrate and to estimate the resource constraint (the latter excludes labor) Loanable funds, government created if created at all.

28  Introduction

1.10 Mitchell-style business cycle indicators, the accounts and DSGE models In 1947 Tjalling Koopmans published a famous article titled ‘Measurement without Theory’ (Koopmans 1947). In economic discussions this article is often understood to be a criticism of the national accounts. It isn’t. It criticizes a book written by Arthur Burns and (again) Wesley Mitchel, Measuring Business Cycles (Burns and Mitchell 1946, also Mitchell 1913, 1927). This book described how monthly data on heterogeneous individual economic variables can be combined into a synthetic variable to measure business cycles, using cycles instead of quarters or years as the central measuring unit. It kind of precedes present day VAR and especially principal components analysis (see Andrle, Brůha and Solmaz 2017). It can be understood as another kind of macroeconomics, less occupied with the level of output, sectoral production, sectoral distribution and accounting relations and more with cyclical swings and the empirical propagation of financial and other shocks. Important is the distinction between leading, coincident and lagging indicators. Some variables tend to lead business cycles, some tend to lag. These business cycle series were often based upon monthly sub-series. Since, quarterly accounts have been developed, which has diminished the difference between the two approaches as this enhanced the possibilities to use national accounts data for business cycle analysis. At the other side, the business cycle analysts developed indices of leading, coincident and lagging indicators as well as indicators used to measure economic sentiment which eases the accounting identities inherent in the national account and the flow of funds. The development of business cycle analysis is still closely associated with the National Bureau of Economic Research and methods are used by statistical institutes the world over, one example being the grey bars in the famous ‘FRED’ graphs, indicating the lengths of downturns. Figure 1.2 shows that the cyclical indicators neatly coincide with declines in GDP – but do not capture the growth – the change in level – of GDP. Neither do they capture relations between sectors of the slow build-up of financial vulnerabilities. On the other hand, the national accounts data neatly show the volatility of investment as compared with consumer and government spending. We will come back to this. But first it is worthwhile to look at the complexity of the data in figure 1.2. Nominal data are gathered, using an extensive network of contacts and based upon explicit laws. Subsequently they are aggregated using the delineations and interrelations of the NA. Next, these aggregated data are deflated using a deflator which estimated in a comparable way. The resulting data are seasonally adjusted and changed into a seasonally adjusted annual rate of growth by multiplying them by four. Second, the role of business cycle indicators in DSGE models is more important than often understood. Lucas, one of the founding fathers of DSGE modelling, was explicit about his endeavor to explain to explain these fluctuations (the grey bars) using a general equilibrium framework and a ‘micro-founded’ methodology while he did (at this time) not try to explain the level or even the movement of GDP or to use information about the interconnectedness of

Introduction 29

Graph 1.2 Real private consumption, real private investment and real government consumption plus investment, USA: year on year change (%) by quarter, chained 2012 USD, seasonally adjusted annual rate Source: U.S. Bureau of Economic Analysis, retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed, GCEC1, GPDIC1 and LB0000031Q020SBEA, accessed 12 May 2019

sectors (Lucas 1977), which means that the theme of this book – comparing the concepts of DSGE variables with the variables of the national accounts – might be perceived as somewhat dishonest: the DSGE project did not start out to explain national accounts and flow of funds developments in the first place. The recent models however do use the national accounts terminology (consumption, investment etc.) and the authors of the models often calibrate the models using national accounts variables as well as, in the background, the national accounts identities that production is equal to the different kinds of expenditure. But in the beginning Lucas focused on another item: business cycles as identified by Wesley Mitchell. Though not all of the variables included in these cycles. It is, at prima vista, not clear why he left employment and involuntary ­unemployment – key variables of Keynesian theory and macro-measurement – out of his description of the stylized facts of the ‘Mitchell’ cycles. He also only focused on the minor cycles as identified by Friedman and Schwartz (who used an explicit Mitchellian framework as shown by Rockoff 2006) and left the mayor cycles identified by these authors and at present better known as financial cycles outside of his scope. The 2008 crisis did not fit the DSGE research agenda for a reason, DSGE models were not meant to explain or encompass such events. But only to explain smaller

30  Introduction up and downturns. On the other hand, the graph does show that Mitchellian business cycle indicators do correlate with the national accounts data on, for instance, private investment even when these have been thoroughly deflated, seasonally adjusted and changed into growth rates. In Chapter 7, we will return to the deflation procedures used.

1.11 The conceptual model of the book (1): five interrelated phases of development of a macro-statistical variable One of the elements which in a loose way will structure the next chapters of this book are the phases of development of a statistical variable. These phases are not applied in a rigorous way but will pop up in every chapter. Frits Bos (2003) distinguishes, when analyzing the development process of economic statistics, four interrelated phases with many forward and backward linkages. All these phases are necessary (albeit not always in a hierarchical way) to enable measurement of an (economic) variable. Here, a fifth one will be added. The first phase: conceptualization. This is a rather philosophical phase which, when restricting ourselves to consumption, functions to answer the question ‘what is consumption anyway?’ An example from the ESA 2010, lemma 3.93: Two concepts of final consumption are used: (a) final consumption expenditure; (b) actual final consumption. Final consumption expenditure is expenditure on goods and services used by households, NPISHs (Non Profit Institutions Serving Households, MK) and government to satisfy individual and collective needs. In contrast, actual final consumption refers to its acquisition of consumption goods and services. The difference between these concepts lies in the treatment of certain goods and services financed by the government or NPISHs but supplied to households as social transfers in kind. Note, again, the importance of government goods and services but also the fact that not all consumption has to be purchased by households. Even then, it is about the acquisition of goods and services, not about their actual use. The second phase: definitions. A shorter, more practical summary of the concept with often clearer delineations: An example from the ESA 2010, lemma 3.94: ‘final consumption expenditure consists of expenditure incurred by resident institutional units on goods or services that are used for the direct satisfaction of individual needs or wants or the collective needs of members of the community’. Note that, in an implicit way (the ‘resident institutional units’ mentioned) national

Introduction 31 boundaries are introduced which, also in an implicit way (but treated in an explicit way in another part of the ESA 2010) means that expenditure of tourists, who are not residents of a country, is counted as an expenditure of the home country of the tourists and hence as an export of the destination country. The third phase: operationalization. Questions like which period, which sources, which products, which method (surveys or administrative date of both), which populations, will be answered. The ESA 2010 advises yearly as well as quarterly measurements and advises when estimating sales use (unlike in the case of unemployment) to rely on business but not consumer surveys as well as, for practical as well as methodological reasons, administrative data like VAT. It provides an exhaustive list of goods and services which ought to be measured, most visible in the list of items included in the ‘basket’ on which the consumer price index is based. The difference between using variable cycles as the prime temporary variable of estimating frameworks or calendar variables is another example. The fourth phase: measurement. Gathering as well as processing the data: Gathering the data is quite a job which requires a lot of people and, hence, a lot of money. It’s the economic equivalent of the Large Hadron Collider. After data gathering, the information has to be processed. Processing often requires a lot of tacit knowledge as well as a plethora of micro-decisions, for instance in the case of quality changes of products, of new products or old beer in new bottles (which, for instance when going from 500 milliliters to a pint, will be smaller than the old ones). In this book, we will add a fifth element: Presentation and publication. As is clear, statistical variables are conceptualized in a precise way while the results of the measurements are often published by statistical institutes in press releases but also in data bases. Importantly, the academic world of the theoreticians literally requires one to ‘make a name’ while the publications mentioned and even the manuals are often anonymous proceedings. These five phases will not be used in a systematic way to structure the book but they will be the background of the discussion as the rift between the statistics and models concerns all these phases.

1.12  The conceptual model of the book (2): cases Not all economists are well-versed in tacit statistics, defined here as the discussions surrounding measurement of variables and the aggregating of micro-results.

32  Introduction To show the importance of this kind of work as well as to enable a deeper understanding of the theory behind measurement, each chapter will contain an example of an application of concepts and definitions or of aggregation methods to sets of data to show that the simple application of these methods can lead to surprising interpretations of existing information. But this is not about these surprising interpretations as such. It’s about the power of the concepts: are the able to enlighten us, or do they just obfuscate?

1.13 The conceptual model of the book (3): meta-formulas The book contains chapters on the monetary nature of transactions, money and debt, labor and unemployment, land and capital, consumption (including government consumption), investment (including government investment) and household purchases of cars, (un)real production as well as ULC, RULC and NULC. Chapters on the interest rate and imports could have been added. The interest rate is however covered in a way highly consistent with this book in Borio, Disyatat and Rungcharoenkitkul (2019). Leaving imports, exports and international financial flows out is an omission. But: why the other chapters? Economists use a few ‘meta-formulas’ which structure economic discourse. The same holds for macroeconomic statistics. These formulas are used to organize the chapters of the book. One famous meta-formula is: (i) Y = f(capital, labor) or Y = f(K, L) Meaning that production is produced by capital and labor. This leads us to chapters about capital (chapter 5) and labor (chapter 4). We will however, in line with the statistics and consistent with classical economics, also add ‘land’ or ‘unproduced assets’ like natural resources as well as ‘legal’ assets like patents to definition of Capital. These are distinct categories in the national account statistics. Also, Land gained renewed significance as a crucial economic variable after 2008. Another point: capital is estimated by statisticians not as a volume of machinery or ‘intangible assets’ but as the value of ‘rights to nominal income’ connected to ownership of these items, production costs of these items or (re)sale value, capital cannot be used to explain production. For one thing, this value of nominal production is often decisively influenced by the development of relative prices, including the interest rate, and not just by productivity. Capital can however be used to explain the distribution of nominal income and wealth. As total production and total income are equal it is possible to replace the Y in the first formula with “Income” (I) and to look at income related to capital. Looking at it from the income side also means that, in the case of labor, we won’t only look at the labor force but also at unemployment. If we do this while also adding Natural Resources (NR) as a distinct category of capital formula (i) changes. We should also add the utilization rate of fixed capital as well as the utilization rate of natural resources to get at a non-neoclassical formula ‘r’ stands for average rent; ‘i’

Introduction 33 stands for the rate of return on capital; and ‘w’ stands for the average wage rate: (ii) Y = I = (ru1NR + wu2L + iu3K) With r for rent, u1 for the utilization rate of natural resources, u2 for the utlization rate of labor, ur for the utlization rate of fixed capital, w for the wage rate and i for the average return on utilized fixed. Consistency with the national accounts would however warrant a distinct category for legal unproduced capital like patents, production rights and bandwith. In this book, attention will be paid not just to labor but also to unemployment and not just to produced capital but also to ‘land’. Another meta-formula is: (iii) Y = C + I + O + {Ex – Im} This, however, is the textbook version. In this book we will follow the national accounts and use the credit enhanced formula: (iv) {Y + net credit} = {C + I + O + {Ex – Im} + net acquisition of financial assets} In this formula, C is private consumption (chapter 6), I is private investment (chapter 7), O is government investment and government consumption (i.e. education, the juridical system and the like; chapters 7 and 6), Ex is exports and Im is imports. As stated, Exports and Imports will be left out of the book but we have to add ‘net credit’ to these categories, as statisticians state that expenditure is financed by income plus net credit and spent on goods and services on one hand and financial assets (including increases in amounts of deposit money or cash), as well as on existing fixed assets on the other hand. Specific chapters on financial assets are absent but they do play a role in the description of consumption and investment. Exports and imports will however be excluded from this book. Money will be discussed against the background of the credit formula the ECB (and all other central banks) use to estimate M3-money. The money meta-formula will be treated in more detail the chapter about money (chapter 2). Aside of these variables, attention will be paid to the way statisticians calculate ‘real’ variables (i.e. variables expressed in fixed prices), the price level, the difference between real variables and physical volumes (chapter 8). We will, however, start with a chapter not about specific variables but about the nature of a monetary economy (chapter 2).

Notes 1 See, for well-crafted and still valid criticisms of this concept, Veblen (1898, 1899a, 1899b, 1900, 1909). When reading consumer behavior text books it is striking how much these are in line with what Veblen considered sound economic science. 2 DSGE models as a rule also exclude purchases of durable consumption goods. In his chapter two Petty states that these are a kind of capital goods and add to the wealth of a nation (ditto Marshall). They are included in the US flow of funds. 3 As stated by the website economicshelp: ‘We can try to measure utility by using a hypothetical unit of measurement’ (Economicshelp 2019). Trying to measure anything with a hypothetical unit of measurement won’t bring you anywhere. The nature of units of measurement is that they are not hypothetical but well specified.

34  Introduction

Literature Andrle, M., J. Brůha and S. Solmaz (2017). ‘On the sources of business cycles: Implications for DSGE models’. European Central Bank working paper no. 2058. Arrow, K. (1950). ‘A difficulty in the concept of social welfare’. Journal of Political Economy 58:4 328–346. Batty, M., J. Bricker, J. Briggs, E. Holmquist, S. McIntosh, K. Moore, E. Nielsen, S. Reber, M. Shat, K. Sommer, T. Sweeney and A. Henriques Volz (2019). ‘Introducing the distributional financial accounts of the United States’. Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs, Federal Reserve Board, 2019–17. Bokan, N., A. Gerali, S. Gomes, P. Jacquinot and M. Pisani (2016). ‘EAGLE-FLI. A macroeconomic model of banking and financial interdependence in the euro area’. European Central Bank working paper no. 1923. Borio, C. (2012). ‘The financial cycle and macroeconomics: What have we learnt?’ BIS working paper no. 395. Borio, C., P. Disyatat and P. Rungcharoenkitkul (2019). ‘What anchors for the natural rate of interest?’ BIS working paper no. 777. Bos, F. (2003). The national accounts as a tool for analysis and policy: Past, present and future. Berkel and Rodenrijs: Eagle Statistics. Bos, F. (2013). ‘Meaning and measurement of national account statistics’. Paper provided at the Political Economy of Economic Metrics conference. Bowley, A.L. (1920). The change in the distribution of national income, 1880–1913. Oxford: Clarendon press. Bowley, A.L. (ed.) (1942). Studies in the national income 1924–1938. Cambridge: Cambridge University press. Buiter, W. (2009). ‘The unfortunate uselessness of most ‘State of the Art’ academic monetary Economics’. Blogpost on VoxEU, 6 March 2009. Burns, A.F. and W.C. Mitchell (1946). Measuring business cycles. New York: National Bureau of Economic Research. Card, C. (2011). ‘Origins of the unemployment rate: The lasting legacy of measurement without theory’. Berkeley and NBER working paper. CBS (1939). ‘Enkele berekeningen over het nationale inkomen van Nederland’. Speciale Onderzoekingen van de Nederlandse conjunctuur no. 2. Clark, C. (1937). National income and outlay. London: Macmillan. Coli, A. and F. Tartamella (2015). ‘The role of micro data in national accounts. Towards micro-founded accounts for the household sector’. Paper Prepared for the IARIW-OECD Special Conference: “W(h)ither the SNA?” Cravino, J., T. Lan and A. Levchenko (2018). ‘Price stickiness along the income distribution and the effects of monetary policy’. Center for Economic Policy Research Report DP 12967. Daragh, C., P. Jacquinot and M. Lozej (2014). ‘The effects of government spending in a small open economy within a monetary union’. European Central Bank working paper no. 1727. Derksen, J. (1940). ‘Het onderzoek naar het nationale inkomen’. De Economist 88 571–594. Derksen, J. (1941). ‘Berekeningen van het nationale inkomen van Nederland voor de periode 1900–1920’. Speciale Onderzoekingen van de Nederlandse conjunctuur no. 4.

Introduction 35 Dimsdale, N., S. Hills and R. Thomas (2010). ‘The UK recession in context – what do three centuries of data tell us?’ Bank of England Quarterly Bulletin 577–591. Economicshelp (2019). www.economicshelp.org/blog/26552/concepts/measur ing-utility/. Accessed 17 March 2019. Eurostat (2013). European system of accounts ESA 2010. Luxemburg: Publications Office of the European Union. Eurostat (2018a). ‘Statistics explained. Building the System of National Accounts – supply and use tables’. http://ec.europa.eu/eurostat/statistics-explained/index. php/Building_the_System_of_National_Accounts_-_supply_and_use_tables Accessed 17 June 2018. Eurostat (2018b). ‘First estimates for 2017. Wide variation of consumption per capita across EU Member States GDP per capita ranged from 49% to 253% of EU average’. News release 101/2018–19 June 2018. Eurostat (2018c). Statistics explained. https://ec.europa.eu/eurostat/statisticsexplained/index.php/Glossary:Actual_individual_consumption_%28AIC%29. Accessed 12 October 2018. Fessler, P. and M. Schürz (2017). ‘The functions of wealth: Renters, owners and capitalists across Europe’. Draft prepared for the first WID World Conference Paris School of Eurostat (2017). GDP per capita, consumption per capita and price level indices. 14–15 December 2017. Friedman, M. and A. Schwartz (1963). A monetary history of the United States 1867– 1960. Princeton: Princeton University press. Godley, W. and M. Lavoie (2007). Monetary economics: An integrated approach to credit, money, income. Antony Row: Chippenham and Eastbourne. Haavelmo, T. (1944). ‘The probability approach in econometrics’. Cowles foundation paper no. 4. Hirschman, D.A. (2016). Inventing the economy or: How we learned to stop worrying and love the GDP. Unpublished PhD thesis, University of Michigan. Iwata, Y (2012). ‘Non-wasteful government spending in an estimated open economy DSGE model. Two fiscal policy puzzles revisited’. ESRI discussion paper series no. 285. Keen, S. (2016). Developing an economics for the post-crisis world. World Economics Association book series. College Publications. Keynes, J.M. (1919). The economic consequences of the peace. London: MacMillan. Keynes, J.M. (1940). How to pay for the war. A radical plan for the chancellor of the exchequer. London: MacMillan. Keynes, J.M. and E. Rothbarth (1939). ‘The income and fiscal potential of Great Britain’. Economic Journal 49:196 626–639. King, W. (1922). ‘All branches of government’ in: W. Mitchell (ed.) Income in the United States volume II: Detailed report 201–222. New York: National Bureau of Economic Research. King, W., F. Macaulay, W. Mitchell and O. Knauth (1921). Income in the United States: Its amount and distribution 1909–1919. Volume 1: Summary. New York: National Bureau of Economic Research. Klamer, A. (2001). ‘Late modernism and the loss of character in economics’ in: J. Amariglio, S.E. Cullenberg and D.F. Ruccio (eds.) Post-modernism, economics and knowledge 77–101. London: Routledge. Koopmans, T. (1947). ‘Measurement without theory’. The Review of Economics and Statistics 29:3 161–172.

36  Introduction Kuznets, S. (1941). ‘Concepts of National income’ in: S. Kuznets (with Lillian Epstein and Elizabeth Jenks) National income and its composition, 1919–1938, volume I. Cambridge (MA): National Bureau of Economic Research Kuznets, S. (1955). ‘International Differences in capital formation and financing’ in: National Bureau of Economic Growth, Capital Formation and Economic Growth pp. 17-110. Princeton: Princeton university press. Lucas, R. (1977). ‘Understanding business cycles’. Paper prepared for the Kiel Conference on Growth without Inflation, June 22–23. Marshall, A. (1920, 1st edition 1890). Principles of economics. 8th edition. London: MacMillan. Mitchell, W.C. (1912). ‘The backward art of spending money’. American Economic Review 2 269–281. Mitchell, W.C. (1913). Business cycles. Berkeley: University of California Press. Mitchell, W.C. (1916). ‘The role of money in economic theory’. American Economic Review 6:1 Supplement, Papers and Proceeding of the Twenty-eighth Annual Meeting of the American Economic Association 140–161. Mitchell, W.C. (1927). ‘The processes involved in business cycles’ in: W.C. Mitchell (ed.) Business cycles: The problem and its setting 1–60. Cambridge (MA): National Bureau of Economic Research. Mitchell, W.C. (1945). ‘A record of 1944 and plans for 1945’ in: W.C. Mitchell (ed.) The national bureau’s first quarter-century 41–72. New York: National Bureau of Economic Research. Mitra-Kahn, B.H. (2011). Redefining the economy. How the ‘economy’ was invented in 1620 and has been redefined ever since. PhD thesis. London: City University. Monacelli, T. (2009). ‘New Keynesian models, durable goods, and collateral constraints’. Journal of Monetary Economics 56:2 242–254. Morreau, M. (2014). ‘Arrow’s theorem’ in: E.N. Zalta (ed.) The Stanford encyclopedia of philosophy (Winter 2016 edition). https://plato.stanford.edu/archives/ win2016/entries/arrows-theorem/>. Office for National Statistics (2018). ‘Economic statistics transformation programme: Enhanced financial accounts (UK flow of funds) – enhancing the UNDERSTANDING of UK household finance’. www.ons.gov.uk/economy/nationalaccounts/uksectoraccounts/economicstatisticstransformationprogrammeenhancedfi nancialaccountsukflowoffundsenhancingtheunderstandingofukhouseholdfi nance/2018-06-26. Accessed 20 May 2018. Patnak, U. (2018). ‘Profit inflation, Keynes and the Holocaust in Bengal, 1943– 1944’. Economic and Political Weekly 53:42. Petty, W. (1690). Political arithmetick. London: Robert Clavel at the Peacock and Hen. Prescott, E. (2016). ‘RBC methodology and the development of aggregate economic theory’. NBER working paper no. 22422. Reinhart, C.M. and K.S. Rogoff (2009). This time is different: Eight centuries of financial folly. Princeton: Princeton University press. Rockoff, H. (2006). ‘On the origins of “A Monetary History” ’. NBER working paper no. 12666. Russel Kincaid, G. (2008). ‘Adjustment dynamics in the Euro area: A fresh look at the role of fiscal policy using a DSGE approach’. European Economy. Economic paper no. 322.

Introduction 37 Ryan-Collins, J., R.A. Werner and J. Castle (2017). ‘A half-century diversion of monetary policy? An empirical horse-race to identify the UK variable most likely to deliver the desired nominal GDP growth rate’. Journal of International Financial Markets, Institutions and Money 158–176. Samuelson, P. (1954). ‘The pure theory of public expenditure’. The Review of Economics and Statistics 36:4 387–389. Sims, Eric (2015). ‘Graduate macro theory II. Notes on medium scale DSGE models’. Unpublished lecture notes. https://www3.nd.edu/~esims1/preliminaries_ 2015.pdf Accessed 9 November 2018. Smith, A. (1784, 1st edition 1776). An inquiry into the nature and causes of the wealth of nations’ book I, II, III, IV and V. 4th edition. London: Methuen. Stähler, N. and C. Thomas (2011). ‘Fimod – A DSGE model for fiscal policy simulations’. Banco de España Documentos de Trabajo 1110. Stapleford, T.A. (2009). The cost of living in America. Cambridge: Cambridge University Press. Stevin, S. (1608). ‘Van de vorstelicke boukhouding in domeinen en financie extraordinaire’ in: S. Stevin (ed.) Wisconstige gedachtenisse inhoudende ‘t gehene daer hem in gehoeffent heeft den doorluchtigsten hoochgeborene vorst ende here Maurits prince van oraengien part 5. Leiden: Jan Bouwensz. Tily, G. (2009). ‘John Maynard Keynes and the development of national accounts in Britain, 1895–1941’. Review of Income and Wealth 55:2 331–359. Tinbergen, J. (1939). Statistical testing of business cycle theories: Part II: Business cycles in the United States of America, 1919–1932. New York: Agathon Press. Veblen, T. (1898). ‘Why is economics not an evolutionary science?’ The Quarterly Journal of Economics 12 373–379. Veblen, T. (1899a) ‘The preconceptions of economic science I’. The Quarterly Journal of Economics 13 121–150. Veblen, T. (1899b). ‘The preconceptions of economic science II’. The Quarterly Journal of Economics 13 396–426. Veblen, T. (1900). ‘The preconceptions of Economic Science III’. The Quarterly Journal of Economics 14 226–240. Veblen, T. (1909). ‘The limits of marginal utility’. Journal of Political Economy 17 151–175.

2 Money, prices and pricing

2.1 Introduction Key aspects of the monetary nature of our economy will be discussed.

2.2 Transactions The national accounts and the flow of funds track transactions. Monetary transactions. Monetary transactions involve money. Sometimes existing money. Sometimes new money. Quite often, a transaction creates the money necessary to execute it like when a man uses a credit card to buy Valentine’s Day flowers. The credit provided by the bank issuing the card creates the deposits used to pay for the transaction. Before the transaction these did not exist. Now, they do. And the buyer of the flowers is even more indebted than he already was. The monetary debt can and generally will be paid down. But this is a second transaction, separate from the buying of the flowers. The second transaction will more often than not use deposit money created by other transactions instead of the money created to buy the flowers. Paying down the debt will not only extinguish the debt but also these deposits. Transactions create as well as destroy money. As transactions are institutionally embedded they shape a large part of our relations and even ourselves, be it as wage laborers, entrepreneurs, debtors or creditors or all of these at the same time. One relation is with the government. The debt to the credit card company has to be paid. If it’s not paid the law stipulates that, ultimately, assets can be seized. And they will be seized, which is why the bank lets me use a credit card in the first place. Credit cards are relatively modern. But the customer credit system isn’t. The last vegetable and fruit shop in Leeuwarden, where I live, using a book with handwritten debts of well-known customers – the peer-to-peer version of a credit card – closed down only in 2018. In Europe, this peer to peer consumer credit system dates back at least to the 15th century. Most day-to-day transactions in Western Europe before 1800 were based on consumer credit, not cash. Even when the debts were, eventually, settled with cash or cleared (Vermoesen 2011; Ronsijn 2014; Borghaerts and Knibbe 2017; Meertens Instituut 2019). Credit cards are modern, consumer credit isn’t. Debts have permeated society for centuries. They still do. It is impossible to talk about transactions

Money, prices and pricing 39 without taking debt into account. It is impossible to talk about us, debtors in a monetary economy, without doing this. Monetary transactions, including money creating transactions, are the very thing which macro-statisticians measure. Even when it comes to the most important direct volume estimate of the macroeconomic statisticians, ‘gainful labor’. This is measured in people (including the unemployed), jobs, hours as well as activities and delineated according to age and sex and sometimes education and economic sector. But it has to be ‘gainful’. Employment and unemployment are conceptualized and defined in relation to the money economy (International Labor Office 2013a). This is not to say that non-monetary labor is not important. To the contrary. Wesley Mitchell wrote in 1912 about the difficulties housewives encountered when needing to spend the household money wisely and effectively (Mitchell 1912). Nowadays the Levy institute publications about ‘time poverty’, which investigate the very same difficulties, state: The predominant framework for measuring poverty rests on an implicit assumption that everyone has enough time available to devote to household production or enough resources to compensate for deficits in household production by purchasing market substitutes. (Levy institute 2017) The publications measure if this indeed is the case. It isn’t (Antonopoulos et al. 2016). Household accounts – a much better method to estimate household monetary income and spending than surveys or experiments – show that monetary saving and investment happens and is of crucial importance even for the poorest households (Collins et al. 2010). Paid and unpaid labor are intertwined. But the macro-accounts only measure remunerated labor – including the ‘mixed income’ of the self-employed. And the related monetary value of production, saving and investment. The production, income and expenditure boundary of the national accounts and the flow of funds is determined by monetary transactions, even when it comes to labor. Monetary transactions do not exist in a vacuum. They exist in a monetary society. This means that to understand the statistics we have to devote attention to the nature of a monetary society. What do we define, measure, delineate and embed, and why? This is not just about the height or the weight or the tail or the trunk of an elephant – it’s about its nature of the beast. Can the models help us out when we want to know more about this nature? Nope. At least not when they use a single representative consumer and are basically only pondering the question if this consuming now or in the future, using a psychological interest rate, is the question. As Bokan et al. (2016, p. 12) state: The representative patient household, labelled ‘saver’, gets utility from consumption of the nondurable composite good. . . . and from housing services and gets disutility from working. The household maximizes her lifetime utility subject to the budget constraint taking all prices but wages as given. All nominal

40  Money, prices and pricing variables in the budget constraint are expressed in ‘real’ terms by dividing them by the consumption price deflator. It’s all about lifetime utility – ‘lifetimes’ lasts forever while consumption is defined as acquiring a physical item and not as a monetary transaction, while the budget constraint is also defined in physical terms and not as transaction. Robinson Crusoe and the coconuts. Money is, at best, tacked on to this society but is supposed not to influence preferences, institutions or relative prices of goods and services, at least not in a permanent way. The basic questions are how many coconuts will be harvested today, how many will be planted today and how many will be consumed, today and tomorrow and forever after. That does not teach us much about a monetary society. As transactions are, by definition, between more than one person. Money is, as Samuelson famously stated, a social contrivance, not an individual one. Bokan et al. do model more than one consumer, which opens up the possibility of modeling monetary transactions. This, however, does not happen in the model. There is, contrary to the macro-measurements, no complete set of balance sheets with money creating debt added to the intertemporal aspects of model. But it still does help us to define the nature of a monetary economy. A monetary economy is not about an individual but about a network of multiple persons and companies and division of labor and different categories of spending and income and different products and services and long and short-term transactions and geographical differences. Also, accepting debt based transactions from people who you do not know are crucial. Banks play a pivotal role in such a system. But there is no homunculus in this network. Or to borrow a phrase from the models, there is no social planner. In reality there is a social monetary network, embedded in legal and cultural institutions and constantly recreating itself in ever changing patterns. A nice example of some of these patterns can be found in the interactive UK flow of fund (FOF) graphs as published by the ONS (Office of National Statistics 2019), which shows historical changes in sub-sectoral debts and assets, using a whole array of financial instruments. It clearly shows the historical changes in intersectoral debt – the ‘from whom to whom’ table of the FOF in a historic-dynamic framework. In the real world, governments and central banks are, to an extent, expected to steer the aggregates. Maybe that’s slightly successful. Maybe not. But that’s not the point. Regardless of the intentions and actions of these entities the flows can be measured. And they are measured. The amazing maze of the socially and politically embedded micro transactions in aggregated into a still complicated but manageable number of accounts. The transactional monetary economy, including debts and credits, can be measured, and aggregated. We do have to be careful with this aggregates, however. They are historically contingent. Which makes for a difference with the way the models treat the economy. An example. The Bokan e.a. quote above mentions that the quantity of production is calculated by dividing the monetary value of production by the price index. This is not much of a problem when there is one good, and one price, as is the case in the models. In reality there is a plethora of goods. And a plethora of prices. Which change all the time. But a deflated nominal

Money, prices and pricing 41 budget constraint is only time consistent when relative prices and quantities do not change. But they do change, meaning that the models have to model a one good economy. Also, new products are introduced all the time, which means that deflated or ‘real’ variables are, strictly speaking, not historically comparable even when monetary budget constraints, taking lines of credit into account, are binding. New products do make a difference. Antibiotics are an example. In 2014 John Kay wrote that, in 1836, ‘Nathan . . . Rothschild died of septicaemia following an abscess, and in spite of buying the best medical attention available in Europe at the time . . . he was dead at the age of 58 from an illness that could today be cured by an antibiotic costing a few pence’ (Kay 2014). But nowadays, the exploding price of insulin in the USA is an example – it costs more than a few pence. People are dying because of this. Nathan Rothschild would have been able to pay for this – others aren’t. Products and relative prices change, in unpredictable and sometimes fundamental ways and because of a multitude of causes. At the same time the monetary transactions world of monetary income and expenditure as well as lending and borrowing seems to have quite lasting characteristics, just like the inherently social nature of these transactions and their often long-term consequences. Modeling one representative consumer disables the modeling of the social nature of these transactions and their long-term consequences. The statistics do take the social nature of our monetary system into account and in a fundamental way as they are based on transactions, which are social by nature, as well as monetary – as they are about money prices, which leads us to the following question: what is in a price?

2.3  Prices and pricing Transactions define quantities, qualities and prices. What is the nature of prices? The subsequent paragraphs and chapters will to quite an extent be about prices and we will need some idea of their nature. Let’s first look at markets. A market is an institution were people supplying a good or service and people demanding a good or service negotiate a monetary price, quantities and qualities and only subsequently execute a transaction. Negotiating can be simple and short and lead to short fast transactions. In the supermarket one looks at the price tags and pays – or not. Negotiating can be elaborate and lead to contracts lasting many decades as in the case of the famous Millau bridge in France, which is based on a 75-year contract. But in all cases market prices are ex-ante prices – or at very least there are ex-ante agreements how to deal with pricing in future situations. Explicit, known prices exist before a transaction is finalized. Think of the monthly rent of a house, which is often fixed for a year ahead. Also, prices are explicit. They exist – even when no transaction is finalized and people walk by the detergent or the peanuts in the supermarket without making a purchase. Economists have established the concept of ‘shadow prices’. Shadow prices are prices inferred by economists. They do not exist like ex ante or ex post market prices do. They do, however, exist in the minds of economists. These sometimes argue that the possibility to envisage and even calculate a shadow price infers the existence of a

42  Money, prices and pricing market for the goods or services in question. This is not true. Looking at the definition of ‘shadow prices’ one encounters phrases like ‘the estimated price of a good or service for which no market price exists‘ or ‘the assignment of a dollar value to an abstract commodity that is not ordinarily quantifiable as having a market price’ (Investopedia 2018). Shadow prices are not market prices. Also, a shadow price can often only be satisfactorily calculated ex post. These are hence not market prices, as there is no market. Monetary transactions, and surely market transactions and the ex-ante nature of market prices enable but do not necessitate rational calculation in the neoclassical sense (Mitchell 1916). As they involve money they also do, necessarily, impose a brutal monetary budget restriction, even when people have a credit line. Money creating transactions, be it in the old fashioned village grocery shop or using a credit card or mortgaging a house or emitting a ‘payable’, lift or at least ease the constraints. But even then, there is a price to be paid which has to be known before credit will be provided. It is for a reason that the shenanigans in the Merchant of Venice still strike a chord. Situations were no money, no market and market prices or government exist but exchange does happen are rife. Many an institution exists to facilitate ‘gift’ exchange or ‘reciprocal’ exchange. An example are the Western Christian wedding vows which contain clauses like ‘to have and to hold, from this day forward, for better, for worse, for richer, for poorer, in sickness and in health, to love and to cherish, till death do us part’ (The Knot 2019). In economese: ‘I’ll put up with you whatever the ex-post shadow price’. Western marriage is, hence and from the contractual point of view, not a market transaction and not even reciprocal exchange but gift exchange. Interestingly, non-Western rituals often do not contain individual vows but entail couple and society oriented phrases, while in some non-Western societies, ex ante dowries between families are important – those are market transactions between families which brings us to prices again. Prices (or, sometimes, information on how prices will be set) have to exist before transactions can be finalized while transactions often have a long-term nature. This holds for market prices. Prices set by the government (‘administered prices’ as statisticians call them, regulated prices as economists call them) basically have the same ex ante nature. As transactions often have a long-term nature, these prices restrict future budgetary possibilities and shape our world. The following sections will to quite an extent be about different prices and the world shaped by monetary transactions. First, however, a little bit more about money.

2.4 Money What is money? As stated, monetary relations shape our society. Like family ties, some transaction based monetary bonds are binding in the long term. Ask the Greek. Also, monetary transactions are the very thing which macro-statisticians measure – income, expenditure and production related transaction in case of the national accounts and financial relations and (changes in) wealth and debts in the case of the flow of funds. To really understand these statistics we have to devote attention to money and the monetary society (another effort: Mitchell 1927). So,

Money, prices and pricing 43 what is money? First, money is not just coins or cash or deposit money. It’s much more (Mehrling 2017). To get this across I will discuss three kinds of private money not issued by the state or banks: stamps, 17th- and 18th-century bills of exchange and modern receivables and payables. First, stamps. Figure 2.1 shows a stamp with Dutch super model Doutzen Kroes photographed by Dutch super photographer Anton Corbijn (who shot the album cover of U2’s Joshua Tree). Using the widely accepted (but incomplete) ‘money is a means of exchange, a store of value and a unit of account’ definition of money to postal stamps, this stamp is money. It is a means of exchange: it can be exchanged for postal services. It is a store of value: it can be used next year. And ‘forever stamps’ even have their own unit of account. Also, like many kinds of historical and present day cash, stamps often bear the picture of a well-known person. It is a sign of the times that in this case this is not a living Caesar, a dead president or a historical scientific or artistic celebrity. It’s not just money that changes. But stamps are money. Important: there are government guarantees that using this stamp will get your letter delivered (and other ones which exempt a picture of a stamp, within limits, from copyright laws). Sometimes, a fourth element is added to the definition of money: ‘money is the standard of deferred payments’: debts are quoted using the unit of account and means of payment. As unused stamps are liability on the balance sheet of the postal service, there is an element of this, too, to stamps.

Figure 2.1  A Dutch stamp from 2017: private money

44  Money, prices and pricing The second example: Bills of Exchange, at least from the time on when, in 17th-century France, the joint liability rule was introduced. According to Santarosa (2015) this rule ‘specified that every party who used a bill of exchange to pay for goods or settle a debt was liable for the face value of the bill if it was not paid at maturity’. People used these bills to pay. They functioned as money. And therewith were money. In the case of stamps as well as in the case of bills of exchange, rules set and enforced by the government were pivotal, while, as in the case of stamps, the amount of money was related to the volume of trade. The third example is Irish. In recent years, multinationals like Microsoft have established headquarters in Ireland. For reasons related to tax arbitrage, many ‘intangible assets’, such as patents, were sold by Microsoft entities in the USA or other countries to the Microsoft subsidiary in Ireland. Basically, these transactions were financed by creating and accepting intercompany payables and receivables. Microsoft Ireland issued an IOU to Microsoft USA and Microsoft USA accepted this. These IOUs are on balance sheets and are based upon the law: they are legally binding contracts. The transactions are also legally binding. Ownership of property does legally change hands. When it comes to money the government always is within shouting distance. In the national accounts (NA) these transactions showed up as an extreme spike in the macroeconomic Irish investment rate. Also, according to the flow of funds data of the Central Bank of Ireland and the Central Statistical Office of Ireland they also contributed, between the last quarter of 2014 and the first of 2015, to a €230 billion deterioration of the Irish Net International Investment Position (Central Statistical Office Ireland 2018). One part of a company becomes indebted to another part of this company as assets and liabilities are transferred and a whole country becomes, according to the statistics, deeply indebted. As stated: these liabilities are legally binding. When Microsoft USA goes bankrupt, creditors can and will ask for this money. Also: the statistics are ‘granular’ which enables us to break down the international debt by financial instrument as well as by sector (Office of National Statistics 2019). The sectoral break down shows that Irish ‘non-financial companies’ like Microsoft had by far the largest negative international investment position of all Irish institutional sectors, including banks, the government and households. On a conceptual level there is nothing special to be seen here. Companies emit and accept payables/receivables all the time, as their balance sheets show. These payables/receivables are promises to pay. But unlike in the case of deposit money, you generally can’t use your receivable to pay down a payable owed by you. But they are a kind of money, too. When a ‘payable’ is emitted by the buyer and the seller accepts it, it leads to a legally binding economic transaction. The payables serve as a means of exchange, they are a store of value (they are on the balance sheets of companies as well on the balance sheets of nations) and their value is expressed in the official unit of monetary account. The bills of exchange – which were ‘receivables’ for the person owning the bill – show that they can even finance transactions by third parties. Buying petty ‘receivables’ at a haircut in 19th century New York and gathering payments in money was the way Marcus Goldman, of Goldman Sachs fame, started his banking career. For the

Money, prices and pricing 45 record, not the entire amount of payables/receivables shown in Graph 2.1 was created by Microsoft Ireland, though they seem to have been the most important player.1 The influence of the government in all these cases is paramount, not only as the contract enforcer of last resort but also as the organization which sets the rules. As it shows as the official unit of account is used as the standard for these deferred payments. Technical addendum: The Bank of Ireland data on these flows are not as granular as data on Irish (sub-)sectoral loans, probably because the magnitude of the changes would enable people to identify individual companies. What does Graph 2.1 teaches us about the nature of a monetary society in relation to the statistics? First, the idea that payables and receivables are a kind of money is, even when not in textbooks on general economics, widely accepted by economists. Just consider any book on either business economics or transaction law. The data of Graph 2.1 are also routinely available as part of Central Bank of Ireland financial statistics (table 3.1a of these statistics) as well as the Eurostat flow of funds data (Eurostat financial balance sheet items F7 and F71, respectively). If you’re skeptical about the idea that receivables and payables are ‘money’ you should be skeptical about our monetary statistics, as well as about balance sheets of companies which list them as liquid assets (even when selling them requires a haircut). Clearly, many kinds of money exist. This insight into the essence of money is embodied in business economics textbooks and

Graph 2.1  Other accounts receivable and payable, flows, Ireland Source: Central Bank of Ireland, financial statistics

46  Money, prices and pricing macro-statistics manuals and articles (Copeland 1949, 1952; ECB 2012b). It’s the received wisdom of the science of economics. Remarkably, and this is a great achievement of the statisticians, the sectoral delineations as well as the definitions of financial instruments are internationally ‘normalized’ by the statisticians in the same sense as weights and measures are normalized. A French bolt fits into a Japanese nut as both countries use the metric ISO standardization system instead of the British imperial system (still in use). Comparable standardization has been accomplished with the national accounts, a drawback however being that this leads to lock in and hysteresis. As the economy is a moving target the statisticians try to circumvent this by regularly updating the manuals. The accounts, even when not expressed in units of GDP, also show the importance of receivables and payables as a means of payment and a debt instrument. Ireland has a population of 4 million, the payables accepted by creditor companies in the fourth quarter of 2015 alone already amount to almost €20,000 per inhabitant. This quarter is a black swan – but this swan seems to have had a lot of ugly ducklings, which shows that economies are moving targets. It is also important to note that these data come from somewhere. Companies are obliged to provide them, statistical institutes and central banks gather and classify the micro-data. The sheer magnitude of the developments shown in Graph 2.1 forced the Irish Central Statistical Office to calculate (after consulting organizations like Eurostat) a corrected measure of GDP (OECD 2016). Returning to payables/receivables as money – directly and indirectly state involvement in all the monies mentioned thus far is considerable. Postal services enjoy a state monopoly (or at least often did), the joint liability rule was an official law and receivables and payables are, as stated, a legal means of paying a transaction, ‘legal’ meaning that transfer of ownership related to the transaction is final but also that when the debtor does not pay action can be taken to seize assets.2 Even private money relies on the state – developments in Ireland are related to its status as a tax heaven. But deposits and ‘legal tender’ are different. In the case of legal tender the state does not only back the transactions but also the money itself, in multiple ways. It accepts it as a means to pay taxes. It guarantees a 1:1 exchange rate between cash emitted by the state and deposit money emitted by designated banks – system banks or, as they are called in the statistics, monetary financial institutions (MFIs). As all deposits originated by all MFIs enjoy this state guarantee which also guarantees a 1:1 exchange rate between money emitted by these banks and which is the backbone of our fiat money system. In the eurozone, the European Central Bank provides ‘emergency liquidity assistance’ (ELA) to banks running into trouble when depositors change their deposit money for cash (i.e. use bank money to buy government money) to guarantee this exchange rate. These additional loans from the ECB enable these banks to buy additional cash from the ECB which can be sold to customers. The government also permits banks to use the particular brand connected to a national unit of account, like ‘dollar’ or ‘euro’, for the money emitted by banks – for free! Money is a shapeshifter, a social contrivance as well as a means of exchange, a store of value and it has a unit of account. It’s more often than not based on debt.

Money, prices and pricing 47 And it’s backed by the state. Aside: 18th- or 17th-century probate inventories are often stated in the unit of account, like the guilder of twenty stuivers (a stuiver was a Dutch pre-decimal currency), but often also list considerable amounts of different kinds of means of exchange like Thalers or Doubloon’s or Pieces of eight. Nowadays and thanks to the state the unit of account and the means of exchange are joined (at least within nations). This does not have to be the case. See Koning (2019) for the example of the Haitian dollar, which only exists as a unit of account. A 16th-century example are auctions ‘by the burning candle’ of farmhouses (not the land, which at that time was not juridically attached to buildings) in 16th-century Friesland. All of these were in guilders of 20 stuivers – but the debt invoked might be paid in totally different coins (generally, around 100 guilders had to be paid directly while the rest of the debt had to be paid in installments. The buyer also had to pay to tavern costs or ‘gelach’, if he came short this could halt the final transfer of property) (Borghaerts and Knibbe 2017). After these musings we arrive at the central theme of this chapter. How does money and transactions define our lives and how does this relate to the measurements? What do economists say about prices and transactions? This is the object of the next section.

2.5  The nature prices and pricing in a monetary society In 1944, the same year when Wesley Mitchell assigned Morris Copeland with the task to develop the FOF (Mitchell 1945a, p. 61), two books about the nature of our monetary society were published. Both books were written by survivors of the Austro-Hungarian empire. Both authors took part in World War I, one at the Russian side, the other at the Austro-Hungarian side. Both books would rapidly get and still have an iconic status. In the UK. Friedrich (von) Hayek, born in Vienna in 1899, published The Road to Serfdom (Hayek 1944). In the USA, Karl Polanyi, born in Vienna in 1886, published The Great Transformation (Polanyi 1944). The books have a quite different but related messages and the same theme: taking the road pointed at by Veblen (1898) they investigate how evolutionary developments lead to a monetary market society including the culture and politics and people living in these societies. Both books investigate how money, prices, monetary structures, institutions and coordination systems guide and misguide these developments from a moral and social and economic point of view. Important: both authors wrote about really existing monetary prices – not about psychological ‘neoclassical’ prices derided by Veblen or shadow prices calculated by the economist. These books can and have been criticized and rightly so. For an insightful and erudite discussion of the ideas and approach of Hayek and the circle of people around him as well as Ludwig von Mises see Slobodian, 2018. For an insightful and erudite discussion of the ideas but also the approach of Polanyi see Hejeebu and McCloskey (1999). The discussion that follows is, however, not so much about the specifics of the books but about their overarching theme: the nature of our monetary society and the direction of the evolutionary change of monetary societies – as guided by prices and money and monetary

48  Money, prices and pricing coordination systems like market contracting and government expenditure and rules. It does not aim at providing an extensive description and analysis of monetary societies. It also does not pretend to be a lengthy, thorough and critical assessment of the ideas of Hayek and Polanyi. But it does try to investigate the implications of these ideas for conceptualizing and measuring a fundamental economic variable: transactions – not just when it comes to measure these transactions but also when it comes to understand the nature of the results of the measurements and estimations. To clarify the concepts and ideas, some empirical examples are used. These are mainly based on historical transactions in the Dutch dairy chain – as those are the transactions I have the most in depth knowledge of while, as milk and dairy are historically relatively homogenous products, these also enable longer term comparisons. I will have to discuss the ideas of Hayek and Polanyi at some length as I might understand Hayek and Polanyi in a different way than others. As the discussion will be based upon my understanding, it is necessary to make this explicit. But enough caveats: let’s go on with what still is the immodest endeavor of characterizing our monetary economy in one short paragraph. As a kind of counterpoint (or synthesis, if you want) I will also discuss the ideas of Van Bavel (2016), who looks at the influence of a monetary market system on wealth and inequality. Hayek stressed the role of monetary prices in economic coordination in a market system (Hayek 1944, 1945). Prices of inputs and outputs are, for producers and households, a piece of information which enters in a quantitative way into the complex, dynamic and uncertain monetary organization of these companies and households. Outside these organizations and households nobody exactly knows what these prices mean for these households and organizations, which means that outsiders lack the tacit and extensive knowledge needed to enable companies and households to adapt and thrive or even survive given exogenous price changes. In this framework, the informational content of a price is not mysteriously embedded in the price itself but only exists in relation to the production or consumption structure of the company or household which has to deal with this quantity. Prices influence, depending on technology, tacit knowledge and even the nature of the accounting system like direct costing versus absorption costing decisions about present and future production and consumption. Of course, people make mistakes and sometimes price changes will make some enterprises simply not viable anymore. Market forces will weed out such companies. As companies are assumed to be largely price takers, they will use prices as a fact and not so much as a goal or a means and adapt production and the production process as good as it gets to whatever prices society comes up with. Higher output prices will generally lead to more production and less demand while, as ‘markets’ enable and reward active search behavior, the price mechanism will, even when everybody only minds his or her own business and does not steer a strictly monetary-rational course, lead to an efficient allocation of production and consumption. Overproduction will simply not be sold (at an efficient price) while underproduction will lead to higher prices and higher production in the future. It will not necessarily be existing companies who decide about higher or lower production – it might well be

Money, prices and pricing 49 that existing companies go bust or that new companies enter the market. Hayek (1944) also argued that direct government interference in markets (as opposed to setting the rules or providing basic social security) would lead to ever more interference. Not because prices would contain less or the wrong information for individual companies paying or receiving them – for these, it would still be a piece of information entering the monetary production system. But because it would disable the market mechanism to work and would hence lead to shortages or overproduction, which would require even more government interference. And in the end a kind of economic dictatorship and suboptimal outcomes. Part of the market coordination stressed by Hayek is however dependent on ‘bourgeois’ social values and rules which have evolved in the course of centuries.3 Markets do not operate in a vacuum but are a social enterprise, which requires trust and trusted institutions – habits and structures and cultures which are continuously evolving. This process had and has no clear predefined goal but would, as history showed, lead to an increase of prosperity and a more civil society. Importantly, Hayek (1945) stresses the importance of actual monetary costs and monetary, not psychological, prices and of tacit knowledge of minutiae. There is no magic in markets. There are just people struggling to make both ends meet in a complex, dynamic, social and monetary environment – which sometimes works wonders. In a sense, the producers and consumers in Hayek (1945) are blind when it comes to the rest of the supply chain and are guided by their prices, market transactions and the structure of their own production process only. This blindness is not a universal characteristic of modern production. Coase (1937) stated that companies often have to abstain from market contracts as these were, exactly for the reasons mentioned by Hayek, not efficient. Adapting your company to ever changing prices invokes costs while, in the real world, many business to business prices, quantities and qualities are negotiated – a time consuming process. A better or at least cheaper solution than ever lasting negotiations and blindly stumbling along is to incorporate multi stage production processes in a firm – or, as for instance in the case of cooperative and private dairy factories, within an organized supply chain – and to manage them. This also means using inter- and intra-factory administered prices guiding the production chain, which requires detailed knowledge of the production processes. In the case of a dairy factor it makes little sense to negotiate with farmers about the prices of perishable milk on a daily basis. The perishability in combination with high transport costs raised, once dairy production moved to the factory, the specter of asymmetric market power. Using a balance sheet phrasing: inter and intra company administered price contracts and planned management of complex production processes become incorporated in the ‘goodwill’ and hence the equity value of a company. Coase, however, did not initiate this discussion. Means (1935) already showed, based on extensive research, that ‘administered prices’ (which are, contrary to the administered prices of present day statisticians, not prices set by the government but prices set by companies), based upon longer term contracts and relations between companies were omnipresent. And behaved quite a bit different than ‘pure’ markets like the stock market, to an extent because entrepreneurs

50  Money, prices and pricing anticipate the problems mentioned by Hayek and tackle these. Ideas like those of Coase and Means led to the development of the branch of economics nowadays known as Transaction Cost Economics (Ketokivi and Mahoney 2017), which stresses the importance of fixed prices to organize firms as well as value chains and the importance of, in case of more complex products and product processes, embedding crucial activities within the firm instead of outsourcing them. The most basic difference between Hayek’s view and the business-economics related views of people like Means, Ketokivi and Mahoney might be this: in the world of Hayek ‘blind’ people adapt production and purchases to prices, going forward in a tentative way. In the world of Means it pays off going forward while organizing an adjacent part of the production/consumption chain. To do this and to invest and organize the company, producers as well as suppliers need longer term contracts. An interesting historical example is Tussenbroek (2013) which studies building contracts in the Netherlands before 1650. These contracts – connected to the rise of independent contractors – are clear examples of what Skitmore and Smyth (2007) call ‘traditional contracting’ (TC) instead of ‘design and construction’ or ‘speculative building’. TC is a system of pricing which, according Ketokivi and Mahoney is not consistent with ‘blind’ neoclassical pricing and is quite often even characterized by ‘perverse’ demand curves, i.e. higher prices are sometimes empirically associated with higher demand as they are associated with higher quality. As construction projects are often one of a kind, quantity and quality and price tend to intermingle. But even when the project price is known, costs are only known with any kind of certainty after building has started (elements of design and construction prices as defined by Skitmore and Smith are visible in the historical contracts but the frequent use of competitive tender by clients made me opt for TC). For mass production processes, Frederick Lee investigated pricing systems in economic theory (Lee 1999). He stresses the importance (and omnipresence) of cost-pricing technology based on ‘normal’ costing and direct and indirect costs in a multiple product company, pricing systems, which by their very nature have a long-term character as they are based on tacit knowledge about ‘normal’ production as well as on the quasi-stable relation between the different products and services produced by a company. ‘Normal’ should be understood as ‘normative’ in relation to the size of the production process and the capital invested. These pricing technologies take long-term changes in capital goods used and the organization of firms as a multiple goods producers into account and by necessity contain choices about the distribution of costs over time and between products. It’s a more complex society than the romantic one of small, independent, price taking producers sketched by Hayek. But monetary prices are as important as in the world of Hayek or even more so as long-term price contracts about pricing, quantities and qualities become part of the production structure itself. Interestingly, recent big data research basically yields the same results as the work of Means (Hiroshi et al. 2015). An example would be farm prices for milk in Friesland.4 Were these prices just quantitative pieces of information only getting a meaning in relation to the farm or were these prices part of an longer term strategic organization of the

Money, prices and pricing 51 production chain? After 1880, already specialized Friesian dairy farmers switched from selling butter, low fat cheese and animals to national and international resellers (the London market!) to selling milk to dairy factories while they continued to sell animals. Butter and cheese production was shed from the farms which as a consequence became more specialized. Quite fast, the private and cooperative factories switched, using cheap modern technology to estimate the fat content of milk, from paying farmers per kilograms of milk to paying for the amount of butterfat. In the case of cooperative factories, this was endorsed by the members of the societies. This process has continued. Nowadays, the milk price is based upon content of fat, protein, lactose (milk sugar), time of the year as well as bacterial quality, efforts to introduce measurement of phosphorus to be able to mitigate phosphorous outflows of farms are ongoing. Again and again, this was enabled by using scientific methods to estimate these variables, the methods getting spectacularly cheaper whenever they were up-scaled. This was not unique to Friesian farmers and companies but they were either early adopters or (as far as we know) initiators (protein), working together with research institutes like the Wageningen agricultural university. Interestingly cooperatives were in the lead when it came to innovative pricing. The largest private company in the north of the Netherlands, Lijempf, was a clear and explicit follower. The same scientific methods used to estimate fat and protein and bacterial quality at factories were used to provide farmers with tacit knowledge about their own productive cows and production processes, to enable them to increase production, content of fat and protein and, especially, quality (Knibbe and Molema 2018; Knibbe, Molema and Plantinga, in progress). The basic formulas to calculate milk prices were quite stable over the decades even when ever more variables were added while prices of butter, cheese prices, whey and milk powder were imputed and factory processing costs were subtracted (the rationale behind high winter prices was a wish to make optimal use of spare capacity during this slack season). Needless to say, members of the cooperatives had long-term contracts with the factories or the other way around, partly also because of the extreme perishability and bulky nature of milk which, up to about 1970, severely restricted long range transport of milk. On the farm level, a genuine Taylorist mindset focusing on frequent objective measurements of contents of milk per cow and purposeful breeding took hold.5 Data on fat content of milk were used for breeding purposes even when ‘nobility’ of cows stayed an ill-defined but explicit and important breeding purpose.6 As early as 1910, the Achlum cooperative factory even inspected farms with on average low bacterial quality milk which led to the replacement of bacteria invested wooden buckets by seamless metal buckets and introducing lower prices for low quality milk. Comes in the government. During the Great Depression prices paid to farmers included an indirect government subsidy of about 30 to 60 percent, the money coming from, basically, a blend of consumer price levies, forced substitution of margarine by butter and import and export controls. As there are no free government subsidies factories and farmers had to adapt to higher standards of quality for milk. Including, from 1926 onward, ever higher sanitary standards focusing on combatting tuberculosis of humans as well

52  Money, prices and pricing as bovine. One telling example: in the west of the Netherlands and 25 years after cooperative factories made this mandatory in Friesland, government stipulations led to the disuse of wooden, bacteria infected buckets while sale of tuberculosis infected cows from the butter and cheese districts to the consumption milk districts in the west (where less calves were reared) became prohibited. During the war prices and production as well as transport of milk became even more regulated while after the war, government control continued, including export subsidies while import quote from countries like Germany – the most important trade partner – would last far into the 1960s even when it was clear from the beginning of that decade onward that the EEC was rapidly becoming a surplus area. Only in the 2006, after seventy years of heavily regulated prices, European governments loosened price controls (or, in fact, tied controls to relatively small ‘world markets’ in dairy products). After about 1975 surpluses became an ever larger problem and governments tried to curb spending first by restricting production and later by taking less responsibility for production qua prices which led to a marked change in the pricing pattern (Graph 2.2). As farmers were bound to maximum quota of nitrogen and phosphorous this did not lead to a sustained increase in production. Just like farmers, more than a century before, switched from producing and selling butter to producing and selling milk, they now – reluctantly but unavoidably – became producers of a combined milk and manure composite. The conscious connection of milk prices to factory costs, the season, multiple dairy products and multiple

50 45 40 35 30 25 20 15 10 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Standardized 3,7% milk fat

12 per. Mov. Avg. (Standardized 3,7% milk fat)

Graph 2.2 Farm gate milk prices, the Netherlands, before and after globalization of the market Source: Wageningen University and Research, Agrimatie https://agrimatie.nl/Prijzen.aspx?I D=15125, accessed 11 November 2018

Money, prices and pricing 53 elements of milk answer the question posed at the beginning of this paragraph: there are many ways to establish a price and many methods to establish a price. In this case, pricing and hence prices were part of long-term strategic management and economists should think more about pricing instead of prices.7 Output and input pricing as well as ownership structures and price setting of products but also of waste flow becomes part of the long-term dynamic cost and production curve (De Haas and Knibbe 1993). In the end this led in combination with technological developments to a process of increase of scale of farms (a five to tenfold increase), factories (a ten to hundred fold increase) and also supermarket retailers (on the level of negotiating contracts with suppliers: a hundred to thousand fold increase), all of them increasingly embedded in production chains which require ever closer ‘commodification’ of products enabled by ever more high tech methods of measurement and ever more precise contracts, while government demands regarding flows of nutrients and discharges (including dead animals containing phosphorous) and of sanitary as well as veterinary ‘quality’ of milk, farming and factories has enormously increased – just like the number of products. When it comes to labor market rules, companies were generally benign (though less so once scale increased and increased) ‘rule takers’, if it was about the 8-hour working day around 1919 (introduced in these companies, too, according to the annual accounts) or collective bargaining or pensions or safety. On the other hand the relentless increase of efficiency which started around 1905 when the first ultra-small factories started to merge continues, unabated, until today. The hand guiding production has become very visible and contractual. Remarkably, moral elements are, to an extent, also visible in contract and price guided international production chains and becoming part of what marketeers call the ‘augmented product’. A non-trivial example is in the case of slavery and Chocolonely chocolate. The Chocolonely ‘no slavery’ statement requires extensive knowledge of the production chain. In the case of animal welfare, ‘Beter leven’ is a successful certification mark. The major European supermarket chain Lidl is planning to only sell ‘Beter Leven’ meat anymore, which requires extensive knowledge of the production chain in combination with adding some additional clauses to the already highly specific long-term contracts with suppliers. In the case of biodiversity and margarine Unilever tried and probably tries to become an example when it comes to palm oil production. These are not niche products. In 2017 Chocolonely had a 17 percent market share in the Netherlands (Tony Chocolonely 2017). Lidl will only sell Beter leven meat and other super market chains are following at least in the Netherlands and Unilever is a major supplier of margarine. At this stage, Chocolonely can guarantee that the cacao it uses is slave labor free, but not the cacao butter – Lidl is still in a transition process and Unilever surely does not yet know everything which happened on the (only!) 332 supplying palm oil farms in 2017. Global supply chains still have to make a larger mark here! Also, at the time of writing it just sold its margarine division to archcapitalist KKR. On the other hand, the Unilever (Ben and Jerry’s) / Cono (a dairy cooperation) ‘Caring Dairy’ initiative really is cutting-edge when looking at the sustainability details of the integrated management system. But the point:

54  Money, prices and pricing administered prices are not just about anonymous market forces. They are also about organizing global value chains and measured qualities and quantities of products, sometimes even including ethical aspects of production. They are consciously used as a tool to influence production processes of suppliers. Products augmented with market based moral considerations and measurements bring us to Polanyi. Like Hayek, Polanyi looks at coordination and more specifically at the requirements for market coordination. He stresses the dynamic, global nature of the modern economy as much as Hayek and other members of the famous Mont Pélerin Society do (Slobodian 2018). But his main argument is that these global markets lead to a kind of ‘dictatorship’ of markets and price oriented goals of companies which changes people into a factor of production and hence to dehumanization, commodification and alienation of people and nature. As can be seen from the economic measurement units in the 3,800-yearold Code of Hammurabi to the labels on super market products, markets crave measured quantities and qualities. This comes back in prices, from bacteria, fat, protein and lactose in producer prices for milk to the time your car sands in a paid parking place. Just read any label on a supermarket product. This craving extends to the measurement of people too. Time and production is measured and, to an extent, defined as the outcome of these measurements. Hours and days of work in combination with job descriptions, coaching trajectories and flow sheet based procedures abound. Just like in the case of products people and human actions have to be commodified just to fit the market mold. People are increasingly defined and understood as parts of this monetary market machine – anonymous and replaceable cogs which are trained and incentivized to be as cog-like as possible. Also, unfettered markets will lead to the depletion of natural resources as these are no longer defined by their intrinsic value or beauty but by their monetary value. The phrase ‘natural capital’, used to classify and sometimes value ‘eco-system services’, is a clear example. But this is not the end of the story. As history showed, spontaneous action had led to the growth of non-market countervailing power as well as countervailing government rules and interference which prevented the dehumanization and alienation of society and people at least to an extent. Polanyi was massively wrong on the periodization of the rise of wage labor and comparable institutions (Hejeebu and McCloskey 1999; Van Bavel 2016). He states that the rise of a labor class started around 1800. For Europe data on wage labor like money wages already abound for the post-1400 period. Large private companies employing hundreds of laborers were rare, government projects like maintaining and improving waterways or harbors did often employ a multitude of workers, which surely in combination with seasonal labor flows suggests the existence of something like a labor class. On the other side – the databases of the International Labor Organization, at present a Polanyian countervailing institution established in 1919 by the Treaty of Versailles, shows that it is only quite recently that wage employment has become more important than self-employment. The historical processes Polanyi hinted at seventy years ago are still playing out – even when measuring labor as the ILO does is a step toward commodification, too (Graph 2.3).

Money, prices and pricing 55 1800000 1600000 1400000 1200000 1000000 800000 600000 400000 200000 0

Employees

Own account workers and contribung family members

Employers

Graph 2.3  Workers of the world Source: ILO, labor force statistics, status in employment – ILO-modelled estimates, Nov. 2018

The ideas of Hayek and Polanyi did not come out of thin air. To name only one development: both must have been very aware of the sudden introduction of the 48-hour work week in many countries during and shortly after World War I – a long standing aim of labor. This truly was a watershed. A bit of the gist of the time can be gleaned from the first annual report of the International Labour Office (ILO), established by the 1919 Versailles peace treaty. According to the report: It would therefore be almost impossible to exaggerate the truly revolutionary character of the events which during the last years of the war or after the war took place in the sphere of the regulation of hours of work. . . . During the years 1918–19 the 8-hour day has, either by collective agreements or by law, become a reality in the majority of industrial countries. . . . Before the war, whenever even a minimum or gradual reduction of the working day was proposed, foreign competition was one of the arguments employed in the universal controversy concerning the working day. (ILO, annual report 1920, paragraph 125 and 126)8 The last sentence might ring a bell – it mentions the same argument which is used to force ‘program countries’ on eurozone countries to cut real wages and workers’ rights. This ‘truly revolutionary’ event must according to the ideas of

56  Money, prices and pricing Polanyi as well as Hayek have been guided by global competition and strive and government rules and rulings and a change of culture and habits. Not coincidentally, the International Chamber of Commerce, which just like the ILO after World War I set up shop in Geneva and which employed, among others, Ludwig von Mises, vehemently propagated the idea of global competition. At the end of the 1920s both Friedrich von Hayek as well as Ludwig von Mises were involved in these efforts (Slobodian 2018) which contributed to what’s nowadays called the ‘Polanyi problem’ or ‘how the current tendency towards the creation of a global free-market economy can be reconciled with a degree of stability and cohesion in society’ (Munck 2004) – a problem which around 1918 was solved, to an extent, by the introduction of the 8-hour day. The change in labor market institutions around 1918 was not the result of rational government or private policies aimed at improving live by using the spoils of increases of productivity to increase family time but a wave which started in Uruguay in 1915 and engulfed the globe in subsequent years. It was no doubt stimulated by the labor shortages caused by World War I and by the turmoil and fear instilled by the Russian Revolution of 1917. There is more to the commodification of labor than selling your labor by the hour. The hour itself is an invention. And so is ‘social time’ or keeping people, directly or indirectly, to time schedules like eight hour working days. An eight hour working day only makes sense in an industrial monetary society with a large number of wage laborers and complex production systems which require exact coordination of people, stocks, machines and buildings, where hours are counted and remunerated and where labor commodified at least to an extent. Note that long ago, every city had its own time and it’s not a coincidence that it were the railroads who started with the project which would result in national time. See Thompson (1967) for what’s supposed to be one of the first studies about this in the Anglo-Saxon realm, Mayall (1942) however seems to predate him. Every time you hear a church bell ring think about it that this was the way pre-watch societies kept track of prayer time. Boerner and Severgnini (2016) show that medieval cities who were among the first to introduce clocks knew higher rates of growth than other cities – tough at this time every city still had its own time. It would take some centuries, the growth of the nation state and the meticulous time tables of the 19th century railways before the church clocks were nationally synchronized, hours were counted an wages were converted from daily wages to hourly wages (Mayall 1942). This process still goes on. To quote a Daily Mail headline: ‘Amazombies: Seven seconds to find an item, every move filmed and blistering 12-hours shifts with timed toilet breaks . . . what YOUR Christmas order does to your “worker elves”’ (Daily Mail 9 September 2018). Nudged by low Internet costs and rapid deliveries we’re increasingly living in the epoch of camera discipline 24/7 rhythms now – and foreign competition is still used as one of the arguments in the struggles about this. The monetary economy is a measurement economy – and we’re getting better at measurement all the time. There are strong alienating pressures which press us to think and feel like an Amazombie and to change our behavior accordingly. In one of the subsequent chapters, I’ll laud the invention and introduction of the ‘weekend’, which can be understood

Money, prices and pricing 57 as a Polanyian adaptation to high productivity production processes, even when there are strong forces to erode such institutions. An author taking on the same challenge as Hayek and Polanyi but adding ‘capital’ to the equation is Bas van Bavel (2016). He analyses the long run development of four monetary market societies: Mesopotamia between 500 and 1100, North Italy between 1100 and 1600, the Low Countries between 1500 and 1800 and (in a limited way) the present-day ‘modern economies’. Time and again the same patterns are visible. Vibrant market economies create a class of rich people who can invest a larger part of their larger income for a longer time in riskier projects than other people which enables them to have a higher rate of return over a larger part of their income than poor people while they can afford to reinvest a higher share of these already higher earnings than small savers can. This leads, using a mathematical expression, to an ‘evolving power law’ pattern of the distribution of wealth. Over time, ever more wealth is concentrated into the hands of ever less people. It is a main contribution of Van Bavel that he shows this for a whole number of civilizations. His idea is also corroborated by recent data on long run returns on capital: the return to different kinds of capital (ownership of real estate, different kinds of bonds and equity) is in the long run quite a bit higher than the rate of economic growth while long-term ‘risky’ rates are higher than short-term rates (Jordà et al. 2017, also Piketty 2014), which means than when these returns are not consumed or expropriated – and rich people can afford not to consume it – wealth of rich capital owners tends to grow faster than the economy. We can give this an interesting twist by relating it to recent research by Fix (Fix 2018) and Liu, Mian and Sufi (Liu, Mian and Sufi (2019). Fix shows that, within firms, wages are related to ones position in the hierarchy, not to whatever direct or indirect metric we have of productivity. Firms as a whole may of course, because of competition, be forced to choose an productive combination of labor and capital for the entire firm but this is clearly not true within the firm. Liu, Mian and Sufi show that to gains for industry leaders to invest in productivity enhancing technology are larger than for laggards, especially in low interest surroundings like those prevailing after 2008. When interest is low for a somewhat longer time, it becomes less profitable for low productivity firms to catch up, which increases productivity differences. These differences might reach a magnitude where it isn’t interesting to invest anymore even for industry leaders, leading to a lower level of overall investment and productivity growth. Apple Inc. and comparable companies, with tens of billions of idle cash on their balance sheet, come to mind. Acemoglu and Robinson (2008) show that the rich, as there are few of them, as they have money and as they have more to lose than their chains, might tend to dominate informal networks as well as structures yielding economic power. Aside of the economic processes already leading to an ever more unequal distribution of wealth eventually, the rich will gain political influence and use this to rig the system in their favor, while ever more of their income will come from rents. To state this differently: an ever larger part of their income is based upon legal, political and ownership rights favoring the rich and not on labor or enterprise or productive

58  Money, prices and pricing investments, which leads to the demise of societies. Van Bavel differentiates between financial investments and fixed investments and states that, over time, a larger amount of total investments will tend to go to financial investments which, unlike fixed investments, do not contribute to the expansion of the productive capacity of societies. The increasing preponderance of financial wealth – and who will deny that our present day society is characterized by an increasing preponderance of financial wealth – also contributes to the rise of a class based society with ‘ownership’ as the prime demarcation of class. He pushes back the origins of the ‘great divergence’ as explained by Polanyi from around 1800 to at least before 500. By extending his analysis from economic exchange proper to wealth accumulation enabled by the system and to the influence of ‘the rich’ on politics he puts the insistence of Hayek of the civilizing aspects of market exchange, stressed by McCloskey (2006), in a dismal perspective. Markets will be rigged – by the rich. Though it has to be stressed that the productive forces of our modern societies, which according to Van Bavel show the same processes of petrification and ever increasing rent incomes instead of profit seeking, are unimaginable larger than those of the preceding societies, which might counteract such tendencies. After 1880 better and cheaper transport and higher agricultural production and productivity in combination with the low price elasticity of demand for agricultural products led to lower agricultural prices. Lower prices not only increased ‘real’ manufacturing and construction wages but also led to an epic decline of the income, wealth and influence of the ‘landed interests’ and increased dynamism of modern economic societies. This might have been a one off event. A neoclassical model like Acemoglu and Robinson (2008) doesn’t explain such events as it’s about a one good economy without changes in relative prices. Nowadays, a secular increase of rent incomes is visible leading to a new ‘landed class’. The influence of technology itself on the capitalist system should not be underrated but even then the present increase of inequality does bear resemblance to the processes described by Acemoglu and Robinson (2008), Piketty (2014), Van Bavel (2016), Liu, Mian and Sufi (2019) and Fix (2018). See also Ager, Boustan and Eriksson (2019) for how sons of families from the southern states of the USA which after the civil war lost their slave wealth quickly managed to bounce back, capitalizing on their connections and land ownership. To paraphrase Scarlett O’Hara: for the rich ‘tomorrow was another day!’ Or as Acemoglu and Robinson phrase this: ‘slavery was replaced by monopsonistic arrangements, policies designed to impede labor mobility, political disenfranchisement, intimidation, violence and lynching’ (Acemoglu and Robinson 2008, p. 269). This, however, is all related to prices as elements of contracts. On another level, the accounting level, and it is this level which brings us to the statistics, they are related to money flows. Flows of income, profits, expenditure and the monetary value of production. This, not prices as incentives or as inputs into production processes, is what the national accounts and the FOF measure. These categories are remarkably resistant to historical change: a profit or a wage of a hundred years ago are still recognizable as a profit or a wage, today.

Money, prices and pricing 59 The underlying reality does change, because of increasing inequality, ever more complicated and integrated production chains or changes in the nature of the working week and working year or because activities cross the border of the production boundary and they become (de)monetized. The government is always within shouting, or even whispering, distance, setting the rules, enforcing contracts and investing in the technological and physical infrastructures. Next to this the rise of institutions which aim to subjugate this system to the interest of rich owners, therewith thwarting dynamism, skewing incomes and wealth and corrupting the political process can be witnessed. Ideas like those of the authors mentioned in this paragraph underscore the importance of measuring monetary flows using real prices, which is what the measurements do. But they also underscore the importance to understand the flows of income not just as a consequence of growth or ‘benign’ monetary market exchange but also as the result of supply chain transactions, social struggles and evolving power laws and rent extraction at work. The next section contains a case which aims to show how the measurements relate to this.

2.6  The nature of the statistical production boundary The statistics are, as we’ve seen, transaction based. To be able to do this, a boundary demarcating monetary from none monetary activities is drawn. The size of the area within these borders is measured. This size (and what we measure) is dependent on where the border is drawn. Ideally, the borders – not just the production/consumption boundary but for instance also the borders between sectors – are drawn in a way and place which enables rigorous analysis of the evolvement of our economies. Is this the case? We’ve seen that monetary transactions are astoundingly social. Is it, against that background, right to demarcate the monetary realm from the non-monetary spheres of life? Is it meaningful to draw a border between monetary and non-monetary activities? What is the concept and operationalization of this border? We will investigate this using a casus: domestic labor. It will, using a historical approach, investigate six aspects of the border and the area: • Where is the conceptual boundary between economic variables and other events drawn? What is, when we look at actual measurement, included and what isn’t? • Do the definitions of the production/consumption boundary change? • Does the size of the area delineated by a variable change because of (de) monetizing of activities? To state this otherwise: do activities sometimes cross the border? • Does the nature of the area delineated by a variable change? Are activities like labor or consumption nowadays comparable with these activities in the past? • Are the definitions themselves part of the social changes and social and political struggles mentioned? • And, not entirely unimportant, can we measure it at all?

60  Money, prices and pricing We will focus the on (de-)monetization of household labor. The points discussed will be if concepts shifted, if definitions shifted, if quantities shifted, if activities were (de-)monetized, if the nature of household work changed and how this relates to statistical politics. In the meanwhile, I will also try to crush the meme that when a men marries his maid measured GDP will shrink. But first, some basics. The accounts are and were clear about unpaid household labor. In 1921 King, Macaulay, Mitchell and Knauth published their milestone ‘Income in the United States: Its Amount and Distribution, 1909–1919’, which estimated time series of nominal and real income in the USA. The authors were well aware of the limited nature of their estimates of the money economy and stated: ‘Following common practice once more, we do not count as part of the National Income anything for which a price is commonly not paid. On this score we omit several of the most important factors in social well-being, above all the services of housewives to their families’ (they do present a very rough monetary estimate of the imputed monetary value of these services).9 About twenty years later Elsas wrote an article titled ‘The definition of national income’ (Elsas 1944). The article was based upon the results of a survey of leading national accounts economists from around the world. Some names: Simon Kuznets, Morris Copeland, Gerard Verrijn Stuart, Arthur Bowley, Corrado Gini, Wesley Mitchell. Based on this survey and related literature Elsas states (reminding us of the present state of DSGE models): ‘The main authors do not even all agree on the chief points’ (Bowley 1944, p. 6). But about the value of the work of housewives he is clear and in line with King et al. (1921) he stated: ‘the whole concept of national income is based on the principle that only those goods and services are to be taken into account which are bought and sold, or can be exchanged for money, but on the other hand to omit all those which are not’ (Elsas 1944). After World War II the United Nations initiated a project, initially headed by Dutch economist Jan Derksen, to establish international guidelines about the ‘chief points’ of the national accounts which turned out to become very influential (Bos 2003, 2013). One aspect of these guidelines: the accounts exclude unpaid household labor. Even the present ESA 2010 guidelines (Eurostat 2013) still mimic the quotes in the previous paragraph: Production excludes the production of domestic and personal services that are produced and consumed within the same household. Examples of domestic services produced by households themselves that are excluded are: (a) cleaning, decoration and maintenance of the dwelling as far as these activities are also common for tenants; (b) cleaning, servicing and repair of household durables; (c) preparation and serving of meals; (d) care, training and instruction of children; (e) care of sick, infirm or old people; (f) transportation of members of the household or their goods.

Money, prices and pricing 61 Domestic and personal services produced by employing paid domestic staff and the services of owner-occupied dwellings are included in production. (Eurostat 2013, pp. 54–55) Clearly, the discussion about unpaid domestic services is not new, it’s ongoing and the answer does not change: unpaid household labor is not included. But the underlying reality did change and activities did cross the boundary even when the boundary itself did not change. Elsas (1944) cites an 1922 article from Bowley: The ignoration of the value of women’s domestic services has been less plausible in recent years. During the war the domestic staffs of many houses decreased and well-to-do women rendered more services to their own households. If the housemaid left and made munitions and the housewife did her work, the total of goods and services was increased by the value of the munitions, but part is cut of the reckoning because no longer paid for. If in 1920 the former servant helped in her own home, when her wages were no longer needed, the total of services is still greater than in 1914 if her former mistress is still doing the housemaid’s work. Elsas goes on to state that this was the exception to the rule. But was it? Bowley’s example strikes the modern reader at first sight as slightly trivial, a technocrat pushing a technocratic detail for the sake of consistency only, not unimportant but arcane. A paid domestic servant which takes on another job, how important can that be? The short answer: very. As the investigation of domestic labor and the production boundary will show. A marked characteristics of the economic development of ‘the west’ is the gradual disappearance of the ‘housemaid’ or, to use ESA 2010 parlance, ‘paid domestic staff’. It is difficult to overrate the importance of this sector in the past. In 1910, a full 9 percent of the non-agricultural workforce of the USA consisted of domestic workers (Boyd Leon 2016). Or to quote an excellent NBER publication by Stigler about domestic workers in the USA and some other countries (Stigler 1946, p. 2): ‘in 1939 there were as many domestic servants as employees of the railroads, coal mines, and automobile industry combined’. Stigler also shows that a majority of the USA domestic servants were either black or immigrant women. The number of working hours was high, sometimes even 80 hours a week. Think also of the often dismal living quarters of servants in the USA around 1880 (May 2011). Think of the comparable situations nowadays. As the ILO stated in 2013 in the preface of the most ambitious report on global domestic labor up till then: In an unprecedented manner, this report attempts to capture the size of the domestic work sector and the extent of legal protection enjoyed by domestic workers on the basis of a verifiable and replicable methodology. Its findings contribute to overcoming the invisibility of domestic workers and carry a powerful message: domestic work represents a significant share of global wage employment, but domestic workers remain to a large extent excluded from the scope of labour laws

62  Money, prices and pricing and hence from legal protection enjoyed by other workers. Marginalization and exclusion is a theme that runs through the findings of this report. (ILO 2013b, p. 5) The ILO did a good job just like Stigler did 1946. See also the foreword of International Labour Conference (2011), signed by many ILO members, for measures to be taken a companion booklet about measures to be taken: ‘for the first time, international instruments are applied to an essentially informal segment of the global workforce’. Measuring won’t stop marginalization. But it has to happen. Measuring is however not easy. May states ‘Paid housework is thus perhaps even less visible to the public and to researchers than it was at the turn of the century or during the Great Depression’ (May 2011, epilogue, p. 8). As stated, the ILO tries to improve this situation. It needs clear definitions to do this. It discusses five alternative definitions of domestic workers and their (dis)advantages, in the end choosing The Domestic Workers Convention, 2011 (No. 189) . . . defines ‘domestic workers’ in Article 1: (a) the term ‘domestic work’ means work performed in or for a household or households; (b) the term ‘domestic worker’ means any person engaged in domestic work within an employment relationship; (c) a person who performs domestic work only occasionally or sporadically and not on an occupational basis is not a domestic worker. The simple, but very distinctive feature of being employed by and providing services for a private household is therefore at the heart of the Convention’s definition. It is narrow in scope. (ILO 2013b, pp. 7–11) Note that this definition coincides with the broadly defined production boundary of the NA: paid work. Using this definition and excluding the occasional baby sitter the ILO investigated labor statistics and censuses to gauge worldwide domestic labor, it arrives at a minimum estimate of 67 million domestic servants in 2016, which enables us to update Stigler’s remark about the number of domestic servants in the USA being larger than employment in the automotive industry, coal mining and railroads combined. Is this still the case on the global scale? According to a rough check it still is, taking account of employment in supply chains except retail (Organisation International des Constructeurs Automobiles 2019; Union international des Chemins de Fer 2017; Coderre-Proulx, Campbell and Mandé 2016, p. 5). But change does take place and in rich countries paid domestic labor has been on the wane. The same censuses and labor statistics which were used by Stigler are the main source for Table 2.1. No efforts have been taken to make the international data comparable. Despite this it is clear that after 1950 domestic servants as an occupational group dwindled. What caused this decline? Anderson and Bowman (1953) point out, for the Deep South of the USA and stressing the fact that the domestic workforce there consisted predominantly of black women, that this decline was supply induced. Alternatives opened up. Just like in the case of

Money, prices and pricing 63 Table 2.1  Number of domestic servants, selected years, 1899–1981 (thousands)   United States of America 1900 1930 1940 1950 1960 1970 1980 Great Britain 1901 1931 Germany 1895 1933 1939 1956 1961 1982 The Netherlands 1899 1930 1947 1960 1981

Number (Thousands)

Percentage of the Labor Force

Number per 1,000 of the Population  

1,509 2,025 2,098 1,492 1,817 1,204 1,062

 

1,344 1,484

  1,434 1,269 1,561 661 492 178   197 244 183 116 8

5.1 4.2 4.1 2.4 2.6 1.5 1.0   7.2 6.8   6.5 3.9 4.5 3.2 1.9 0.7   9.6 7.7 4.7 2.7 0.2

19.8 16.5 15.9 9.7 10.0 5.9 0.5   41.3 37.1   31.2 19.5 13.4 1.3 0.9 0.3   39.0 31.1 19.2 10.2 0.6

Sources: USA: Stigler (1946). ‘Domestic servants in the USA, 1900–1940’. NBER; Grossman (1980). ‘Women in domestic work: yesterday and today’ in Monthly labor review 103–8 17–21, p. 18. Great Britain: Stigler (1946). Netherlands: CBS (2001). Tweehonderd jaar statistiek in tijdreeksen. Voorburg/Heerlen; www.volkstellingen.nl/nl/index.html; CBS (1981). Beroepstelling 1981. Voorburg/Heerlen. Germany (after 1945 West-Germany): https://histat.gesis.org/ histat/de/table/details/373538AB7EBE48F0710D1EAC38300005#tabelle, accessed 17 July 2018.

a comparable decline in agriculture in the Dutch province of Groningen around 1900, as Collenteur and Paping show in an unusual detailed study (1997). And just like comparable changes in New York were supply induced (May 2011). Other job and educational options became available, not just in the market but, possibly because of rising male incomes (not just of spouses but also of fathers), also outside of the market. Contrary to the foundational fantasy of a popular GDP-meme, which states that measured GDP will decline when a maid marries her employer, maids did not do this. Too many socio-ecoomic differences. They left them. Which made total production, including unpaid household production, go up. It is clear that Bowley’s statement about shifts between paid domestic labor and other sectors of the economy was not trivial and arcane. We can look at this from a Hayekian/Polanyian perspective. To do this, we will first ask the question:

64  Money, prices and pricing who was this maid? May (2011) stresses that, up to 1914, in New York domestic servants as a rule were young, female and disproportionally Irish. Young women from southern European descent typically did not enter this vocation. After 1914, when first World War I and, after the war, stricter US immigration laws led to much lower rates of immigration these immigrant women and girls were replaced by black women from the South of the USA (this was the situation measured by Stigler). As these black women often had families of their own, they refused living in, changing the nature of domestic work while, nowadays and in a global perspective, the international immigrant woman has taken center stage again, especially when global economic and political developments lead to an increase in the supply of women seeking work. According to May (2011, epilogue p. 4): ‘Hiring a household worker therefore often means participating in a system of global labor exploitation’. This remark tallies with the findings of ILO (2013). Like in 1939 and even more strongly so before 1914, immigrant women are disproportionately involved in this sector. She also points out that, in New York between 1880 and 1940, domestic servants were in frequent contact with each other via family or neighborhood contacts or via churches or organizations aimed at furthering their interests. At the same time, a whole cottage industry of brokers (often female) emanated from the possibility to earn a little by matching supply to demand. A market demand for information about prices and quality of jobs clearly existed. As servants regularly changed jobs this did lead, despite the fact that the work as such was carried out in the secluded space of middle class households, to some standardization of (low) wages. Despite the often informal contracts, there are also numerous instances in which servants successfully claimed back wages via the courts which all fits in a Hayekian perspective where individual behavior is guided by (non-perfect) information on non-perfect prices and qualities and an imperfect personal accounting system. May (2011) also states that immigrant women from southern and central European descent did not want to work as life in servants, which fits the idea that cultural norms are as important for guiding behavior as prices. Looking at developments from an Polanyian perspective, the international and national flows of labor are part of the commercialization and commodification of life, people left their traditional communities behind to take part in the market economy. And to care for the kids of somebody else. There is a twist to this. According to Polanyi (and adding the labor market to this quote), While the organization of world commodity markets, world capital markets, and world currency markets under the aegis of the gold standard gave an unparalleled momentum to the mechanism of markets, a deep-seated movement sprang into being to resist the pernicious effects of a market-controlled economy. Society protected itself against the perils inherent in a self-regulating market system – this was the one comprehensive feature in the history of the age. Was this the case, too, for domestic workers? May (2011) explicitly addresses this problem. She notices that educated middle class women often played important

Money, prices and pricing 65 roles in the progressive politics of this area, pushing for restriction of hours of industrial workers and worker safety. Frances Perkins, arguably the most influential and, eventually, the most powerful of these women, even was a witness to the fire in the Asch building where 146 mainly female textile workers lost their live. The impact of this fire can be compared to the impact of the 2013 collapse of the Savar building in Bangla Desh, where 1,146 mainly female textile workers lost their life. It made Frances Perkins quit her job and start a distinguished career aimed at worker safety, employment and hours of (female) workers. It seems that the original Fair Labor Standards Act of 1938 for which she was responsible did include provisions for domestic workers; opposition from southern democrats gutted these (AAregistry 2019). But there was attention for their plight from high places. Remarkably, it was only in 1974 and to an extent even in 2015 that the US domestic workers eventually got their standards (US Department of Labor 2013). Perkins seems to have been the exception. Many of the activist women mentioned declined to extend their ideas to the work of domestic servants and did, unlike grass roots organizations of the servants themselves, not push for shorter hours and a minimum wage for domestic workers – even when they did care about education and more formal market structures. When it came to domestic services, US society did, for a century, not protect itself against the perils of a self-regulating market even when social struggle was clear and prevalent. Fast forward to today: the ILO does an outstanding job by trying to make domestic service and its perils more visible (International Labour Organization 2011; 2017). It has to be said that since May published her book (2011) some advances have been made. On the EU level, ‘In January 2014 the EU Council adopted a decision authorising Member States to ratify the ILO Convention. Currently, EP FEMM Committee is preparing a Report on Women domestic workers and careers in the EU (2015/2094(INI))’ (EUR-lex 2019). This all serves to underscore the countervailing forces stressed by Polanyi – and the role played by statisticians and economists. Domestic labor is at least a bit more visible. But it can still be doubted if it’s also more protected. Changes in the nature of labor and labor flows have not been the only changes in households. Technological developments were revolutionary, too. About these technological developments: let’s first realize that houses are, by far, our most valuable piece of fixed capital. The value of houses dwarfs the value of machinery and even roads and harbors. And do not underestimate ‘domotechnology’. Robot lawn mowers are the most numerous kind of robot even when robot vacuum cleaners are starting to threaten their position. But the relation between technological developments and household production is complicated. Chang (2011) states that the washing machine and comparable innovations enabled women to work outside the household. Chang, however, forgets about domestic workers. Arguing along his lines it seems probable that these inventions also diminished demand for domestic servants. Woersdorfer (2017) tests the Chang hypothesis but finds that having a washing machine had and has no predictive value for the amount of market hours worked by married women, for one thing because the introduction of washing machines led to higher norms for

66  Money, prices and pricing cleanliness. Maybe households bought washing machines to replace the vanishing maids, at least to an extent. But washing machines and robot lawn mowers and vacuum cleaners are far from the only kinds of household technology. We can think of the introduction of stoves in the 19th century, no-iron shirts, vacuum cleaners, readymade food (bread replacing porridge is an old example) or cars. An oldy: the replacement of the distaff by the spinning wheel, which tripled spinning productivity which, consequently, gave a boost to commercial weaving even when, eventually, mechanized spinning made spinning disappear as a household activity. All these innovations made and make the production boundary shift even when its definition does not change. But it’s not just that monetary activities like paid domestic work are leaving the household, to make room for homeliness. Airbnb is a recent example of an innovation leading to a shift of the monetary boundary back into the household just like the spinning wheel commercialized domestic spinning. With regard to domestic labor: prices cause and coordinate (inter)national flows of labor, social strive is prevalent and technological developments disrupt traditional patterns of production. Monetary accounts measure transactions and aggregate these into the national accounts but do not measure strive and only indirectly show shifts between paid and unpaid productions. But the transactions are based on an historical and global system of relations, administered prices and value chains, rules, power and technological change, which influence prices and flows of labor. Just think of the US immigration restrictions and the introduction of 8-hour working days after World War I. Or the other way around, the lifting of all legal immigration restrictions within the EU, while cultural patterns influence supply and demand for (domestic) labor, too. Just think of the southern European women who, around 1900, contrary to Irish women and for whatever reason refused to work as live in domestic servants. And of the importance of middle class values influencing standards of homeliness and therewith the work and workload of servants. Summarizing: the boundary surely can serve as an analytical device. To do this, we have to understand that it is a well-defined dependent variable showing historical changes. Returning to the production boundary: this boundary is clear when it comes to the definitions. It’s about the difference between monetary transactions and other kinds of production. If this little history shows anything, it’s that even when it comes to domestic chores, there sure is a difference between the same chores as performed by paid labor or unpaid labor (or robots). Also, the boundary shifts. Not too long ago this border cut right through middle class Western households but it nowadays is almost completely outside it, while it still cuts right through the middle of many non-Western households. It shifted out of middle class households, because of market, social, technological and political processes but seems to have mainly been a supply side process. At least in rich countries. The amount of domestic servants didn’t dwindle because single men married their young Irish or black or, nowadays, immigrant female housekeepers. They didn’t. But because of global economic and (bio-)technological social developments and migration flows. The servants had better things to do.

Money, prices and pricing 67 Returning to the six points raised at the beginning of this paragraph: • Where is the conceptual boundary between economic variables and other events drawn? When it comes to paid domestic labor, the boundary has not really changed. The national accounts measure paid domestic labor. They do not measure unpaid domestic labor. But considering the answer to the second point it’s not ‘where is the boundary drawn’ but ‘how is the boundary defined’. • Does the size of an ‘area’ delineated by a variable change? When it comes to paid domestic labor, the ‘area’ definitely changed. It shrank in Western countries tough though there were sizable differences between countries while the decline seems to have been a post-1960 development. Globally paid domestic labor is still a major occupation. • Does the nature of the area delineated by a variable change? When it comes to paid domestic labor the nature of the work changed tremendously in one way but much less in another way. When it comes to technology, changes are huge with the washing machine and running water as the most important examples of this. Also, living in servants have become less common. It still is, however, a subordinate kind of work often performed by ‘newcomers’ or ‘marginalized groups’ on the labor market and still connected to global flows of overwhelmingly female labor. Dignity, gender and social aspects seem to have been remarkably stable. • Do the definitions of a variable change (one can think of the new fixed capital definitions of the national accounts, which nowadays comprise military equipment while, in the past, did not). The definition has not changed. The ultimate goals of the activities of domestic of one hundred years ago are still the same as today, even when standards of cleanliness may have increased, partly because of domotechnological advances like the introduction of washing machines or also electric light. • Are the definitions themselves part of the social changes and social and political strife mentioned? Repeatedly, ILO reports or reports of the European commission are mentioned, the ILO sees it as its task, for social and political reasons, to measure it and provides precise definitions. The answer is, hence, yes. Monetary transactions are an important part of these definitions. • Can we measure it? Paid domestic labor can be measured in a way consistent with the definitions of national accounting. Interpretation: using a well operationalized monetary transactions based production shows that in the case of domestic labor the fact of payment leads to totally different flows of labor and coordination mechanism than is the case for unpaid domestic labor. Money is not neutral, in this sense. A well-crafted boundary enables us to show that activities enter as well as leave the production boundary, for economic, legal, cultural and technological reasons which means that it is a powerful analytical device, which means that it makes great sense to demarcate monetary transactions from non-monetary ones. Monetary transactions are social by nature and definition. They are the backbone not just of the magnitudes of the flows of income, production and

68  Money, prices and pricing expenditure but also organize these flows. The social realm encompasses relations, the legal system, power and politics. Transactions are not just made against the background of such a system but are also part of it. Every new transaction recreates and changes this system, sometimes also by creating and destroying money. But also by enabling and giving rise to national and global flows of labor and the construction and use of fixed capital, be it for monetary or non-monetary purposes. In this transactional sense, the traditional economic question if money is neutral is meaningless.

2.7  National accounts as an instrument of control A final point: ideas like those of Polanyi can also be invoked to point out that the availability of national account statistics is a sign of but also enables and facilitates the monetization and commodification of life. Think of the criticisms mentioned in the previous section or about how the 60 percent of GDP threshold for eurozone government debt is used to discipline entire countries – as the word ‘domestic’ in GDP already implies, the accounts are designed with national political policies in mind. Also, macro-statistics definitely got a boost during wars, when politicians needed information about the productive capacity of the economy as well as about the distribution of income (Keynes 1940). The macrodata, which includes data on unemployment and private debt, can however also be used as part of Polanyian ‘network of measures and policies . . . integrated into powerful institutions designed to check the action of the market relative to labor, land, and money’. Stapleford (2009) argues that the US Bureau of Labor Statistics and its predecessors as well as comparable organizations which, after 1885, were established in different countries had, as one of main tasks, to assemble statistics about wages and the cost of living. Assembling such data was considered one of the most important and consistent demands of US labor organizations at the end of the 19th century. And not just in the USA. Gathering such data can surely be seen as part of a strategy to resist the ‘pernicious effects’ of marketization, urbanization and the increasing importance of wage labor. As a consequence of such dialogues at present extremely detailed ‘baskets’ used to calculate the consumer price index have been calculated which even include the price of prostitution. This can to an extent be understood as the result of ‘class struggle’. Wage earners wanted a price index based on the recent price of groceries as they paid them, not of the prices paid by jails or orphanages. Some of the first national income statistics were even compiled with such ideas about social transformation in mind

Notes 1 The balance sheet of Microsoft shows an increase of around USD$50 billion for short-term debts (assets) as well as capitalized leases (liabilities) for these years. 2 It is noteworthy that postal companies earn a seigniorage income on the emission of stamps: the interest earned on money paid for stamps not yet used as well as the

Money, prices and pricing 69 profit on stamps forgotten somewhere in your drawer or bought by collectors (on private seigniorage: Knibbe 2015 and (more radical) Bossone and Costa 2018). 3 Reading (von) Hayek as well as Ludwig (von) Mises I always get the impression that they wanted to restore the pre-war social and monetary order of the AustriaHungary which enabled the upward mobility of their lesser-gentry families. See also Slobodian (2018). 4 Theoretically this story is consistent with Murray and Markley-Towler (2019) especially of the consistent strategy of factories to expand production to cut costs and to produce at the minimum cost level even when they do not cover changes in product specifications as part of the evolvement of the pricing system. See also De Haas and Knibbe (1993). 5 At the 2017 Leuven rural history conference three of the four presenters of the 8.1 Across the sectoral divide: the development and use of agrarian-industrial knowledge session independently explicitly mentioned the introduction of Taylorist methods in agriculture around 1900. To these measurements, scores of others (feed, soil, fertilizer) can be added. 6 The most remarkable feat was the eradication of bovine tuberculosis by the members of the cooperative dairy factory in Kimswerd in 1900/1901. The members obliged themselves and new members to sell all cows tested positive for tuberculosis – at a time when the contagious nature of it was still contested. At this time, tuberculin had only recently hit the market while using the services of professional veterinarians was also not completely common. 7 In 1927 the cooperative factories in Friesland (who worked together when it came to this) used a formula to calculate the price of milk, to understand the formula it is necessary to know that the milk was used to make butter as well as hard cheese while butter contained 15 percent of water while hard cheeses lost 14.5 percent of their weight during the ageing process: In the case of cheese with 20 percent fat per 100 kg of milk: butter value of milk = (fat in fresh milk – 0.88*fat in cheese milk)*1.15*price of butter: A cheese value of milk = 0.855*production of cheese*price of cheese production costs = 1.6 cents per 100 kg – price per 100 kg of milk (A + B – 1.6) This means that the price of milk is tied to technological aspects of the products (as opposed to consumer preferences), market prices and factory costs, which means that, contrary to the vision of Hayek, these prices do contain explicit information and are not just quantities which obtain informational content only in relation to the company of the producer. 8 According to the 1919 annual report of the Lijempf dairy company of the graph: Echter niet alleen het productievraagstuk baarde ons dit jaar zorgen ook de loonkwesties en arbeidersbewegingen vroegen veel van onze krachten en hoewel wij in deze zware tijden de exploitatiekosten graag zoo min mogelijk wilden opvoeren aan den anderen kant moeten wij toch ook met onzen tijd meegaan en hebben getracht in dezen zoo veel mogelijk den gulden middenweg te bewandelen. Achturige werkdag en vrije Zaterdagmiddag werden door ons ingevoerd en werd er tevens een vaste pensioenregeling aan onze zaak verbonden. (Short translation: we have to swim with the tide and introduced an eight hour working day and a free Saturday afternoon while also introducing worker pensions.) 9 The depletion of national resources is also discussed: ‘No systematic deduction from the National Income is made in our estimates to cover depletion of natural resources. Doubtless this item is of considerable size as well as of peculiar interest’.

70  Money, prices and pricing

Literature AAregistry (2019). https://aaregistry.org/story/fair-labor-standards-act-becomeslaw/. Accessed 23 February 2019. Acemoglu, D. and J.A. Robinson (2008). ‘Persistence of power, elites, and institutions’. American Economic Review 98:1 267–293. Ager, P., L.P. Boustan and K. Eriksson (2019). ‘The intergenerational effects of a large wealth shock: White southerners after the civil war’. NBER working paper no. 25700. Anderson, C and M.J. Bowman (1953). ‘The vanishing servant and the contemporary status system of the American south’. American Journal of Sociology 59:3 215–230. Antonopoulos, R., V. Esquivel, T. Masterson and A. Zacharias (2016). ‘Measuring poverty in the case of Buenos Aires: Why time deficits matter’. Levy institute working paper no. 865. Bavel, B. van (2016). The invisible hand? How market economies have emerged and declined since AD 500. Oxford: Oxford University press. Boerner, L. and B. Severgnini (2016). ‘The impact of public mechanical clocks on economic growth’. Blogpost on Voxeu, 10 October 2016. https://voxeu.org/ article/time-growth Bokan, N., A. Gerali, S. Gomes, P. Jacquinot and M. Pisani (2016). ‘EAGLE-FLI. A macroeconomic model of banking and financial interdependence in the Euro area’. European Central Bank working paper series no. 1923. Borghaerts, P. and M. Knibbe (2017). ‘A capital market without banks. Lending and borrowing in Hennaarderadeel, Friesland, 1537–1555’ in: Faruk Ölgen (ed.) Financial development, economic crises and emerging market economies 126–141. London: Routledge. Bos, F. (2003). The national accounts as a tool for analysis and policy: Past, present and future. Berkel en Rodenrijs: Eagle Statistics. Bos, F. (2013). ‘Meaning and measurement of national account statistics’. Paper provided at the Political Economy of Economic Metrics conference. Bossone, B. and M. Costa (25 June 2018). ‘Monies (old and new) through the lenses of modern accounting’. Blogpost at Voxeu, 25 June 2018. https://voxeu.org/ article/monies-old-and-new-through-lenses-modern-accounting Bowley, A.L. (1944, 1st edition 1942). Studies in the national income 1924–1938. Cambridge: Cambridge university press. Boyd Leon, C. (2016). ‘The life of American workers in 1915’. Monthly Labor Review February 1–21 U.S. https://doi.org/10.21916/mlr.2016.5. Centraal Burau voor de Statistiek (2001). Tweehonderd jaar statistiek in tijdreeksen. Voorburg/Heerlen: CBS Centraal Bureau voor de Statistiek (1981). Beroepstelling 1981. Voorburg/Heerlen, available at www.volkstellingen.nl/nl/index.html Central Statistical Office Ireland. www.cso.ie/en/releasesandpublications/er/iiped/ internationalinvestmentpositionandexternaldebtmarch2016/. Accessed 6 January 2018. Chang, H.J. (2011). 23 things they don’t tell you about capitalism. London: Penguin books. Coase, R.H. (1937). ‘The nature of the firm’. Economica 16:4 386–405. Coderre-Proulx, M., B. Campbell and I. Mandé (2016). ‘International migrant workers in the mining sector’. ILO report.

Money, prices and pricing 71 Collenteur, G.A. and R.F.J. Paping (1997). ‘De arbeidsmarkt voor inwonend boerenpersoneel in het Groningse kleigebied 1830–1920’. NEHA-Jaarboek voor de economische, bedrijfs- en techniekgeschiedenis 60 97–136. Collins, D., J. Morduch, S. Rutherford and O. Ruthven (2010). Portfolios of the poor: How the world’s poor live on $2 a day. Princeton: Princeton University press. Copeland, M. (1949). ‘Social accounting for moneyflows’. The Accounting Review 24:3 254–264. Copeland, M. (1952). A study in moneyflows in the United States. New York: National Bureau of Economic Research/Arno Press. Daily Mail (9 September 2018). www.dailymail.co.uk/news/article-3997864/ Amazombies-Seven-seconds-item-filmed-blistering-12-hours-shifts-timed-toiletbreaks-Christmas-order-does-worker-elves.html#ixzz5FST7M79U. Accessed 9 September 2018. Elsas, M.J. (1944, 1st edition 1942). ‘The definition of National Income’ in: A.L. Bowley (ed.) Studies in the national income 1924–1938 1–52. Cambridge: Cambridge university press. EUR-lex (2019). https://eur-lex.europa.eu/legal-content/GA/TXT/?uri=CELE X:52016IP0203. Accessed 23 February 2019. Eurostat/European Commission (2013). European system of accounts 2010. Brussels. Fix, B. (2018). Economics from the top down: Does hierarchy unify economic theory? Unpublished Ph. D. Thesis. Grossman, A.S. (1980). ‘Women in domestic work: Yesterday and today’. Monthly Labor Review 103:8 17–21. Haas, J. de en M. Knibbe (1993). ‘De concurrentiepositie van de Nederlandse kaasindustrie’. Economisch Statistische Berichten no. 3906 332–336. Hayek, F.A. von (1944). The road to serfdom. Chicago: University of Chicago Press. Hayek, F.A. von (1945). ‘The use of knowledge in society’. The American Economic Review 35:4 519–530. Hejeebu, S. and D. McCloskey (1999). ‘The reproving of Karl Polanyi’. Critical Review 1999:13 284–314. Hiroshi, Y., A. Hideaki, I. Hiroshi and F. Yoshi (2015). ‘Deflation/Inflation Dynamics: Analysis based on micro prices’. RIETI discussion paper series 15-E-026. International Labour Conference (2011). ‘Decent work for domestic workers Convention no. 189 and recommendation no. 201’. Geneva: International Labour Organization. International Labour Office (1921). Annual report for the year 1920. Geneva: International Labour Office. International Labor Office (2013a). ‘Statistics of work and of the labor force. Report for discussion at the Meeting of Experts in Labor Statistics on the Advancement of Employment and Unemployment Statistics (Geneva, 28 January – 1 February 2013)’. Department of Statistics MESEU/2013. Geneva: International Labour Office. International Labour Office (2013b). Global and regional statistics and the extent of legal protection. Geneva: International Labour Office. International Labour Organization (2011). ‘Global and regional estimates on domestic workers’. Domestic work Policy brief 4. International Labour organization (2017). www.ilo.org/global/topics/domesticworkers/who/lang – en/index.htm. Accessed 24 July 2017. Investopedia (2018). www.investopedia.com/terms/s/shadowpricing.asp. Accessed 20 October 2018.

72  Money, prices and pricing Jayadev, A. and J.W. Mason (2014). ‘Fisher Dynamics in US Household Debt, 1929– 2011’. American Economic Journal: Macroeconomics 6:3 214–234. Jordà, Ò., K. Knoll, D. Kuvshinov, M. Schularick and A.M. Taylor (2017). ‘The rate of return on everything, 1870–2015’. Federal Reserve Bank of San Francisco working paper series 2017–25/. Kay, J. (4 November 2014). ‘Nathaniel Mayer Rothschild. The second richest man of all time was poorer than us’. www.ft.com/content/2d1bb8ca-641211e4-bac8-00144feabdc0 Ketokivi, M. and J.T. Mahoney (2017). ‘Transaction cost economics as a theory of the firm, management, and governance’. Oxford research encyclopedias – business and management. http://oxfordre.com/business/view/10.1093/ acrefore/9780190224851.001.0001/acrefore-9780190224851-e-6 Keynes, J.M. (1940). How to pay for the war. A radical plan for the chancellor of the exchequer. London: Macmillan. King, W., F. Macaulay, W. Mitchell and O. Knauth (1921). Income in the United States: Its amount and distribution 1909–1919, volume 1: Summary. Cambridge, MA: National Bureau of Economic Research. Knibbe, M. (2007). ‘Geen lezers maar Schrijvers. Uitingen van verschriftelijking van de cultuur in Hennaarderadeel rond 1560’. Fryslan 13:4 10–13. Knibbe, M. (2015). ‘Private seigniorage, defined and estimated’. World Economics Association Newsletter 5–4 8–10. www.worldeconomicsassociation.org/files/ Issue5-4.pdf Knibbe, M. and M. Molema (2018). ‘Institutionalisation of knowledge-based growth: The case of the Dutch-Frisian dairy dector (1895–1950)’. Rural History. Economy, Society, Culture 29:2 217–235. Koning, J.P. (2019). ‘The Haïtian dollar’. Blogpost. http://jpkoning.blogspot. com/2019/01/the-haitian-dollar.html. Accessed 1 May 2019. Lee, F.S. (1999). Post-Keynesian price theory. Cambridge: Cambridge University press. Levy institute (2017). ‘How time deficits and hidden poverty undermine the sustainable development goals’. Policy note 2017/4. Liu, E., A. Mian and A. Sufi (22 January 2019). ‘Low interest rates, market power, and productivity growth’. Unpublished paper. May, V. (2011). Unprotected labor. Household workers, politics and middle-class reform in New York, 1870–1940. Chapel Hill: University of North-Carolina press. Mayall, R.N (1942). ‘The inventor of standard time’. Popular Astronomy 50 204–209. McCloskey, D. (2006). The Bourgeois virtues: Ethics for an age of commerce. Chicago: University of Chicago Press. Means, G.C. (1935). 74th congress 1st session j document No. 13. Industrial prices and their relative inflexibility. Washington: United States Printing Office. Meertens Instituut (2019). (www.meertens.knaw.nl/boedelbank/zoekvoorwerp.php ?treffers=392&offset=20&act=zoek&sort=&sortorder=ASC&max_pagina=10&sp ecial=&code=1781&plaats=Doesburg&. Accessed 12 May 2019. Mehrling, P. (2017). ‘Financialization and its discontents’. Finance and Society 2017 1–10. Mitchell, W.C. (1912). ‘The backward art of spending money’. The American Economic Review 2:2 269–281. Mitchell, W.C. (1916). ‘The role of money in economic theory’. American Economic Review 6:1 Supplement, Papers and Proceeding of the Twenty-eighth Annual Meeting of the American Economic Association 140–161.

Money, prices and pricing 73 Mitchell, W.C. (1927). ‘The processes involved in business cycles’ in: W.C. Mitchell (ed.) Business cycles: The problem and its setting 1–60. Cambridge, MA: National Bureau of Economic Research. Mitchell, W.C. (1945a). The national bureau’s first quarter-century. Cambridge, MA: National Bureau of Economic Research. Mitchell, W.C. (1945b). ‘A record of 1944 and plans for 1945’ in: W. Mitchell (ed.) The national bureau’s first quarter-century 41–72. Cambridge (Massachusetts): National Bureau of Economic Research. Munck, R. (2004). ‘Globalization, labor and the Polanyi problem’. Labor History 45:3 251–269. Murray, C.K. and B. Markley-Towler (April 2019, forthcoming). ‘A theory of return-seeking firms’. Australian economic papers. https://doi.org/10.1111/ 1467-8454.12152 OECD (2016). ‘Irish GDP up by 26.3% in 2015?’ www.oecd.org/sdd/na/IrishGDP-up-in-2015-OECD.pdf. Accessed 1 May 2019. Office of National Statistics (2019). www.ons.gov.uk/economy/nationalaccounts/ uksectoraccounts/articles/economicstatisticstransformationprogramme/ukflow offundsexperimentalbalancesheetstatistics1997to2015. Accessed 21 March 2019 Organisation International des Constructeurs Automobiles (2019). www.oica.net/ category/economic-contributions/auto-jobs/. Accessed 14 March 2019. Piketty, T. (2014). Capital in the twenty-first century. Harvard: Harvard University press. Polanyi, K. (1944). The great transformation. New York: Farrar & Rinehart. Robé, J.P. (2011). ‘The legal structure of the firm’. Accounting, Economics, and Law 1:1. www.bepress.com/ael/vol1/iss1/5 Ronsijn, Wouter (2014). Commerce and the countryside. The rural population’s involvement in the commodity market in Flanders, 1750–1910. Gent: Academia press. Santarosa, V. (2015). ‘Financing long-distance trade: The joint liability rule and bills of exchange in eighteenth-century France’. The Journal of Economic History 75:3 690–719. Skitmore, M. and H. Smyth (2007). ‘Pricing construction work: A marketing viewpoint’. Construction Management and Economics 25:6 619–630. Slobodian, Q. (2018). Globalists. The end of empire and the birth of neoliberalism. Harvard: Harvard University Press. Stapleford, T.A. (2009). The cost of living in America. Cambridge: Cambridge University Press. Stigler, G. (1946). ‘Domestic servants in the United States, 1900–1940’. NBER uncatalogued publication. www.nber.org/books/stig46-1. The Knot (2019). www.theknot.com/content/traditional-wedding-vows-from-vari ous-religions Accessed 11 March 2019. Thompson, E.P. (1967). ‘Time, work-discipline, and industrial capitalism’. Past and Present 38 56–97. Tony Chocolonely (2017). https://tonyschocolonely.com/nl/nl/onze-missie/ jaarfairslag/feiten-en-cijfers-over-2016-en-2017. Tussenbroek, G. van (2013). Also zult gijlieden dat maken. Gebruik en ontwikkeling van bouwcontracten en bestekken in de Noordelijke en Zuidelijke Nederlanden tot 1650. Meppel: Ten Brink. Union International des Chemins de Fer (2017). ‘UIC statistics – Synopsis 2017’. https://uic.org/IMG/pdf/uic-statistics-synopsis-2017.pdf. Accessed 23 February 2019.

74  Money, prices and pricing USA Department of Labor (2013). ‘Fact sheet: Application of the fair labor standards act to domestic service, final rule’. www.dol.gov/whd/regs/compliance/whdfsFi nalRule.htm Accessed 23 February 2019. Veblen, T. (1898). ‘Why is economics not an evolutionary science?’ Quarterly Journal of Economics 12 373–379. Vermoesen, Reinoud (2011). Markttoegang en commerciële netwerken van rurale huishoudens. De regio Aalst 1650–1800. Gent: Academia Press. Woersdorfer, C.S. (2017). The evolution of household technology and consumer behavior, 1800–2000. Abingdon: Routledge.

3 Money and how it’s estimated

3.1 Introduction What is money? Money – modern fiat money – is the child of transactions. All of us participate in this transaction game. We borrow from a bank, to buy a house. We use a credit card. A credit card is a token which ensures a seller – or which enables a seller to check this – that we’re trusted by a bank. These banks have to be special. They have to be Monetary Financial Institutes (MFIs) as statisticians call them. Money is only created only when we borrow from a MFI. MFI banks are different. They have a government given right to create money and the government guarantees this money. When you borrow from a pension fund or a non-MFI bank no money will be created. These financial institutions first have to gather money to be able to lend. Only the MFIs have a license to create. Well, that’s not exactly true. Every bank or financial institution or business or even household can create money. Think of frequent flier miles, or stamps. But only MFI money has the guarantee that you can use it to pay taxes due and that it can be exchanged, using a guaranteed 1:1 exchange rate, for government money, aka cash. The money creating MFI transactions are measured, by central banks, as a matter of routine, to estimate the amount of money. And the amount of debts. And financial flows. We will look at how this is done.

3.2  Money and its measurement 3.2.1  Monies, manuals and measurement What is money and how do we estimate it? Let’s restate this question: what do central banks consider to be money and how do they estimate money? Or: how do they estimate monies? The ECB distinguishes eight kinds of different monies to estimate the monthly published ‘M3-money’ aggregate plus the changes in debt of different borrowing sectors connected to changes in the amount of money. Measuring M3 money is part of a larger system of measurements aimed at mapping flows of different kinds of money and credit. This system, the flow of funds, is consistent with the national accounts, which means that the measurement of money is based upon an overarching system of macro-measurement aimed at

76  Money and how it’s estimated measuring income, production and expenditure but also at measuring money and credit. And debts. This system demarcates a sector of MFIs, which includes the central bank. The MFIs are called ‘monetary’ as they are allowed to create some of the eight kinds of money included in the M3 aggregate and as they have special lines of credit with the central bank to be able to do this which provide the central bank with some leverage over money creation. But the banks do not create money on their own. They need borrowers – us – to do so. Aside – credit card debt and mortgages get a lot of attention. But commercial credit in a broad sense might be the most important short-term business to business financial instrument when it comes to enabling the transactions which make up our economy! The Google quarterly balance sheet of the fourth quarter of 2018 showed a stock of USD$24 billion of current liabilities: deferred payments, wages due, payables and the like. This amount might not change too much from quarter to quarter (see, however, the Irish situation). But it does enable USD$24 billion of transactions and gross flows are even larger! Let’s return to MFI credit. The money for these loans is created by a keystroke, not out of thin air but out of a kind of ‘thin trust’ vested on impersonal network relations, to use Putnam’s phrase, while receivables and payables are slightly (especially in case of small companies) more based upon Putnam’s ‘thick trust’, vested upon personal relations. It will be clear that as banks are central in the case of ‘thin trust’ and money creation, the trustworthiness of banks is key. Central banks watch this process on a monthly basis and estimate the total amount of money and debt created – for one thing to be able to guard the trustworthiness of the system. An important source we will use to investigate this money measuring process will be an arcane but official European Central Bank (ECB) manual stocked with definitions on how MFIs in their obligatory monthly reporting to the ECB have to estimate money and other financial instruments based on balance sheet data of the banks and other financial flows as officially defined (ECB 2012b). The definitions are technocratic but also based on clear conceptualization (ECB 2012a). We will also widen our scope and look at the flow of funds system on which the manual is based as well as at alternative definitions of money as an aggregate. The ECB manual is, taken at face value, what economists call a ‘heterodox’ publication. ‘Heterodox’ is a phrase used to denote theories and measurements which are to some extent at odds with ‘mainstream’ economics. A publication by the Bank of England about the measurement of money (McLeay, Radia and Thomas 2014) consistent with the manual earned wide acclaim among ‘heterodox’ economist, who felt vindicated. But ‘mainstream’ models are used by central banks, too. An extension of (part of) the ECB flagship mainstream DSGE ‘EAGLE’ model, like that of Bokan et al. (2016), is at odds with the manual, as we will discuss in more detail in the paragraphs that follow. The manual puts transactions between households and non-financial companies on one side and the MFIs at the other at the center of the monetary stage. It uses balance sheets to do this which show that borrowing leads to matching changes on the liability and the asset side of balance sheets of households or non-financial companies on one side and the banks on the other sides: quadruple accounting. To paraphrase

Money and how it’s estimated 77 a common expression: ‘The law enables loans to create state backed deposits’. This is not just accounting. As we all know, borrowing ties you to the bank. The fine print of many mortgage contracts in the Netherlands stipulates that the bank can sell the house even when the borrower has always faithfully and completely payed his or her dues. The ties are binding. The borrower borrows money and pledges to pay it back, the lender is allowed to create this money and lend it at the borrower, the state takes care of enforcing contracts. The loan provided is an asset on the balance sheet of the bank, the money created is a liability. The last aspect might seem strange – but it means that when the borrower pays back the loan, the bank has a ‘debt’ to itself, can’t use this money to spend and has to destroy the money by using it to annihilate the balance sheet liability. As stated, the central bank estimates of money creation are part of a larger statistical system, the flow of funds. We can’t really understand money without understanding these statistics or, to be more precise, the world conceptualized by these statistics. Hence, the next paragraph will discuss this world of debtors and creditors.

3.2.2  The flow of funds as an overarching model The flow of funds are estimated by central banks the world over. See Research and Statistics Department of the Bank of Japan (2017) for Japan, Bê Duc and Le Breton (2009) for the ECB, Bowens (2016) for the Bank of England, Reserve Bank of India (2000) for the Reserve Bank of India and Jayadev, Narayan and Mason (2017) for a clear and concise contemporary description of these flow of funds as well as some first efforts to couple the Indian flow of funds to the Indian national accounts. Table 3.1 shows a part of the official quarterly presentation of US flow of funds, published by the FED, a number of sectors like the government and financial institutions have been left out. These accounts do not just measure ‘M3’ money like cash and deposits but also other flows financing transactions, like payables and receivables. Note the quantitative importance of trade credits (item 36). As stated, trade credits are a legal way to pay and even a cornerstone of capitalist trade and transactions. Note the importance of purchases of household durables (item 6, treated as a kind of fixed capital in these accounts). Note the importance of government investment. Note the accounting consistency: for every debit of one sector there is a credit in another sector. When I pay something, somebody else receives something. When I pay down credit card debt, the money disappears – but so does the debt. Note the difference between flows and stock: though the magnitude of the flow of commercial credits is comparable with the flow of mortgages the stock of mortgages is larger as they have a much longer shelf life. Note also the granular nature of the data. While Table 3.1 is about transactions Table 3.2 is about balance sheets. It’s called the from whom-to-whom table and is available for financial assets as in the case of these assets the debt of one ‘counterpart’ is the asset of another part. The data exists for different kinds of financial assets, including stocks, Table 3.2 only shows long-term loans (albeit not yet including pension funds).

1 Gross saving less net cap. transfers 2 Capital consumption 3 Net saving (1 less 2) 4 Gross investment (5 plus 11) 5 Capital expenditures 6 Capital expenditures; Consumer durables 7 Capital expenditures; Residential 8 Capital expenditures; Nonresidential 9 Capital expenditures; Inventory change 10 Capital expenditures; Non–produced, non–financial assets 11 Net lending (+) or net borrowing (–) 12 Net acquisition of financial assets

Description

– –

–

–

–

–

–

–

653.3

210.7

–

–12.6

1550.5

2060.1

802.2

–255.2

–1.2

56.5

2310.5

138.6

2504.4 –

–

–

–

–

–

–

– –

1899.3 717.4 –

– 1639.8 – – 1322 – 3761.8 – 2249.3

2211.4 1359.9

2616.6

2961.7

–

–

Source

Use

Use

Source

Business

–

–755

Source

267.9

–988.7

–0.7

–

291.2

–3.7

286.8 –

–

–

–

–

–

–

– –

– 283 – –1038 –701.9 –

Use

House­ Nonfinan­ Nonfinan­ Federal Federal holds cial cial Gc Gc

NPISH NPISH business

House­ holds

–

111.7

Source

State and

Use –

28.8

–264.9

14.5

–

383.9

7.3

405.6 –

–

–

–

–

–

–

– –

3159

41.7

0

56.5

3196.4

795.4

5408.3 1359.9

–

–

–

–

–

–

– –

4100.3 834.7 –

4935

Source –

1673

249.8

–

–

279.1

3.4

282.5 –

– – 532.3

Use

Use –

–

–

–

–

– –

626.4

289.1

–

–

–

–

– –

240.4 – 72.6 – – 289.1

313

Source

Use –

All Sectors

5741.1

Source

All Sectors

–

Instru­ ment

–

–

–

–

–

–

– –

5458.4

580.6

–

56.5

3475.5

798.8

5690.8 1359.9

–

–

–

–

–

–

– –

–

–580.6

–

–

–

–

50.3 –

– – 4340.7 – 493.1 – 1400.4 – – 6271.4 – –530.3

493.1

Source

Domes­ Domes­ Domes­ Domes­ Rest of Rest of tic tic tic tic the the

– 278.3 – – –166.6 – 140.7 – 5450

Use

State and

Table 3.1  An excerpt from the flow of funds, US Federal Reserve Board (use and source table)

13 Net increase in liabilities 14 Net increase in liabilities; U.S. official reserve assets 15 Net increase in liabilities; SDR certificates 16 Net increase in liabilities; Treasury currency 17 Net increase in liabilities; Foreign deposits 18 Net increase in liabilities; Interbank claims 19 Net increase in liabilities; Checkable dep. and currency 20 Net increase in liabilities; Time and savings deposits 21 Net increase in liabilities; Money market fund shares 22 Net increase in liabilities; Fed. funds and security repos 23 Net increase in liabilities; Debt securities

509.6

–

–

–

–

–

–

–

–

–

–0.3

–

–

–

–

2.2

–

154.5

122.7

–

425

684.5

–5.7

–19.6

36.2

52.4

10.5

–

–140.1

–

–

–

–

86.8

–

–

–

–

–

–

–

–

–

1057.3

–

–

–

–

–

–

–

0.1

173.6

6

–

1409.8

–

–

–

–

–

–

–

–

–

1256.5

–1.2

–1.4

1.3

9.8

–3.2

–

–

–

–

–

–

–54

–

–

–

–

–

–

–

–

–

293.8

–

–

–

–

–

–

–

–

–

3117.3

–

–

–

–

–

1423.2

449.7

16.4

0.1

–90.3

13.3

41.2

76.6

34.3

–

–

–

–

–

379.9 167.8

2.8 –53.7

189.9

528.6

321

–388.1 –376

–2.2

0.4

0

–

677.6 1442.3 1118.4

–21

160.2

487.3

335.3

–137.9

6

–

–

–

– 0.4

6

–

–

–

375.1

189.9

528.6

321.6

(Continued)

–

–

–0.6

–22.1

205.2

–0.4

–

–

–

33.1 –342

189.9

528.6

321

–353.8 –376

65.1

6

4877.8

141.6 1963.8 1963.8

30.4

–

–

–

–

65.1 –140.1

6

337.3

24 Net increase in liabilities; Debt securities; Open market paper 25 Net increase in liabilities; Debt securities; Treasury securities 26 Net increase in liabilities; Debt securities; Agency– and GSE–backed 27 Net increase in liabilities; Debt securities; Municipal securities 28 Net increase in liabilities; Debt securities; Corporate and fgn. Bonds 29 Net increase in liabilities; Loans 30 Net increase in liabilities; Loans; Depository inst. loans n.e.c. 31 Net increase in liabilities; Loans; Other loans and advances

Description

–

–0.3

–

488.2

113.1

–72.5

64.3

–18.5

–11.4

–26.9

34

–

579.6

–

–

–

–

2.3

–

–4.1

–1.8

1.8

–1.6

15.7

163.7

444.7

116.1

–3.9

–

–

–25.5

Source

Use

Use

Source

Business

–

–

–

–

–

9.7

–

104.5

Use –

–151.7

–

–151.7

–

–

–1.5

1411.2

Source

House­ Nonfinan­ Nonfinan­ Federal Federal holds cial cial Gc Gc

NPISH NPISH business

House­ holds

Table 3.1 (Continued)

–

–1.9

–1.8

–0.1

0.6

0.8

–0.7

Use

State and

–

0.9

–

0.9

–54

–

–

–

Source

State and

–25.5

Source

197.7

782.1

116.1

–58.2

–1.5

–1.7 –162

–

86.3

62.5

–76.7

111.8

582.2 1411.2

–2.2

Use

–206.2

278.7

719.5

276.2

18.4

38.7

745.1

40

Use

–16.2

14

–13.6

70.5

–

246.5

–

62.9

Source

27.9

–

27.9

–3

0.1

94.5

83.9

–7.7

Use

30

Source

All Sectors

278.7

833.7

335.7

–58.2

245.1

–180

278.7

833.7

335.7

–58.2

245.1

1411.2 1411.2

30

Use

All Sectors

–1.7 –180

67

65.2

149

–

–

–

–7.4

Source

Domes­ Domes­ Domes­ Domes­ Rest of Rest of tic tic tic tic the the

–

–

–

–

–

–

–

–

Instru­ ment

–103.3

5.5

358.6

–

–

–

–

–160.3

–

765.7

367.4

–

18.9

–

–

–

–

–

–

2.9

–

–

187

–

2.3

294

–

–

–

337.5

262.4

–

128.7

–14.3

325

–

–513.5

–

265.4

–

–53.2

15

–

–

–

–39.1

–

–

3.8

4

91.5

3.3

–

–26.5

–

–

–

–

–

0.2

24.7

–

–

–

–

–

–

–

–29

–3.8

8

–

–

22.8

–0.5

–0.9

–

–1.9

–

300.1

–

–

–

–

–

–

46.8

–

–

–

– 187

559.4 99.9

547.1

–514.9

844.1

–

–160.3

145.5

–31.1

312.6

29.1

393.5

–58.5

–

614.1

262.4

–

128.7

–14.3

–

0.2

415.4

–

–219.3

–94.1

–

–

–

–

–

0.2

0.6

12.3

–

204

2.7

29.5 291.9

–

16.8

–2

315.9

44.7

–19.7

–64.4 105.3

218

–

–11.4

51.6 –148.3

–

28.9

–

–

3.8

17.4

22.7

–111.2

90.2 –513.5 –164

87.1

1 187

548 187

548

145.5

–31.1

316.5

47.1

428.5

–64.4

145.5

–16.3

316.5

47.1

397.1

–64.4

–

28.6

–

–530.3

898.5

291.9

–

–

–

14.8

–

–

–31.4

–

–

–

–

–

–530.3

494.3 –404.1

291.9

–131.4 –131.4 –131.4

–

–

0.6

2.2

1.4

–

127.6 –167.9 –167.9

–

–

Notes: U = use of funds; S = source of funds. Domestic nonfinancial sectors (columns 9 and 10) are households and nonprofit organizations, nonfinancial business, state and local governments, and federal government.

Source: US Federal Reserve Bank, financial accounts of the USA, table Z1

–2.7 32 Net increase in liabilities; Loans; Mortgages –4.3 33 Net increase in liabilities; Loans; Consumer credit 190.5 34 Net increase in liabilities; Corporate equities –63.5 35 Net increase in liabilities; Mutual fund shares 36 Net increase in liabili­ 8.3 ties; Trade credit 29.1 37 Net increase in liabilities; Life insurance reserves 312.6 38 Net increase in liabilities; Pension entitlements – 39 Net increase in liabilities; Taxes payable 145.5 40 Net increase in liabilities; Equity in non–corp. business – 41 Net increase in liabilities; U.S. direct investment abroad – 42 Net increase in liabili­ ties; Foreign direct investment in U.S. 67.2 43 Net increase in liabilities; Miscellaneous 44 Sector discrepancies –800.1 (1 less 4)

82  Money and how it’s estimated Table 3.2 From whom to whom matrix, long-term loans (billions), 2017 end of third quarter ex. pension assets and liabilities The Netherlands    

Sectors,   Liabilities

Sectors: Assets

 

 

 

 

 

 

NonFinancial Govern­ Households Pension Rest of Total Financial companies ment funds World companies

Non-financial 39 companies Financial 280 companies 11 Government Households, 2 NPISH Pension funds   Rest of World 188 Total 520

 

 

 

 

 

 

12

2

9

0

196

258

230

47

700

0

960

2218

2 0

10 0

30 5

0 1500

15

 

68 1508

  621 865

  15 75

  3 747

  0 1500

  0 1172

  827 4879

Source: Centraal Bureau voor de Statistiek, Van wie aan wie rekeningen

All these aspects are measured by economists as a matter of routine, using these flow of funds accounts as the overarching system to classify sectors and transactions and to aggregate individual data. It is the system used by central banks to map flows of credit and spending throughout the economy and has during the last decades increasingly been integrated with the national accounts. The FOF is also used to measure ‘money’, or at least the M1, M2 and M3 definitions of money also used in Friedman and Schwartz (1963b, on this also Keynes 1930 pp. 10–11; M3 more or less equals his current money minus savings money). The integration of the national accounts with the flows of credit and lending was the explicit goal of the FOF project initiated by the NBER in 1944 together with multiple partners, among whom the US Fed. The project was assigned to Morris Copeland who (with lavish help from the US Fed) succeeded, as Tables 3.1 and 3.2 and graph 3.1 show, in a glorious way (see Copeland 1952) while Copeland became the 1957 president of the American Economic Association. Everybody who has ever investigated money growth or (sectoral) changes in debt and credit is indebted to this 1944 project. It’s the major accomplishments of macroeconomic monetary statistics. Comparing the measurements in ECB (2012b) with the diagrams in Chapter 1 of Keynes (1930) is like comparing a sundial with an smartphone. Unlike the national accounts (but see Keynes 1940) the flow of funds of for instance the USA also contains distributional accounts (albeit only since 2019, see Batty et al. 2019) showing wealth differences. The ownership of different kinds of assets and liabilities as well as net worth of different wealth groups since 1989 are brought to the fore (Graph 3.1). This data does not fit with the ‘representative

2018:Q4 "Top1"

2018:Q4 "Next9"

2018:Q4 "Next40"

2018:Q4 "Boom50"

Source: US Federal Reserve Bank, distributional financial accounts

Graph 3.1 Distributional accounts, USA, top 1%, next 9%, next 40% and bottom 50%, assets and liabilities (selected assets and liabilities as well as totals), USD millions

0

5000000

10000000

15000000

20000000

25000000

30000000

35000000

40000000

45000000

84  Money and how it’s estimated consumer’ idea of many of the DSGE models but comparable information is used by economists like Piketty (Piketty 2014).

3.2.3 The monthly monetary press release of the ECB and the macroeconomic formula of everything Central banks estimate monetary aggregates and present these estimates in a high frequency, multi-dimensional and detailed way, showing money creating debts per kind of debt as well as per main sector of the economy. These estimates are connected to the national accounts (which use the same economic sectors) using what can veritably be called ‘the macroeconomic formula of everything’. To show this, we will discuss the monthly monetary press release of the ECB. In this release, it distinguishes three kinds of money (M1, M2 and M3) which consist of eight subcategories of money as spelled out in Table 3.3. About the subcategories it states that they are: Monetary liabilities of MFIs and central government (post office, treasury) vis-à-vis non-MFI euro area residents excluding central government. In August 2012, the ECB has amended its statistical measurement of broad money to adjust for repurchase agreement (repo) transactions with central counterparties. The quote shows that the ECB operationalization of money changes when the facts change. And a ‘repurchase agreement’ is a tradeable financial pawn agreement which pawns financial assets instead of consumer durables. This ECB operationalization of ‘money in circulation’ leaves us in the dark about what happened to ‘Deposits with an agreed maturity of more than 2 years’ (longer term saving deposits or more or less Keynes’ ‘savings money’). Loans create deposits, but not all these deposits are included in the M3 monetary aggregate. These long-term deposits are not considered to be part of ‘money in

Table 3.3 The ECB definition of the eurozone monetary aggregates M1, M2 and M3 (including deposits which are left out of these aggregates)  

M1

M2

M3

Not ‘money’

Currency in circulation (cash, notes) Overnight deposits Deposits with an agreed maturity of up to 2 years Deposits redeemable at notice of up to 3 months Repurchase agreements Money market fund shares/units Debt securities with a maturity of up to 2 years

X X          

X X X        

X X X X X X X

             

Longer term deposits and debt securities

 

 

 

X

Source: www.ecb.europa.eu/mopo/strategy/monan/html/index.en.html

Money and how it’s estimated 85 circulation’ as, conceptually, the distinguishing aspect of money is according to the ECB the ability to use it as a means of exchange at short notice and against little cost (ECB 2012a), which is generally not true for longer term save deposits. These long-term deposits are however created in the same way as ‘M3’ deposits: ‘loans create deposits and some of these may be channeled to long-term saving accounts’. Anyway – they are on the balance sheet of MFIs as well as households and non-financial companies. It is deposit money but it is supposedly not used for current transactions and hence not included in ‘M3’ money. The flow of funds, the basic system of measurement, uses ‘quadruple accounting’. Quadruple accounting means that a new loan is a liability on the balance sheet of the lender as well as an asset on the balance sheet of the borrower while the amounts on both balance sheets have to match. The money borrowed and shifted to the bank account of the borrower, which is an asset of the borrower, has to match with the amount of money created by the bank (a liability of the bank). And the liability of the borrower, or the debt to the bank, has to fit with the new asset on the balance sheet on the bank (the promise to pay down the debt by the borrower). Even when the borrower shifts the money to a long-term savings account. The perceptive reader will have noticed that depositors can’t take their deposits out of the financial system. They can shift deposits between different bank accounts. But they stay inside the banking system. They can buy something abroad via electronic transfer to a bank abroad which means that the domestic amount of deposit money will shrink. But it stays in the (international) banking system. They can use an ATM to buy banknotes, paying these with deposit money. But this deposit money will be transferred to an account owned by the bank selling the notes and might end up at the central bank when this bank buys the notes from the central bank. But the central bank is part of the banking system, too. Notice that because of the definition and operationalization of M1, M2 and M3 deposits on bank accounts owned by MFIs or the central bank are not part of M1, M2 or M3. Money created by one MFI can be used to pay down a debt at this or another MFI which is guaranteed by central bank guarantees and government rules. Doing this means that the money disappears, ‘into thin air’. But it can’t be taken out of the banking system. When, as happened after the ECB lowered interest rates after 2008, there is a shift money from accounts with an agreed maturity of more than 2 years to for instance overnight accounts, this shows up as an increase of the amount of M1, M2 and M3 money while the shifts from short-term savings accounts to overnight accounts increased M1 at the cost of M3. But the total amount of deposits did not change. The monetary press release of the ECB connects money creation to lending on a monthly basis. Money creation is in an explicit and granular fashion tied to changes in credit as well as money flows. This approach is based upon a long standing Bundesbank tradition, as the Bundesbank annual accounts show. To be able to do this well the ECB uses on a monthly basis statistical data about gross and net lending by MFIs using the flow of funds / national accounts sector classifications. The results are statistics about, among many other things, the amount of M3 money micro-founded in the statistical sense. The graph on Irish

86  Money and how it’s estimated lending shown in the first chapter is part of such statistics. Balance sheets of banks however do not just change because of money creating lending, which means that other factors have to be considered, too, when the raw data are aggregated. These factors are operationalized as (ECB 2012b): D M3 ≡ (Current and capital account balance) – (External financial transactions of resident non-MFIs) + (D credit to euro area residents) – (D longer-term financial liabilities) + (D other counterparts (net)) In this formula D M3 means the change in M3 money. The ‘current and capital account balance’ and ‘External financial transactions’ phrases mean that if money is shifted from non-eurozone countries to accounts from eurozone residents at eurozone banks this shows up as an increase in M3 and vice versa. ‘Longer term financial liabilities’ are the long-term deposits of Table 3.2 not included in M3. As the current account is the Exports minus Imports or (Ex – Im) part of the macroeconomic ‘Y = C + I + G + (Ex – Im)’ the formula connects the estimate of deposit money to the national accounts and the flow of funds, which makes it a ‘macroeconomic formula of everything’ as the emphasis on change means that stocks have to be known, too. We do know lending and borrowing and value added and profits and wages and production per subsector of the economy and it all adds up to nominal GDP. It also shows that the amount of money is a consequence of net ‘MFI’ credit to households, non-financial companies and pension funds and other non-bank financial institutions as well as by the net flow of money from sight to long-term deposits in (in this case) the eurozone and net international flows. ‘Other counterparts’ is a residual post.1 As the formula connects credit to the national accounts it also means that accounting identities like ‘income is production’ have to be adapted to take account for net increases or decreases of credit. As the modern national accounts indeed do. Textbooks have to follow. The monthly press release graph (based on comparable graphs of the Bundesbank annual reports of the 1980s) is based on the M3 formula. Graph 3.2 shows the contributions of several developments to M3 money change for a longer period. In combination with the tables of the release it enables an analysis of which sector and credit flows were instrumental in making the monetary aggregate change. Translating this to theory: the monthly press release is consistent with heterodox economics. Graph 3.2 shows different kinds of credit, different borrowers and different financial instruments. It also shows that domestic deposit creation can be counteracted by money flowing over the borders of a currency union and to long-term savings deposits or vice versa. On an economic level, the increase of lending to unsustainable levels before 2008 sticks out, as does the fact that only part of this lending showed up as M3 deposits mainly because of inflows into long-term savings deposits. Money was created to buy houses, the sellers of these houses basically saved the money. We will see that the ‘Divisia’ definition of money

Money and how it’s estimated 87 20

15

10

5

0

–5 Sept 1998 Sept 2000 Sept 2002 Sept 2004 Sept 2006 Sept 2008 Sept 2010 Sept 2012 Sept 2014 Sept 2016 Sept 2018 D plus 'remaining counterparts' = M3 C minus deposits transferred 'abroad' = D B minus shi of deposits into long term savings accounts = C A+Credit to general government = B A: Mortgages, loans and consumer credit to households and non-financial companies

Graph 3.2  From MFI credit to M3 money Source: ECB database

shows and even lower growth of the amount of money before the financial crisis, which means that M3 money is an imperfect gauge of the state of the monetary economy, a statement underscored by the flow from long-term savings into M3 money after 2008, which was probably incentivized by low interest rates. It just didn’t pay anymore to put your savings deposits on a long-term account anymore. Graph 3.3 shows Irish households as actors creating deposit money which shows the granular nature of the data: these data are available for other eurozone countries, too. But there is more to this. As can be seen, mortgage debt in Ireland increased head over heels, before 2008. As in the Netherlands, this was enabled by securitization of mortgage related assets on the balance sheets of banks and shifting these to special purpose vehicles. Graph 3.2 also shows that financial institutions provide quite a lot of credit to the government. To understand this, one has to realize that the MFI sector includes the ECB. Whenever the ECB buys bonds from MFI banks, the money the banks receive is not added to the stock of M3 money as deposit money owned by banks (‘in the electronic vaults of the banks’) is not considered to be money in circulation. However, pension funds and insurance companies also sell existing government bonds to the ECB. And money owned by non-MFI financial institutions is, by money accounting definition, added to the stock of money. The bonds purchased by the ECB move from non-MFI banks to the MFI sector which, hence, provides more ‘credit to the government’ which should be understood as owning more government liabilities. And part of the money created ends up in the vaults of pension funds

88  Money and how it’s estimated 250000

200000

150000

100000

50000

0

–50000 2002

2004

2006

2008

2010 Stock

2012

2014

2016

2018

Transac ons

Graph 3.3  Stock and flow of ‘long-term loans’ (mainly mortgages) of Irish households Source: Central Bank of Ireland financial accounts table 9.a and 9.b

and insurance companies, which, when looking at the balance sheet of the large Dutch Algemeen Burgerlijk Pensioenfonds (ABP), quickly use this money to purchase other financial assets but as stated earlier, the aggregate data are consistent with such micro-data. Clearly, the ECB is on the right track when publishing granular quadruple accounting estimates of the changes in the amount of money, even when the interpretation of these data requires a discussion of relevant circumstances. Economists, however, tended to forget about all this and to focus on M3 money, without the proper quadruple accounting and flow of fund context even when the Bundesbank and, later, the ECB continued to publish flow of fund related money data. The resulting disappointments subsequently led them to neglect money. Nowadays, economists are occupied with reintroducing money into theory. The next paragraphs will be about such developments.

3.2.4 Single accounting concepts of the account of money (1): Friedman and Schwartz Money, nowadays including credit, was and is the target of macroeconomic central bank policies. In the eurozone, this tallies with the two analytical pillars of monetary policy. The first of these is an analysis of economic developments, which are largely outside of the policy grasp of the bank. The other pillar is an analysis of monetary developments, which are, to an extent, within the grasp of the bank.

Money and how it’s estimated 89 As the ECB states: ‘Monetary analysis consists of a detailed analysis of monetary and credit developments with a view to assessing their implications for future inflation and economic growth . . . particularly those of the broad aggregate M3, based on information stemming from their components and counterparts’ (European Central Bank 2019). This has been different in the past. For some time, central banks targeted the growth of a single accounting concept of the stock of money instead of the quadruple accounting concept from the quote. This single accounting approach eventually made way for policies aimed at targeting inflation instead of money, let alone money and credit. But for some time, money without credit took central stage. Nowadays, central banks still state that they pursue inflation targeting policies. These are based on the assumption that if a central bank only makes credible its promise to prevent high inflation, i.e. to promises to tank the economy when inflation is considered to be too high by increasing interest rates fast and furious until this rattles the economy, rational actors would believe this. And it would not be necessary to do this as rational actors would not increase prices consistent with a high inflation future, the stock of money will passively adapt to this (Knibbe 2013). But as after 2008 stimulating the economy proved, for a central bank, more complicated than tanking it, a shift away from ‘credibility’ to money and credit based policies is visible. Why such changes? We have to ask why the policy focusing on money only made its way for inflation targeting. To understand this, we have to know about its origins and why this policy had fatal flaws. One paragraph is not enough to give a precise answer but we can investigate the relation between the concept, definitions and measurement of the ‘single accounting’ measurement of money and its interpretation by economists and if these apply, too, to quadruple accounting concepts. An important part of this story is related to the work of Milton Friedman, Anna Schwartz and (again) Wesley Mitchell. In 1949 Arthur Burns, who became the spokesman of the US National Bureau of Economic Research after Wesley Mitchell died in 1948, wrote in the 1948 annual account: Work on another monograph, dealing with the cyclical behavior of the money supply, its rate of turnover, and the condition of banks in different parts of the country as well as in the aggregate, will get actively under way this year. The study will go back to the Civil War, but will give special attention to developments in the sphere of money and banking since 1914 when the Federal Reserve System was instituted. Milton Friedman, Associate Professor of Economics at the University of Chicago, is in charge of the study. (Burns 1949) This work culminated in Friedman and Schwartz (1963b) which, using a Mitchellian business cycle methodology (Rockoff 2006), put time series analysis of monthly monetary series in the center of the economic stage (Burns and Mitchell 1946). Friedman and Schwartz (1963b) is a rare example of synthesis of a neoclassical inspired study (Friedman 1956) with a strong empirical, historical and institutional bend which conceptualizes, defines, operationalizes, measures

90  Money and how it’s estimated as well as analyzes an economic variable in a macroeconomic setting: money. It is a laudable example of the integration of theoretical and empirical concepts. For a time the study, which ascribed a central place to the role of a kind of M2 money aggregate in some economic cycles, carried the day. As a result the stock of money became a prime target of central bank policies. At the same time quadruple accounting flow of funds and credit analysis became ever more important as a statistical system and also started to influence monetary policy. But official targets increasingly became focused on a ‘single accounting’ aggregate of money: M2 or M3 or sometimes even M4. This was based upon the idea that as there presumably was a stable relation between the development of business cycles and the amount of money stabilizing the growth of the amount of money would stabilize the economy (Belongia and Ireland 2016). Remarkably, Friedman and Schwartz (1963a, 1963b) do not contain references to the rapidly developing flow of funds, even when the godfather of this system, Morris Copeland, had been a colleague of Friedman at the NBER for quite some time, which meant that the stock of flow of credit disappeared from the official radar screen, including the inherently unstable nature of credit and hence the stock of money. Even before Friedman and Schwartz (1963b) was published this was already mentioned by Minsky (1963). The question of credit had disappeared from the screen – even when more and more data on credit, integrated in the overarching frame of the flow of funds, became available. Friedman and Schwarz were acutely aware of the lack of an overarching theoretical framework, like the FOF, for their single accounting data on money used in their attempt at business cycle analysis. They at least suggested that (instead of using a kind of ‘periodic table’ as set out by the FOF) a choice theoretical framework should be used to analyze the data on minor business cycles. Later these would be called microfounded models. They however also identified larger downturns of the business cycle. According to them these were characterized by a wholly different behavior of the single accounting monetary aggregates. And hence, by accounting necessity albeit not mentioned by Friedman and Schwartz, by a different behavior of the credit aggregates (Minsky 1991, footnote 12). These might be prevented when the central bank guaranteed a stable development of the stock of money (and hence, in a quadruple accounting perspective but not mentioned by Friedman and Schwartz, of credit). Nowadays, such downswings are called financial cycles, who follow another path than ordinary business cycles (Borio 2012). The already mentioned ECB monetary press releases show that this is entirely possible to use consistent and coherent FOF based data which show an integrated and theoretically sound view of credit as well as money. Such an approach was the road not chosen by Friedman and Schwartz and economists in their wake. In hindsight, this set back academic monetary macro by decades. Belongia and Ireland, following the wake of Friedman and Schwartz and focusing on single accounting estimates of money, have little more to say than: ‘Taken together, all of these empirical results suggest that additional research might be directed fruitfully to theoretical analysis of how the private banking system acts alongside the Federal Reserve to create liquid assets’ (Belongia and Ireland 2016, p. 1362). Such

Money and how it’s estimated 91 analyses have, however, already been carried out (about these and their history Minsky 1991). Friedman and Schwartz (1963b) is brilliant. But as it used a single accounting concept of money already outdated when it was written. Which put back monetary analysis for decades to come. Quite soon, however, neoclassical ‘micro-founded’ studies started to erode the efforts of Friedman and Schwartz and stated that time series on the ‘monetary aggregates’ did not contain serious information (only one example: Estrella and Mishkin 1996). In doing this, they fatefully ignored the distinction made by Friedman and Schwartz between small and major economic cycles which according to them were characterized by a quite different behavior of the monetary aggregates. This idea that money (and hence in a quadruple accounting perspective: credit) didn’t matter was based on a small number of relatively benign downturns only. So, even when these economists discarded the attention paid to larger economic downturns and money they did follow Friedman and Schwartz by ignoring debt and credit and focusing on a single side of an aggregate monetary balance sheet without sectoral divisions only which, predictably, was not very interesting. Increasingly, estimated money and the monetary sector was left out of the neoclassical macro-models (Belongia and Ireland 2016, p. 1236). It is possible to state that this was related to the rise of DSGE models: what’s the use of money when there is only one ‘representative’ consumer? The last thing a Robinson Crusoe needs is money and banks. This approach to money made Goodhart, stressing the importance of debt, desperately exclaim: ‘Whatever became of the monetary aggregates’ (Goodhart 2007), while Minsky stated (in 1964): ‘Although the evidence for the monetary explanation of mild depression cycles is admitted to be tenuous by Friedman and Schwartz the evidence that money is a significant part of the mechanism generating a deep depression is strong.’ (Minsky 1964). After 2008, this changed again. At this moment, economists try to introduce institutional detail as well as better data into the models. But even before 2008 neoclassical economists not satisfied with the a-monetary models tried to remedy this situation by looking at other definitions and operationalizations of ‘money in circulation’, while another approach was to look at credit, once again. Both approaches will be discussed in the next section.

3.2.5 Single accounting concepts of the amount of money (2): Divisia indices Divisia indices are a welcome neoclassical effort to change the operationalization of money in circulation (Darvas 2014; Belongia, Smith and Ireland 2017). They are a clever proposal to enhance the informational content of the noncredit monetary aggregates by constructing a (neoclassical) theory consistent weighted aggregate of the amounts of different kinds of money (cash, different kinds of deposits etc.). Ideally, the weights are, using interest rates, based upon the ‘medium of exchange’ content of a particular kind of money. Ideally, money which circulates a lot, like €20 notes, gets a higher weight than other kinds of money, like €500 notes. This is not an entirely new idea. John Stuart Mill (1848)

92  Money and how it’s estimated already noted that different kinds of money had different origins and uses and, hence, not the same kind of influence on the price level and the economy. Irving Fisher (1920) basically made the same point with his replacement of the ‘vulgar’ equation of exchange {MV = PT} by {MV + M’V’ = PT}, M and M’ denoting different kinds of money with different kinds of velocity. From a more recent and practical angle: the ECB will slowly phase out the €500 note as this is mainly used for hoarding and not for exchange. Divisia indices of a money-aggregate can be understood as a modern version of such ideas which takes the differences in velocity between different kinds of money into account. It tries to construct an estimate of money in which Fisher’s M and M’ get different weights. It basically tries to measure the amount of ‘exchange’ money, a concept which is more consistent with neoclassical micro than the well-known ‘unweighted’ M1, M2 and M3 aggregates of money used by Friedman and Schwartz and defined, measured and presented by the ECB and many other central banks. This idea gets even more traction when we realize that M1 and M2 are actually not unweighted at all. The construction of these aggregates are based on an implicit weight of 1 for all kinds of money – without anyone arguing that this is the right weight! Though neoclassical, the idea behind Divisia money is however still at odds with neoclassical macro as the single ‘representative consumer’ which inhabits the models doesn’t need money at all, not even to hoard.2 As a Mitchellian business cycle indicator a Divisia index is in theory preferable to the unweighted money aggregates used for this purpose by for instance Friedman and Schwartz (1963a, 1963b). But the proof of pudding is in the eating. We can ask the question how the ECB 1:1 weighted monetary press release aggregates compare with its reweighted version, the Divisia index. Such an index is published for the eurozone by Bruegel (data drawn from the database published by Bruegel, for a description Darvas (2014). There are clear differences (graph 3.4). Do these diverging developments of the Divisia index teach us more about cyclical developments than the M3 data? According to Darvas (2014) the answer is yes: ‘we find sensible and statistically significant responses to Divisia money shocks, while the responses to simple-sum measures of money and interest rates are not statistically significant, and sometimes even the point estimates are not sensible’. But Darvas is framing the discussion by excluding many of the aggregates included in the standard ECB press release. When we follow the lead of the ECB and take a joint look at money and credit in a quadruple accounting perspective another picture emerges. The recent divergence between M3 and the Divisia index can be explained by shifts in money holdings – there has been quite a shift of money stacked away in long-term savings accounts (not included in M3) into short-term savings accounts ( 0 and no labor income.’ Unemployment is leisure and people choose unemployment because it’s fun. As we will see, this difference of conceptual perspective matter a lot for our views of economic history as we know it.

4.3.3 The ILO ‘periodic table’ of monetary and non-monetary work, unemployment and leisure The ILO unemployment definition fits into a framework which encompasses ‘gainful employment’ as well as unpaid work (and remember, from the preceding paragraph, that there is no unpaid work in the world of Ljunqvist and Sargent 2016). Diagram 4.1 gives the idea, note that this labor statistics diagram uses the production boundary of the national accounts but also looks at the world at the other side of the boundary. Note, again, that it’s based upon activities of individual people. The statistics of work and unemployment are micro-founded in the statistical sense (albeit not in the new classical sense). Note that two of the three categories of ‘broad’ unemployed are not included in the labor force while ‘normal’ unemployed as well as ‘part time unemployed for economic reasons’ are included. Note that the unemployed might do unpaid household work.

Total population Engaged in productive activities (may engage in non-productive activities) Persons who work

Engaged in productive activities within SNA

(may also engage in productive activities beyond SNA)

In employment to generate income i.e. paid employment and market self-employment (may also engage in own-production, trainee and/or volunteer work

Persons in employment

exclusively in employment

Also did ownproduction, trainee, or volunteer work

Exclusively in own-production, trainee and/or volunteer work*

Seeking and available for employment Persons in unemployment

Exclusively unemployed

Also did ownproduction, trainee, or volunteer work

Exclusively in nonproductive activities

Engaged exclusively in productive activities beyond SNA, within the general production boundary

Not seeking or not available for employment Persons outside the labourforce

Exclusively outside the labourforce

Also did ownproduction, trainee, or volunteer work

*New treatment based on proposed revied scope of employment.

Diagram 4.1  Proposed classifications of people in the labor force framework Source: ILO (2013)

Labor and unemployment 117 Technical addendum: SNA: System of national accounts. Productive activities beyond SNA are mainly household work. ‘Own production’ relates to production of goods and services which are generally within the SNA production boundary but sometimes outside of it, like a vegetable garden or building your own house (important in quite some countries!). Starting from this concept of work unemployment is defined by the ILO as labor underutilization: Labour underutilization refers to mismatches between labour supply and demand, which translate into an unmet need for employment among the population. Measures of labour underutilization include, but may not be restricted to: (a) time-related underemployment, when the working time of persons in employment is insufficient in relation to alternative employment situations in which they are willing and available to engage; (b) unemployment, reflecting an active job search by persons not in employment who are available for this form of work; (c) potential labour force, referring to persons not in employment who express an interest in this form of work but for whom existing conditions limit their active job search and/or their availability. (ILO 2013, paragraph 40) Many aspects of this definition require operationalization. People not having a job and trying to emigrate are for instance classified as ‘unemployed’ (Diagram 4.2.) The definition is also quite broad, which is on purpose as the economic and institutional situation of the member countries can show large differences. It actually covers the biblical workers in the vineyard (Matthew 20 1:16), which would be classified as part time workers for economic reasons with their search activities classified under point (v), while somebody without a job changing his LinkedIn profile is also classified as ‘searching’(point (vii)).

(i) (ii) (iii) (iv) (v)

arranging for financial resources, applying for permits, licenses; looking for land, premises, machinery, supplies, farming inputs; seeking the assistance of friends, relatives or other types of intermediaries; registering with or contacting public or private employment services; applying to employers directly, checking at worksites, farms, factory gates, markets or other assembly places; (vi) placing or answering newspaper or online job advertisements; (vii) placing or updating résumés on professional or social networking sites online. Diagram 4.2  Examples of search behavior Source: International Labor Office (2013), paragraph 47

118  Labor and unemployment

4.3.4  The difference between people and jobs An important distinction is between people and jobs. Based upon this distinction, statisticians have, next to statistics about employment and unemployment, crafted job creation/destruction statistics as well as labor flow statistics. These statistics empirically show the intensely dynamic nature of the labor market. Job creation statistics look at the number of jobs created and destroyed every period. labor flow statistics looks at flows between the main categories. In the Netherlands, about 15 percent of all jobs lasting longer than one year are destroyed and replaced, every year (which includes a teacher who retires and is replaced by a younger one). For the UK, Table 4.1 shows inflow and outflow into and out of unemployment measured in people. The decline of UK unemployment after November 2011 (figure 4.2) was not caused by a higher flow from unemployment to employment. but by lower inflows into unemployment, while the increase of UK unemployment after August 2008 was not caused by a lower flow from unemployment to employment or inactivity. To the contrary. The number of unemployed getting a job increased. The driver of increasing unemployment after 2008 were larger inflows into unemployment – from activity as well as from inactivity. This development is consistent with the idea that firms keep hiring and prefer recently fired people to new entrants. The table does not show all flows. There are for instance also flows from employment to inactivity (mainly pensions, a larger flow than the flow into unemployment) and flows from inactivity to employment. Interestingly, job-to-job flows also increased considerably after 2015, which is consistent with the lower inflow into unemployment. But here, it’s about the idea of these flows, which show amazing dynamics on the labor market. Despite these dynamics, increases and decreases in total unemployment are limited. It’s like a lake with sizeable inflows and outflows but a level which only changes slowly. The quarterly increase in unemployment in the 13 quarters between august 2008 and November 2011 was only 10 percent of the flows. The most remarkable aspect of Table 4.1 might well be that the flow of unemployment to employment is the most stable one, which is consistent with the idea that even in a depressed economy, companies which do not falter still need labor and destroy and create jobs all the time, for instance to replace people going into retirement or to occupy space left by bankruptcies of other companies (Mortensen and Pissarides 1994). Comparing the data for the UK with those of Spain (which in a relative sense has much higher flows than the UK) or France (which has roughly comparable rates as the UK) shows that the size of the flows is not related to net increases of the labor force or declines of unemployment: more dynamism, as in Spain, does not seem to lead to more jobs on a net basis (Bentolila, Ignacio-Pérez and Jansen 2017). Also, it turns out that even when compensating for age, education, gender or the sector of the economy it is long-term unemployed who have the largest problem to re-enter the labor market (Bentolila, Ignacio-Pérez and Jansen 2017, p. 24), even when an depression also has quite an influence on the change to stay unemployed at all spells. Job

Source: ONS, table X02

340 436 341

444 509 514

287 384 347

100 134 121

Nov 2001-May 2008 Aug 2008 – Nov 2011 Feb 2012 – Feb 2018

100 114 116  

100 128 100 Persons * 1,000

inactivity

Nov 2001-May 2008 Aug 2008 – Nov 2011 Feb 2012 – Feb 2018  

employment

unemployment

Unemployment to inactivity

Index, Oct-Dec 2001 – April-June 2008 = 100   

Unemployment to employment

Employment to unemployment

 

   

377 507 454

100 134 120  

unemployment

Inactivity to unemployment

Average Quarterly Labor Market Flows: A Partial View (UK, October to December 2001 – January to March 2018)

Table 4.1  Flows of labor, UK, 2001–2018

–14  51 –66

       

unemployment

Net change in unemployment

120  Labor and unemployment turnover might be compared to ‘musical chairs’. Increased skills might increase your change of getting a chair but this will decrease the chances of others. And higher turnover leads to more people getting jobs but also to more people losing jobs. Faster music and more frequent stops do not increase the number of chairs, while people who just rose from their chair have much better chances than people who have been standing for some time. One person can have multiple jobs or work multiple jobs within a year. ‘Jobs’ are hence not the same thing as ‘employment’. Data on jobs created and destroyed show the same pattern as employment but have the advantage of being somewhat simpler as they are not about three categories (employed/unemployed/inactive) but only about jobs. The first thing which strikes the eye when looking at these data for the USA is the impressing amount of yearly job creation and destruction. But as in the case of employment and unemployment, net changes are less spectacular. As in the case of employment job gains (which can be compared with flows from inactivity or unemployment to employment) are (cutting out seasonal patterns) more stable than data on job losses. Again, relatively short bouts of high job destruction (and relatively low gains) lead to a long period of high unemployment – a development which, when it comes to unemployment, is much more important than the secular decrease of the levels of gains and losses. On top of the variables discussed in this chapter, statisticians also measure, among other aspects, duration of unemployment as well as the age and sex of the unemployed and the number of hours worked. In the next section, this will turn out to be important.

9500 9000 8500 8000 7500 7000 6500 6000 5500 5000

1992

1994

1996

1998

2000

2002

2004

Gross Job Gains

2006

2008

2010

2012

2014

2016

Gross Job Losses

Graph 4.4  Gross job gains and losses, USA, quarterly data, 2Q1992–2Q2018. Source: Bureau of Labor Statistics

Labor and unemployment 121

4.4 Neoclassical ideas about labor and the working of the labor market 4.4.1 Concepts and definitions: leisure and the difference between people and hours In the previous section, we’ve seen how statisticians define and measure labor. This section will investigate the labor concepts of the neoclassical DSGE model. In a sense, the neoclassical concept is clear and simple: work sucks. People do not like to work. Even people searching for a job derive utility from their workless situation and experience it as leisure. This is not a fringe idea. A whole number of winners of the SRPiESiMoAN are explicit about this (Lucas 1976; Mortensen and Pissarides 1994; Prescott 2016; Ljunqvist and Sargent 2008). So, people have to be paid to work. If they are paid more or if their job becomes more productive, they will work more. If they are paid less or if their activity becomes less productive, they will work less. When Robinson Crusoe discovers a faster way to harvest coconuts, he will spend more time doing it. When some kind of bug makes the coconuts grow less large he will spend less time harvesting. The representative consumer/producer might be mistaken about productivity and think that, in the future, it will be higher than it actually will be. This will result in something equal to unemployment: he will work not ‘enough’ as he thinks that in the future there will be more food than there actually will be. And vice versa. The concept can be exemplified with a formula from Bokan et al. (2016). It shows the behavior of the ‘representative household’. The formula looks complicated. But it only has three basic variables, two positive ones (C, consumption of non-durable consumption goods and services except government goods and services), H (consumption of housing services) and a negative one (N, ‘labor services provided’). Here, we will focus on N: what is it? Utility. The representative patient household, labelled ‘saver’, gets utility from consumption of the nondurable composite good, CI,t (subject to external habit formation) and from housing services HI,t and gets disutility from working NI,t ∞ 1 −κ Et  ∑ (β I )k   k = 0 1 − σ

 C I ,t + k − κ C I ,t + k -1    1−κ

1− σ

+ ιI ln H I ,t + k −

 1 N I1,+tζ+ k   , 1+ζ     (12)

where 0 < βI < 1 is the discount factor, 0 ≤ κ ≤ 1 measures the degree of external habit formation in consumption, σ > 0 denotes the inverse of the intertemporal elasticity of substitution. ιI > 0 is a parameter for utility from housing services and ζ > 0 is the inverse of the elasticity of work effort with respect to the real wage (Frisch elasticity). Bokan et al. (2016) p. 10. Looking at the labor part of this formula (the part at the right with ‘N 1+ ζ; I,t+k ’) we see a number of complicating conjoints: the phrase (1/(1-ζ)), the variable N

122  Labor and unemployment as well as a variable ‘I’. What do these variables mean? The subscripts of N mean that it’s about labor provided by a certain kind of household (I) at a certain time (t + k) and a certain household specific relation between real wages and labor sold (when wages are high, more labor will be sold); neither time nor the kind of labor (hours, persons?) are specified. More precisely: • First: ‘N’. ‘N’ stands for ‘labor’. Bokan et al. are not clear about the definition of this variable. It is defined as ‘labor services, these services are not defined as hours or persons or even jobs. The paper does not contain the word ‘hours’ once while ‘employment’ surfaces only one time. • The minus in front of N means: work sucks. Labor services, as the model calls them, are a negative, measured in the same dimension as utility. The larger N is, the larger the subtraction. As a smaller N leads to a smaller utility subtraction the model implicitly assumes that less labor is better. Unemployment basically does not exist, remember that the model is about a representative consumer who works a little more or a little less but is never wholly unemployed. As we know, ‘measured unemployment’ does exist. It is measured in persons. We might try to include it in the models – but the problem is that the model does not contain a labor variable measured in people. It’s measured in utility. Unemployment measurements in people seem at odds with the DSGE models at a deep level. De Vroey’s book Involuntary Unemployment (De Vroey 2004) investigates old and new classical ideas about this concept at great length and comes to the conclusion that involuntary unemployment is not consistent with classical economics. Tellingly, the book does not spend one title or iota on the concept and definition of measured unemployment and does not discuss any of the clear differences between ‘labor services’ as used in the models and labor and unemployment as measured by statisticians. Lawrence Christiano, a DSGE modeler himself, tackled the problem head on. In a comment on a DSGE model written by among others the head economist of the ECB which, 35 years after Lucas (1976), bluntly ignored measured unemployment and equated the concept of unemployment with the concept of leisure, he stated: ‘First, I am skeptical that the people designated as unemployed in the model satisfy the official United States definition of unemployment. Second, the model implies that the unemployed are happier than the employed’ (Christiano 2011). Christiano goes on to show in a detailed way that what’s called unemployment in such models is not the same as the definition and operationalization of unemployment by economic statisticians. We can investigate if the unemployed measured in people are indeed happier than the employed. Eurostat disagrees: ‘Nearly half (48.7%) of unemployed persons aged 16–64 in the European Union (EU) were at risk of poverty after social transfers in 2016. In other words, the risk of monetary poverty was five times greater than for those in employment (9.6%). Over the past 10 years, the proportion of unemployed persons at risk of poverty has risen continually, from 41.5% in 2006 to 48.7% in 2016’ (Eurostat 2018). This means that the unemployed are poorer than the employed, a fact ignored

Labor and unemployment 123 in the models. Somewhat comparable statements are made by Lindner, Mitchell and Nichols (2013), about long-term unemployment: ‘Being out of work for six months or more is associated with lower well-being among the long term unemployed, their families, and their communities. Each week out of work means more lost income. The long-term unemployed also tend to earn less once they find new jobs. They tend to be in poorer health and have children with worse academic performance than similar workers who avoided unemployment. Communities with a higher share of long-term unemployed workers also tend to have higher rates of crime and violence’. Clearly the unemployed are not just economically but also socially and psychologically worse off than their gainfully employed counterparts. These are just two examples of a momentous literature about such issues. The juggernaut take away of these papers and articles: ‘Work doesn’t suck. Unemployment sucks – it’s not leisure, it’s not pleasure but it’s pain. Great and lasting pain’. As the concept of unemployment is part and parcel of the concept of gainful employment of the statisticians, not recognizing unemployment as a negative is a fundamental difference between the statistics and theory. As stated earlier, modern attempts to introduce unemployment into the models had to get rid of ‘–N’ to be able to do this (Haan, Rendahl and Riegler). 2017 Which as the consumption formula is the heart of the DSGE models is a paradigm change when it comes to the models (such models hence won’t get specific attention in this book, bar the epilogue). • The term (1/(1-z)) between the minus and N is called the Frisch elasticity, which indicates how many more ‘labor services’ people are going to provide when wages increase. A higher wage is supposed to compensate for more disutility of working. There is a problem with this variable. We’ve already seen that labor supply in Greece and Spain didn’t decline despite brutal declines of the real wage. This seems part of a pattern. Using measured labor variables like ‘employed people’ Martinez, Siegenthaler and Saez (2018) states about this elasticity ‘Indeed, the Frisch elasticities that we estimate are orders of magnitude smaller than the Frisch elasticities that many macro business cycle models require to match the employment reductions during recessions’. Translated: declines in real wages do not explain the (lack of) decline of the number of employed people during an economic downturn. Along comparable lines, Attanasio et al. (2018) finds about this elasticity of labor supply measured in hours with regard to wages: ‘The most important conclusion from our analysis is that the macro elasticity is not a structural parameter, it is simply the result of highly nonlinear aggregation which depends on demographic structure as well as the distribution of wealth and the particular point in the business cycle. This implies, for instance, that to understand the consequences of income tax changes, we need to be explicit about whose tax is changing: for some, such as women who are working few hours or on low wages, responses to changes in wages are substantial’. Translated: the models use a one person description of society, the representative consumer. In reality, many people exist. Some people work, some are unemployed, some do not take part in the gainful labor economy.

124  Labor and unemployment In the course of time, the demographic balance changes, participation rates change and whatever. This means that a change in the wage level in one year might have wholly different consequences than changes in another year, because of underlying social and demographic developments not captured by the one person model. It’s not a stable parameter as it’s, on the macrolevel, prone to fallacies of composition as well as influenced by changes in the business cycle or the very thing the models try to explain, meaning that the models have to specify a dynamic and estimated (not calibrated) relation between the variable and the phase of the business cycle which the models which I’ve seen do not do. This leaves us with two problems. One is definitional in nature: are the labor services as used by Bokan et al. (2016) in hours or persons? And can, in a world with income inequality as well as households and a gendered division of labor the assumption of a ‘representative consumer/producer’ be upheld? Christiano et al. (2011)) states in an overview of the state of the DSGE art in which Ht denotes the amount of ‘market work’ sold by households: ‘Under one interpretation, Ht represents the amount of hours worked by a typical person in the labor force. . . . An alternative interpretation of Ht is that it represents the number of people working’. This is a highly remarkable statement. There is a model – and the scientific community using this model has no clearly stated what the variables are about. Christiane et al. admit these are, from a macro- and from a micro-perspective, quite different concepts. And models are fuzzy about it. As stated, Bokan et al. (2016) circumvents this discussion by mentioning ‘labor services. But as we have seen it is very well possible that the number of hours declines while the number of working persons increases. This even happens on a regular basis (Ohanian and Andrea 2012). In Germany after 2008 the average number of hours worked per person dropped considerably which strongly contributed to a mitigated increase of U-3 unemployment measured in persons, while it also happens that employment measured in persons declines with as much as 20 percent while average hours per working person do not drop, which of course makes headline unemployment soar, like in Spain after 2008 and in Finland after 1991. So, Christiano, Trabandt and Walentin (2011) is too US-centered when it states: ‘It is well known that much of the business cycle variation in employment reflects changes in the quantity of people working, not in the number of hours worked by a typical household’ and imply that DSGE models are about ‘persons’ instead of hours. Ohanian and Raffo (2012) provide an overview of the differences of other countries compared with the USA. Also, Cole and Ohanian (1999), Prescott (1999) and Prescott (2017) explicitly use hours instead of persons while as stated Bokan et al. use undefined ‘labor services’, which simply means that on top of the underlying changes in the labor force there are also large differences between DSGE models. A problem which is not only not settled but which often even goes unrecognized. Christiano et al. (2011) also assumes that especially ‘marginal’ workers like the elderly and ‘spouses’ will enter the labor market when real wages increase, which is a mistake. First, spouses and elderly persons are not marginal. Second, during the Great

Labor and unemployment 125 Depression ‘family members’ entered the labor market when the breadwinner was unemployed and because wages declined – not because real wages increased (Mathy 2018). Low employment made real wages decrease but also made more people enter the labor market. Attanasio et al. (2018) states that such relations pervade the entire labor market and are not an exception or even the rule but the core of household behavior. Third, one of the most consistent labor market events after the 2008/2009 crisis (except for the USA) was a fast increase of the employment and participation rate of older female spouses, caused by cultural, institutional and demographical changes and not so much by lower (or higher) wages. One might call this a shock to the system. But it is the historical evolution of a gendered labor market where people take care of families. Also, the Frisch elasticity should of course be measured net of people getting unemployed. These are estimated as part of the labor force which is not recognized by the models. Summarizing: There is no clear, authoritative concept and definition of ‘N’ and the relation of N with wages and demographical and economic changes in the models.

4.4.2  Deconstructing the future DSGE models assume that households look ahead toward an infinite future. Their behavior with regard to this future is completely market regulated: there are ‘complete’ markets between the present and the future. It is interesting to look at what happens when the assumption of ‘complete markets’ is abandoned. Hagedorn, Manovskii and Mitman (2019) do exactly this, in an otherwise impeccable DSGE model. Immediately a non-classical, fundamentally Keynesian world of involuntary unemployment emerges. They do not mention ‘unemployment’ even once in their article and do not model it. But the temporarily high rate of leisure in their model is not a choice. It’s involuntary. People are not working and want a job. Let’s call it by it’s true name. The model restricts consumption to non-durable consumer private goods: it’s not a complete macro-model. But it does show that the idea that involuntary unemployment or even unemployment as such does not exist depends on strict assumptions.

4.4.3 Fundamental and non-fundamental ways to make the models consistent with high unemployment Purging unemployment from a model does not mean that one does not have to come to grips with it. Unemployment was and is a highly visible and highly cyclical measured variable. Neoclassical macroeconomists have often tried to understand labor market developments (not just unemployment), since the financial crisis in 2008 also by tweaking the models. Some examples: after 2008 real wages in many European countries decreased which should have led to a lower participation rate. Quod non. Nucci and Riggi (2016) explains the absence of a decline of the participation rate in the EU despite what they call ‘additional looseness in the labor market’ and lower real wages by introducing the idea that home

126  Labor and unemployment production (household work) is important, by assuming ‘habit persistence’ (‘the rent has to be paid by the unemployed, too’) and the idea that wages are sticky and hence, during downturns, too high which lures people into the labor force who shouldn’t be there in the first place. Stating it like this sounds snarky, but the comparable Casares (2010) article is titled ‘Unemployment as excess supply of labor: implications for wage and price inflation’. This title literally means that there are too many people on the labor market and the unemployed shouldn’t have been there in the first place. Despite this, such articles are a positive. Household work matters. Consumption patterns do not and cannot change overnight. But the Nucci and Riggi household economy is still not the dynamic, investing sector incorporating mayor (bio)technological and social progress and change which contributed so much to the rise of the female participation rate. The complex interaction between increasing levels of education, lower fertility rates, household technology and cultural changes is not taken into consideration, which, especially for Spain where the rise in the female participation rate has been exceptionally swift, is not satisfying, which means that for these models the absence of the expected decline of the labor market participation rate following wage cuts is a wage problem which has to be solved by introducing ad hoc explaining variables. To underscore this: the idea that high wages lure people into the labor force can in no way explain events like those in the UK where wages declined with 10 percent while participation increased. Or events in Greece, where real wage decreases of 40 percent did not lead to a decline in participation. Another problem with neoclassical ideas, extreme unemployment in countries like Spain and Greece, is tackled by Casares and Vázquez (2016), see also Casares (2010). In Spain and Greece after 2008 unemployment soared from an already high 6 to 7 percent to around 25 percent. Such rates of unemployment were literally normalized by the economists of European Commission (2014), whose estimates of 22 percent ‘natural unemployment’ in Spain have not only been ridiculed on ‘heterodox’ economics blogs but also by the Wall Street Journal ( Dalton 2016) and Voxeu (Fioramanti and Waldmann 2016). Casares and Vasquez are also not satisfied by the idea that 22 percent unemployment is ‘natural’ and explain extreme unemployment in Spain using ‘habit persistence’ as well as wages which are set too high by the households (the models are not explicit about the coordination mechanism), which lures too many people to the labor market. Such studies however still use the general equilibrium framework and the assumption that the ‘right’ level of real wages will quickly solve the problem of unemployment, either by driving people out of the labor force or by increasing production. (Un)employment and wages statistics are not kind to this idea, surely not as real wages in countries like Spain and Greece and the UK (Tily 2016) showed considerable decreases without a concomitant drop in the participation rate. The basic ideas are not changed. A more fundamental change is proposed by Roger Farmer, who sticks to the use of microeconomic utility functions to derive intertemporal optimization of consumption (perhaps the most accessible of his articles: Farmer 2013). But he cuts the consumption-wage-employment nexus out of the models by introducing

Labor and unemployment 127 a subtle change. Expectations are not, by assumption, consistent with the unique equilibrium, anymore. They are, in a sense, part of the cultural technology of a society and drive the level and nature of investments in a certain direction. This changes the path the economy takes – and alters the final equilibrium. It’s the difference between water going down a drain (DSGE models) and a marble rolling over an uneven pavement (the Farmer model). Despite possible initial disturbances water in a bathtub will go down the drain in a predictable way (the equilibrium situation). But a seemingly small difference in direction might make the marble run into a small object, change course and end up in a hole – the hole being high unemployment. Households do not by assumption equilibrate consumption, investment and work in the long run in this model. Once in a hole, they will know that they can’t get out individually and act accordingly by cutting spending, even when they would be able to get out together, by conscious coordination. By severing the long run equilibrium relation between consumption, wages and labor supply he also destroys the general equilibrium and ‘natural unemployment’ setting of these models, which enables a model with ‘multiple equilibria’. Such a model enables a better description about what happened in Finland and Sweden in 1991, in Spain and Greece after 2008 or in Poland and Bulgaria and Eastern Germany after 1992: very long lasting extreme levels of unemployment: more than 15 or even 20 percent for the better part of a decade. Farmer is also much more explicit about the differences between hours, people and the participation rate and does not shy away from a slightly inductive approach, taking (un)employment statistics at face value. Instead of a DSGE model (dynamic stochastic general equilibrium) such models might be called DSIE models, with ‘I’ for ‘indeterminate’. At this moment, Farmer is however not a representative DSGE economist. As he states it: ‘Persistent expectations is a strike against rational expectations PLUS the uniqueness [economies are like bathtubs, M.K.] assumption. It is the uniqueness assumption that needs to go; not the rational expectations assumption which simply reflects a fact that we have known for a long time: Expectations are incredibly persistent’ (Farmer 2014). Ideas like his might carry the day – but that day still has to come.

4.4.3  Models which take the statistics at face value There are also less model and more statistics centered ideas used by neoclassical economists to come to grips with ‘measured unemployment’. One example is ‘search theory’ (Mortensen and Pissarides 1994). Search theory accepts the fact that in the real economy large amounts of jobs are destroyed and created every year (see Table 4.1 and Graph 4.4). In fact, the number of jobs created is even larger than the flows of persons. Many jobs, for instance in tourism or agriculture, do not last for a year. When a job is destroyed this means that people are dismissed and have to search for a new job. It takes them time to find such a job: unemployment results. The same holds for companies trying to find an employee: vacancies result. Finding a suitable job gets, according to these ideas, easier when the number of vacancies increases during a boom. Hence, during a boom unemployment

128  Labor and unemployment will tend to be lower. It becomes less easy during a downswing of the economy, when more jobs are destroyed and less vacancies are opened: unemployment increases. Job seekers might also think they can do better than the first offer they get and continue seeking, which will raise unemployment as more people keep seeking. Remarkably, for people without a job searching as such is costless while it isn’t for people with a job (Mortensen and Pissarides 1994). Mortenson and Pissarides deserve praise for taking the empirics serious and for trying to explain the patterns. But a pure neoclassical aspect of their model is equating unemployment with leisure and implicitly stating that seeking a job is as pleasurable as a Mediterranean holiday. Search theory can be part of DSGE models. Gertler, Sala and Trigari (2008) at first even seem to allow multiple unemployed and employed people to inhabit their model: a genuinely micro-founded model. But following Mortensen and Pissarides (1994) the variables they use are fractions, which describe populations. This means that consistency with the model is still obtained by assuming a representative homo economicus, which does not work for a fraction of the time and tries to work a little more. This is not the same thing as measured headline unemployment which is about people working not at all. Search theory and related concepts like ‘hazard rates’ (the chance that an unemployed person becomes unemployed and vice versa) are highly useful. See Mathy (2018) and Bentolila, Ignacio-Pérez and Jansen (2017) for important (and non-neoclassical) ways to use it. It is an excellent tool to gain deeper insight into labor markets, for instance by estimating the influence of different economic circumstances on different groups on the labor market and by looking at the consequences of a changing composition of the labor force. But search theory (more precise: ‘searching’) only explains structural or frictional unemployment of maybe 2 percent. To this, maybe 1 percent of unemployment can added caused by optimistic searchers (or an employer who mistakenly think he can find somebody better than you). It does not explain 25 percent unemployment in Greece and Spain post 2008. Labor flows show that the reservoir lake of unemployed does not consist of the same people all the time. But as there (in normal times) is a high correlation between unemployed getting a job and unemployed people becoming unemployed, the level of the lake only changes quite slowly – at least downwards. Sudden storms might cause a fast rise of its level. Aggregate unemployment seems quite independent of the age, skills and education of the labor force or even the willingness of labor to accept much lower re-entrance wages (Bentolila, IgnacioPérez and Jansen (2017)) which shows the limits of search theory. It also does not explain years and decades of employment of over 10, 15 or even 20 percent as experienced in countries like Spain, France, Italy, Slovakia, Slovenia, Latvia, Lithuania, Estonia, Poland, Finland, Bulgaria, Ireland or ‘East’ Germany. Mathy (2018) shows that the chances of people getting a job rapidly decline when their spell of unemployment gets longer while at the same time new people get unemployed (also, using search theory, Sushant et al. (2018). The empirics showed (as is also stated by Mortenson and Pissarides) that job creation is not very sensitive to the business cycle, unlike job destruction. (This might however be different during financial crises (Graph 4.4)). ‘Sudden storms’ leading to a fast increase

Labor and unemployment 129 of unemployment in combination with a gradual decrease mean that when a storm hits the economy and many people are fired, their chances of getting a job will diminish which, arithmetically, means that many of these unemployed will become long-term unemployed, which will greatly diminish their chances of getting a job. It is even conceivable that their chances dwindle to nil, which means that economies are stuck with permanent high long-term unemployment not because of legal rigidities but because of rigidities engendered by high unemployment itself. Hendry and Mizon (2014) shows that this idea is consistent with the time series. Unemployment itself is, in such a case, the largest rigidity of the labor market while Mathy (2018) argues that, during the Great Depression, it took the unlimited demand for labor of the war effort to break this spell in the USA. Lowering wages does not work. When in Greece nominal wages are decreased with 40 percent this did not mean that the labor force or unemployment declined. Search theory is a useful addition to the toolkit of economists. But it does not explain persistent high unemployment. To get a better idea of the interplay of the concepts of the models and the actual estimates when investigating real life events, the next paragraph will analyze two episodes which both saw a massive decline of hours worked: the USA in 1929–1933 and the USA in 1945–1949. The declines had, however, very different consequences. Unemployment and suffering in the first case, shorter working weeks and paid vacations in the second case. At least, according to the statistics. But not according to the models.

4.5 Voluntary and involuntary declines in hours of labor and unemployment during the Great Depression and beyond 4.5.1  The anomaly DSGE models do not accept the idea of ‘involuntary unemployment’ or even the idea of involuntary unemployment at all. Statisticians however measure a situation where people do not have a job and actively try to escape, efforts which, considering unemployment rates which are over 15 percent on a regular basis, not always successful. Investigating the situation of these people it turns out that they feel bad and have low incomes. The DSGE economists counter the challenge of these statistics (if at all) by stating that people just should try harder, for instance by accepting lower wages. Or they state as the macroeconomists of the EU do, that 20+ percent unemployment is needed to stabilize the economy. Who’s right? I will discuss the statistical concepts and measurements as well as the neoclassical ideas using the period of the Great Depression plus the Second World War and the subsequent ‘age of Doris Day’ (say: 1947–1957) in the USA as an casus. I will however show that other countries around the North-Atlantic experienced similar developments. In this period, two comparable declines in USA hours worked, one between 1929 and 1933 and the other between 1944 and 1949 (Figure 4.4) led to very different outcomes. The first decline, taking place after 1929, led to high

130  Labor and unemployment unemployment and misery, the other was associated with relatively low unemployment and a period of unprecedented prosperity (Figures 4.5 and 4.6); the data for Germany are added for comparison. Focusing on the USA we see that the 1929–1933 decline in labor demand led to unemployment skyrocketing to, eventually, around 25 percent of the labor force, just like in Germany and, as we’ve seen, the UK and the Netherlands. In the USA this led to a fall in nominal wages with between –3 percent for workers in public utilities and –45 percent for workers in bituminous coal mining (Wolman 1933, also US Department of Commerce 1975, D739–D754). Surprisingly, the post 1944 decline in hours did not lead to such an extreme rise in unemployment or to wage deflation. Also, in 1945–1948 nothing like the post-World War I spike in unemployment can be witnessed in the USA. Until 1957 unemployment would stay reasonably low, even when nominal wages increased with around 5 to 10 percent a year (U.S. Department of Commerce 1975, series D739–D754). What was the difference between these periods? There are two answers to this question, an institutional/statistical one and a sharply different neoclassical economics one.2 Comparing these stories will highlights the rift between the statistics and neoclassical theory.

4.5.2  The institutional/statistical explanation of the anomaly The rapid rise of US unemployment after 1929 was not a unique phenomenon (Graphs 4.2 and 4.5). Unemployment in Germany, the UK and the Netherlands

1929-1933

1944-1949

1944-1949; -21% 1929-1933; -23%

Graph 4.5  Declines in number of hours worked, USA, 1929–1933 and 1944–1949 Source: Kendrick and Pech (1961)

Labor and unemployment 131 and many other countries skyrocketed, too. In some countries, especially those who abolished the gold standard, it also fell quite swiftly. The UK, which left the gold standard after September 1931, still witnessed a mild increase in 1932 but in 1933 employment fell swift and decisively. The USA left gold in June 1933; unemployment already showed a mild decline in the same year. Unemployment in the Netherlands, the last country to leave the gold standard in 1936 (a few hours later than France) proved tenacious until that very moment. The timing of the increase was synchronous, the timing of the decline wasn’t and this difference which was related to national monetary policies. This convinced economists that unemployment was manageable (or at least influenced by national policies) and, during the war, made them fear for a large increase of unemployment once countries started to wind down the war effort. This increase did not happen even when the number of hours worked did decline dramatically – the only country with high unemployment after the war was ‘Germany’, which literally was a new country with a dislocated society and population. Why were economists wrong about their idea that unemployment would skyrocket, again? To be able to give an answer, we first have to ask the question if the historical data are comparable with present day data. Unemployment data for the prewar years is not entirely representative of the entire economy. It tends to overemphasize developments in manufacturing, data are prone to revisions and also not totally consistent with modern concepts. The massive and unprecedented increases after 1929 are undisputed (Galenson and Zelfter 1957; Darby 1975; Den Bakker en van Sorge 1991), just like the larger declines. But are the levels also comparable with present day levels of unemployment? Modern definitions of unemployment require that somebody has zero hours of gainful employment during the last week and is actively seeking gainful employment before he or she is counted as ‘unemployed’. Pre-1970 or 1980 or, for the USA, pre-1955 requirements were less strict, in line with the recommendation of the International Labor Organization in 1925: ‘that the necessary and sufficing condition for being enumerated as unemployed is that the individual must have been not at work for one day at least’ quoted in Galenson and Zellner 1957, p. 441; for practical examples see also pp. 442–443). This means that the historical data include at least some people who were ‘employed part time for economic reasons’ and who are at present included in ‘broad’ unemployment as defined in Graph 4.3 instead of in ‘U-3’ or headline unemployment. As we’ve seen differences between headline and broad unemployment are substantial for the USA but nowadays also for other countries (Knibbe 2011). Post-2008 ‘underemployed part time workers’, as Eurostat calls them and which are only a part of total broad unemployment, amount to between 3 and 4.5 percent of the labor force of the EU as a whole, almost 6 percent in Germany and almost 7 in Ireland and Spain (data for 2017). This suggests that pre-war data are, compared with today’s headline unemployment, on the high side, even when modern researchers do try to make them comparable (Den Bakker and Van Sorge 1991; Den Bakker and De Gijt 1994). It might have been that part time workers for economic reasons hardly existed in the 1930s, which would mitigate the problem.

132  Labor and unemployment To get closer to an answer to this conundrum we need to know more about the unemployed during the Great Depression. We won’t try to make an estimate of underemployed part time workers – we will just, in a qualitative way, investigate if they existed, to answer the question if comparability of the series is a problem. To do this, we will investigate a particular kind of unemployed which will also enable us to shed some light on the question if being unemployed is a voluntary situation and comparable with ‘leisure’, as neoclassical economists suggest. In the USA of the Great Depression, a particular kind of underemployed part time workers (and I doubt if they were counted as ‘unemployed’, despite the recommendations of the ILO) were the so-called ‘hobos’. Hundreds of thousands of them, mainly young men and teenage boys had to leave their families as these had to save on household expenditure. They started seeking jobs, free riding on trains. On the Internet, letters and oral history documents of hoboes are published which enable us to gain some idea of the nature of their quest (Erroluys.com 2017). According to these documents they were often cold and hungry and occasionally did not eat for days. They slept wherever it was dry (or not); places to sleep take up a remarkable amount of space in the oral history documents. Hopping trains was dangerous and many former hoboes tell of people who were injured or died during their free rides. Importantly, they literally crossed the country seeking work and considered working for about one dollar a day well paid. This wage can be compared with hourly wages for common labor which in 1932, after the wage declines taking place between 1929 and 1932, were as low as 32 or even 25 cents per hour for mainly female labor in hotels (Wolman 1933); Wolman states that the hotel workers might have had some perks like meals. Domestic workers earned around 25 to 35 cents a day, but had free meals. Clearly, during the Great Depression and as a consequence of the decline of the amount of work available, life of many laborers in the United States was brutish, mobile and cheap. Using modern definitions, these hoboes, ‘employed part time for economic reasons’, would not have been included into ‘headline’ or ‘U-3’ unemployment as calculated by the Bureau of Labor Statistics in the USA and all kind of other national statistical agencies in dozens of other countries. But they were part of broad unemployment. ‘There is no doubt whatever about that’. Using an institutional/ Marxist instead of a statistical framing: the hoboes can be looked at as a reserve army of labor which could be paid below subsistence wages as free riding and ‘free sleeping’ trains while seeking a job can be understood as subsidized traveling and lodging while they also often relied on begging for a living. Being an unemployed hobo was nor a leisurely life. It was hard work even when some do mention a sense of freedom. But that’s not the point. They preferred work over being a hobo and did whatever had to be done to get work. Underemployed part time workers sure were a big thing, during the Great Depression, which, though it is not clear if they are included in the unemployment data, makes it possible that the level of unemployment in the 1930s, as the operalization of unemployment was broader than today, might have been higher than ‘headline’ unemployment. On the other side – when the definition of broad unemployment would have been used it probably would have been higher.

Labor and unemployment 133 This leaves the question if unemployment was voluntary and a kind of leisure. The hoboes enabled a decline of weekly wages of regular employees of their households as their ‘taking the road’ enabled households to shift down the household cost curve. The reservation wage, as economists call it, could decline and, considering the nominal decreases, it probably did. But this did not solve unemployment. After about 1933 US unemployment did decrease rapidly, also because of New Deal ‘workfare’ jobs, but it was only the war effort which really solved the problem. Clearly, desperately seeking work and accepting low wages was not enough to get unemployment down – even despite the millions of New Deal ‘workfare’ jobs which were created after 1933 and which rapidly changed the infrastructural face of the USA. Mathy (2018) shows that the aggregate USA ‘Beveridge curve’, which shows the relation between unemployment and vacancies, returned to normal only after the war. Until the war, the long-term unemployed clearly were less employable as employers preferred other people. Everything shows that unemployment was, in different senses of the word, involuntary. A methodological note: following Darby (1975) and consistent with the modern concept of unemployment as developed by the International Labor Organization the data in Graph 4.6 do not consider people working workfare

30

25

20

15

10

5

0

1910

1920

1930

1940

1950

1960 Germany

1970

1980

1990

2000

2010

USA

Graph 4.6  Unemployment, USA and Germany, 1910–2018 Source: USA: Darby 1975; US history.com (2017). Technical addendum: historical unemployment data are not as dependable as modern data and are based upon slightly different concepts. The data are often based upon unemployment of members of labor unions and are often about what nowadays is called ‘broad unemployment’ as they also include people who in a certain week had some days of work but wanted to work more. The data in this graph exclude people on ‘workfare’ which means that for instance the USA data show a faster post 1932 decline than the more often used data from Lebergott (1957) which include people on workfare as ‘unemployed’.

134  Labor and unemployment jobs as unemployed. But even when these workfare jobs are subtracted, unemployment stayed high and only replacing a reserve army with a real one did the trick of getting unemployment really down and making the hoboes disappear, which shows continuous searching for whatever job available, accepting whatever wage offered and leading a lifestyle which did not even guarantee regeneration of the body was not enough to cure unemployment. Unemployment was not voluntary. And it was not leisure. And desperately seeking work did not solve the Great Depression. When post and pre-war unemployment are compared it might be better to compare the level of Great Depression unemployment not just with headline unemployment but also with ‘broad’ unemployment. The low level after World War II might to an extent be a statistical fluke. But even then the very limited increase is, considering the massive decline of hours, spectacular while the hoboes did not return and Mathy (2018) shows that the post war Beveridge curve had returned to ‘normal’, which is spectacular. Between 1944 and 1949 the winding down of war expenditure led to a decline of hours worked comparable with the Great Depression. But after 1944 (and these data are comparable) unemployment only increased from 1 percent in august 1944 to around a still reasonably low 4 percent in 1946–1948, despite the massive 12 percent decline of GDP in 1946. Taking broad unemployment into account the increase might have been a percentage point more. But nothing resembling the increase of 1929–1933 and the change of the Beveridge curve in these years can be witnessed. During the war, economists were worried that the end of the war would lead to an economic downturn and a large decline in demand for labor, which is what happened. The economists were right. But this did not lead to a return of depression like unemployment. Something had changed. In one way or another, economic policies and historical developments prevented the rise of mass unemployment, poverty and suffering and even enabled Americans to adopt or at least to aspire a new, more playful lifestyle: ‘the age of Doris Day’. Tellingly, Doris Day’s first movie, Romance on the High Seas (1948) is foreshadowing one of the growth sectors of 20th and even more the 21st century, ocean cruises. What was the difference between these two periods? Why did a decline in hours of over 20+ percent lead to poverty and unemployment in one period while a comparable decline in hours two decades later led to happy days, iconic prosperity – including paid vacation – and reasonably low unemployment? The starting situation of the two periods was very different. The US economy during its involvement in World War II, 1942–1945, was and is the epitome of what economists call an overheated economy.3 The decline of hours worked after 1944 was, after almost two decades of abnormalcy, to an extent a return to (or: the creation of) ‘normalcy’ The ‘hobo’ oral history documents mentioned clearly show that ‘taking the road’ was not considered ‘normal’ by the hoboes. Also, the female participation rate declined from 34 to 30 percent in one year (Acemoglu, Autor and Lyle 2004 but already noted by Lebergott 1957). The average number of hours per week declined to about 41 or with about 10 percent compared with the war years and with about 2.5 percent compared with 1937

Labor and unemployment 135 (Kendrick and Pech 1961, p. 316). Sunday became a day of rest, again. Instead of some people becoming entirely unemployed, all people started to work shorter hours – while paid vacations decreased the number of hours actually worked and hence increased the demand for labor. Also and despite the end of the war a much larger percentage of the population than before served in the army (a staggering 3 percent, at that time largely draftees). And while production took a large hit after 1944 it stayed, partly thanks to a lightning fast transition from war production to mass production of consumer goods, considerably higher than before the war. The combined effect was high and relatively stable employment and hence low and relatively stable unemployment. How did this all happen? Did ‘New Deal’ rules, higher wages and more union power (as well as a much larger army than before the war) enable this development? Probably. But this paragraph does not aim to explain the influence of institutional and economic background of these events. It only wants to show that the labor side of the events of the Great Depression, the war and post-war prosperity can be described using standard definitions and measurements about employment and unemployment while also showing that there is no mechanical relation between these variables. Often, large declines in (gainful) hours worked lead to mass unemployment and many people actively seeking for work. Sometimes – not. It surely helps to understand the 1944–1947 event in the USA against the background of the general decline of the length of the working week which in the Western world took place between in the century between 1880 and 1980. Higher productivity translated into less hours worked per person employed. The statistical data enable to carve out a coherent macroeconomic story, which can be fitted to the ‘micro’ lives of individual people, such as the so-called ‘hobos’. The fact that they all but disappeared after 1941 and did not resurface after 1945 is a tell-tale sign that a different society had emerged, characterized by low unemployment and more widely shared prosperity. Employment for all was possible. Against a background of technological change and progress, new policies led to unprecedented prosperity and a booming economy. According to the (unemployment) statistics. We’re reminded of the words of the first president of the ILO: ‘It would therefore be almost impossible to exaggerate the truly revolutionary character of the events which during the last years of the war or after the war took place in the sphere of the regulation of hours of work’ (International Labor Office 1921). The USA, which to an extent had not taken part in these vents after World War I, was catching up. But was this also the case according to neoclassical economics?

4.5.3 The neoclassical explanation of labor during and after the Great Depression The neoclassical story of the Great Depression is based upon ideas put forward by Edward Prescott, Robert Lucas, Dale Mortenson and Christopher Pissarides, all of them Nobel Prize winners. At least Edward Prescott actively propagates this view. It might be a minority view. It has been ridiculed. But the authors themselves call it the neoclassical story. The story has a problem: ‘unemployment’ is

136  Labor and unemployment not a variable in neoclassical macro. Unemployment does not fit into this thinking other than as the short-term consequences of unanticipated shocks or frictional unemployment. A long-term or large increase in measured unemployment is, according to these models, considered to be an increase in leisure and ‘leisure’ is the phrase which is used to denote such events. During the depression, unemployment in the USA suddenly increased to about 25 percent. And stayed high for quite a period. It is difficult to explain this as a sudden preference for leisure, which is a problem for neoclassical authors. In the next paragraph, we will investigate two ways neoclassical economists cope with this. The first neoclassical story is mainly associated with the names of Nobel prize winner Edward Prescott as well as with Harold Cole and Lee Ohanian (Prescott 1999, 2017; Cole and Ohanian 1999). This last study (1999) explicitly uses the word ‘neoclassical’ in the title while Prescott links the story explicitly to neoclassical growth theory. Prescott (2017, p. 3) summarized the problem they had to solve in his chapter of the new Handbook of Macro Economics: Cole and Ohanian . . . initiated a program of using the theory [the idea that society can be modelled as one representative homo economicus, M.K.] to study great depressions. They found a big deviation from the theory for the 1930– 1939 US Great Depression. This deviation was the failure of market hours per working-age person to recover to its pre-depression level. Throughout the 1930s, market hours per working-age person were 20 to 25 percent below their predepression level. The reasons for depressed labor supply were not financial [here, Prescott changes, while stating this only later, from the 1929–1941 period which he wrongly labels the 1930–1939 period to the 1934–1939 period, M.K.]. No financial crises occurred during the period 1934–1939. The period had no deflation, and interest rates were low. This led Cole and Ohanian to rule out monetary policy as the reason for the depressed labor supply. Neither was the behavior of productivity the reason. Productivity recovered to trend in 1934 and subsequently stayed near the trend path. Note that Prescott does not mention unemployment. Note also that Prescott frames the situation of ‘depressed labor supply’, not demand, and excludes unemployment from supply. Note that he writes about ‘market hours per working-age person’ and not about working hours per person of the labor force, which saves him the problem that the labor force consists of the employed as well as the unemployed. He also uses a fraction, which describes aspects of a population and not individual people – he really looks at the entire population as a holistic representative consumer. A remarkable aspect of the excerpt is the treatment of financial crises. It is a proven fact that financial crises lead to prolonged and slow recoveries (Reinhart and Rogoff 2014). This was not yet common knowledge when Cole and Ohanian published their ideas in 1999. But it was common knowledge when Prescott wrote his chapter in the 2017 Handbook: financial crises cast a long shadows. Even neoclassical models yield this result once the assumption of rational expectations is traded in for measured real time real expectations (Hollmayr and

Labor and unemployment 137 Kühl 2016). Prescott should have known this – he clearly didn’t. But the main point: unemployment – as well as employment measured in persons – is defined away. Using, I have to say, a complacent kind of cunning writing smacking of academic group think. The inclusion of the ideas of Cole and Ohanian in the Handbook serves to show that their ideas are, as yet, part of the neoclassical canon.4 There is therefore reason to focus on their ideas, also as comparable neoclassical ideas are still background of the idea that ‘equilibrium’ unemployment rates of about 25 percent are meaningful concepts. Cole and Ohanian (1999) looks at the average number of hours worked by the ‘representative consumer’. This shifts their focus from an analysis of on the one side the employed and the unemployed to a world where unemployment is basically meaningless and not of no interest to economists. As another proponent of using this methodology, Robert Higgs, bluntly states: ‘By using hours worked as our measure of employment, we avoid the necessity of distinguishing who is employed and who is unemployed’ (Higgs 1997, p. 152). Cole and Ohanian (1999) in effect state that measured US unemployment during the 1930s is not an economic interesting variable. It is defined away. Despite the fact that we measure it. Despite the fact that during the Great Depression an entire sector of the spending economy – expenditure on investment goods – had fallen of a cliff (here defined as a decline of 10 percent of more of GDP) and had left a hole in aggregate spending that could not be filled by either consumption (surely not as unemployment increased) or peace time government expenditure. The implosion of the investment rate, central to the work of Keynes, rendered the representative consumer unable to keep consumption up. It’s not that a decline in average hours worked wasn’t important during the 1930s. The average working week of people with a job declined spectacularly. According to Whaples, working hours for employed males (non-farm) declined with about 25 percent, from 48 hours to around 36 hours (Whaples 2001). But despite this, unemployment stayed higher than ever and did not decline to earlier rates. Failing to distinguish between declines in the actual working week (‘part time workers for economic reasons’) and unemployed people disables an analysis of the importance of either development, as well as convolutes our understanding of the period. Another strategy to un-see unemployment during the Great Depression can be found in Cole and Ohanian (2003). The problem Cole and Ohanian (2003) tries to solve is low growth of total hours after 1934. Despite an increase in productivity (which, according to the model, makes the representative consumer work more as production per hour is higher which will mean that more hours of leisure are traded for hours of work) 1939 production per capita was still way below the 1929 level and average hours were still low. The problem it does not try to solve is the decreasing but still high level of ‘measured unemployment’ throughout the 1930s. ‘Unemployment’ is only mentioned twice in the 50 pages of text and this only to debunk the concept of the statistic: Our model predicts the fraction of individuals in the market sector who search for a job. The number of searchers in our model, divided by the number who are

138  Labor and unemployment either working or searching, is 11 percent during the early part of the transition, and then declines to about five percent. The unemployment rate is one empirical measure of the fraction of individuals searching. Measured unemployment fluctuated between 16 percent in 1934 to about 11 percent in 1939. It’s of course not the fraction of individuals but the fraction of the labor force, which consists of individuals. And considering the importance of workfare projects and government employment in general, employment was not just about the market sector. But aside of this: they totally redefine unemployment. I couldn’t find anything in the text which explains how the ‘searchers’ were measured or which empirical metric they use to derive their guesstimate – it seems to be just an artefact of their model. And they simply disregard the difference between (much higher) measured unemployment and their hypothetical ‘searchers’. It can be added that in 1932, unemployment was 23 percent, according to the source they use (Darby 1975) – quite a bit higher than the 16 percent of 1934, which means that workfare, going of gold and more expansive government strategies had solved part of the problem. They also state that in their models ‘Individuals work either full-time, or not at all’ which means that next to the large chunk of headline unemployment that they threw out of the window the ‘part time unemployed for economic reasons’ disappear from the radar. No so-called ‘hobos’ there. While Mathy (2018) shows that such variables can be measured. They solve the problem of the existence of their hypothetical searchers by blaming it on the New Deal. This set of new rules and priorities enabled, in their view, some sectors of the economy to cartelize. Together with more rights for labor and unions, this New Deal caused (according to Cole and Ohanian) a situation in which these sectors were starting to act as if they were a kibbutz or a Yugoslav workers cooperative. Theoretically, these organizations do not try to maximize profits but try to maximize wages of existing employees, if necessary by restricting production and charging monopoly prices. Such high wages were of course attractive, which made people quit their low paying jobs in other sectors and made them spend time to search for these higher paying jobs, which resulted in high unemployment. According to Cole and Ohanian. This has been ridiculed by calling it the ‘soup kitchens caused the Great Depression’ view of the crisis. There is an excellent reason for this. The ideas of Cole and Ohanian (2003) are clearly inspired by search theory. But they cut out an important part (a part which, indeed, is not consistent with DSGE modelling). Mortensen and Pissarides (1994) is explicit about the fact that changes in firing are much more important than changes in hiring (see also Graphs 4.4 and 4.5). But it is only in the second paragraph of their appendix G that Cole and Ohanian mention the theoretical possibility that employees are fired. Basically, their model of the Great Depression is a world were no people were fired. But there are problems with their idea of hiring too. Mathy (2018) shows that rigidities were associated with the long-term nature of an increasing fraction of total unemployment. In a sense, these people, while still seeking, disappear from the screen of employers – even when events during World War

Labor and unemployment 139 II showed that they were not just seeking but also willing and able to work. The authors also cover the 1939–1950 period albeit in an extremely sketchy way, not explaining why the ‘measured unemployed’ discarded by them did not seek work suddenly turned up as workers in the war industry. The large decline in hours after 1944 or changes in participation rates are not even mentioned. Whatever Cole and Ohanian (2003) does, it does not explain measured unemployment. It does not even try. But it is the canonical neoclassical analysis of the Great Depression. Nobel prize winner Edward Prescott looks at the 1939–1949 period in a less sketchy way. A quote: It’s interesting that market hours in 1949 were only slightly higher than market hours in 1939, while the investment share of output had returned to normal. Growth theory predicts the return of the investment share to normal, because by 1949, the economy had essentially converged to its new lower constant growth path. However, given that market hours were still low in 1959, the U.S. economy was still depressed in 1959. Between 1931 and 1959, only during wartime when public consumption is temporarily high, was the U.S. economy not depressed. Prescott (1999, p. 28) His fundamental mistake is looking at a precise measurement of hours worked instead of a rough idea of total hours available, including the unemployed. Correcting for this grave error shows that unused potential in 1939 was way higher than in 1959. Instead, Prescott surreptitiously uses elevated unemployment and hence relatively low hours worked in 1959 to characterize the entire 1945–1959 period, to state that the entire period up to 1959 was ‘depressed’. But it wasn’t. It was only in 1958 and some later years (after 1980: periods) that unemployment rates of over 5 and 6 percent became a regular occurrence, once again. One wonders why Prescott surreptitiously tries to characterize the entire period up to 1958 as depressed (the NBER data shown in Graph 1.1 show short recessions just before 1950 and in 1954, not during the rest of the period). Looking at unemployment there are reasons to regard to 1975–1995 period as a permaslump (Figures 4.3 and 4.5). But there is no reason to look at the period 1946– 1957 as such. In this case, the decline of average hours did indicate an increase in leisure (and probably a substitution of paid for unpaid hours spend on home production) as people weren’t unemployed. This, however, was not related to the wage level but to institutional change aka the New Deal. Also, his idea that the growth path (not the rate of growth, but the growth path) was, in 1949, permanently lower is than before is, after the explosive growth of the US economy during the 1941–1945 period which, despite the 12 percent set-back in 1946, led to a permanently higher level of economic production outright silly. Ignoring unemployment and confusing it with leisure led neoclassical economists not just to totally misrepresent the Great Depression but also to misunderstand post-war growth and prosperity.

140  Labor and unemployment

4.6 Summary Labor and unemployment are carefully measured and defined in relation to on the one side the transaction economy of the national accounts and on the other side non-monetary work. DSGE models have difficulties keeping up with this and tend be fuzzy about hours, persons, employment and unemployment, sometimes not even using these words but instead phrases like unspecified ‘labor services’ for work or ‘leisure’ for unemployment.

Notes 1 Actually, Friesian butter traders buying butter from farmers on the butter market in the beginning of the 19th century kept their ‘shop’ open longer than usual on low price days as they knew that farmers had to go home again and at a certain point had to accept bargain prices. On high price days, the market of course closed early as soon as everything was sold. Just the opposite of what Lucas states. 2 The phrase ‘neoclassical’ is included in the title of one of the articles used, calling it the neoclassical view is hence right. 3 The war effort started earlier and was in full swing in 1941. The overheating was however only characteristic for 1942–1944 and a part of 1945. 4 In Cole and Ohanian (1999) the text as well as the literature states that Keynes’ General Theory appeared in 1935, not 1936. This must have been missed by them as well as by reviewers.

Literature Acemoglu, D.A., D. Autor and D. Lyle (2004). ‘Women, war, and wages: The effect of female labor supply on the wage structure at midcentury’. Journal of Political Economy 112:3 491–551. Ahn, H.J. and J.D. Hamilton (2019). ‘Measuring labor-force participation and the incidence and duration of unemployment’. Finance and Economics Discussion Series 2019–035. Washington: Board of Governors of the Federal Reserve System. Attanasio, O., P. Levell, H. Low and V. Sánchez Marcos (2018). ‘Aggregating labour supply elasticities: The importance of heterogeneity’. Blogpost on Voxeu, 10 November 2018. https://voxeu.org/article/aggregating-labour-supply-elastici ties#.W-fzpkNWvtI.twitter Bakker, G.P. and W. van Sorge (1991). ‘Het onbenut arbeidsvolume in het Interbellum’. Economisch en Sociaal Historisch Jaarboek 54 212–241. Bakker, J.P. and J. de Gijt (1994). ‘Labor force data in a national accounting framework. Estimation of the Dutch interwar labor force’. CBS report NA-072. Bank of England (2016). www.bankofengland.co.uk/statistics/research-datasets#. Accessed 1 April 2016. Bank of England (2018). ‘A millennium of macro economic data’. www.bankofeng land.co.uk/statistics/research-datasets. Accessed 16 March 2018. Bentolila, S., J. Ignacio-Pérez and M. Jansen (2017). ‘Are the Spanish long-term unemployed unemployable?’. SERIEs 8 1–41. Bie, R.J. van der (1996). Een doorlopende groote roes’. De economische ontwikkeling van Nederland, 1913–1921. Rotterdam: Tinbergen Research Institute.

Labor and unemployment 141 Bokan, N., A. Gerali, S. Gomes, P. Jacquinot and M. Pisani (2016). ‘EAGLE-FLI. A macroeconomic model of banking and financial interdependence in the Euro area’. European Central Bank working paper series no. 1923. Card, C. (2011). ‘Origins of the unemployment rate: The lasting legacy of measurement without theory’. Berkeley and NBER working paper. Casares, M. (2010). ‘Unemployment as excess supply of labor: Implications for wage and price inflation’. Journal of Monetary Economics 57:2 233–243. Casares, M. and J. Vázquez (2016). ‘Why are labor markets in Spain and Germany so different?’ Documentos de Trabajo Lan Gaiak Departamento de Economía – Universidad Pública de Navarra 1602. Christiano, L.J. (2011). ‘Comment on Gali, Smets and Wouters “Unemployment in an estimated New Keynesian model” ’ http://faculty.wcas.northwestern.edu/~lchrist/ research/Gali_Smets_Wouters/manuscript.pdf. Accessed 28 June 2019. Christiano, L.J., M. Trabandt and K. Walentin (2011). ‘DSGE models for monetary policy analysis’ in: Handbook of monetary economics, volume 3A 286–384. Amsterdam: Elsevier. Centraal Bureau voor de Statistiek (2018). www.historisch.cbs.nl/detail.php?id= 117382485 Accessed 7 March 2018. Cole, H.L. and L.E. Ohanian (1999). ‘The great depression in the United States from a neoclassical perspective’. Federal Reserve Bank of Minneapolis Quarterly Review 23:1 25–31. Cole, H.L. and L.E. Ohanian (2003). ‘New deal policies and the persistence of the Great Depression: A general equilibrium analysis’. Federal Reserve Bank of Minneapolis research department working paper no. 597. Dalton, M. (2016). ‘Questions about Europe’s ‘natural’ unemployment-rate estimates’. Blogpost on the Real Time Brussels blog, 30 September 2014. https://blogs.wsj.com/brussels/2013/09/30/questions-about-the-eus-natu ral-unemployment-rate-estimates/ Darby, M. (1975). ‘Three-and-a-half million U.S. employees have been mislaid, or, an explanation of unemployment, 1934–1941’. NBER working paper no. 88. Dimsdale, N., S. Hills and R. Thomas (2010). ´The UK recession in context. What do three centuries of data tell us´. Bank of England Quarterly Bulletin 2010 Q4 277–291. Erroluys.com (2017). http://erroluys.com/greatdepressionarchive10.html Accessed 17 august 2017. European Commission (March 2014). Quarterly report on the Euro Area 13:1. Brussels. Eurostat (2016a). ‘Labor market and labor force survey (LFS) statistics’. Statistics explained. https://ec.europa.eu/eurostat/statistics-explained/index.php/ Labour_market_and_Labour_force_survey_(LFS)_statistics Eurostat (2016b). ‘Underemployment and potential additional labor force statistics’. Statistics explained. https://ec.europa.eu/eurostat/statistics-explained/index. php/Underemployment_and_potential_additional_labour_force_statistics Eurostat (2018). ‘Almost half the unemployed at risk of monetary poverty in the EU’. http://ec.europa.eu/eurostat/web/products-eurostat-news/-/DDN-201 80226-1?inheritRedirect=true&redirect=%2Feurostat%2F. Accessed 26 February 2018. Farmer, R. (2013). ‘The Natural Rate Hypothesis: An idea past its sell-by date’. Bank of England Quarterly Bulletin 2013 Q3 244–256.

142  Labor and unemployment Farmer, R. (2014). ‘Rational expectations and animal spirits’. Blogpost at Roger Farmers Economic Window, 4 February 2014. www.rogerfarmer.com/roger farmerblog/2014/02/macroeconomics-and-animal-spirits.html Ficket, J.W. (2003). ‘Part time for economic reasons background’. Blogpost at Clear on Money, 9 December 2009. www.clearonmoney.com/dw/doku.php?id=public:part_ time_for_economic_reasons_background. Accessed 13 August 2018. Fioramanti, M. and R. Waldmann (2016). ‘The stability and growth pact: Econometrics and its consequences for human beings’. Blogpost at Voxeu, 19 November 2016. https://voxeu.org/article/econometrics-and-its-consequences-human-beings Frumkin, N. (1998). Tracking America’s economy. New York: Armonk. Galenson, W. and A. Zelfter (1957). ‘International comparison of unemployment rates’ in: National Bureau of Economic Research (ed.) The measurement and behavior of unemployment 439–584. Princeton: Princeton University Press. Gertler, M., L. Sala and A. Trigari (2008). ‘An estimated monetary DSGE model with unemployment and staggered nominal wage bargaining’. Journal of Money, Credit and Banking 40–8 1713–1764. Goldberg, J. and W. Moye (eds.) (1985). The first hundred years of the Bureau of Labor Statistics. Washington: US Government Printing Office. Haan, W.de, P. Rendahl and M. Riegler (2017). ‘Unemployment (fears) and deflationary spirals’. Journal of the European Economic Association 16:5 1281–1349. Hagedorn, M., I. Manovskii and K. Mitman (2019). ‘The fiscal multiplier’. NBER working paper no. 25571. Hendry, D and G.E. Mizon (2014). ‘Why DSGE models crash during crises’. Blogpost at Voxeu, 18 June 2014. www.voxeu.org/article/why-standard-macro-mod els-fail-crises. Accessed 14 September 2015. Higgs, R. (1997). ‘Regime uncertainty: Why the Great Depression lasted so long and why prosperity resumed after the war’. The Independent Review 1:4 561–590. Hollmayr, J. and M. Kühl (2016). ‘Learning about banks’ net worth and the slow recovery after the financial crisis’. Bundesbank discussion paper no. 39/2016. International Labor Office (1921). Report of the director-general. Geneva: ILO. International Labor Office (2013). ‘Statistics of work and of the labor force. Report for discussion at the Meeting of Experts in Labor Statistics on the Advancement of Employment and Unemployment Statistics (Geneva, 28 January – 1 February 2013)’. Department of Statistics MESEU/2013. Geneva: ILO. International Labour Organization (2019). www.ilo.org/safework/info/publica tions/WCMS_113134/lang – en/index.htm. Accessed 27 February 2019. Kendrick, J.W. and M.R. Pech (1961). Productivity trends in the United States. Princeton: Princeton University Press. Keynes, J.M. (1936). The general theory of employment, interest and money. London: Macmillan. Knibbe, M. (2011). ‘U3 or U5: A note’. Real World Economics Review 56 151–154. Knibbe, M. (2016). ‘Unemployment around the North Atlantic, 1948–2014’ in: V. Beker and B. Moro (eds.) The European crisis 251–276. London: College publishers. Lebergott, S. (1957). ‘Annual estimates of unemployment in the United States, 1900–1954’ in: National Bureau of Economic Research (ed.) The measurement and behavior of unemployment 211–237. Princeton: Princeton University Press.

Labor and unemployment 143 Lindner, S., J. Mitchell and A. Nichols (2013). ‘Consequences of long-term unemployment’. Urban institute. www.urban.org/sites/default/files/publication/23921/ 412887-Consequences-of-Long-Term-Unemployment.PDF Ljunqvist, L. and T.J. Sargent (2007). ‘Understanding European unemployment with a representative family model’. Working paper. www.sciencedirect.com/science/ article/pii/S0304393207001262 Ljunqvist, L. and T.J. Sargent (2008). ‘Two questions about European unemployment’. Econometrica 76:1 1–29. Ljunqvist, L. and T.J. Sargent (2016). ‘The fundamental surplus’. Working paper. www.tomsargent.com/research/FundSur30.pdf Lucas, R. (1976). ‘Understanding business cycles’. Paper prepared for the Kiel Conference on growth without inflation, June 22–23. http://icm.clsbe.lisboa.ucp.pt/ docentes/url/jcn/mabes/LucasUnderstanding.pdf Martinez, I.Z., M. Siegenthaler and E. Saez (2018). ‘The myth of intertemporal labor supply substitution: Evidence from tax holidays’. Blogpost at Voxeu, 22 August 2018. https://voxeu.org/article/myth-intertemporal-labor-supply-substitution Mathy, G.P. (2018). ‘Hysteresis and persistent long-term unemployment: The American Beveridge Curve of the Great Depression and World War II’. Cliometrica, Journal of Historical Economics and Econometric History 12:1 127–152. Mortensen, D.T. and C.A. Pissarides (1994). ‘Job creation and Job destruction in the theory of unemployment’. The Review of Economic Studies 61:3 397–415. n.a. (1963). Measuring employment and unemployment. Hearings before the subcommittee on economic statistics of the joint economic committee of the congress of the US 88-th congress. June 6 and 7 1963. Washington: U.S. Government Printing Office. Nucci, F. and M. Riggi (2016). ‘Labor force participation, wage rigidities, and inflation’. Banca d’Italia, Temi di Discussione 1054. Ohanian, L.E. and A. Raffo (2012). ‘Aggregate hours worked in OECD countries: New measurement and implications for business cycles’. Journal of Monetary Economics, Elsevier 59:1 40–56. Prescott, E. (1999). ‘Some observations on the great depression’. Federal Reserve Bank of Minneapolis Quarterly Review 23:1 25–31. Prescott, E. (2017). ‘RBC methodology and the development of aggregate economic theory’. Chapter 22 in Handbook of Macroeconomics, 2 1759–1787. Amsterdam: Elsevier. Reinhart, C.M. and K. Rogoff (2014). ‘Recovery from financial crises: Evidence from 100 episodes’. NBER Working Paper 19823. Stapleford, T.A. (2009). The cost of living in America. Cambridge: Cambridge University Press. Stein, R. (1967). ‘New definitions for employment and unemployment’. Monthly report on the Labor Force February 1967 1–25. Sushant, A., J. Bengui, K. Dogra and S.L. Wee (2018). ‘Slow recoveries and unemployment traps: Monetary policy in a time of hysteresis’. Federal Reserve Bank of New York staff reports 831. Tily, G. (2016). ‘UK real wages decline of over 10% is the most severe in the OECD (equal to Greece)’. Blogpost at TUC, 27 July 2016. www.tuc.org.uk/blogs/ uk-real-wages-decline-over-10-most-severe-oecd-equal-greece Treaty of Versailles. 1919. https://archive.org/stream/treatyofversaill00knox/treatyofversaill00knox_djvu.txt. Accessed 28 October 2018.

144  Labor and unemployment U.S. Department of Commerce (1975). Bicentennial edition: Historical statistics of the United States, colonial times to 1970. Washington: US Department of Commerce. US history.com. www.u-s-history.com/pages/h1528.html Accessed 14 August 2017 Vroey, M. de (2004). Involuntary unemployment. New York: Routledge. Whaples, R. (2001). ‘Hours of work in U.S. history’ in: R. Whaples (ed.) EH.Net encyclopedia. http://eh.net/encyclopedia/hours-of-work-in-u-s-history/ Wolman, L. (1933). Wages during the depression. New York: National Bureau of Economic Research.

5 Capital (and land)

5.1 Introduction What is Capital? This chapter will try to answer the question how economists answer this question. Alas, some of the answers will turn out to be mutually exclusive. Economists struggle with sub-questions like: ‘is capital is a physical or a monetary item or is it a social and legal and hence a political variable?’ and ‘is it a produced item or can it also be a non-produced natural or legal entity’ or ‘are capital and consumer goods basically alike or fundamentally different?’ Next to this they tend to understand Capital as a factor of production but sometimes (and in my view rightly so) also as a factor of distribution. Also there are questions if Capital, whatever it is, relates to wage labor dominated economies with large scale production facilities taking part in global exchange only (Marxists), if it relates to the market economy including the self-employed but without durable consumer goods (DSGE models) or to activities within the national accounts production boundary, i.e. including the self-employed but also the government (the national accounts). Questions can be posed if consumer durables should be added to our estimates of the stock of capital (to an extent this happens in the flow of funds). Also various economists, using the idea of capital as a factor of production but not distribution, are elaborating ideas of ‘natural capital’ or the value of aspects of nature yielding services to humans. First, however, we will discuss the vision of the historian: Capital as an ever changing non-homogenous monetary, legal, social and physical factor of production, distribution and power. The chapter will, consistent with the national accounts, be based upon on a quadruple accounting balance sheet representation of ‘fixed capital’ or the value of all kinds of items used by households, the government, companies and non-profits. It will argue that looking at both sides of a balance sheet will solve some of the conundrums mentioned. But: what do we put on the balance sheets and what do we exclude?

5.2  The (r)evolutionary nature of capital So, what is Capital? To answer the question we have to make it more precise as the meaning of Capital is ambiguous. In this chapter, we will largely but not totally exclude financial capital like bonds, bills, loans and equity and deposits and

146  Capital (and land) coins and mainly but not solely investigate the question: ‘What is fixed capital?’ This boils down to questions like ‘which items are considered to be fixed capital’, ‘why are such items considered to be fixed capital’, ‘how is the value of these items estimated’, ‘who owns and/or uses these items’ and ‘what is the influence of the law and power on ownership and the value of ownership?’ To get a satisfying we also have to pose a companion question which will be discussed in this paragraph: ‘what used fixed capital to be?’ One of the characteristics of ‘fixed capital’ is that it changes over time. Even more so than labor or consumption, capital changes. To an extent this historical character of capital is reflected in the changing official definitions of fixed capital in the national accounts. Nowadays these include software, some decades ago they didn’t. To understand such changes it is important to grasp that the Capital with a C includes, looking at balance sheets, non-produced assets like land and patents and even ‘goodwill’ or the excess of the money paid to purchase the asset or business over the total value of the assets and liabilities – which can be embedded in ‘administered prices contracts’ which guide future transactions. Capital, even fixed capital, is not a physical item. It’s the aggregate of the monetary value of ‘fixed assets’ owned by a person, a household, an organization or even a nation which is used in some kind of monetary production process and can either be sold or reproduced or used for gainful purposes by the legal or economic owner. As the statisticians say: ‘an economic asset is a store of value representing the benefits accruing to the economic owner by holding or using the entity over a period of time. It is a means of carrying forward value from one accounting period to another’ (ESA 2010 lemma 7.15, Eurostat 2013). Think of the value of a bridge, a house, tools, jewelry or robots. This definition brings different kinds of ‘ownership’ into the discussion. ‘Economic owner’ means that the legal owner and the user do not have to be the same person or organization. A company leasing cars is the economic owner while they are leased from the ‘legal owner’ (ESA 2010 lemma 15.05, Eurostat 2013). This distinction – well known from the difference between a renter and a landlord – is crucial. It indicates that there is a difference between the use of ‘Capital’ as a factor of production and ownership of capital as a factor of income, distribution and: power. And the value of capital is based on the balance sheet of the legal owner (even when legal and economic ownership can sometimes get blurred: strong economic rights tend to get a market value which, eventually, may show up on the balance sheet of the economic owner). It is stressed by classical economists like Smith, Ricardo, Mill or Marx and is related to the left side of the balance sheets (assets used for present production) as well as the right side of the balance sheets (who provided the different funds to finance these items?). As balance sheet values are by default monetary values they tie ownership and value of capital to the monetary economy of prices and transactions. The value estimated by statisticians is a monetary value, which means that ‘free’ natural resources, like clean air, aren’t included in the concept as these are not owned and hence do not enter the monetary transaction economy, even when for instance oxygen and clean air are crucial to a multitude of production processes. There are borderline cases: catalytic converters, and there are

Capital (and land) 147 hundreds of millions of these, can be understood as producers of clean air which are to a considerable extent used within the transaction economy. But while the production costs of these last convertors and their exchange value do enter the capital calculations the production of clean air is outside of the national accounts production and consumption boundary. Economists have grappled with the question why a particular item is considered to be ‘fixed capital’ or not. It’s basically a question of the production boundary, which changes in nature as well as position. It can, as Marxists do, be restricted to a wage labor economy in which the concentration of ownership in an economy characterized by vehement economies of scale and in combination with legal rules enables owners of fixed capital to exploit others. To quote Marx: A negro is a negro. In certain circumstances he becomes a slave. A mule is a machine for spinning cotton. Only under certain circumstances does it become capital. Outside these circumstances, it is no more capital than gold is intrinsically money. . . . Capital is a social relation of production. And a historical relation of production. (Marx 1887 [1867] chapter 33 footnote 4) As plantations functioned in a global wage labor economy, slaves were capital in the Marxist sense, even when slaves in the Greek and Roman sense weren’t. The modern SNA mimics this Marxian idea, albeit in more technocratic prose and a production boundary which also includes the self-employed. An example is the definition óf natural non-produced assets: The choice of which natural assets to include in the balance sheet is determined, in compliance with the general definition of an economic asset, by whether the assets are subject to effective economic ownership and are capable of bringing economic benefits to their owners, given the existing technology, knowledge, economic opportunities, available resources, and set of relative prices. Natural assets where ownership rights have not been established, such as open seas or air, are excluded. (Eurostat/European Commission 2013, hence: ESA 2010, lemma 7.26) So, what is considered to be capital changes. And capital itself changes. A contemporary example relating to the value of capital: house prices in parts of cities popular with tourists have increased quite a bit (+14 percent in Los Angeles) because of platforms like Airbnb (Koster, Van Ommeren and Volhausen 2018), turning house owners into entrepreneurs. Clearly, changes – in this case a change in value – in what is capital have taken place and are taking place. Including concomitant changes in the social structure of our societies. Essentially: these changes are part of the nature of capital. Nowadays, capital can even be created ‘out of thin air’: ‘The classification of non-produced assets is designed to distinguish assets on the basis of how they come into existence. Some of these assets exist naturally,

148  Capital (and land) while others, which are known as constructs of society, come into existence by legal or accounting actions’ (Eurostat (2013)). Examples are production rights for phosphate in manure from farm animals as well as contained in animals leaving the farm for the slaughterhouse. See Fosfaatrecht.nu (2019) for a site where phosphate production rights are traded. A well-known phrase about production and consumption is: ‘By the sweat of your brow you will eat your food until you return to the ground, since from it you were taken for dust you are and to dust you will return’ (Genesis 3:19). When we continue to read in Genesis we encounter division of labor, iron and bronze tools, fields (lots of them), musical instruments and even a large wooden boat. But where these items ‘Capital’? We can use modern definitions as a lens to look at Genesis. Using a Marxist framework none of these would classify. Using the national account definitions, the tools and the fields might, if used for monetary production or when they had a market value. The boat surely classifies as a consumer durable. But would these distinctions appeal to contemporaries? Probably not. But today, even ‘dust’ is a monetized commodity and balance sheet itrem (see 5.3.5). Owning fixed capital endows people and organizations with ownership rights and to flows of income connected to letting or using it. Such rights as a rule legally codified, which makes capital a social and political and not just an economic variable. Being social and political variables, these rights tend to change, sometimes in an evolutionary but occasionally also in a punctuated and even revolutionary way. A contemporary example is the debate on international agreements about rights of capital owners on one side and national governments on the other side, like TTIP. A historical example is slavery. One of the remarkable data in the famous Piketty book Capital in the Twenty-First Century is a graph of the value of pre-Civil War fixed capital in the slave states of the USA, expressed as a percentage of regional GDP (Piketty 2014a, p. 161). It shows the total value of fixed capital before the civil war and includes estimates of the value of houses, land, means of transport, cattle and machinery. And the value of slaves. The total balance sheet capital-value of slaves (based upon market prices, not production costs) was over 50 percent of the total value of fixed capital – or more than the value of all land, houses and other kinds of fixed capital combined. After the US Civil War (1861–1865) the capital-value of slaves dropped to zero as slavery was abolished. Over half of the balance sheet wealth of southern society disappeared into thin air. This of course influenced the distribution of income and wealth. Remarkably, the negative influence of this revolutionary change in property relations and the organization of production on the amount of production and economic activity seems to have been limited. Cotton production quickly rebounded, based upon a production structure not so much based on plantations using slave labor but on share croppers, small farmers and also plantations using oppressed wage laborers (Acemoglu and Robinson 2008). In 1871 production of cotton as well as exports to the UK matched the record level of the 1859 again and the fast growth of the 1820–1860 period resumed. According to Melvin Copeland (1912, p. 179): ‘The cotton crop of the United States averaged 1,749,000,000 pounds per year from

Capital (and land) 149 1856 to 1860 and 6,157,000.000 pounds per year in 1906–1910’. Also, despite the abolishment of slavery the families of former slave owners did quite well the civil war, capitalizing on educational advantages, financial and social networks and also outright suppression ( Ager, Boustan and Eriksson 2019). The example shows that what’s called ‘fixed capital’ is not homogenous over time and not just an economic but also a legal, social and political variable and as such a factor of distribution, sometimes even more than a factor of production. The abolishment of slavery is far from the only example of a revolutionary change of ownership of ‘fixed capital’. For protestant Europe, the expropriation of all kinds of ecclesiastical organizations connected to the reformation in the 16th century comes to mind. Land was transferred from cloisters (not from local churches) to the government and charities. In this case, fixed capital itself didn’t change but ownership did, again with profound consequences for distribution. After the expropriations land rents did not flow to the monks anymore, but to governments as well as to education and to orphanages. This changed the course of history (Cantoni, Dittman and Yuchtman 2017a, 2017b). Remarkably, despite the expropriations (agricultural) production was often hardly affected by the change in landlords or, if so, in a positive way for instance because of an increase in investment and scale (Knibbe 2006 for Friesland in the 16th century; Finley, Franck and Johnson 2017 for the expropriation of ecclesiastical lands in France during the French revolution). Remarkably, ownership of Capital seems to have functioned mainly as a factor of distribution, not of production. But fixed capital changes in an evolutionary way, too. Different classes of fixed capital age at different rates. Combined with the changing level and composition of aggregate investment as well as changes in technology and bankruptcies leading disastrous micro-write-downs this leads to continuous changes in the balance sheet composition of the stock of fixed capital. One of the reasons housing and other buildings dominate the stock of fixed capital is simply their low rate of depreciation while, according to the rules of accounting, land (including land underlying houses) does not depreciate at all. Another reason is that, mainly because of the greying of societies, in many countries the average number of inhabitants per house shows a secular decline. Also, house prices increased relative to other prices, partly because of increases in quality and partly because of cheap and easy mortgage credit. Taken together such events continuously change the composition of the stock of capital which makes it tricky to estimate it and which, of course, influences production as well as distribution. Only one example: the secular increase in house prices has transferred enormous wealth to the baby boomers owning houses as well as their descendants. All this mean that there is no simple answer to the question ‘what is fixed capital?’ Questions like ‘why are patents included but privately owned cars excluded?’ or ‘are slaves still included in national balance sheets in societies were slavery still exists?’ cannot easily be answered without taking circumstances into account. Apps like Uber might well lead to an enhanced capital content of privately owned cars. Also, as far as I know slaves aren’t included into the national accounts but as

150  Capital (and land) the production boundaries of the SNA 2010 include criminal production they maybe should. More on the demarcation lines in the next paragraph.

5.3  Capital and the statisticians 5.3.1  Concepts and definitions The system of national accounts uses balance sheets as its main tool to systematize capital ownership. The value of fixed assets is related to ownership per sector and even sub-sector of the economy in a quantitative and measurable way. Balance sheets have two sides. One side shows the value of assets: buildings, land, cash, machinery, receivables. These assets are owned by a company or a person. The other one shows the funds used to finance these assets: retained earnings (or losses), loans, shares emitted, payables, items connected with ownership rights of different counterparties – or in the case of retained earnings also the company or individual owning the assets. Balance sheets clearly and deliberately show the dual nature of capital which is emphasized by, among many others, Clarence Ayres (1944), especially chapter 3, but which has also bedeviled quite some economists putting undue emphasis on capital as a factor of production. When it comes to the statistics it’s clear, however. The dual and legal nature of is part and parcel of the statistical concept. The asset or ‘factors of production’ side is shown together with the liability or ‘factor of distribution’ side. On top of this a quadruple accounting system is used. Financial assets of one sector or subsector are liabilities of another. The debts of the Irish construction sector in Chapter 1 are lent by somebody. These relations are spelled out in the ‘from-whom-to-whom’ tables which, per financial asset, spell out who’s indebted to whom, as shown for one sector for one kind of financial instrument in Graph 5.1 (this data is not readily available for the construction subsector of Graph 1.1). The graph shows only ‘long-term loans’ but data for other debt instruments and stocks are available, too, and the website of the Central Bank of Ireland contains interactive ‘network’ graphs which shows assets as well as liabilities for the entire economy per debt instrument. Important: the changes in financial positions are not only due to financial transactions but also to upward or downward revaluations of these instruments (bankruptcies) which shows that there is no 1:1 relationship between transactions (borrowing or paying down loans or debt) and changes in gross and net wealth. A sign of the times is of course the extreme increase of the amount of long-term loans between the last quarter of 2015 and the first of 2016, which is not a statistical fluke but which is related to international shifts of assets and liabilities or, to state this otherwise, to Irelands position as a tax arbitrage heaven. It will be clear that quadruple accounting is a check on the system: total assets do have to match total liabilities. These are however financial assets and liabilities. The main theme of this chapter is gross fixed capital: ownership titles of machinery, buildings, sub-soil gas and oil, patents and the like. A simplified but still extensive list of produced and nonproduced assets from Eurostat/European Commission (2013) (there Table 7.1

Total

Non-financ. Corp.

Total

Monetary Financial Instuons

Other financial instuons

2014 Q4

Insurance Pension funds corporaons

2015 Q1

Financial Corporates

Non-MMF investment funds

General government

Households Rest of World and NPISHs

Source: Central bank of Ireland, interactive from-whom-to-whom graph www.centralbank.ie/statistics/data-and-analysis/financial-accounts/ financing-and-investment-dynamics-interactive-web-app, accessed 06–05–2019

Graph 5.1  Resident non-financial companies, Ireland, long-term loans (liabilities): total and per lending sector, EUR€ millions

0

100000

200000

300000

400000

500000

600000

700000

152  Capital (and land) p. 172/173; here Table 5.1), the list as shown here is still extensive to be able to pit it against the much less granular neoclassical definition of capital which will be discussed in a subsequent paragraph. Assets are divided into financial and non-financial assets. Following the ideas of classical, 19th-century economists (in fact the ideas of economists up to the neoclassical take over in the 1960s) the national accounts divide non-financial assets into produced and non-produced assets (‘Land’) while the non-produced assets Table 5.1  Fixed capital items listed in the SNA 2010 AN.1 Produced non-financial assets AN.111 Dwellings AN.112 Other buildings and structures AN.113 Machinery and equipment AN.114 Weapons systems AN.115 Cultivated biological resources AN.117 Intellectual property products AN.1171 Research and development AN.1172 Mineral exploration and evaluation AN.1173 Computer software and databases AN.1174 Entertainment, literary or artistic originals AN.1179 Other intellectual property products AN.12 Inventories AN.121 Materials and supplies AN.122 Work-in-progress AN.123 Finished goods AN.124 Military inventories AN.125 Goods for resale AN.13 Valuables AN.2 Non-produced non-financial assets AN.21 Natural resources AN.211 Land AN.2111 Land underlying buildings and structures AN.2112 Land under cultivation AN.2113 Recreational land and associated surface water AN.2119 Other land and associated surface water AN.212 Mineral and energy reserves AN.213 Non-cultivated biological resources AN.214 Water resources AN.215 Other natural resources AN.2151 Radio spectra AN.2159 Other AN.22 Contracts, leases and licenses AN.221 Marketable operating leases AN.222 Permits to use natural resources AN.223 Permits to undertake specific activities AN.224 Entitlement to future goods and services on an exclusive basis AN.23 Purchases less sales of goodwill and marketing assets Source: Eurostat/European Commission (2013) table 7.1, p. 172/173

Capital (and land) 153 are divided into ‘natural resources’ and ‘contracts, leases ad licenses’. Neoclassical economics, traditionally uses a production and distribution function only based on labor and produced capital (Solow 1956) and not on unproduced capital even when this is changing again (Rognlie 2015). Statisticians, using sectoral balance sheets as well as a sectoral approach need a complete set of assets and liabilities, are forced by the logic of their chosen model in combination with a focus on distribution on income to use the classical trichotomy instead of the neoclassical dichotomy. An example: subsoil assets in Alaska are owned by the state of Alaska, which distributes government rent income from these assets as a ‘citizens dividend’ over all Alaskan citizens. In Texas, subsoil assets are owned by the owner of the ‘space’ and any (rent) income from oil or natural gas does not go to the state of Texas but to the owner of the land. In Europe, the code Napoleon established that, like in Alaska, subsoil assets are owned by states. Contrary to Alaska, these states as a rule use rent income related to the exploitation of sub soil assets as a source to finance government expenditure. Being forced to value subsoil assets and to allocate ownership rights to sectors like households or the government makes it impossible for statisticians to follow the guidance of neoclassical economists and to sidestep the classical Land, Labor and Capital trichotomy. As at this moment land is the most important single item of ‘fixed capital’, it can’t be left out. Rognlie (2015), a neoclassical study trying to deconstruct the estimates of Piketty, is forced by the same logic to re-introduce ‘land’ into the neoclassical edifice. On the other side neoclassical economics does encompass a concept like ‘human capital’ (skills, strength, education, health, the color of your skin) which is not defined as capital by the statisticians as it does not have a cost or market price as it can’t be traded or even collateralized. Statisticians basically experience it as a metaphor, not a variable which can be consistently considered to be ‘capital’. As we will see, ‘dust’ did turn into capital, however. Modern ideas to equity-finance higher education of students might do the same for the education part of human capital. At the moment of writing this still is dystopia. Consequently, Piketty, who focused on the distributional consequences of ownership of capital and based himself on national accounts data, had to use the concepts and data of the statisticians and thus to exclude human capital (Piketty 2014a). A point of possible contention about these definitions: ESA 2010 defines the clearance of pristine forests as an addition to the stock of capital (ESA 2010 lemma 3.128-b (Eurostat/European Commission 2013). Pristine forest can be defined as ‘natural capital’. But in many countries which the ESA 2010 guidelines have to cover too, clearances are still an important source of new agricultural land and will continue to be so for quite some time (Koning 2017). Clearance does bring land from outside the production boundary as defined by the accounts into this boundary. Summarizing: statisticians conceptualize and define ‘fixed capital’ in a granular way and, as they look at the total economy and use balance sheets as their main organizing principle, also look at unproduced assets as well as at the interrelation between ownership and debt. They also make a distinction between economic and legal ownership, which enables them to define ‘rent’ as a legal distributional variable. Keeping the true classical tradition alive they also pay ample attention to the (quantitatively) most

154  Capital (and land) important asset of them all: Land, which leads us to the question why statisticians value land so highly. That question will be discussed in the next paragraph.

5.3.2 Measurements Official statistical institutes are treasure troves of data as well as information on how data are measured. Fortunately and contrary to the days of yonder, a lot of meta-data is published on the Internet which enables us to investigate how capital is valued. Some examples for the Netherlands: Tinbergen (1942); CBS (1947); Korn and Weide (1960); Verbiest (1997); Bergen et al. (2009); Veldhuizen et al. (2009). An overview: Leenders (2016). What are the results of these measurements? Graph 5.2 shows Dutch ownership of assets per sector. More detailed

1800000 1600000 1400000 1200000 1000000

800000 600000 400000 200000 0

Financial companies

Government

Non-financial companies

Households

Mineral and energy reserves (net)

Inventories (net)

Intellectual property products (net)

Machinery and equipment and weapons systems (net)

Other buildings and structures (net)

Dwellings (net)

Land (net)

Graph 5.2  Ownership of fixed assets, households, the Netherlands, 2016 Source: Centraal Bureau voor de Statistiek, nationale rekeningen.

Capital (and land) 155 breakdowns are possible. Several things stand out. First, households, and not companies or the government, are the main owners of capital (more detailed data in the national accounts, Knibbe 2014 and, an extension of the Piketty database, Leenders 2016. Second, The most important fixed capital good in the Netherlands is land, including land underlying houses and other structures. Third, dwellings come second. Fourth, other buildings and structures like bridges and roads come third which underscores the importance of infrastructure including houses. This should not come as a surprise. Just look around you in whatever city you are. The Netherlands are no exception when it comes to the overwhelming importance of land and buildings (Piketty 2014a). The high value of land (especially land underlying houses) is caused by the extra-ordinary increase of house prices after around 1965. As, in the national accounts, houses as such are valued at historical or actual construction costs minus depreciation, most of this rise is ascribed to the value of land. It shows that the old adage that the value of a building is caused by three factors: ‘location, location and location’ is wrong for buildings. But right for land. In a growing world where ever more people want larger houses and cars, more roads, more meat, more office space and nearby soccer fields and parks as well as places to put their caravans and boats, where cheap solar cells in combination with batteries are revolutionizing transport (an electric car only needs about 4.000 kwH per year) and where a greying population leads to a lower average number of inhabitants per house and hence relatively more houses, an increase in the value of land is bound to happen. Price increases induced by easy credit play a role, too (Hudson 2012; Borio 2012). These inflationary increases are as a rule mainly imputed to the value of land and not to the value of the buildings. But the total construction value of houses increased too as the number of houses increased just like the quality and size of houses. Think of insulation, solar panels, better heating systems and (a little earlier) running water, indoor toilets and electricity. Fifth, the share of machinery (including trucks and company owned cars) is remarkably small for several reasons. The millions of cars owned by households are not included in the production boundary. For the world as a whole these would add several trillions (Euro) to the stock of capital (increasing the already preponderant position of households). Also, depreciation rates of machines are much higher than rates for structures or land (which is not depreciated at all). A one euro investment in a house simply lasts longer than an investment in a truck. Lastly: machinery is basically not that expensive. Stories about ‘the industrial revolution’ with their pictures of smoke stacks and steam trains have led us to believe that investment is all about machinery. But a smoke stack is part of a building and the train in these pictures typically drives over a new bridge. True, it was and is largely thanks to machines including vehicles that productivity increased. Physical labor productivity of harvesting potatoes has increased about one hundredfold thanks to the development of potato harvesters. Or compare the physical labor productivity of a man with a dragline with a man with a shovel: it is off the scale.1 And it has even become hard to imagine that trucks replaced horses and carriages and even humans carrying stuff. Machines also widened the

156  Capital (and land) scope of our activities. Using an MRI machine extends the realm of medical opportunities and enables physicians to save lives. But all these items are, compared with labor, in rich countries relatively cheap per hour used. Or to state this otherwise: we do not need that much ‘machine’. This may run counter to our intuition. But our intuition might be based too much on household behavior, not on the market use of machinery. Household used cars are typically idle for no less than 97 percent of the time (CBS 2012), contrary to taxi cabs. Even MRI machines in hospitals are used (taking opening hours into account) about at least 50 percent of the day. Anybody who has ever visited a modern cheese or dog food or tobacco or waste treatment factory will have been impressed by the relentless speed at which relatively small factories spit out enormous amounts of products. When we think of machines we do not have to think of the price per machine but of the price per hour used and product produced. Investments in machinery are, compared with investments in construction, per capita relatively limited and depreciation rates are high – which is a reason why, in Figure 5.2, they do not loom as large as ‘real estate’ and ‘land’ even when they are highly productive or at least effective. Summarizing: the statisticians use a granular multi items, multi-sector, quadruple accounting system to classify, value and embed the value of fixed assets. Valuating fixed assets is, even when clear rules exist, an art as well as an science. But art and science point out: land is the most valuable asset of them all.

5.3.3 Valuations How do statisticians value capital? Statisticians are sometimes called bean counters. In this case they try to count jumping beans. Or to be more precise: to condense the value of capital over time in one point estimate. The ESA 2010 manual proposes to use market prices as much as possible. But possibilities to do this are limited as markets are limited which means that other methods have to be used, too. Planes have a second hand market selling price while there also is a lively second hand market in planes or, for that matter, military equipment, which enables the use of market prices to value such items. But how about roads? Or bridges? Clearly, these have a value. But there is no second hand market. Also, many bridges are unique as well as historical: prices for modern bridges with different specs can’t be easily used to value historical bridges. At the same time, interest rates change, which influences the value of capital. Still, we do value it. How? That will be the subject of this paragraph. As I’m most familiar with the way capital is measured in the Netherlands (Knibbe 2014) we will take a closer look at Dutch measurement – which is consistent with ESA 2010. A quote from the trenches (Bergen et al. 2009, p. 4): In the past, the purpose of measuring capital at Statistics Netherlands was mainly restricted to the calculation of consumption of fixed capital. Following recommendations of the European System of Accounts (ESA-1995, Eurostat, 1996) the consumption of fixed capital was estimated on the basis of a linear

Capital (and land) 157 age-price profile. In addition, mortality functions were used to correct for the probability of premature asset discards. A disadvantage of this method was that, although premature discards were taken into consideration, no adjustments were made for assets with longer than average service lives. Valuation methods clearly have a relation with why capital is valued at all, while multiple methods exist. Consumption of fixed capital is important to be able to calculate net incomes, balance sheet analysis however requires methods which yield data consistent with balance sheet methodology. Aside of this, there are problems with data as well as short cuts to cut out short-term volatility from the estimation of long-term assets. Nieuwkerk and Sparling (1985, 1987) point out the problems associated with the measurement of international fixed capital; problems with international statistics do not have to seem receded (Linsi and Mügge 2019). Rossum and Swerts (2011) state that three year running averages of market prices are used to gauge the value of extractable natural gas in the Netherlands. There are, however, some central tenets to measuring the value of capital. First, capital is owned, by a person, an institution, an organization or the government. This may seem trivial, but it isn’t. Non-owned natural resources, like clean air or whales in the ocean, are not considered capital – hence, they are not valued. According to the most recent Eurostat national accounts manual (Eurostat/ European Commission 2013, hence: ESA 2010), consistent with the UN guidelines, ‘Natural assets where ownership rights have not been established, such as open seas or air, are excluded’ (ESA 2010 7.26).2 There are efforts going on to measure the economic value of ‘natural capital’ which will be discussed in the following paragraphs. Here, it is important to note the crucial importance of ownership. Combining ownership with economic sectors it is crucial that fixed capital owned by households (legal plus economic ownership) is not used in the same way as capital used by companies. Companies will always have a tendency to use fixed capital as efficiently as possible. A hotel owner want its rooms to be occupied. For households, this is different as can be witnessed by the increasing number of single elderly living in houses fit for an entire family or the extremely low utilization rate of private cars, compared to for instance the use of cars by a taxi company. To go a bit meta: in markets, prices are agreed upon before transactions are finalized – market prices are basically ex-ante prices. As these prices have to engender a flow of monetary income enabling the company to pay for costs (including wages and imputed ‘mixed income’ of the self-employed) companies do have to consider the expected utilization of fixed capital (a central point of Lee 1999). In the household economy there are no such market prices and people do not take these into account. Most families which go on an extended vacation do not rent their house out. Monetary costs are in such a case to an extent what economists call ‘sunk costs’ or ‘spilt milk’, as one may consider gasoline costs of a trip to his or her parents-in-law. But do we really calculate money foregone because we use our private cars only 3 percent of the time while we could have rented them to other people for a part of this time? Do we really estimate depreciation?

158  Capital (and land) Companies have to do this. Households don’t. Or maybe they have to do this but they don’t. In such a situation, market prices may not always be the best way to gauge the value of capital and consumer durables hence excluded from the production boundary. Cars however do have a second hand value – an strong argument to include them in the production boundary, which does not happen in the national accounts but which does happen in the flow of funds. Also, Airbnb is changing things. As stated: Capital changes. An important point is the value of land. Back in the 1950s, when Solow formulated his famous neoclassical version of growth theory (Solow 1956), the value of land had, driven by a decline of relative prices of agricultural goods, reached a historical low (Knibbe 2014), which enabled Solow to explicitly state that his growth theory, consistent with neoclassical ideas but in stark contrast to the ideas of the classical economists of the 19th century, was about produced capital only and not about land and other unproduced capital. This (plus assuming that capital goods and consumer goods are only different ways of using essentially the same universal good) enabled him to link the total stock of ‘fixed capital’ in a seamless way to the rate of investment and depreciation. But ‘land’ is back. The increase in the value of land since around 1965 singlehandedly drove the large long-term swings in the capital to GDP ratio central to the analysis of Piketty (Piketty 2014a, 2014b; Knibbe 2014; Rognlie 2015). This means that, at least when it comes to distribution, we are living in a classical, ‘Ricardian’ economy again, where banks providing mortgages have taken the role of the 19th-century land owners (Hudson 2012). Economic statisticians have extended the concept of natural unproduced capital to ‘human’ unproduced capital items, like patents, production permits and even research & development (R&D): ‘nonproduced non-financial assets . . . are economic assets that come into existence other than through processes of production. They consist of natural assets, contracts, leases, licences, permits, and goodwill and marketing assets’ (ESA 2013 7.24)3. But how are these valued? The basic answer is: in a multitude of ways, which leads to the following question: which are these ways and why are they used? To answer this question, a little digression is needed. In growth theory the amount of fixed capital is theoretically related to nominal GDP via investments and depreciation. The value of capital is ‘stock-flow consistent’ – assuming that all fixed capital is alike. We’ve already seen that, because of technological change, different rates of depreciation in combination with shifts in the composition of investment and the unique nature of at part of fixed capital as well as the fact that land does not depreciate, this is not the case, which, as not all fixed capital is alike, leads to remarkable developments. In the Western world, the level of investment (expressed as a percentage of GDP) has been declining for decades, often by as much as 8 to 10 percent of GDP (Knibbe 2014). Even when fixed capital is not entirely stock-flow consistent one would expect that this decline in the rate of investment would have led to a gradual decline of the value of fixed capital (expressed as a percentage of GDP). Piketty pointed out that (until 2008) the opposite happened (Piketty 2014a). The value of capital increased. A lot. This increase was not caused by a high level of investments but by price increases of

Capital (and land) 159 houses and especially land underlying houses (Knibbe 2014; Rognlie 2015). But the value of a dike may to the contrary be measured using historical production costs corrected for depreciation (the perpetual inventory method) or present day replacement costs – whatever comes in handy (Groote 1995). Different kinds of estimation procedures are used for a reason: different kinds of capital know different use and production processes and are used or constructed and owned by different actors which means that the ‘second hand market price’ model, which can be used to gauge the value of planes or the present ‘location’ value of houses, is not a good template to gauge the capital value of coastal levees which are not only not traded on markets but also not used in a market context. But even where a market exist using market prices sometimes leads conceptual problems. Often some kind of quality and location adjusted average of selling prices of houses is used to value all houses. But can we really value houses which are not ‘on the market’ with the price of houses for sale? At the time of writing, our family house was not ‘on the market’. But theoretically it was ‘for sale’ – if anybody offered a price high enough (in this case considerably above ‘market value’) we would have sold it. Shouldn’t we ideally use these ‘off market’ prices to value houses? Or should we indeed use market prices as this is related to the monetary value of a house as collateral? Or should we use market prices as we can estimate selling prices while a theoretical selling price as mentioned does not have a clear price tag attached? We won’t discuss these questions – but it’s not unimportant to know that statisticians do like things they can measure and will have a habit to take any cost or market or rebuilding price they can obtain. To an extent, this is an example of the idea that when one has a hammer, every problem is a nail. On the other hand – there are reasons why sometimes no market prices exist while we do have information about reconstruction costs.

5.3.4 Volumes Despite the monetary and changing nature of ‘Capital’ economists often write and talk about the ‘volume’ of capital, as if it’s a stock of machinery and equipment which increases or decreases in size but not in composition or with regard to ownership. But ‘volume’ is in such cases a metaphor, meaning that fixed prices are used to construct estimates to enable a specific kind of historical or international comparisons. The word still denotes a monetary balance sheet value and prices are still used to value and enable aggregating of individual items of fixed capital. These prices are sometimes cost prices, sometimes market prices and sometimes even other kinds of prices. But they are monetary prices which, like the composition and ownership of Capital, change over time. The use of fixed relative prices of different kinds of fixed capital to make multi-year estimates of the aggregate volume of capital may obfuscate these changes especially as the value and the use of capital is influenced by changing relative prices. A case in point is the decrease in land rents and the price of agricultural land in Europe after around 1880 when imports of grains drove down agricultural prices of land in Europe, which led to a historical decrease in the income and

160  Capital (and land) social position of land owners (Knibbe 2014; for a larger number of countries Leenders 2016). Technological induced changes in the stock of fixed capital can also lead to social changes. A classic example is the connection between the steam based factory and railway and shipping system, the rise of wage labor and the growth of large cities in the 19th century (Heblig, Redding and Sturm 2018); for famous contemporary observations Engels (1952 [1845] and the first chapters of Dostojevsky (1917 [1866]). Nowadays, in at least some popular destinations for tourists, Airbnb seems to lead to comparable changes. Using fixed prices in combination with the, often implicit, assumption that the composition of the stock of capital is stable might therefore misguide the economist when interpreting the data. This is true in a technocratic sense, too. Using relative prices of 1960 (when prices of land and structures were relatively low) will yield a different estimate of the change of the volume aggregate capital than using prices of 2018 (when prices of land and structures were relatively high). Using low prices for land and high prices for produced items will lead to a relatively high growth of the aggregate and vice versa. Also, new items may, on average, have shorter or longer life spans than the typical fixed capital goods of the past, changes which might not be reflected in ‘mortality functions’ used to gauge the total stock of capital. On top of this changes in interest rates but also in the business or the financial cycle or technological or legal developments can influence the value and composition of aggregate fixed capital. An academic example: quite soon, rules of ownership with regard to articles appearing in scientific journals might radically change. This leads us to the question if we really can aggregate heterogeneous capital goods with different ages and rates of depreciation and different production processes and technologies and which are used in different institutional surroundings into one over time and internationally homogenous ‘volume’, necessarily using ‘fixed prices’ which are only representative of relatively short spans of time and for an individual country, taking discontinuous technological change as well as changes in demand, availability of credit, interest rates and legal systems into account. In the case of the flow of value added choosing a price to value transactions is not too difficult: we use the very prices of the transactions. But in the case of existing fixed capital there often are no representative transaction prices. One might use the average price series of the flow of investments to overcome this problem. Or as in Bokan et al. (2016), even the consumer price index. There are conceptual problems with this procedure – unless we assume, like DSGE models do, that there is a homogenous ‘jelly’ stock of capital which does not know any historical changes in its composition and which can be readily and at no costs substituted into consumer goods: a stock of coconuts. Alas, we’re not living on a tropical island. We’ve seen that the composition of the flow of investment as well as the composition of the stock of capital changes, which means that some of the conceptual problems are comparable to those of the consumer price level, though they loom even larger because of the larger variability in prices of investment goods as well as the larger variability of the set of products purchased (partly

Capital (and land) 161 because of the often large sums involved). Valuations are influenced by, among other things and comparable to products included in the consumer price index: • • •

Changes in relative prices; Changes in the set of products purchased (more planes or more bridges?); Changes in quality (computers are the obvious example, genetically changed farm animals another).

But in the case of fixed assets some additional problems have to be added, like • •

Large differences in depreciation rates; Changes in prices of existing assets not due to depreciation (houses, but also assets which become worthless because of technological change or changes in for instance environmental rules).

Also, when we deflate the total stock of assets with the investment price series we will overstate the importance of items with a high rate of depreciation which, exactly for that reason, have, compared with their importance for the rate of investment, a relatively low value in the stock of fixed assets. In a comment on a blogpost in which Groote (1995) was mentioned as an example of this procedure Groote however stated: A remark on the suggestion that in national accounting stocks of fixed capital are often deflated with an investment deflator (which would indeed be problematic). Normally, however, one would use the perpetual inventory method for building stocks of capital (net or gross) from flows (of investment, scrapping and depreciation) in constant prices. So the flows are deflated, but the stocks do not need to be as they are by definition in constant prices. This is exactly what I did in for the Netherlands capital stock in infrastructure 1800–1913, as referenced in the text. So not a problem if stocks are built the way they are supposed to be. (World Economics Association 2019) This only looks at infrastructural investments which already mitigates the problems. But when relative prices change, we still have a relative prices bias (it is somewhat equal to using a superlative price index, see Chapter 7). But this is not what the models do. These deflate the nominal value of a 2019 bridge with a price index based upon for instance 2015 computers, which is not a good idea – for one thing as it often takes years to build a bridge. Next to this, house prices show large upward and downward credit related swings. Do we really want such booms and busts, which affect relative prices of fixed investments and hence the investment price series, to influence our estimates? Also, we know that international financial flows or credit creation can fuel house price booms. Do we really want monetary flows influenced by international differences in macroeconomic regulation to lead to vehement changes in relative prices used to construct

162  Capital (and land) volume series of capital? Constructing a deflator for the flow of investments is difficult; constructing a deflator for the entire stock of fixed assets is much more difficult. To state this in another way: it is really hard to calculate a meaningful volume index for the stock of fixed assets. It’s sometimes stated by economists that assets should be valued as bonds, i.e. by discounting the expected flow of future monetary benefits. In ESA 2010, this method is also mentioned – but only as a method of last resort. Statisticians deplore this method as sudden changes in expectations of future income flows as well as in interest rates lead to mayor changes in estimations (Knibbe 2014). Discounting just does not deliver the kind of intertemporal stable estimates which statisticians adore. Maybe people have expectations of stable future flows of income and stable future interest rates. But statisticians simply do not measure the future. They measure the past and dislike estimates which are influenced by debatable and, in the real world, ever changing expectations. This does not mean that the other methods used deliver any kind of estimate of the ‘true’ value of fixed capital. The balance sheet value of fixed assets is a rather crude guess of some kind of value which changes from year to year (this contrary to the value of the debts on the liability side). This does not mean that those estimates are worthless. There are clear long-term patterns. The post 1965 increase of the ratio of the (nominal value of) capital to (nominal) GDP in almost all Western countries, mainly driven by increasing house and land prices (Knibbe 2014; Rognlie 2015) did influence intergenerational distribution of wealth as well as borrowing and lending – borrowing and lending which contributed to the 2008 crisis. The breathtaking increase of household debt alone warrants some kind of estimate of the value of the assets at the other side of the balance sheet. But the estimates are shots from the hip at a large but moving target.

5.3.5  Natural capital? Which leaves us with the problem of ‘natural capital’. Some natural assets have a clear market value or, like the Zuid-Kennemerland National Park in the Netherlands, which will be analyzed more fully in the chapter about consumption, a clear cost of production. Others haven’t. It is of course possible to call nonowned non valued items, like whales or even a sunset unspoiled by windmills standing between the beach and the horizon, capital. Basically, the word ‘capital’ is used as metaphor based on the idea of capital as a factor of production, not of distribution, to denote the production of use values by ‘nature’. Sometimes, it is possible to put a market price on these use values. Efforts to do this abound, which is remarkable. The word capital, as used by economists, has an impeccable monetary background and was used, in the middle ages, to denote the principal sum of a money loan (compare: ‘raising capital’). The concept has been extended to all kinds of ‘principals’ which yield returns of a monetary or even non-monetary nature. The best known example is ‘human capital’. People invest in themselves, for instance by having an education. Should we proceed in this direction with the idea of ‘natural capital’? In the case of natural capital, often

Capital (and land) 163 the concept of monetary value is used as the yardstick by excellence to measure the value of something which is often understood to be quintessential nonmonetary and even non-human: nature. In an excellent overview of ‘natural capital accounting’ the (Australian) Bureau of Meteorology (BM), highly interested in water accounts, defines natural capital, as: ‘The stock of living and nonliving components of the earth that provide a flow of valuable ecosystem goods or services’ (BM 2013). It is even possible to find scientific articles which) include the sun in our concept of ‘capital’ (Monfreda, Wackernagel and Deumling 2004). Despite Diogenes and his use of the exclusivity argument: I do not approve. The national accounts restrict ‘capital’ to private, government or institutional ownership and to identifiable future monetary benefits or historical production costs while ownership is associated to certain rights to these benefits. Whatever you think of it – it’s a monetary concept. This also means that ‘capital’ (and therewith the distribution of income) has a clear legal and political side to it – a point which might not always be appreciated enough by people defending the idea of ‘natural capital’. Putting a price on these flows of ecosystem goods and services might soon lead to the discovery that the most profitable way to exploit nature is to build nice houses. Or as it is 2019, ‘natural’ graveyards. Like the burial ground of the Irish ‘Green Graveyard Company’, people who ‘hope to open more Natural/Woodland burial grounds in the coming years’ (Greengraveyard.com 2019). As stated earlier, nowadays even dust become capital (or at least a land value enhancing commodity). The ecosystem flows of goods and services are often also defined in non-monetary ways, for instance related to biodiversity or the amount of walking trips. But there always is a reason why there is no price for these items. Putting a price on these non-monetary variables might be a step to changing the very thing which is crucial to such use values: non-monetary valuation. It is important to define ‘national accounts’ capital and especially unproduced natural resources with clear ownership in relation to the total stock of ‘living and nonliving components of the earth’. The ESA 2010 might well be rewritten in this regard as we do spend a lot of money to (re-)create (or at least to try to re-create) non-monetary assets, like clean air and clean water. But there are also very good reasons to separate ‘monetary’ assets, natural or not, from non-monetary assets. Money matters – which is a reason to distinguish monetary assets from nonmonetary variables. Capital is property. Calling a beach sunset ‘natural capital’ is a step toward commodifying sunsets. Returning to natural capital, this of course also means that depletion of stocks of the natural assets, like clearing of pristine forests to turn them in agricultural land or to build lodges and burial grounds for rich tourists, which are included as a positive in the accounts, should be a negative when we calculate eco-enhanced GDP. Likewise with the depletion of natural resources with a market price, like natural gas, which would also enhance the stock-flow consistency of the balance sheets. This is not a new point. King et al. (1921) already stated that the depletion of natural resources should be included in estimates of national income. There is, however, a difference between the monetary and the non-monetary realm. A blunt example: a well-defined market price for sexual intercourse exists. In theory, it should be included in the

164  Capital (and land)

Ecological theory: biodiversity, nonmonezed ecosystem characteriscs and values as well as services and use values; monezed ecosystem services and use values (cost price, market price or esmated price). So and hard individual and collecve ownership and use rights of different parcipants, clear difference between produced and unproduced ‘natural’ resources

Environment

Society

Economy

Economic accounts: naonal accounts, Flow of Funds, Balance Sheets, Input Output models for physical inputs and outputs including ‘waste’

Environmental accounts, in case of clear ownership rights and monezed values of either flows of income or costs of maintenance esmates of capital and ‘land’ and sectoral balance sheets

Figure 5.1  The relation between the national accounts and ‘natural capital’ Source: Adapted from Bureau of Meteorology 2013

eurozone consumer price index. But should we really use this price to value nonmarket, non-monetary intercourse and add this estimate to GDP?4

5.4  Capital and the neoclassicals 5.4.1  The concept, implicit or otherwise We’ve seen how the statisticians define and value capital. Is this consistent with the way neoclassical macroeconomists do this? One difference, the stock of capital is treated as a physical and not a monetary object. Sims (2016), a teaching text aiming to teach graduate student the basics about DSGE models with capital, is explicit about this. Second, it is supposed to be a ‘perpetual inventory’ item, not

Capital (and land) 165 in the monetary sense of accounting but in a real sense. The stock of capital is a remnant of past decisions and will linger on to influence our decisions in the future. It is equal to the old stock minus depreciation plus investment while heed is taken of ‘investment costs’ needed to change the consumer good into a capital good (we have to plant the coconuts). Sims writes about the accumulation of capital and mentions the perpetual inventory formula. I could not find it in his text. But an advanced one, which boils down to a perpetual inventory formula, can be found in Burriel, Fernandes-Villaverde and Rubio-Ramirez (2009, p. 10). Central to the value (or should I write: the amount) of capital in the models is the interest rate in combination with productivity. Or maybe I should put this the other way around: the stock of capital basically evolves via a ‘balanced growth path’ which is set out by the interest rate, consumer preferences and the productivity of capital. Burriel, Fernández-Villaverde and Rubio-Ramirez assume a steady increase in productivity of the physical capital good but do not specify what this means for the consumer good (which basically is the same good as the capital good). The next excerpt from Stack Exchange, a site where people can ask questions about (neoclassical) macroeconomics, explains this. First, the question (source of the quotes: Stackexchange 23 April 2018): In New Keynesian models, like the ones in Gali’s simple New Keynesian model or even Mankiw-Reis NK model on sticky information, capital is often not included. Now people do say that there are modeling difficulties and that’s why capital (K) is not included, but is there another justifiable reason? Part of the answer was, ‘Capital is included in all the big estimated New Keynesian models’. But also: ‘you’re absolutely right that the stylized core NK model does not have ­capital – which is hard to defend on empirical grounds, since capital investment is a very important part of business cycle fluctuations and the response to monetary policy’. Note that this answer is about the importance of investment, not about the importance of capital itself. The reasons provided to include it are enlightening: ‘the two core equations (the intertemporal Euler equation and New Keynesian Phillips curve) of the ordinary log-linearized NK model are completely forward-looking. Adding K to the mix eliminates this nice analytical feature’ and ‘seemingly small changes in the real interest rate must be accompanied by massive swings in the capital-output ratio, which we never see in practice’ and ‘Capital adjustment costs are needed to avoid absurd results’. What does this all mean? Let’s first compare capital in the models with a fixed rate bond. When the interest rate declines, the price of a fixed rate bond goes up. Something comparable will happen with the price of houses. When mortgage interest rates decline and mortgage credit is unbounded, the price of houses will tend to increase as borrowing becomes cheaper. Capital goods in the model

166  Capital (and land) however do not have a monetary value, they are an amount of a good. The same holds for the interest rate: it is a physical interest rate. Even then, it is supposed that the relative amount of capital will behave as if capital is a bond: when interest rates decline, its value will go up which, as this cannot come about by price changes, will have to happen by overnight swings in volumes. These swings are huge. A decrease of the interest rate from 3 to 2 percent is a decrease of on third or 33 percent. As the interest rate is supposed to be connected to the productivity of capital, this boils down to, in the models, a decrease of the productivity of capital of 33 percent. That’s where the capital adjustment costs come in: these prevent this from happening overnight. The volume of capital behaves more or less like the nominal value of bonds. One could think that the very act of investing (making plans, realizing these) already takes time. Time preferences (habit formation in the models, we will come to that) instantaneously change. Even then, such processes take time – often decades. Aside from this the concept of capital in the models is more limited than the statistical concept. The concepts assume a general unspecified good which at no cost can be turned into fixed capital or consumption goods. However – fixed capital owned by the government is understood to be wasteful by definition and hence not included in the stock of capital. Also, most models exclude consumer durables from consumption and, hence, from their concept of capital. Should durable goods be included in the neoclassical concept of capital? Models with durable goods exist (Monacelli 2009; Patroba 2015). The reason for developing these is the well-known fact that expenditure on durable goods like cars is much more pro-cyclical than expenditure on food or local home-work travel. The models state that the utility of durable goods is equal to the depreciation rate of these goods and state that durable goods can be used as collateral for borrowing – which in the cases of especially cars and houses is true. It is not clear if houses are part of the stock of durable goods in the articles mentioned, it is also not clear if the authors take account of the implication that when the consumption boundary is extended to durable goods they have to shift the production and investment boundary, too. But it is a good thing that models which aim to explain the business cycle also incorporate the most volatile part of consumer spending. It is however not clear from the models if consumer durables are ‘capital’ or not. Summarizing: the ‘capital boundary’ of the models is not entirely clear. The very nature of capital in the models is also not entirely clear. The models treat fixed capital as a kind of ‘jelly’ which can be readily substituted into other kinds of fixed capital. Sometimes, this makes sense: swords into plowshares or clunkers into scrap into military equipment. Sometimes, it doesn’t: software into ‘Land’? It is also assumed that fixed capital can be changed into consumer goods without effort. Sometimes this makes sense. But not everything is malleable. Cars can be used by companies and by consumers. Companies like Uber enable a relatively smooth transition of cars from the household realm to the transactions space. But sometimes, it doesn’t. Large planes are not easily converted into solar cells. Ideas about the malleability of capital were for quite a time in vogue in macro when ‘vintage models’ of ‘jelly-clay’ capital goods were used in for instance growth

Capital (and land) 167 models. Such ideas were vehemently criticized – even when physical fixed capital stays the same its role can be changed because technological progress and changes in interest rates and prices may change its economic value and function. Responding to this, Solow et al. (1966) shows that neoclassical economic models do not necessarily need the malleability assumption. Nowadays, these ideas have been discarded. Fifty years after the Solow article DSGE models still use the malleability assumption. Why? Anyway, returning to cars Graph 5.3 shows that the worldwide number of cars has reached almost one billion, while the logistic pattern of the curve seems to indicate that the car industry is in the growth and not the maturity phase. Not all of these cars are owned by households but it does indicate that consumer durables should be taken seriously as capital.5

5.4.2  Back to the way ahead We can question where the ideas that fixed capital is malleable come from. To answer the question we have to go back in time. Neoclassical economists purged ownership based ‘class’ and therewith distribution from their models. A first defining moment is the work of John Bates Clark who, to actively counter the ideas of Henry George (Gaffney 1994), purged ‘land’ and unproduced inputs from the neoclassical concept of capital by focusing on the liability instead of the asset side of the balance sheet and the fact that the total value of liabilities was ‘eternal’ and not dependent on the fixed assets in question, which (except land)

1,400,000

1,200,000

1,000,000

800,000

600,000

400,000

200,000

-

1960

1970

1980

Car registraons

1990

2000

Truck and bus registraons

2005

2010

World total

Graph 5.3  Worldwide vehicle registrations, 1960–2016 Source: https://en.wikipedia.org/wiki/Motor_vehicle, accessed 06–05–2019

2016

168  Capital (and land) would wither away anyway (Clark 1899, IX.7). As he also introduced a kind of representative liabilities owner, differences in ownership of capital and the obvious consequences for distribution were also purged from the model. Clark was criticized for using a ‘jelly’ concept of capital already in 1907 by Böhm-Bawerk (as quoted in Cohen and Harcourt 2003). Reading him this does not seem just: subsequent economists, not Clark, mixed up the asset and the liability side of the balance sheet. He is only guilty of using the marginalist idea that economic classes (for instance: land owners) do not matter and replacing owners and non-owners with one representative capital owner while also stating that it is the liability side which really matters but without totally discarding the assets side. And the posts on the liability side are of course highly substitutable and ’eternal’ (also a main point of Piketty 2014a). Distribution becomes by definition a non-issue when one, like Clark, defines society as a kind of ‘representative consumer’. When a kind of ‘representative consumer’ which stand for society by definition owns all the capital he/she also is, by implication, as well the economic as the legal owner while the distribution of ownership does not matter. Also, debts and financial assets are purged from our economic universe as we can look at the consolidated balance sheet (as all debts are a debt of the representative consumer to the same representative consumer). But Clarks’ concept of capital still enabled one to pose questions about distribution as it was still based on a crude idea of a balance sheet. The idea of different sectors or classes was purged from the model but the idea of ownership wasn’t. I could also find no mention of the ‘universal asset’ idea of fixed capital used by Solow (1956) in the work of Clark (which I did not read in its entirety).6 But Clark’s idea that, as firms can substitute fixed assets for others, society can do so too is a fallacy of composition. Firms can divest or purchase existing assets (though even these possibilities are limited) but on a national scale this does not lead to a change in the physical composition of the stock of capital. Changing your analysis from ownership of specific assets to the grand total of the liability side doesn’t change this. We’ve already seen that, taking a historical view, it is quite complicated to make an estimate of the volume of fixed assets. Only looking at the liability side doesn’t solve this problem. Clark did however enable the neoclassical community to look away from differences between owners of certain kinds of capital and non-owners. Interestingly, authors like Bokan et al. (2016) and Rognlie (2015) reintroduce differences in ownership (of houses as well as fixed capital) and hence economic class respectively land into neoclassical macro-discourse. The idea of the representative consumer is scuttled. As soon as this is coupled with the use of balance sheets, as in Piketty (2014a), this will bring questions of distribution and about the nature of ‘Land’ and ownership to the fore, again. A second defining moment was the publication of the growth theory of Solow (Solow 1956). He did look at fixed assets. Unlike Clark, Solow explicitly rejected ‘land’ from his analysis and understood fixed capital as a ‘one good’ concept to solve the problems of composition mentioned.7 He had some empirical reasons to do this. His article was published when ‘land’ had reached a historical minimum as part of the stock of capital (see the data in Piketty (2014a)). A theoretical reason to do this was however that one of the basic rules of accounting is that

Capital (and land) 169 land itself does not depreciate. Including a non-depreciating non-produced kind of capital in his model would have played havoc with the way his model crucially relates the stock of capital and the capital/output ratio to investment and depreciation. Another reason to exclude ‘Land’ is that supposing the existence of one good which can be both an investment and a consumer good enables the economist to deflate the stock of fixed assets with the consumer price index. None of these reasons is very convincing. Solow however enabled economists to look away from Land and fallacies of composition, while the ever larger importance of legal unproduced assets was less of an issue in 1956 than it is today. A third defining moment was the fall out of the ‘battle of the Cambridges’. After World War II economists increasingly focused on estimates of the stock of fixed capital and growth theory. A side show of this tendency was this confusing discussion which in the end was about the obvious (but not very neoclassical) fact that when you have an existing stock of fixed capital and interest rates change the existing stock of fixed capital (with concomitant capital:labor ratios) won’t change immediately while existing assets can also be used in another way. The composition of the existing stock of capital does matter in that case. Even when oil prices decline and interest rates rise, many existing oil wells will keep producing even when new comparable rigs are not profitable anymore. The practical solution to this problem was to get rid of ‘jelly’ and to use so called putty-clay models, which used different vintages, every vintage with its own labor/capital ratio (see also footnote 17 in Stiglitz 2006). Solow et al. (1966) shows that it was possible to adapt neoclassical ideas to this criticism. DSGE models do however not take account of this research and still use one kind of good which is fit to be a consumer good as well as a consumption good and which does not change over time. A fourth defining moment was Samuelson’s definition of public goods and, hence, the introduction of government fixed capital into the corpus of neoclassical economics. In the course of the 20th century, government investment and government owned capital goods had become ever more important. The Hoover dam in the USA, the Millau bridge in France (built, financed and exploited by the Eiffel company but owned by the government), the highways in Germany but also the stock of nuclear weapons are iconic examples. Samuelson’s theory of public goods incorporated such investments into the corpus of neoclassical ideas. DSGE models as a rule however exclude them. A fifth defining moment was the introduction of neoclassical DSGE macromodels (first originated as real business cycle models) which in the cases in which they model capital at all discarded government owned produced and non-produced capital, did not use the putty-clay models, discarded double accounting methods (let alone quadruple accounting methods) and did not look at durable consumer goods. In almost all DSGE models the government is an expensive set of monetary rules and a redistribution mechanism and government expenditure, including government investment in dams, dikes and roads, is considered to be wasteful by definition. As it diminishes the amount of goods available for consumption or private investment. The ‘putty clay’ models of existing capital as well as the idea of government capital were abandoned without even discussing them. And models using the

170  Capital (and land) single representative consumer of course exclude capital as a factor of distribution by definition. This leaves modern neoclassical macro-models with a limited set of capital goods: private, produced fixed capital with a ‘jelly’ structure which can be rented. To quote Bokan et al. (2016): ‘Final goods can be used both for private consumption and investment’, not that final goods exclude consumer durables and government owned investment goods while the model at the same time states that all final goods are the same. There is no real difference between a plane and a haircut. This enables ‘total substitutability’ between investment goods and consumer goods. Bokan et al. however do introduce two kinds of fixed capital into their model and have to be lauded for this: real estate as well as ‘other fixed capital’. ‘Households and entrepreneurs demand real estate, which is assumed to be nontradable across countries and in fixed (per capita) aggregate supply’ (Bokan et al. 2016). The non-tradability of houses across countries limits the substitutability, which enables Bokan et al. to model asset price bubbles caused by international flows of capital. A most remarkable aspect of Bokan et al. is that introducing ownership of capital into their model also necessitates them to introduce economic classes – called households (which have nothing to sell but their labor) and entrepreneurs (which own all fixed capital as well as a lot of real estate and which, as they own all the fixed capital, ‘capitalists’). As real estate is distinguished from other kinds of capital this has a ‘land’ element to it. They hence and as far as I know unwittingly reintroduce two ideas that Clark and other marginalists had purged from economics. It makes their model clearly less neoclassical and more old-classical.8 Households are subdivided into ‘patient’ households (which lend deposits (not money?) to the banks) and ‘impatient’ households (who are borrowing existing deposit from the banks). Aside from this there are some bankers. Capital as well as real estate is used in a Cobb-Douglas production function which relates the model to growth theory. In technical terms the authors seem to have a putty-putty as well as a putty/clay model of capital in the sense that normal fixed capital seems to be totally substitutable while real estate has a fixed relation to labor and does not seem to be substitutable at all.

5.5  A comparison In 1966 Goldsmith wrote: After considerable hesitation by many of our traditionally inclined and appropriately cautious colleagues, the desirability and feasibility of an integrated comprehensive system of national accounts, which includes a balance sheet as a necessary component, seems now to be generally accepted. (Goldsmith 1966, p. 96) Fifty year later, his prediction came true when Piketty and Zucman wrote ‘Building upon recent progress made in the measurement of wealth, and pushing forward Goldsmith’s pioneering attempt, Piketty and Zucman (2015) construct aggregate wealth and income series for the top eight rich economies’ (Piketty and Zucman 2015, p. 1308). The next comparison might yield some idea why it took so long to move the ideas of Goldsmith forward:

National Accounts Basic method of aggregating data

Basic method of valuation

Contains not owned natural capital (including human capital)

Contains owned ‘unproduced’ natural capital, like land and subsoil stocks of oil Contains owned ‘unproduced’ legal capital, like production permits Contains government owned produced capital, like coastal defenses Contains specific balance sheets for all economic sectors Contains accounts for financial monetary institutions and Non Profit Institutions Serving Households (NPISH) owned produced capital

Neoclassical Macro-Models

Quadruple accounting for a complete set of sectors and fixed as well as financial assets.

Basically single accounting. Inclusion of a financial sector by necessity leads to more emphasis on double and even quadruple accounting even when this is not yet economy wide regarding assets and sectors Model consistent Using market prices, valuations, preferably historical cost prices discounting of expected minus depreciation or future financial returns or estimates of replacement future utility returns. costs. Discounting future financial flows is the method of last recourse. No No. Efforts are made to define human centered ‘ecosystem services’, to add a shadow price to these services and to discount them. Ecosystem services can be negative, as in the case the climate effects of rising CO2 Yes No, efforts are however made to do this. Yes

No

Yes

Generally: no.

Yes

No

Yes

No

(Continued)

(Continued)

A distinction between capitalist owners (‘entrepreneurs’) and labor, or households which only sell labor, exists Nature of fixed capital

National Accounts

Neoclassical Macro-Models

No

Yes (in some models)

Heterogeneous with regard to composition and depreciation rates, detailed classification of items exists. Capital can only in special cases be used for household consumption. Defining criterion is ‘possible future economic benefits’ which include resale value and production costs foregone. Measured or derived Measured amalgam of price of existing capital cost prices, perpetual inventory methods, replacement prices and the market price of items which are sold on the second hand market. Only a limited relation with investment prices. Partly. Autonomous Stock-flow consistent price changes of assets with production (houses, sub-soil assets) accounts? are excluded and declines of the stock of natural assets are not subtracted from GDP. Price changes sometimes overwhelm investment as in the case of houses. Fundamental measurement issues when intertemporal comparisons are made. Yes (but measurement Sectoral consistent problems with net (sectoral balance sheets match with each international investment other) positions, which fall outside the scope of this book) Nature of financial Money creating banks plus market loanable funds in the case of non-monetary financial institutions

Except for distinction in some models between real estate and other capital: homogenous. All capital can be used for household consumption instead of production. Capital yields rents, no other monetary benefits are acknowledged.

Derived mark-up on consumer prices (mind that capital and consumer prices are basically the same good) influenced by borrowing behavior and rules; consistency with discounted flow of utility. To a limited extent. There is no unproduced capital, but prices changes because of exogenous changes in the financial market do exist.

Theoretically: yes. But sectoral division is incomplete (government, NPISH and monetary financial institutions are generally excluded). Loanable funds, in rare cases including international flows of capital.

Capital (and land) 173

Notes 1 A quick scan of websites on which second hand agricultural machines are sold showed that the often the amount of ‘machine hours’ or ‘hectares harvested’ were mentioned. 2 The ‘7.26’ is a lemma of the ESA 2010 manual. 3 As the national accounts basically define fixed assets as a factor of distribution it seems right to me to include production permits and the like into the concept. R&D is however a ‘sunk cost’ as well as, in business accounts, not treated as an investment. It might yield patents which can be included in our concept of capital. But R&D itself should be excluded even when it yields an array of small but significant improvements in quality or productivity. 4 These problems might also be understood as basic characteristics of the dynamism of our economy, which to me seems a more fruitful way to think about them. A good example of the insights an analysis of these developments yields: Lafrance 2016. 5 Consumer durables do have a second hand value and can be included, as happens in the USA flow of funds statistics. 6 He introduced or at least used the idea of the representative consumer to be able to disregard the distributional consequences of unequal ownership of wealth. 7 The concept of ‘land’ sometimes leads to confusion. It relates to ‘unimproved’ land and is the ‘location, location, location’ value of real estate or the value of real estate without the value of the building 8 Classical economists, including Marx and Mises (in his PhD thesis), used an economic definition of classes. Your economic position (laborer, capital owner) and not for instance your education, profession and income define your class.

Literature Acemoglu, D. and J.A. Robinson (2008). ‘Persistence of power, elites, and institutions’. American Economic Review 98:1 267–293. Ager, P., L.P. Boustan and K. Eriksson (2019). ‘The intergenerational effects of a large wealth shock: White southerners after the civil war’. NBER working paper series no. 25700. Ayres, C.E. (1944). The theory of economic progress. Chapel Hill: The University of North Carolina Press. Bergen, D. van den, M. de Haan, R. de Heij and M. Horsten (2009). ‘Measuring capital in the Netherlands. Paper prepared for the OECD Working Party on National Accounts, Paris, 11–14 October 2005’. CBS discussion paper no. 09036. Bokan, N., A. Gerali, S. Gomes, P. Jacquinot and M. Pisani (2016). ‘EAGLE-FLI. A macroeconomic model of banking and financial interdependence in the Euro area’. European Central Bank working paper series no. 1923. Borio, C. (2012). ‘The financial cycle and macroeconomics: What have we learnt?’. BIS working paper no. 395. Bureau of Meteorology (2013). ‘The environmental accounts landscape’. Environmental information program publication series no 1. Canberra. Cantoni, D., J. Dittman and N. Yuchtman (2017a). ‘The religious roots of the secular West: The Protestant Reformation and the allocation of resources in Europe’. Blogpost on Voxeu, 31 October 2017. https://voxeu.org/article/ protestant-reformation-and-allocation-resources-europe Cantoni, D., J. Dittman and N. Yuchtman (2017b). ‘Religious competition and reallocation: The political economy of secularization in the protestant reformation’. Working paper.

174  Capital (and land) CBS (1947). ‘Uitkomsten van enige berekeningen betreffende het nationale vermogen van Nederland in 1938’. Statistische en econometrische onderzoekingen no. 3. CBS (6 April 2012). ‘Personenauto’s rijden gemiddeld 37 kilometer per dag’. Press release. www.cbs.nl/nl-nl/nieuws/2012/10/personenauto-s-rijden-gemi ddeld-37-kilometer-per-dag Clark, J.B. (1899). The distribution of wealth. A theory of wages, interest and profits. New York: The MacMillan Company. Cohen, A. and G. Harcourt (2003). ‘Whatever happened to the Cambridge capital theory controversies’. Journal of Economic Perspectives 17:1 199–214. Copeland, M.T. (1912). The cotton manufacturing industry of the United States (2nd impression). Cambridge: Harvard University Press. Dostojevsky, F. (1917 [1866]). Crime and punishment. New York: P.F. Collier & Son (original Russian edition 1866). Bartleby.com, 2000. www.bartleby.com/318/. Engels, F. (1952 [1845]). Die Lage der arbeitende Klasse in England. Nach eigner Anschauung und authetische Quellen. Berlin: Dietz. Eurostat/European Commission (2013). European system of accounts ESA 2010. Luxembourg: Publications Office of the European Union. Finley, T., R. Franck and N.D. Johnson (2017). ‘The effects of land redistribution: Evidence from the French Revolution’. George Mason University department of economics working paper nos. 17–29. Fosfaatrecht.nu (2019). https://fosfaatrecht.nu/?gclid=Cj0KCQiAzePjBRCRAR IsAGkrSm5tax-Yy51QJqDPhlIHYvaSWI21N5d2f0foPKCokT7zVCSDa3JmgGQaAlrJEALw_wcB. Accessed 2 February 2019. Gaffney, M. (1994). ‘Neoclassical economics as a stratagem against Henry George’ in: M. Gaffney and F. Harrison (eds.) The corruption of economics 29–164. London: Shepheard-Walwyn. Goldsmith, R.W. (1966). ‘The uses of national balance sheets’. Review of Income and wealth 15:2 95–133. Greengraveyard.com (2019). www.greengraveyard.com/. Accessed 4 February 2019. Groote, Peter (1995). Kapitaalvorming in infrastructuur in Nederland 1800– 1913. Capelle aan den IJssel: Labyrinth publication. Heblig, S., S. Redding and D. Sturm (13 October 2018). ‘The making of the modern metropolis: Evidence from London’. Voxeu. https://voxeu.org/article/ making-modern-metropolis-evidence-london Hudson, Michael (2012). Finance capitalism and its discontents. Dresden: Islet. Jordà, O., M. Schularick and A.M. Taylor (2014). The great mortgaging: Housing finance, crises and business cycles. NBER working paper no. 20501. King, W., F. Macaulay, W. Mitchell and O. Knauth (1921). Income in the United States: Its amount and distribution 1909–1919, volume 1: Summary. New York: National Bureau of Economic Research. Knibbe, M. (2006). Lokkich Fryslân, Landpacht, arbeidsloon en landbouwproductiviteit in het Friese kleigebied, 1505–1830. Groningen: NAHI. Knibbe, M. (2014). ‘The growth of capital: Piketty, Harrod-Domar, Solow and the long run development of the rate of investment’. Real-world Economics Review 69 100–121. Koning, N. (2017). Food security, agricultural policies and economic growth. Longterm dynamics in the past, present and future. London and New York: Routledge. Korn, B. and Th. D. van der Weide (1960). ‘Het nationale vermogen van Nederland, 1948–1968’. Statistische en econometrische onderzoekingen 1960:3 63–92.

Capital (and land) 175 Koster, H., J. van Ommeren and N. Volhausen. ‘Short-term rentals and the housing market: Quasi-experimental evidence from Airbnb in Los Angeles’. Blogpost on Voxeu, 20 December 2018. https://voxeu.org/article/shortterm-rentals-and-housing-market Lafrance, A. (2016). ‘A history of technology, via the consumer price index. The story of America’s relationship with appliances and gadgets, as seen in seven decades of government statistics’. The Atlantic, 5 April 2016. www.theatlantic.com/ technology/archive/2016/04/bls-data-tech/476763/ Leenders, W. (2016). Wealth income ratio’s in the Netherlands. Unpublished bachelor thesis. Linsi, L. and D.K. Mügge (2019). ‘Globalization and the growing defects of international economic statistics’. Review of International Political Economy. doi:10.1080 /09692290.2018.1560353. Luijt, J. and M. Voskuilen (2009). ‘Langetermijn ontwikkeling van de agrarische grondprijs’. LEI nota 09–014, LEI Wageningen UR, Den Haag. Marx, K. (1887 [original German edition 1867]). Capital. A critique of political economy volume I. Book one: The process of production of capital. Moscow: Progress Publishers. www.marxists.org/archive/marx/works/download/pdf/Capital-Volume-I.pdf Monacelli, T. (2009). ‘New Keynesian models, durable goods, and collateral constraints’. Journal of Monetary Economics 56:2 242–254. Monfreda, C., M. Wackernagel and D. Deumling (2004). ‘Establishing national natural capital accounts based on detailed ecological footprint and biological capacity assessments’. Landuse Policy 21 231–246. Nieuwkerk, M. van and R.P. Sparling (1985). ‘De internationale investeringspositie van Nederland’. DNB, Monetaire monografieën nr. 4. Nieuwkerk, M. van and R.P. Sparling (1987). ‘De betalingsbalans van Nederland: Methoden, begrippen en gegevens (1946–1985)’. DNB, Monetaire monografieën nr. 7. Patroba, H. (2015). ‘New-Keynesian DSGE model, durable goods and collateral constraint in a small open economy’. Working paper. http://2015.essa.org.za/fullpa per/essa_2815.pdf Piketty, T. (2014a). Capital in the twenty first century. Cambridge (MA): Harvard University Press. Piketty, T. (2014b). Technical appendix of the book ‘Capital in the twentieth century. Harvard: Harvard University Press. http://piketty.pse.ens.fr/capital21c’. Accessed 10 July 2014. Piketty, T. and G. Zucman (2015). ‘Wealth and inheritance in the long run’. Handbook of Income Distribution 2 1306–1364. Amsterdam: Elsevier. Rognlie, M. (2015). ‘Deciphering the fall and rise in the net capital share’. Brooking papers on economic activity. Rossum, M. van and O. Swerts (2011). ‘De Nederlandse aardgaswinning’ in: CBS (ed.) De Nederlandse economie in 2010 232–254. http://silverberg-on-meltdowneconomics.blogspot.nl/2014/06/nitpicking-piketty-productively-part-i.html Solow, R.M. (1956). ‘A contribution to the theory of economic growth’. The Quarterly Journal of Economics 70:1 65–94. Solow, R.M., J. Tobin, C.C. von Weizsacker and M. Yaari (1966). ‘Neoclassical growth with fixed factor proportions’. The Review of Economic Studies 33:2 79–115. Stackexchange. http://economics.stackexchange.com/questions/1658/why-is-cap ital-often-not-included-in-new-keynesian-models-is-there-a-reason-oth. Accessed 23 April 2018.

176  Capital (and land) Stiglitz, J. (2006). ’Samuelson and the factor bias of technological change: Toward a unified theory of growth and unemployment’ in: M. Szenberg et al. (eds.) Samuelsonian economics and the twenty-first century 235–251. Oxford: Oxford University Press. Tinbergen, J. (1942). ‘De groei van den voorraad van eenige kapitaalgoederen in zes landen vanaf omstreeks 1870’. Maandschrift van het CBS 113–121, 296–299, 497–509. Veldhuizen, E., C. Graveland, D. van der Bergen and S. J. Schenau (2009). Valuation of oil andgas reserves in the Netherlands 1990–2005. Voorburg/Heerlen: CBS Verbiest, P. (1997). ‘De kapitaalgoederenvoorraad in Nederland’. CBS Nota M@O 007. Verstegen, S. (1996). ‘National wealth and income from capital in the Netherlands 1805–1910’. Economic and Social History in the Netherlands 7 73–108. World Economics Association (2019). www.worldeconomicsassociation.org/news letterarticles/models-and-measurement-3/ Accessed 5 February 2019.

6 Consumption

6.1  Introduction: the concept Meet the Zuid-Kennemerland National Park (Nationaal Park Zuid-Kennemerland 2018), or ‘Kennemer Duinen’. A nature reserve and Dutch national park of 3.800 hectare, consisting of dunes, woods and a tiny bit of beach located about 20 kilometers west of Amsterdam, stretching along the Dutch coast. It is easily accessed by bike, car, foot and even horse. Some patches of it are owned by central, provincial and local governments, some others by individuals and quite a chunk by the largest Dutch private land owner, the charity ‘Natuurmonumenten’. It is managed by a consortium of Staatsbosbeheer (a state entity managing wildlands with a legal obligation to provide access to the public), Natuurmonumenten and ‘PDW’, the land management organization of North-Holland (a Dutch province). People go for a walk, a bike ride, horse driving, mountain biking or swimming. It’s free (of charge). According to the rules of national accounting, visiting this place is consumption. But as people do not have to pay a ‘meaningful economic price’ to get in visitors are not counted and market prices can’t be used as a yardstick. So, how do we value this kind of consumption? As a rule, the national accounts use maintenance costs and not for instance the happiness or sense of place it provides to value it. This contrary to procedures used in recent endeavors to calculate the value of ‘natural capital’ (ONS 2018) which do try to put a monetary value upon happiness, love and purpose. To be more precise, the national accounts stipulate that maintenance costs of these lands are part of ‘actual individual consumption’ (AIC) of households, a variable which lumps household expenditure on consumption goods and services together with expenditure of the government and Non Profit Institutes Serving Households (NPISH), like Natuurmonumenten, on public goods and services which are individually used. This shows that, according to the statisticians, consumption is more than buying a sausage in a supermarket. Of course, people love areas like the ZuidKennemerland National Park as well as other government owned nature reserves. When the Rutte II government of the Netherlands tried to privatize (read: enable real estate development) part of the Dutch dunes along the shore massive popular protests caused a major setback for the government. Across the North Sea, massive popular protests also caused the first mayor political setback of the

178  Consumption conservative Cameron government of the UK when they launched plans to privatize government owned nature. And recall the fuzz created, across the Atlantic Ocean, by the endeavors of the Trump government of the USA to sell the larger part of the 547,000 hectares of ‘Bears Ears’. This is a nature reserve in the USA co-managed by the Bureau of Land Management and United States Forest Service, along with a coalition of five local Native American tribes; the Navajo Nation, Hopi, Mountain Ute, Ute Indian Tribe of the Uintah and Ouray Reservation, and the Pueblo of Zuni. At the moment of writing, law suits have been filed to stop the sale: don’t Privatize My Nature! But national accounts don’t care about emotional attachments. They care, when no ‘meaningful economic prices’ are charged, about the actual cost incurred in maintaining the area and classify these costs as: AIC. Just as the national accounts would classify real estate development in the Dutch dunes and mineral extraction in Bears Ears as investment and production, activities which both would increase GDP a lot more than what would be lost as AIC. ‘Consumption’ as defined in the national accounts is about monetary transactions, not about welfare, nature and prosperity. But these transactions also comprise health care, education and even sermons by priests and pastors – as long as these are paid. They do acknowledge the existence of nonmarket monetary consumption activities – even when the many volunteers working in the Zuid-Kennemerland National Park are not counted or acknowledged by the accounts. Did I mention that Bears Ears is about 150 times larger than the Zuid-Kennemerland National Park and twice as large as the combined surface of all the lands of the National Trust, the largest nature charity in the UK? Consumption (AIC) is quantitatively the most important expenditure category of the national accounts. Also, modelers and statisticians alike seem to agree about the dictum of Adam Smith: ‘consumption is the sole end and purpose of all production’. But in most DSGE models public goods and services and durable goods are excluded. And there are more differences between the statistics and the models. The national accounts have a well-articulated, granular and estimated definition of consumption. The models have an implicit and fuzzy ‘one good’ concept of consumption. Also, ‘household consumption’ as defined by the national accounts can, as far as it constitutes the purchase of market goods, be considered to be the transaction which transfers goods and services from the monetary economy to the non-monetary realm of home, household and family – a sphere absent from the models. The crossing of this border is less clear for public goods and services, which include items like defense and education which are not directly used in the household. These items are part of the monetary economy. How does this all come together? How do statisticians operationalize and measure it and how do the modelers look at this? These questions are central to the next paragraphs.

6.2. Consumption in the national accounts: definitions and operationalizations The national accounts define consumption as: ‘final consumption expenditure consists of expenditure incurred by resident institutional units on goods or services that

Consumption 179 are used for the direct satisfaction of individual needs or wants or the collective needs of members of the community’ (ESA 2010 lemma 3.94). • This community can be the residents of a country, even when consumption of people on vacation abroad count as an export of the destination country and consumption of country of origin. When people are permanent residents of another country, their consumption is supposed to be part of the consumption of the host country. The community can however also be members of a church or a sports club. • Consumption is all about expenditure. Use-value of goods and services is central to the concept of consumption. But what’s measured is not the use or use value of goods and services but monetary transactions: purchases of cars, government outlays for the provision of education or maintenance costs of national parks. • The resident units mentioned in the definition can be households but also companies (for instance when these provide meals for employees which, in low wage countries, can be a considerable part of total remuneration), NPISH or the government. • When analyzing or interpreting such data it is important to note that the production/consumption boundary changes all the time, the surge in room and house letting facilitated by Airbnb being an example. It increases on the production and income side household production and income (both items in the household sector accounts) as well ass, when the people renting a room are foreigners, on the expenditure side national exports but, in the case of permanent residents, household consumption. • The main ‘residential institutional units’ purchasing goods and services for the benefit of households are households themselves, the government and NPISH. In Graph 6.1, the NPISH have been left out. As can be seen, historical changes are large. Remember: this is about food, rent, healthcare, vacation and education. Collective government consumption, like defense and the providing of law and order. Individual government consumption (called expenditure on goods and services here) includes education and large parts of healthcare. Household consumption is expenditure of households on durable and non or semi-durable goods (these last two categories are distinguished in the national accounts. One large item which is not a monetary transaction but an imputed post is the imputed monetary value of services of owner occupied dwellings, by far the largest exception to the rule that the accounts only measure the monetary economy. If possible, ‘equivalent market rental prices’ are used to value the service of these owner occupied houses. As there are several reasons why people do not rent a house, price only being one, a case can also be made that production costs like maintenance costs and interest paid or foregone or a ‘hedonic price’ should be used to value the consumption value of owner occupied houses. A ‘hedonic price’ means that prices should be adapted to account for different characteristics of owned houses, including the fact that they

180  Consumption 140 130 120 110 100 90 80 70 60 50 40

Germany, individual government goods and services

Greece, individual government goods and services

Germany, public government goods and services

Greece, public government goods and services

Germany, household purchases of goods and services

Greece, household purchases of goods and services

Graph 6.1 Historical development of different components of household consumption, Germany and Greece Source: Eurostat database, ‘GDP and main components (output, expenditure and income) [nama_10_gdp]’, accessed 8 May 2019

are owner occupied. ‘Equivalent market rental prices’ however have the overriding important advantages that they are available, do not require too much tinkering and are intuitively easy to understand. Anyway, looking at the aggregates it shows that consumption is the mainstay of the macroeconomy. The border between household and government expenditure shifts over time and differs between countries, which complicates historical and international comparisons. This led statisticians to develop the concept of AIC, which adds individual public and individual household consumption together: ‘Actual individual consumption . . . refers to all goods and services actually consumed by households. It encompasses consumer goods and services purchased directly by households, as well as services provided by non-profit institutions and the government for individual consumption (e.g., health and education services)’ Eurostat (2019). On top of this, there is consumption of public goods provided by the government which are not individually consumed, like ‘law and order’ and the proverbial streetlight. A snapshot from table 3.2 from the ESA 2010 explains the ESA 2010 definition of acquiring goods and services definition more detail, note that actual final consumption is considered to be consumption by households, of services like education but also of defense. Some might not define this last kind of consumption as ‘optimizing utility’. But that’s not the point. According to the definitions, the government pays and households hence consume it, by definition. A private company

70

60

50

40

30

20

10

0

Individual consump on expenditure of general government

Collec ve consump on expenditure of general government Germany

Final consump on expenditure of households

Greece

Graph 6.2  Components of household consumption (% of GDP), Germany and Greece Source: Eurostat database, ‘GDP and main components (output, expenditure and income) [nama_10_gdp]’, accessed 8 May 2019

Diagram 6.1  Sectoral consumption flows according to the SNA  

Sectors

 

 

 

NPISH

Households

Total

X (= social transfers in kind)

X

Collective X consumption

 

 

Total

Government’s final consumption expenditure

NPISH final consumption expenditure

Household final consumption expenditure

X: applicable

 

 

 

Household actual individual final consumption Government’s actual collective final consumption Actual final consumption = Total final consumption expenditure  

Kind of Government Expenditure Individual X (= social consumption transfers in kind)

Source: Eurostat/European Commission (2013), p. 72

182  Consumption army (like, in the olden days, the armies of the British East India Company and the Dutch Vereenigde Oostindische Compagnie) would however be considered ‘intermediate demand’ and would, hence, not be added to consumption. With regard to Diagram 6.1: Airbnb increases the ‘own production’ element of the X below ‘Households’; government paid education is an example of a social transfer in kind. To give an idea of the granular nature of the national accounts some detail about the national accounts definition of individual and public government consumption (Eurostat/European Commission 2013) is provided here: Alternatively, individual consumption expenditure of general government corresponds to division 14 of the classification of individual consumption by purpose (Coicop), which includes the following groups: 14.1 Housing (equivalent to COFOG group 10.6) 14.2 Health (equivalent to COFOG groups 7.1 to 7.4) 14.3 Recreation and culture (equivalent to COFOG groups 8.1 and 8.2) 14.4 Education (equivalent to COFOG groups 9.1 to 9.6) 14.5 Social protection (equivalent to COFOG groups 10.1 to 10.5 and group 10.7). 3.106 Collective consumption expenditure is the remainder of the government final consumption expenditure. It consists of the following COFOG groups: (a)  general public services (division 1); (b)  defence (division 2); (c)  public order and safety (division 3); (d) economic affairs (division 4); (e)  environmental protection (division 5); (f)  housing and community amenities (division 6); (g)  general administration, regulation, dissemination of general information and statistics (all divisions); (h) research and development (all divisions). Household purchases are defined and estimated in an even more granular way. Consumption includes some but not all household production for own use. A clear example is production from vegetable gardens (insignificant in rich countries but crucial in some African ones) and as stated imputed rents for owner occupied dwellings. One reason to include vegetable gardens is practical. It is often possible to make an estimate of total physical agricultural production of a farm but quite difficult to estimate the amount of it which is retained for own consumption. Unpaid housework is excluded. Aside: counting houses as capital and not counting unpaid house work means that, whenever estimates of aggregate labor/capital ratios are made, either houses and paid domestic workers have to be excluded or unpaid house work has to be added.

Consumption 183 Graph 6.2 shows that at present both ‘individual government consumption’ (education) as well as ‘collective government expenditure’ (building and maintaining roads) are non-trivial. But the inclusion of public goods within the consumption and production/consumption boundary deserves additional attention as they are, as a rule, excluded from DSGE models. Inclusion of ‘government provided consumption’ in the consumption boundary can be traced back to at least Adam Smith (Smith 1784 [1776]). The entire fifth and last part of The Wealth of Nations is devoted to the necessity of public goods and services, with a sterling emphasis on evolving institutions and the changing the growing importance of what nowadays is called ‘government consumption’. Smith, of course, also excluded some items and activities like personal services from consumption, which was criticized by later economists. As is often stated in textbooks, personal services were after the marginal revolution in economics considered productive after all and therewith included in the production/consumption boundary of economics. But less often stated in textbooks, the ‘marginal revolution’ also led to a narrowing down of the production/consumption boundary. An example is Alfred Marshall who included personal services in his concept of consumption, but also unlike Smith excluded individual and collective consumption expenditure by the government (Marshall 1920). To be more precise: he does mention government factories and mentions that ‘Prominent among the occupations which have increased rapidly since 1851 in England at the expense of agriculture are the service of Government, central and local; education of all grades; medical service; musical, theatrical and other entertainments, besides mining, building, dealing and transport by road and railway’ (p. 159) and writes about ‘the ever growing activity and wisdom of Government in all matters relating to health’ (p. 117). He does not neglect government products and services. But they are not included in his discussion of the nature of consumption in the first paragraph of his chapter II while he, somewhat against the grain of marginal analysis, also makes a distinction between ‘productive’ or necessary consumption and superfluous consumption. The clearest statement I could find from Marshall was (Marshall 1920): But it is best here to follow the common practice, and not count as part of the national income or dividend anything that is not commonly counted as part of the income of the individual. Thus, unless anything is said to the contrary, the services which a person renders to himself, and those which he renders gratuitously to members of his family or friends; the benefits which he derives from using his own personal goods, or public property such as toll-free bridges, are not reckoned as parts of the national dividend, but are left to be accounted for separately. Consulting Elsas (1944) as well as Mitchell (1922) suggests that around 1920 common practice had already moved beyond the position of Marshall. Economists understood that free bridges did not exist. Even when DSGE economists still follow Marshall by excluding them as well as other government items not directly paid for by consumers from the models.

184  Consumption Consumption, as conceptualized and defined in the national accounts, is a monetary flow concept based upon transactions a restricted period of time. It is not about the use of goods but about acquiring goods and services. But there are also notable imputations of non-monetary, non-market transactions like imputed rent of owner occupied dwellings as well as the imputation of financial intermediation services indirectly measured (FISIM), which tries to divide net interest paid by households into a fee for financial services and ‘pure’ interest. See Coyle (2014) for a brief and not very affectionate investigation of this concept. These imputations are not without consequences for understanding the national accounts. Recently, the ONS has stripped national accounts household income from all kinds of imputations to estimate a monetary concept which is closer to the actual experience of monetary income and consumption by households (Curtis, Davies and Weston 2016). This stricter monetary concept of income and consumption better matches income in all kinds of income surveys and is closer to the experience of households. Especially in 2008–2009 a much stronger increase in the monetary savings rate of households than the national accounts concept can be witnessed. This means that the 2008 crisis was for a much larger part caused by balance sheet effects than indicated by the national accounts income concept. Restricting consumption to acquiring instead of using goods and services also has drawbacks. Purchasing potatoes is not the same thing as cooking and eating homemade Belgian fries, buying a Dacia Logan is not the same thing as driving it. A convincing case can be made that to estimate welfare or prosperity an estimate of using goods and services, like driving your car and eating homemade meals, is more important than an estimate of the acquisition of goods and services. It is clear that at any given moment the stock of cars and roads is more important to consumers than the flow of new cars and the building of new roads. An analysis of this surely worthwhile. This is however not what the national accounts intend to estimate – these were established to estimate the circular monetary flows of production, income and expenditure (and have been extended to include information about (un)employment, assets, debts, employment and money as well as granular data on sub-sectoral flows). To highlight the complexities it is interesting to compare a stock, a use and a flow concept of the most important durable consumption good, next to houses: cars. Graph 6.3 shows global car production, which will be more or less equal to sales. The graph is highly interesting. The preponderance of the USA up to 1955 as well as the sudden growth of China (which needed only 17 years to produce as much cars as the entire world did in 1970) are stunning. Anyway, these sales of, at present, about 100,000,000 cars a year are considered to be part of AIC. When investigating historical developments we should account for changes in quality of cars as well as roads but as far as I’m concerned these data do show the global spread of the automobile society. We can also lock at stock data – at present, there are about one billion motorized vehicles (graph 5.3). These data are, when privately owned, not considered to be part of the stock of capital (though they are included in the flow of funds data). Aside from this, use data are available. This might be considered ‘real’ consumption but is not covered in the accounts at all, except for purchases of gasoline.

Millions of units of annual producon

Consumption 185 100 90 80 70 60 50 40 30 20 10 0 China Africa United States

Rest of Asia, & Australia South America

Europe and Russia Rest of North America

Graph 6.3  Worldwide production of cars, regional data

Summarizing: the accounts have a broad, detailed, well defined and estimated and backward looking monetary flow concept of consumption which includes non-market transactions. The DSGE models use a restricted, non-empirical and totally forward looking concept of consumption which only looks at market transactions. Neither the accounts nor the models look at the use of (durable) consumer goods. But there is reason to keep the use of all kind of imputations to the accounts as limited as possible and to show purely monetary income and expenditure next to concepts of income an expenditure which include all kind of non-monetary production.

6.3  Consumption in the DSGE models The whole edifice of DSGE models starts and ends with consumption. Consumption is defined as ‘permanent consumption’ as modelled with the all-important Euler equation. As Muellbauer (2016) states: The Euler equation for consumption . . . is the centre-piece of DSGE models. It connects the present with the future, and is essential to the iterative forward solutions of these models. It is based on the assumption of additive consumer preferences, defined over consumption both now and for each future period. Additive means that it is possible, with proper discounting, to add consumption in period 1 to consumption in period 2 (and even period infinity) to estimate, or at least to model, total life time utility. As such, the Muellbauer quote is incomplete. It is not only about consumption but about the work-leisure and the consumption-investments trade-offs. Higher consumption means, barring productivity increases, less leisure

186  Consumption and more work. And work is a negative (presumably measured in the same unit as consumption). Also, higher investments means more consumption in the future but also, as investment and consumption goods are basically alike, less consumption today. It is of some use to introduce the Euler equation here. We use the one as used in Bokan et al. (2016), cutting out – N (which has be treated in the chapter about labor) as well as housing and replacing the Greek letters (t stand for ‘now’, k stands for the number of periods from ‘now’). ∞ k   1 − a   Ci , t + k   aCt + k − 1  Et {∑ (Bt )        −   1 − 1 1−a − b a    k =0 

(1− e )

}

Utility. The representative patient household, labelled ‘saver’, gets utility from consumption of the nondurable composite good, Ci,t (subject to external habit formation) and from housing services HI,t and gets disutility from working where 0 < Bt < 1 is the discount factor, 0 ≤ f ≤ 1 measures the degree of external habit formation in consumption, b > 0 denotes the inverse of the intertemporal elasticity of substitution. Bokan e.a. 2016 p. 10. We will concentrate on concepts and definitions, not on the math. Starting at the left side and leaving out the part about N we encounter the next variables: The variable: ‘Et’ stands for: the ‘Eternal utility’. Consumption now and discounted utility in future (literally ‘forever after’) is therewith defined as ‘utility’ which, supposedly, can be added together over the years. This assumption assumes that the dimensions of ‘utility’ do not change because of, for instance, demographic, biophysical or cultural/technological changes. The invention of a cure for cancer or depression ten years from now does not affect the possibility to add utility now to utility 80 years into the future. ‘Et’ is not individual utility but utility for the entire society which, as it is an intertemporal variable, means that the utility of Yon Yonson in 2027 in Wisconsin is, using a proper discount rate, added to the utility of Amitola Eagle in Oregon in 2028. In fact it is not so much assumed that individual utilities can be added (the models do not adapt ‘Et’ for changes in the size of the population) but it is assumed that society behaves as if it optimizes consumption as one homo economicus, irrespective of the size of the population. The symbol S is a Greek letter which is commonly used to denote that Et is the sum of utility in a number of periods, in this case the periods t for k = 0 (which is now) to k = 1 (next year) to infinity, i.e. an infinite number of periods (like years). • The variable Bt is a discount factor, a kind of interest rate. It is supposed to be stable in the sense that a change in B today automatically translates to •

Consumption 187

•

changes in all other periods. As this leads to compound interest rate effects it causes ‘Et’ to be interest rate sensitive though intertemporal changes of consumption counteract this, at least according to the model. ‘B’ indicates that consumption next year yields, per unit of consumption, less utility now (say: –2 percent) than consumption of this unit today. To be able to do this, it is necessary to specify that there is only one good or, which is the same assumption, that the relative amounts and the relative prices of consumption good I (say: Roquefort cheese) to consumption good II (say: heparin)1 do not change. The variable b captures how the representative consumer relates consumption and investment now to consumption and investment in the future (called: intertemporal substitution). It is the centerpiece of the Euler formula. Taking account of the real interest rate (which is assumed to be equal to the growth of productivity) the representative consumer decides about how much today and how much tomorrow. Empirically, this variable is important, albeit not in the way the models assume: ‘the permanent income hypothesis that lies at the core of most DSGE models implies that a high real interest rate is associated with high expected rate of growth of consumption. This is at odds with the data. For example, Hall and Mishkin (1982) have pointed out that there have been long periods of time in which average U.S. aggregate consumption growth was positive though real interest rates were very low (close to zero) . . . Over the period 1967:1–2003:4, results obtained from a SVAR model (to be described later) indicate that, following a monetary shock, the conditional correlation between the real interest rate and consumption growth is highly negative, -0.96’ (Auray and Gallès 2008, p. 123). Comparable information can be found in Hall (1987, 1979). A high growth rate of consumption has to be understood as ‘low consumption today’ as total consumption and consumption growth over the epochs is considered to be a bit of a zero sum game. Anyway, there is supposed to be a strong positive relation between real interest rates and consumption growth – but there isn’t. To the surprise of modelers, people consume a lot when interest rates are low. Intuitively, this empirical relation makes sense. It is not entirely impossible that low interest rates leads to higher dissaving or higher borrowing and more consumption, or to higher investment, growth and employment and hence higher consumption in the future. In real life, high interest rates might, not just because borrowing becomes more expensive but also because existing debts have to be rolled over, restrict monetary budgets (Jayadev and Mason 2014). The models tell us otherwise. Surprisingly, there seem to be no studies which compare a variable which is only about non-durable household consumption goods with consumption which includes durable goods as well as public goods and services even when in reality, models inconsistently often use estimates of consumption including durable goods. The real problem is of course that the models abstract from monetary budgets and balance sheets – while

188  Consumption Jayadev and Mason (2014) doesn’t. Even then, the variable is measured. Are there statistical institutes carefully measuring it with tight, controlled and subjective methods? Not really. Academics do. What happens when the measurements are measured? Tomáš Havránek did. What did he find? Academic mismeasurement: ‘The elasticity of intertemporal substitution in consumption, a key input into macroeconomic models, has been estimated by hundreds of researchers. I argue that the findings of the literature are biased upwards because of the tendency to preferentially select positive and significant estimates for publication. The publication bias is so strong that the literature could consistently produce statistically significant estimates even if the underlying EIS was zero’ (Havránek 2013, p. 1). • To solve the anomaly mentioned, the variable ‘a’ or ‘habit persistence’ is introduced. It indicates that the consumer wants to shift consumption from one period to another in the long term but doesn’t do so in the short term because he doesn’t want that. In a technical sense it shows that if consumption in the previous period was high consumption today will yield relatively less utility: the ‘hangover’ theory of utility. As the representative consumer optimizes ‘Et’ over time this will tend to depress consumption in the previous period. In an applied sense, using a {(1 – a)/(1 – b)} formula with an a and b which are not precisely measured but ‘calibrated’ of course enables a modeler to come up with values clearly above or clearly below one, dependent on the a and b chosen. Looking at this from a monetary point of view: as prices like rents are administered prices while moving is very expensive there is truth to the expression: ‘the rent has to be paid’, it’s not just unconstrained choice by the social planner. • The variable C stands for consumption. In Bokan et al. (2016) this excludes consumption of durables goods, houses and public goods and services, as well as consumption provided by NPISH (like sport clubs, choirs, churches and other non-market entities). Part of yesteryears consumption is distracted from this variable to account for ’habit formation’. People of course have consumption habits which are hard to change. But this is not about eating rice or potatoes. This is about total consumption. It means that when yesteryears consumption was very high, it will distract from utility gained by todays consumption, which means that, as the representative consumer optimizes, this dampens consumption today as it will distract from future utility. Where do these concepts come from? Behind the variables mentioned is the meta-concept of ‘life time consumption’ of the individual household and, in a sense, the entire sector households. This idea was put forward by Friedman in 1957 with ‘A theory of the consumption function’ (Friedman 1957). This book puts the life cycle consumption hypothesis central stage, i.e. the idea that people or households rationally plan their income and borrowing in a learning by doing way to optimize consumption over their

Consumption 189 lifetime. Friedman knew the drawbacks of his idea and starts his second chapter with the phrase: The magnitudes termed ‘permanent income’ and ‘permanent consumption’ that play such a critical role in the theoretical analysis cannot be observed directly for any individual consumer unit. The most that can be observed are actual receipts and expenditures during some finite period, supplemented, perhaps, by some verbal statements about expectations for the future. There are good reasons why Friedman wrote down this caveat: the book continually mixes up consumption and behavior of the individual household with consumption and behavior of the sector households, which is different, for only one of many reasons as the sector household does not retire while people in individual households do. As such, the inability to observe the concept introduced by Friedman in a direct way fits in a neoclassical tradition: other key elements of this theoretical edifice (utility, the natural rate of unemployment, the natural rate of interest) cannot be measured in any direct way either. To his credit Friedman goes to great lengths to investigate if time series and budget studies on consumption are consistent with the concept that consumers smooth consumption relative to income (which is what the theory is all about). Without becoming convincing. To be able to define permanent income Friedman had to use the existing empirical definition of consumption: household purchases, either nominal or on credit, excluding houses. He was obliged to do this as he explicitly wants to give a theoretical explanation for the Keynesian consumption function, which was (and is) inherently macro, monetary, flow oriented and statistics-consistent. Friedman does make a useful distinction between durable and non-durable goods stating that durable goods are an asset on the household balance sheet, an idea nowadays incorporated in the US flow of funds as well as into the national accounts of several countries but not yet in the canonical, workhorse DSGE models. Reading the book, the common accusation that Friedman discarded the realism of assumptions has to be tuned down – of all neoclassical macroeconomists I ran across while writing this book he is the only one who gave the question of the realism of his assumptions serious attention, probably a relic of his years at the NBER. He honestly tried to show if neoclassical theoretical clothes could be wrapped around the empirical institutional skeleton of consumption statistics. But his followers were, alas, not too interested less and less in the bare bones of empirical investigation. Hall, who introduced the Euler equation, still tried to put his ideas of modelling permanent income to the test – with devastating results (Hall 1979, 1987). But later economists seem to have totally discarded conceptual discussions. The idea that houses, mortgages but also cars and furniture serve a longterm consumption purpose, central to Friedman (1957), are all but absent in any but the most recent DSGE models while concepts of government consumption are absent altogether in the flagship models and also in Friedman (1957), despite the best efforts of Paul Samuelson (1954). As was the case with the Schwartz and

190  Consumption Friedman book about the history of money, recent concepts like the flow of funds or, in this case, public goods were not incorporated in the theory. The problem was, and is, that the Keynesian consumption function which inspired Friedman was, and is, a monetary function. It’s the amount of monetary purchases financed by monetary income and access to all kinds of credit which makes the economy go round. It’s not the utility related function of the models. Or a use-value related function: use of these goods and services are only loosely connected to purchases, especially in the case of durable goods. Ultimately, we end up with the somewhat ironical situation that exactly those goods which mitigate and stabilize changes in use-value (durable consumption goods, like furniture or cars or clothes) and hence life time consumption as stressed by Friedman are for the very same reason the most volatile when it comes to purchases and the most destabilizing when it comes to the expenditure economy (a problem which, after the 2008 crisis, governments tried to solve with ‘cash for clunkers’ schemes). This neat empirical contradiction between the monetary and the nonmonetary economy is simply left out of the models. A severe drawback of the ideas of Friedman and most of his followers is the lack of an integrated flow of funds approach which looks at income and net credit on the financing side and the purchase of consumer goods, debt service and the net acquisition of financial assets on the purchasing side (this is the formula used in the modern national accounts). Introducing ownership and use of durable consumer goods (and, a fortiori, houses) into such analysis requires the introduction of balance sheets and hence the data as defined by the modern national accounts as well as the flow of funds. Shifts between public and private consumption are also not tackled, even when such shifts are not only large but, in the case of medical care, the focus of intense political attention. And the interesting bit – the influence of unemployment as we measure it and the associated loss of income on consumption, is not mentioned at all. As stated, after Friedman the next big step in the design of the DSGE consumption concept was using the Euler equation to model intertemporal consumption (Hall 1979). Basically, this equation states that people juggle to balance consumption present and consumption yet to come, preferring more consumption today but also preferring more leisure (and hence less production and consumption) today and taking account of investments. A central concept behind this is that people have stable ‘deep’ preferences and though they do adapt their behavior to superficial changes in policy or technology, they do not change their basic preferences or at least not fast. On a nice day in summer, they will go to the beach. When it rains, not so much but their basic beach preference is still present and once the sunny weather returns, they will go. That’s the micro-intuition – which is not too bad. But it is a macro-model. On the macrolevel, this micro-intuition is morphed in the less intuitively appealing idea that people (or: ‘the social planner’) smooth the future growth of total consumption, taking account of future increases in the level of productivity (which are known). More generally, on the margin they prefer more utility today to more

Consumption 191 utility tomorrow and the consumption growth rate which optimizes present utility and discounted utility yet to come is stable but account is taken of the level of investment which, considering increases in productivity, is needed to do this. This totally forward looking approach is quite a rephrasing of the permanent income hypothesis when we compare it to the backward looking empirical and learning by doing behavior which was the focus of Friedman (1957). The zeitgeist behind this temporal orientation shift is the (modernist) wish to get rid of the past – consumption today is solely based upon expectations of the future and preferences which are not history contingent, a modelling strategy chosen to comply with the ‘Lucas critique’ (Hall 1979). But it has to be stated that the introduction of habit formation does allow modelers to mitigate the results of unfettered influence of the interest rate. History has not been kind to such ideas. Everybody (including Hall himself, Hall 1987, 1988) who took a hard look at the idea and the data has dismissed the Euler equation as an apt description of aggregate consumer behavior. For one thing: the equation states that intertemporal substitution is guided by the interest rate – but consumption is rather insensitive to the interest rate (Canzoneri et al. 2007), even when a little less so for rich consumers. The modelers try to solve this with an ad hoc parameter that accounts for behavioral inertia mitigating the influence of the interest rate. A more fundamental critique is voiced by Muellbauer (2016), who criticizes the absence of housing as well as the general absence of debt, credit and well as other assets and liabilities, which brings us to the concept of consumption. Friedman (1957) started out in the monetary world of Keynesian spending which related to the idea of consumption as the purchase of goods and services by households. This household consumption function was always supplemented by a government expenditure function. Putting households center stage enabled modelers to ignore this. Also, by looking more formally at intertemporal consumption, the focus was shifted for monetary purchases to ‘utility’ while introducing the idea of rational choice shifted the focus from measured consumption in the past to consumption in the future. As it is difficult to model the use of (depreciating) durable consumer goods and their future influence on future consumption, these were left out, too. And in reality, there is substitution between provision of consumption by the government and purchases of households themselves, which is not captured when we only look at household purchases. Remarkably, such ideas are not calibrated at the conceptual level (which is what Friedman still did or at least tried to do in 1957). Looking at the conceptual side is however entirely possible. There are studies which look at household behavior including the use of consumer durables instead of just looking at purchases. These seem to be closer to the intertemporal idea of consumption than models focused on purchasing only (Aguiar and Hurst 2005) which show that the idea of the representative consumer does not hold when looking at actual behavior of (groups of) individual households. This does not seem to be incorporated in the models.

192  Consumption Let’s return to the Zuid-Kennemerland National Park. Applying the implicit consumption border of the models only crossing the land owned by individuals private as well as ordering a drink at the teahouse will be considered ‘consumption’. Government owned lands and lands owned by NPISH fall outside of the consumption/production boundary of the models. When interest rates increase (which, according to the models, indicates an increase in productivity) people will start to visit the area more often as they expect to be richer in the future, an expectation which spills over to behavior today (remember that there is only on consumption good – visiting the area is at par with visiting the hospital). The maintenance of government owned lands, roads, bicycle paths and parking places is, according to the models, ‘wasteful expenditure’. Durable consumption goods also fall outside of the consumption/production boundary which means that using a bike or a car to visit them also falls outside the scope of the models. Unless you rent a car or a bike. In that case a car or a bike is considered to be ‘fixed capital’ and it’s use (actually: renting it) will add to utility. As household owned ‘horses’ are, as far as I know, still considered to be ‘capital’, riding a horse (actually: owning a horse) will yield utility. Also, the DSGE concept of consumption is intertemporal. All the money spent on renting bicycles and maintaining the private lands from here to eternity will add to utility, albeit discounted with an interest rate. In econospeak: ‘the models look at a complete set of Arrow-Debreu commodities’. This relates to the reformulation of the general equilibrium hypothesis by Arrow and Debreu which includes all transactions until the end of time, including an apple bought in Rome in 2023. A trade-off between present and future transactions is obtained by modelling the influence the interest rate on stable preferences for intertemporal consumption which are however mitigated by a parameters for habit formation. The classical mathematical formulation of intertemporal utility optimizing is Samuelson (1937) who was however very critical of this idea and formulated it to point out how consistent modelling could lead to absurd consequences. Also, the set of commodities in Arro-Debreu markets is not complete as ownership of durable goods like cars is lacking – the model is in the end about purchasing goods and services, not about using them. As the models only recognize market transactions and do not recognize non-market production of housing services by households living in their own house this is the only way they can model this. The failure to introduce government production and consumption into the production/consumption boundary is puzzling – not to say baffling. As Marshall himself already noticed for the period after 1851, the realm not just of household consumption but also of government activity had greatly increased. Think of the introduction of citywide sewer systems in the latter decades of the 19th century, induced by the London ‘Great Stink’ of 1858 (Halliday 2011). This process of course continued afterwards. Think of electricity grid networks, highway networks or 4G networks. In all of these, the state looms large even when, as in the case of 4G networks, the actual networks are paid for by private

Consumption 193 companies. Also, health care has changed beyond recognition. Motorized ships, trains, bicycles, cars and planes have, together with a total rebuilding of roads, canals and other transport systems, revolutionized travelling. Education has become compulsory and largely a government service. Next to this, the economic concept of consumption has widened too and once one actually tries to measure ‘consumption’ one quickly runs into inconsistencies with historical and international comparisons when government services and goods are excluded. Why should private education be classified as ‘consumption’ while public education should be understood as ‘intermediate consumption’ of the government? Why should the British NHS be excluded while a private hair transplantation should be included? Such inconsistencies make one expect that, post Marshall, public goods and services were increasingly included in the (neoclassical) production boundary, which is what of course happened. At first. In 1954 the classic Samuelson article (7,900 Google citations) firmly introduced (without mentioning Smith but with a profound use of the metaphor of ‘the invisible hand’) the Smithian concept of public provision of goods and services into neoclassical economics (Samuelson 1954). Samuelson seems to rely on Musgrave (1939) for concepts like ‘non-rivalrous’ (the lamp post shines for everyone) and ‘non-excludable’ (you can’t exclude people from the services of coastal defenses). These concepts also resound in Eurostat/European Commission (2013; hence ESA 2010) lemma 3.102. but they are not to be found in earlier discussions of the concepts of the national accounts (see Chapter 1). It seems that at least partly thanks to the Samuelson article there has been a fusion of theory and measurement. It is important to note that Samuelson explicitly used and connected the phrases ‘public expenditure’ and ‘collective consumption’ while also stating: ‘I introduce no mystical collective mind that enjoys collective consumption goods’ (Samuelson 1954, p. 387).2 Clearly, the concept of public consumption as defined is in the national accounts is at least to an extent consistent with public consumption as defined in neoclassical economics. Neoclassical macro is an exception. Public expenditure is included in the DSGE models but it is, ignoring the work of Samuelson, not supposed to add to ‘utility’ while, contrary to the accounts and Samuelson (1954) the concept of the representative consumer excludes the idea also is no idea of individual consumption of government goods and services. Government expenditure is, in the models, as a rule considered to be wasteful by definition (see formula 55 and 56 of Bokan et al. 2016). Let’s have another Great Stink. A forthright consequence of treating government consumption as wasteful by definition is that using DSGE models to gauge the consequences of cuts to the provision of public goods in Greece will not show any decline of ‘utility’ in the models. To the contrary. Sacking educators and healers from government service will enable these people to sell their labor on the market while the hospitals and educational institutes can be used for whatever. Note that, in the models, ‘unemployment is leisure’ which means that unemployment actually increases ‘utility’ of the sacked people as the educators and healers do not have

194  Consumption to work anymore. Remarkably, the assumed wasteful nature of government spending and the exclusion of durable goods are not a necessary element of the models. DSGE models with non-wasteful government or durable goods spending are possible, as indicated by the title of a 2012 working paper by Yasuhara Iwate, ‘Non-wasteful government spending in an open economy estimated DSGE model: two fiscal policy puzzles revisited’ (Iwate 2012). For the geeks: one can either include government consumption directly into the utility function of the representative consumer or use the concept of Edgeworth complementarity, i.e. the idea that a car is useless without (public) roads. But such DSGE articles are black swans. I’m not aware of a model which includes public goods and services as well as durable goods. To return to the questions posed earlier (why the differences between the statistics and the models): excluding government consumption from the models is, considering Smith (1784 [1776]), Samuelson (1954)3 and Iwate (2012), a conscious, rational choice and not a unavoidable necessity. As things stand today (Bokan et al. 2016) the modelers face a choice. Either they concentrate on utility. In that case they will also have to look at real consumption, like consumption of calories in relation to health, or the use and ownership of cars. There is no way that ‘Utility’, whatever it is, is related to purchasing only. This choice implies that public and individual government consumption have to be included into the concept of consumption, as the boundary between public and private consumption of ‘health’ and ‘education’ is continually shifting, just like the boundary between public and private transport (including government owned roads) – mind that a lot of consumption of public goods and services, like education, is non-monetary. The modelers will also have to start incorporating consumer durables and especially the use of houses into the models. Next, they will have to figure out a way to measure ‘utility. As yet, all efforts thus far have failed. If they don’t do this and keep focusing on household expenditure on non – durable goods only they should stop calling this this consumption and relabel it, as it is only part of total AIC. Even in this case they will have to admit the historically non-homogenous nature of series on expenditure on on-durable household goods as well as to historical and institutional changes – and even changes in the body and mind of humans (Floud, Fogel and Hong 2011). And even then the main really intertemporal item – housing including the location of roads and the like, which often survives the centuries – deserves attention of its own. At this moment, the concept of consumption in the models in an unsavory nonhomogenous soup which is served in another way by each researcher without even looking at the consistency between his or her concept and the actual series as we measure them.

6.4  A comparison Which all leads to the next comparison of the accounts and the models:

Consumption 195 SNA

DSGE

Time orientation Basic variable Set up Definitions Sectors Basic variable estimated? Useful public goods and services?

Backward looking Transactions, not use Granular, bottom up Precise, fixed NPISH included Yes Yes

Forward looking Utility, not use Non-granular, top down Non precise, non-fixed NPISH not included No Generally: no

Durable goods

Yes

Generally: no

Use or purchases Discounting? Houses? FISIM?

Purchases, not use No discounting Owner occupied essential Yes

Purchases Discounting is pivotal All rented No

Notes 1 Heparin is a chemical that prevents blood from clotting in the case of for instance dialysis. It’s extremely expensive and extracted from the ‘mucus’ which lines intestines. Extracting the mucus from the intestines still is manual work. Heparin saves lives. There are issues. Intestines of pigs as well as cattle are used, which makes for problems with it being halal/kosher and the Hindu equivalent of this. 2 See Sekera 2016, for a criticism of the rational, neoclassical nature of Samuelson’s consumer of public goods. 3 Samuelson was well aware that including public consumption into neoclassical economics also required a widening of the scope of transactions beyond market transactions, considering the last sentence of his 1954 article: ‘Political economy [i.e. market oriented neoclassical economics, M.K.] can be regarded as one special sector of this general domain [i.e. all transactions, M.K.], and it may turn out to be pure luck that within the general domain there happened to be a subsector with the “simple” properties of traditional economics’.

Literature Aguiar, M. and E. Hurst (2005). ‘Consumption vs. expenditure’. Journal of Political Economy 113 919–948. Auray, S. and C. Gallès (2008). ‘Consumption growth and the real interest rate following a monetary policy shock: Is the habit persistence assumption relevant?’ Recherches économiques de Louvain, De Boeck Université 74:2 121–142. Bokan, N., A. Gerali, S. Gomes, P. Jacquinot and M. Pisani (2016). ‘EAGLE-FLI. A macroeconomic model of banking and financial interdependence in the Euro area’. European Central Bank working paper series no. 1923. Canzoneri, M.B., R.E. Cumby, D.T. Behzad and T. Diba (2007). ‘Euler equations and money market interest rates. A challenge for monetary policy models’. Journal of Monetary Economics, Elsevier 54:7 1863–1881.

196  Consumption Carroll, C.D. (2001). ‘Death to the log-linearized consumption Euler equation! (and very poor health to the second-order approximation)’. Advances in Macroeconomics1:1 Article 6. Coyle, D. (2014). GDP: A brief but affectionate history. Princeton: Princeton University Press. Curtis, N.C., P. Davies and L. Weston (2016). ‘Alternative measures of real households disposable income and the saving ratio’. Available on the site of the Office for National Statistics. www.ons.gov.uk/economy/nationalaccounts/uksectorac counts/articles/nationalaccountsarticles/alternativemeasuresofrealhouseholddis posableincomeandthesavingratiojune2016#authors Elsas, M.J. (1944 [1942]). ‘The definition of national income’ in: A.L. Bowley (ed.) Studies in the national income 1924–1938 1–52. Cambridge: Cambridge University Press. Eurostat (2019). Statistics explained. Glossary: Actual individual consumption (AIC). https://ec.europa.eu/eurostat/statistics-explained/index.php/GDP_per_capita,_ consumption_per_capita_and_price_level_indices. Accessed 8 May 2019. Eurostat/European Commission (2013). European system of accounts ESA 2010. Luxemburg: Printing office of the European Union. Floud, R., R.W. Fogel, B. Harris and S.K. Hong (2011). The changing body: Health, nutrition, and human development in the Western World since 1700 (New Approaches to Economic and Social History). Cambridge: Cambridge University Press. Friedman, M. (1957). A theory of the consumption function. Princeton, NJ: Princeton University Press. Hall, R.E. (1979). ‘Stochastic implications of the life cycle-permanent income hypothesis: Theory and evidence’. NBER working paper no. R0015. Hall, R.E. (1987). ‘Consumption’. NBER working paper no. 2265. Hall, R.E (1988). ‘Intertemporal substitution in consumption’. Journal of Political Economy 96:2 339–357. Halliday, S. (2011). The great filth: Disease, death and the Victorian City. London: History Press. Havránek, T (2013). ‘Publication bias in measuring intertemporal substitution’. IES working paper no. 15/2013. Iwate, Y. (2012). ‘Non-wasteful government spending in an estimated open economy DSGE model: Two fiscal policy puzzles revisited’. ESRI discussion paper series no. 285. Jayadev, A. and J. W. Mason (2014). ‘Fisher Dynamics in US Household Debt, 1929–2011’. American Economic Journal: Macroeconomics 6:3 214–234. King, W., F. Macaulay, W. Mitchell and O. Knauth (1921). Income in the United States: Its amount and distribution 1909–1919, Volume 1: Summary. New York: National Bureau of Economic Research. King, W. (1922). ‘All branches of government’ in: W. Mitchell (ed.) Income in the United States volume II: Detailed report 201–222. New York: National Bureau of Economic Research. Knibbe, M. (2007). ‘De hoofdelijke beschikbaarheid van voedsel en de levensstandaard in Nederland, 1807–1913’. Tijdschrift voor Sociale en Economische Geschiedenis 4:3 71–107. Marshall, A. (1920). Principles of economics. An introductory volume. 8th edition. London: MacMillan.

Consumption 197 Muellbauer, J. (2016). ‘Macroeconomics and consumption’. Oxford University department of economics discussion paper series no. 811. Musgrave, R.A. (1939). ‘The Voluntary Exchange Theory of Public Economy’. The Quarterly Journal of Economics 53:2 213–237. Nationaal Park Zuid-Kennemerland (2018). www.np-zuidkennemerland.nl/1421/ over-het-park. Accessed 12 September 2018. Office of National Statistics (2018). ‘UK natural capital: Ecosystem accounts for urban areas’. www.ons.gov.uk/economy/environmentalaccounts/bulletins/ukna turalcapital/ecosystemaccountsforurbanareas Samuelson, P. (1937). ‘A note on the measurement of utility’. The Review of Economic Studies 4:2 155–161. Samuelson, P. (1954). ‘The pure theory of public expenditure’. The Review of Economics and Statistics 36:4 387–389. Sekera, June (2016). The public economy in crisis. A call for a new public economics. Heidelberg: Springer. Smith, A. (1784 [1776]). An inquiry into the nature and causes of the wealth of nations’ book I, II, III, IV and V. 4th edition. London: Methuen. Smith, N. (2014). ‘The equation at the core of modern macro’. Blogpost at Noahpinion, 14 January 2014. http://noahpinionblog.blogspot.com/2014/01/theequation-at-core-of-modern-macro.html

7 I stands for gross fixed capital formation

7.1 Introduction We will look at investment through the lenses of statisticians and neoclassical macroeconomists. These two views will be confronted analyzing the development of the long-term rate of ‘gross fixed capital formation’, which means that financial investment in bonds and stocks and cash is excluded. The glossary of Eurostat defines gross fixed capital as: Fixed capital is the value of capital assets available for production purposes at a given point in time. All capital goods are included which are accounted for in gross fixed capital formation. This is measured by the value of acquisitions less disposals of new or existing fixed assets. Fixed assets are produced non-financial assets that are used repeatedly or continuously in production processes for more than one year. Fixed assets consist of dwellings, other buildings and structures, machinery and equipment, weapons systems, cultivated biological resources, and intellectual property products. (Eurostat 2018) Clearly, when we’re defining gross fixed capital formation we’re not talking about ‘land’ and other non-produced assets. These do play a role in production and distribution. But as they are not produced they are no ‘fixed capital formation’. Also, notice the world ‘value’ in the definition above. This is not about ethical value. Or about physical values. It’s about monetary value as shown on balance sheets. We’re dealing with a monetary, not a physical variable. Investment by households in cars or vacuum cleaners are not included in Eurostat’s definition. But some information will be provided to proof the point that, conceptually, household purchases of durable goods can be included in a new concept of total investments as purchases of such consumer durables are highly important for a proper analysis of the business cycle as well as for a proper analysis of the changing of modern lifestyles so intertwined with the rate and composition of Gross Fixed Capital Formation and the flow of production.

I stands for gross fixed capital formation 199

7.2 Concepts: the changing nature and definition of gross fixed capital formation If there is one change in the composition of aggregate expenditure which sets apart modern Western economies from the preceding economies in, say, the 1500–1850 period it’s the doubling, tripling, quadrupling and quintupling of the rate of monetary gross fixed capital formation, also known as ‘investment’ (Kuznets 1955; Rostow 1955).1 The same holds for newly industrialized countries like China, south Korea or Ethiopia when we compare the present rate of investment with the pre-1950 or, in the case of Ethiopia where investments rose from a little over 10 percent of GDP to close to 40 percent of GDP, the pre2000 rate. Graph 7.1 shows the long run development of the rate of investment in Sweden (on these data also Kuznets 1955, table 1.3C which however does not show pre 1863 and post 1930 data). It’s Sweden’s economic history in a nutshell. During the first decades of the 19th century, Sweden was a predominantly agricultural country with a low rate of investment. Around 1850 things started to change: ‘In 1853 the Riksdag of the Estates decided that the State would build main line railways, but that other lines would be built by private enterprises (often with cities as main owners), and in

40

35

30

25

20

15

10

5

0 1807

1822

1837

1852

1867

Riksbank data

1882

1897

1912

Eurostat, old definion

1927

1942

1957

1972

1987

2002

2017

Eurostat, ESA 2010 definion

Graph 7.1  Nominal gross fixed investment rate, Sweden, 1807–2018 Sources: see Knibbe (2014), Annex, updated with data from Eurostat, unless otherwise mentioned

200  I stands for gross fixed capital formation 1856 the first stretch, between Örebro and Nora (a private railroad), was opened for traffic’ (Wikipedia 2018). Sweden entered the modern era. Macroeconomically, this modern era was at first characterized by a structural investment rate of between 10 and 15 percent of GDP but also by vehement ups and downs of the yearly rate. It can be argued that in this period in Sweden a ‘classical’ economic system, with a business cycle led by exogenous shocks to agricultural production or trade, made way for a Keynesian system, characterized by a larger importance of wage labor and a high but volatile rate of investment. In this system the economy itself produced ups and downs in the rate of investment. Declines in investment did not only directly lead to lower aggregate spending but also to higher unemployment which propagated these shocks as a badly needed increase in consumption was of course thwarted by high unemployment and could anyway not compensate the decline of the rate of investment. On the contrary: this decline in investment spending induced lower consumption, aggravating the downswing. The modern Swedish economy was born (for a macro-quantitative analysis of the comparable process for the UK see Dimsdale, Hills and Thomas (2010), especially 280–285). After around 1925 a new phase set in which would continue for about fifty years. Not railways this time, but middle class homes, sewer systems, schools, community buildings, Swedish malls, highways and trucks – investments which eventually led to the world described, for neighboring Norway, in the autobiographical books of Karl Ove Knausgaard: children living in middle class houses, transported by public buses crossing new bridges to heated swimming pools or, when these kids weren’t managed, buying stuff at the local FINA petrol station. And while the world of Karl Ove was not too different from that of his parents – which bought the multi bedroom suburban dreamhouse stocked with stereo players in the first place – it’s an epoch away from the somewhat mystical, cramped house and manual labor of his grandparents. This epoch was the 1925–1970 age of the high rate of gross fixed investment. After 1925 swings in the investment rate were still large but between 1950 and 1970 developments seem to have been more stable – possibly because a larger and more active government pursuing longer term goals moderated ups and downs. 1970 was a new watershed. The investment rate decreased again and, as we will see, not just in Sweden. This decline happened relatively fast, at a rate of 0.5 to 1 percent of GDP a year, meaning it must have been a continuous drag on aggregate spending. The decline was temporarily interrupted by a modest investment ‘boom’ around the end of the 1980s related to a financial deregulation and easy credit induced financial and housing boom which ended badly. This however did not stop the long-term downward movement of the rate of gross fixed investment (Jonung 2009). It is only very recently, in an environment characterized by historically low interest rates, that the rate of investment reached pre credit-bust levels again. My interpretation: after around 1970 the ‘structural’ rate of investments decreased. To put this in a more tangible way: the transformative technologies of our age require investments in computers, smartphones and radio masts – which require less money than investments in railroads and trains or highways and trucks, bridges, buses,

I stands for gross fixed capital formation 201 malls and middle class houses. The rate of investment, which increased so much until 1970, declined again. The Swedish story of a secular decline of the investment rate rudely interrupted by a credit boom (and bust) is not exceptional. Comparable stories can be told for Finland, Spain and the USA as well as the Baltic states or Ireland in the beginning of the 21st century. Or Thailand and Malaysia in the first half of the 1990s. The list is longer. Important to note is that during these booms the investment rate was high and unsustainable – but not higher than rates which were sustainable a few decades earlier. Economic epochs differ – even when including hip replacements and other durable high tech medical ‘tools’ with a clear cost of replacements to the rate of investment might jack up gross fixed investment as we measure a bit. Aside: after the banking crisis of 1991 Sweden quickly reformed its banking system. In spite of this, investments stayed remarkably low for a considerable period and a secular stagnation of the investment rate of around ten to fifteen years is borne out by the facts. Clearly the rate of investment is not just connected to interest rate and the growth of the economy but also to changes in lifestyle and the nature of transformative technologies, which is not a new idea (Kuznets 1961). The last decades of the graph show different estimates of the rate of investment. These different rates are not based on revisions of earlier estimates but caused by changing definitions. One of the reasons definitions change is of course the very change of transformative technologies and changes within the production boundary but also shifts of this boundary. But ideas about ‘the economy’ do play a role, too. The most important changes in the definitions are the inclusion or exclusion of certain items. Changes in ideas about ‘the economy’ led to Eurostat / European Commission (2013 hence ESA 2010) national accounts guideline which include, unlike earlier guidelines, military equipment as fixed capital and defines Research and Development as investment spending. Which of course leads to a higher rate of investment. These were, however, not new ideas as we will see. If there has been, after 1970, a switch from ‘traditional’ investments to these new kinds of spending the post 1970 decline may be overrated. But as defense spending as well as company research and development has declined after about 1990, this hypothesis is, at first sight, somewhat far-fetched and nontraditional kinds of investments might have exacerbated the measured declining trend instead of mitigating it. Other changes are the inclusion of intangible assets, which had become ever more important, in the definition of gross fixed capital and hence gross fixed capital formation. Think of the lease of bandwidth (an unproduced natural asset) by governments and investments in educational software like Moodlerooms. The point: neither the definition of investment nor the composition of investment spending is stable. Gross fixed capital formation in 1880 was about quite different items than the rate of investment in 1980, even when its influence on aggregate spending was more or less the same. Items not included in the investment rate are household purchases of durable goods like cars and smartphones. These items complement gross fixed investment as defined by statisticians: one needs a road to drive and a 4G network to use a smartphone

202  I stands for gross fixed capital formation but these roads and networks are redundant without cars and phones. It is useful to estimate the money we spent on roads and networks – but a proper analysis should incorporate household durables like cars and phones, too. Before we can say more about this, we first have to be more precise about fixed capital.

7.3  Statistical definitions Investments are expenditures of either the government or business entities on goods and services which last longer than one year (including stocks of ‘finished products’ and ’work in progress’) plus investments of households in houses. There are a plethora of different items and even more sector-item combinations: companies as well as the government invest in cars even when these are not the same cars and when they are not used for the same purpose. More formally the process of investment is described as: Gross fixed capital formation (GFCF) consists of resident producers’ acquisitions, less disposals, of fixed assets during a given period plus certain additions to the value of non-produced assets realised by the productive activity of producer or institutional units. GFCF includes acquisition less disposals of, e.g. buildings, structures, machinery and equipment, mineral exploration, computer software, literary or artistic originals and major improvements to land such as the clearance of forests. (Eurostat/European Commission 2013 (hence Esa 2010) lemma 3.124) Investments are often produced in the transaction economy. But own account production is also included: Products used for own capital formation can be produced by any sector. Examples of such products are: (a) machine tools produced by engineering enterprises; (b) dwellings, or extensions to dwellings, produced by households; (c) own-account construction, including communal construction undertaken by groups of households; (d) own-account software; (e) own-account research and development (ESA 2010 lemma 3.22). The ESA 2010 also looks at gross fixed investment in more detail. According to the definitions of the ESA, gross fixed investment consists of expenditure on or the acquisition of: (1) dwellings; (2) other buildings and structures; this includes major improvements to land; (3) machinery and equipment, such as ships, cars and computers; (4) weapons systems;

I stands for gross fixed capital formation 203 (5) cultivated biological resources, e.g. trees and livestock; costs of ownership transfer on non-produced assets, like land, contracts, leases and licences; (7) R&D, including the production of freely available R&D (this has not yet been effectuated) (8) mineral exploration and evaluation; (9) computer software and databases; (10) entertainment, literary or artistic originals; (11) other intellectual property rights. ‘Acquisition’ also consist of purchases of capital goods from abroad, like ships or software patents. In Chapter 3 we’ve treated the Irish Microsoft example. In the case of households, spending on durable production goods does not count as ‘investments’, except for building or extending your own house including new kitchens and bathrooms. In the latest NSA guidelines, investments in brands and military equipment are included in investment too. Proposals to do this can already be found in Kuznets (1955). Investments add to the stock of produced fixed capital. The case to exclude household durables from ‘investment’ can be rationalized as household durables are also not counted as gross fixed capital. There is logic to this idea. Companies use capital to produce monetary value, households use it to produce use values. Idle capital is expensive for companies while the costs of purchasing a durable good, like clothes or a caravan, are often experienced as a ‘sunk cost’ by households, not using them is not experienced as a cost. Households often don’t care if their caravan is idle for ten or eleven months a year. Business have to. Especially when the production boundary one uses is the boundary of monetized production, there is a reason to make a distinction between fixed capital used by companies and consumer durables used and owned by households. Even if it is the same car used for basically the same goals. We will however argue that while it is important to make this distinction, it’s not always right to leave consumer durables entirely out of the picture as many pieces of gross fixed capital are used jointly with consumer durables. The case of the government is different again. The government provides useful public goods. ‘Bridges to nowhere’ which only serve to give a temporary boost to construction output are rare. And though a large part of government fixed capital does not serve the goal of making money but produces use value, like roads do for owners of cars, a lot of this use value enables companies to make money or households to get to their caravans. The timeframe of the government is also different from a company timeframe. Explicitly included in capital are items like ‘sea walls, flood barriers and dams’ even when these are not directly used to produce other goods and services. Coastal levees may be idle for years or even decades to come but are built nevertheless, which makes government investments often less volatile than company investment (or purchases of durable consumer goods). We should take heed of the complementarities between the different kinds of investment: smart phones are largely durable consumer goods with however use a lot of technology produced by government research and development (R&D). This R&D is classified as fixed investment while companies like Telfort own the radio

204  I stands for gross fixed capital formation masts but lease the broadband space (an unproduced asset) from the government. And the famous Millau toll bridge in France is built, financed and exploited by the Eiffel company but owned by (a subsidiary of) the French government (the bridge turns out to be more profitable than expected which means that the originally 75 years term of the lease will be cut to about 40 years), which all means that government expenditure should be considered ‘gross fixed investment spending’, too. This seems a moot point and it is. But is has to be made as the models consider government investment wasteful by definition. The new definitions differ from older classifications which excluded weapons systems and intangible assets like software. It can be discussed if weapons are used for productive purposes. But there is a lively market in second hand weapons and weapons systems which means that they can be used to carry monetary value from one period to another. The same holds for intellectual property rights; ‘free’ software is however also included in GFCF. A problem with accounting for software and patents is that R&D is counted as investment but that there is, unlike in the case of houses or trucks, no explicit relation between the value of software and the cost price of research. Once developed, multiplying software is exceedingly cheap while its price depends on monopoly positions and possibilities for rent extraction instead of on costs of production. The same holds for artistic originals like books or music. An interesting borderline case included in the definition of GFCF is: ‘terminal costs, i.e. large costs associated with disposal, e.g. decommissioning costs of nuclear power stations or cleanup costs of landfill sites’. Demolishing items produces fixed capital. Accounting logic requires one to charge the not yet demolished items with a negative price. But as far as my knowledge goes, this does not happen. Changes in inventories are also important as these changes are, with profits and (changes in) unemployment on the income side as well as changes in monetary savings on the production side, one of the accounting identity ‘buffers’ which ensure that, ex post, the national accounting identities are identities. Stated otherwise: if a company does not sell its production, the resulting increase in stocks is categorized as an investment expenditure by the company and is measured ‘by the value of the entries into inventories less the value of withdrawals and the value of any recurrent losses of goods held in inventories’ (Eurostat/European Commission 2013). Acquisitions of existing GFC from other countries, like purchases of second hand weapon systems, second hand planes of software, are recorded as an investment of the units in the purchasing country and a disinvestment of the units in the selling country. It turned out that the ESA 2010 definitions had a problem. The new guidelines expanded the definition of gross fixed capital to include patents and software while at the same time shifting to a less location specific and more ownership based concept of production. Counting spending on unproduced non-natural assets like patents or fixed assets produced in other countries but legally transferred to another country as investment leads to problems with the national accounts, especially in the case of relatively small countries. The new, more ownership instead of location based national accounting rules of ESA 2010 exacerbate this problem (Fitzgerald 2016). Figure 7.4 shows gross fixed investment in

I stands for gross fixed capital formation 205 two comparable economies, Denmark and Ireland. The volatile nature of investments in Denmark is clear. Danish investments declined with almost over 20% after 2008. Danish decline and volatility is however dwarfed by developments in Ireland: -50% in two years. The first Irish decline was related to the housing bust and led to extreme unemployment, a severe slump and a decline of construction of about 90 percent. But the subsequent rise of, ahem, 341% between 2011 and 2016 (largely financed by in company payables) was not caused by the production of items but by the transfer of ownership of fixed capital items (software but also planes) from companies outside of Ireland to their legal headquarters in Ireland. According to the rules of national accounting the transactions would not have been considered to be Irish investment if these companies had been post box companies. But as there are quite of number of people working at these companies they are considered to be genuine companies – this next to this function as a traditional postbox company, which led to a growth of this sector in 2016 of 119 percent (nominal) while the rest of the economy increased with 8.7 percent in nominal and 7.3 percent in real terms, leading to a growth of the total economy of around 36 percent. At the same time, employment increased with about 3.5 percent (Central Statistical Office 2019). Treating the purchase of existing fixed assets as growth in such situations basically wrecks derived indicators like the capital/labor ratio, productivity and growth in such cases. Even the most extreme swing in ‘real’ investments in Ireland after 2007, the decline of construction, is dwarfed by the transferring of ownership deeds from one country to another.

7.4  ‘Investment’ as a variable in the DSGE models The previous paragraphs leave us with a four dimensional definitional grid of investment. Axes of this universe are time, sectors (including the government), kind of assets (including non-produced assets) and changing definitions (for instance with regard to weapons systems). Entering this universe teaches us that over time sustainable levels of investment as well as relative prices of different classes of fixed investment goods and services change. Also, investment spending is volatile in the short run while in the long run the level of gross fixed capital formation is connected to phases of economic development, which mean that investment spending is pivotal to as well the analysis of short run business cycles as long run economic growth. The question is: how relate these time varying investments to ‘investment’ in DSGE models? To be able to state a little about the concept of investment in the models we first have to investigate the neoclassical concept of capital in more detail, as investment and capital are connected. The ‘workhorse’ concept of capital used in mainstream economics is largely based on growth theory and, therewith, limited to fixed depreciable capital. This shows in the development of thinking about investments. In 1946 Evsey Domar stated that a high rate of net fixed investment does not only lead to an increase of the stock of capital and therewith potential supply (Domar 1946). But it also contributes to aggregate demand. In

206  I stands for gross fixed capital formation combination, however, the longer run effect of investments means that aggregate demand has to increase to keep up with the productive capacity of the economy which means in case of a declining rate of investment that household or government consumption does not only has to increase to keep up with supply but also because investment spending declines. Other solutions, like shorter work weeks, are possible; we’ve already seen that this solution was characteristic for the 20th century – albeit not in a gradual way. As such, the Domar article was a clear and conscious attempt to write down an intertemporal stock/expenditure flow consistent model which related the (change of) the stock of capital to expenditure and potential as well as actual production as defined and estimated in the new national accounts. Somewhat earlier Harrod was less explicit about this but he, too, tied investment expenditure to the increase of the stock of capital, therewith connecting the demand side of the economy to the intertemporal development of the supply side (Harrod 1939). In more modern parlance: Harrod and Domar used a ‘perpetual inventory method’ to estimate the macro-stock of capital using the flow of fixed investment to estimate the stock of capital as well as to analyze aggregate demand in relation to potential output – an important theoretical step for economics. On a meta-level such thinking was, as it was phrased in in the terms of the new national accounts, related to Keynes ‘How to pay for the war’ (Keynes 1940). Domar as well as Harrod however used a restricted definition of capital as they only took depreciable fixed assets into account, though their method also holds when non-depreciable and/or unproduced assets like ‘land’ or ‘oil’ or patents’ are added to their stock of capital. They also discarded the liability side of the balance sheet as well as financial flows and stocks, which disabled a genuine analysis of the distribution of capital income as well as of the financing of investment.2 The same restricted concept of capital and the balance sheet was used by Solow who, in his famous 1956 article, however purged flow consistency from growth theory by assuming neoclassical general equilibrium: whatever the flow of monetary investment expenditure, full employment would be maintained as wages and interest rates and employment and profits would change, miraculously, just enough to assure ‘knife edge’ full employment (Fazarri et al. 2013), which meant that not just financial flows but also output gaps, crucial to the work of Harrod and Domar, were assumed away. The decoupling of investments from aggregate spending was a clear case of unmitigated scientific retrogression as it made a lot of questions difficult to pose. This was not the only Solow retrogression. Kuznets (1955) contains excellent data showing the gross and net investment rate for dozens of countries for the pre-1938 and the post-war period as well as some longer term series for a limited number of countries and pays ample attention to the financing of investment. Rostow (1955) also points out the importance of changes in the long run rate of fixed investment and pays ample attention to the importance of the political, historical and cultural evolution necessary to enable this, therewith foreshadowing modern growth theory (Lee and Lloyd 2018). At the time, Kuznets and were both well-known authors. Solow neglected them, including the careful discussion of Kuznets of the definition of gross fixed

I stands for gross fixed capital formation 207 investment and his choice for a relatively broad concept of gross fixed capital. And while Harrod and Domar chose to use a limited concept of capital and to discard unproduced assets, the general equilibrium view of Solow forced him to do this as wages and profits would only change with the right magnitude when no rent incomes would exist. He therefore had to state: ‘The community’s stock of capital takes the form of an accumulation of the composite commodity’ and ‘there is no scarce nonaugmentable resource like land’ (Solow 1956, pp. 66–67). Note that in stark contrast to the definitions of Kuznets, who points out the importance of changing average depreciation rates caused by changes in the mix of fixed assets, in the model world of Solow investment goods do not differ from consumption goods: the coconuts-concept of investment. A coconut can be consumed or stored or planted. In the first instance, it’s an consumption good, in the second and third example it’s an investment good. The very idea behind distinguishing ‘capital’ from consumer and intermediate goods is the fact that the composition, use, span of life and ‘span of production’ of fixed depreciable assets is not equal to the composition, use or span of life and production of either final consumption goods or intermediate inputs. A bridge is not a banana and growing a coconut tree takes time while consuming a coconut doesn’t (or at least only a little). Taking the radical view: the very fact that a car is owned by a company and not by a household makes it different. A palm oil tree on a large scale plantation is, in an economic sense, not the same thing as a palm oil tree on a plot of land used by a semi-subsistence producer or a ‘free’ palm oil tree in the WestAfrican forests. Looking at the land on which the tree grows makes this even clearer. The market price for palm oil means something different when imputed in the accounting, organizational and technological system of the plantation and a global supply chains with administered prices and land as non-produced capital with a well-defined monetary value than when it’s substituted in the accounting and technological system of the subsistence producer. The subsistence accounting system might be non-monetary and in the head of the subsistence farmer only – but that’s the point. Changing the environment might change choices (emphasis added): Under current conditions in south Benin, where it is now possible to own land, these planters purchase parcels of land that they devote specifically to the cultivation of oil palm. These new planters are almost entirely men. Women small-scale producers are rarely able to own their own palm plantations. The oil palm’s status as a cash crop, reinforced by a symbolic aspect (as a ‘symbol of wealth’), has given rise to a process of growing monopolization of the sector by men. These male planters are fully aware of the profits they can make from processing, especially if they have the capacity to stockpile. (Carrere 2013, p. 22) The most fundamental time inconsistency of the valuation of capital is not related to arithmetic. It’s related to the changes in the structure of production and ownership engendered by investments, which influence the economic value of assets.

208  I stands for gross fixed capital formation It’s a situation which, in a financial sense, becomes even more complicated when nominal interest rates changes change the cost of ‘growing a tree’ especially for the plantation but much less for the subsistence producer. It just makes no economic sense to understand fixed capital as being basically the same product as the consumer goods. Ownership itself is a defining characteristic of a good or service. Capital is owned by a household or a company or the government which is a basic characteristic of a fixed asset. This style of thinking might fetch the reader as out of the ordinary. In the DSGE model of Burriel, Fernándes-Villaverde and Rubio-Ramirez (2009) it is stated that ‘the export goods are produced by monopolistic competitors who buy the final domestic good and differentiate it by brand naming’, i.e. by changing what marketeers call the core product or the idea of the product as experienced by consumers. If we follow this line of reasoning it is not too farfetched to, instead of adding a brand to a consumer product, add an owner to a brand, which indeed is one of the central tenets of marketing. But even understanding Solow’s definition of capital as a convenient modelling strategy instead of a serious definition, close reading reveals that it still seems to exclude houses, roads and other buildings, including the land below these structures. More than thirty years after the publication of his article Solow very approvingly cited, in his Nobel lecture, the at that time still preliminary empirical long run estimates of the stock of capital engineered by Wolff (Solow 1987). Checking the work of Wolff it turns out that he uses long-term estimates of capital obtained from Maddison (Wolff 1991, footnote 5). Checking the Maddison estimates it turns out that these exclude houses, land, natural resources, international assets, gold, and farm animals (Maddison 1982, Annex D).3 Restricting our attention to the seemingly trivial item ‘farm animals’ it turns out that, around 1885 and according to Goldsmith, these still made up about 13 percent of the total stock of reproducible assets in the USA (Goldsmith 1985, table 45) – and the importance of the other items not included in the operationalization of capital used by Wolff was often even larger. It’s clear that any long-term estimates of capital and labor productivity, like those of Wolff, should include not just the wagons but the horses and the (land underlying) roads, too. The Wolff estimates don’t. And a quick check of a number of macro-textbooks reveals that the Harrod/Domar/Solow/Wolff approach to tangible capital, i.e. implicitly restricting it to (a subset of) produced depreciable capital and discarding the liability side of the balance sheet, is still dominant in economic thinking. Kuznets (1955) did better (having had a Russian education Kuznets no doubt read German and French – just like his boss Wesley Mitchell). Modern growth theory often does include ‘human capital’ or even ‘health’ in the definition of capital. But the point here is that many kinds of produced gross fixed capital as well as unproduced capital are left out of the equations as well as out of balance sheets, which leaves us with the conclusion that, despite early attempts at stock/flow consistent estimates of capital which combined the demand and the supply side of the economy serious, the concept of ‘capital’ and hence ‘investment’ used in these studies applies to a subset of the total amount of tangible capital only while it’s, despite excellent data on the

I stands for gross fixed capital formation 209 flow of investments and capital income as well as, more recently, the total stock of capital, focused on the supply side only. This was the state of the art of leaving crucial parts of reality out of your models when DSGE modelling took off, which left out even more. The government is, in many of these models, only a quite expensive rule begetting algorithm which means that weapons systems but also roads and bridges are not included. At least – not when these roads, bridges and weapons are owned by the government. Guns and bridges owned by companies are capital, according to the models. Returning to the Millau bridge: DSGE models do not count this bridge as a piece of fixed capital. Basically, the Eiffel company leases it but it is owned by the government, not by a company, which according to the logic of the models makes the lease a kind of waste. Also, non-produced assets (natural as well as legal assets) are not a part of the models while series on investment are, if used at all, smoothed using a HP-filter (a complicated kind of running average) which means that the short time volatility of the series is filtered out. And none of the models I’ve read mentions long-term shifts in the equilibrium rate of investment connected to the historical phases of economic development of a country which were so important to Kuznets and Rostow. A DSGE shot at defining investment more precisely is the 2011 handbook chapter by Christiano, Trabandt and Walentin (2011). This article states that there is only one good in the economy which can be used as a capital good or consumed (a comparable model in Khramov 2012). Capital goods are owned by households and all investments are household investments while households rent capital goods to companies. Households invest to make up for depreciation and to fulfill their expectation of a steady growth path. This habit can, adjusting for the utilization rate of existing capital, be exploited by for instance monetary policy. Land and weapons and houses and bridges are absent from the models, households themselves do not seem to use capital and the government does not seem to own and use capital. ‘In house’ capital accumulation of intangibles or other capital goods by firms does not seem to exist. This situation is of course not very satisfying. A younger generation of DSGE economists does try to add institutional detail to the models. Bokan et al. (2016) defines ‘households’ as a heterogeneous category. Treating them as one single entity might divert attention from differences in ownership between households. Also, ‘housing’ is in this model ‘location’ intensive (houses can’t be transported from one country to another) and hence a fundamentally different kind of capital than items that can be traded an transported. In combination with flows of (existing) money from one country to another this yields the possibility of booms and busts: prices are not general equilibrium prices. They solve the modelling problems connected with these model results by introducing three kind of households: entrepreneurs who own all the capital and live from renting out this capital (let’s call these ‘capitalists’), ‘lazy’ households which like to borrow now and pay later and thrifty households which have the opposite characteristics. Households have nothing to sell but their labor. Alas, their data on housing do not make any use of the treasure trove on data on housing while the (re-)introduction of economic class

210  I stands for gross fixed capital formation does give the model an genuine ‘classical’ flavor but is not consistent with the statistics (which use other, more ‘bourgeois’ definitions of class than Bokan et al. do). But the de facto re-introduction of unproduced fixed capital is a welcome step into the direction of the statistics. They also introduce two distinct kinds of fixed capital, mainly distinguished by the depreciation rate (houses have a much longer depreciation rate in their model). This also means that, albeit in an implicit way, there are different flows of investments which are financed in a different way. Attempts are also made to include government capital and, hence, government investment (Stähler and Thomas 2011; Iwata 2012). The interesting thing about these studies is that they show that government investment is in an implicit way included in Christiano, Trabandt and Walentin (2011) and Khramov (2012) – but the resulting capital is considered to be wasteful by definition. When a company buys a truck it’s a productive investment. When the government does the same it’s wasteful expenditure. To underscore this remarkable aspect of the models: the title of the Iwata (2012) study is: ‘Non-Wasteful Government Spending in an Estimated Open Economy DSGE Model: Two Fiscal Policy Puzzles Revisited’. The default-concept of government spending in DSGE models is: ‘wasteful’. Stähler and Thomas (2011) and Iwata (2012) show that this not has to be the case in DSGE models. It’s is a modeling choice, even within the boundaries of DSGE modelling. But it is the choice generally made. Laudable as these efforts may be – they are still not based on a thorough reading of the statistics and a careful application of the statistical definitions to the models. Looking at the definition and the operationalization of these variables any connection with the statistical realm seems absent – no sources on statistical definitions are mentioned, which leaves us with a situation that already in 1955 investments were well defined and measured using all kinds of sectoral and type-specific subdivisions with ample attention to financing of these investments, which were embedded in an institutional, historical analytical approach. Now, more than sixty years later, economists try find their way back to this situation but have difficulties to do so as they lack the basic skills to investigate the concepts, definitions and operationalizations and to twist their models to enable them to include the basic data. A general confusion in the models is that they are ‘real’ in the sense that they are supposed to be about investment in physical capital. Very explicit is Sims (2016), which is supposed to be an educational resource and which therewith can be understood to bring the essence of DSGE to the fore. But investments aren’t real in the physical sense, not even in the models. The ‘real’ series are derived from deflating the nominal series of investment, which means that it is a nominal series recalculated using some set of basic prices and not a real series in the sense that its measured in whatever kind of physical units. This sounds esoteric – and it is. But it’s also important. In real life, the composition of aggregate investment shows vehement changes as is well known from the construction sector. As the price index of for instance R&D (part of investment spending) is different from the price index of construction, a sudden decline in construction means that the

I stands for gross fixed capital formation 211 weight of construction in the price deflator has to decline, too and even more so as a decline in construction invariably leads to lower relative prices for construction. This however runs into trouble with the stock of existing capital, which despite a decline in construction still does exist of a large amount of real estate and it’s hard to defend that the deflator of the flow of investment spending with, after a construction bust, a low weight of construction has to be used to deflate this stock of capital where construction still looms large. Also, prices of different classes of existing fixed capital also change in different ways, which adds to the confusion, a problem aggravated by changes in depreciation rates between assets as well as changes in technology and scarcity of land or energy, which change relative prices between for instance existing planes and airports. The reader might think that it is possible to come up with solutions for this. None are, to my knowledge, actually used.

7.5  Time consistency Another aspect of the investment series in DSGE models (this also holds for all other series used in these models) is that smoothed series are used. The high variability of investment is smoothed or averaged away which means that one of the marked characteristics of measured investment is not considered. The same holds for the long-term changes in the level of investment! Typically, the models use a so called HP filter to smooth data, the length of the smoothing period however does not seem to comprise the many decades necessary to map the rises and declines which are such a marked characteristic not so much of the investment series but of successful modern economic development. This is consistent with the gist of Solow (1956) which also abstracted from such issues. As we will see, such an abstraction was warranted for the US case. But it is not warranted for whatever other country we have information about, a point already stressed by Rostow (1955) and statistically proven by Kuznets (1955). Aside from this, HP-filters are increasingly criticized for yielding spurious cycles (Schüler 2018; Hamilton 2017). The last thing one wants when calibrating models which are intended to explain business cycles as well as longterm developments is a detrended series which shows spurious cycles caused by the smoothing itself. A somewhat comparable development is caused by changes in the interest rate. Laudable as the efforts to introduce institutional detail to the models may be – they are not based on a thorough reading of the statistics. Looking at the definition and the operationalization these variables, any connection with the statistical realm seems absent, which leaves us with a situation that already in 1955 investments were well defined and measured using all kinds of sectoral and type-specific subdivisions with ample attention to financing of these investments, which were embedded in an institutional, historical analytical approach. Now, more than sixty years later, economists try find their way back to this situation but have difficulties to do so.

212  I stands for gross fixed capital formation

7.6 Measurement and our intertemporal understanding of the macroeconomy: the rise and decline of the investment rate of the Western world, ca. 1807–2018 Whatever happened to the secular rate of investment in Western countries since 1807? In this paragraph, we will try to answer this question. The paragraph does not aim to ‘explain’ the rate of investment or to make a quantitative estimate of the consequences of changes in the secular rate. It wants to show how measurement of crucial variable – the rate of gross fixed investment as a percentage of GDP – can be measured and might, in an inductive way and against the background of some knowledge of the definition of the variable as well as of secular historical developments, influence our view of the historical nature of our macroeconomies. Even when my series stretch further back in time than those of Kuznets (1955) and (obviously) also cover the 1951–2017 period. The contents of this paragraph pale in comparison with the contents of this 65-year-old masterpiece. The reader is also advised to read Rostow (1955) and to take his warnings about the historical contingency of the nature of investing and investments to heart: historical and national differences were commonplace. Important: the paragraph is about gross fixed capital formation which means that an increase in the average rate of depreciation caused by larger prominence of items with a shorter depreciation time as well as, of course, a larger stock of capital might have led to an ever larger difference between gross and net investment. See also Kuznets (1961). So, what did Kuznets state? His only really long-term rate was for Sweden. According to him ((1955, pp. 31–33): only the record for Sweden reaches back into the period of rising capital formation proportions . . . What the few long series do reflect are the high capital proportion levels reached some time after the acceleration in the rate of growth occurred; and the declines, usually gradual but sometimes abrupt, from these levels. In the United States, Canada, the United Kingdom, France, and Denmark the capital formation proportions tend to decline – abruptly in the United States and Denmark – after World War I. In Sweden they tend to rise during the first decades, declining toward the end of the period covered. Also in the United States and Canada the relative importance of depreciation increases, as expected, and causes a greater decline in the net capital formation proportion than in the gross. But the relation of the two series is quite different in the United Kingdom, and we are at a loss to decide whether it is because of some peculiarities of the estimating procedure or whether it reflects genuine differences in durable capital consumption practices or in the composition of the total capital stock. . . . The preceding comments suggest a long secular swing in the capital formation proportion. Kuznets is right about the past and totally right about the occurrence of secular swings – even if he did not presage that the high post war levels of investment

I stands for gross fixed capital formation 213 would continue and would cause and enable a disruptive change to the Western style of live. His line of investigation however did not catch on. Systematic overviews of the rate of investment in the spirit of Kuznets and using longer time series are rare. Even when we have the measurements. Below, series presented in Knibbe (2014) are updated, which enable us to test Kuznets’s conjectures – a test which subsequently will be used to evaluate the statistics and the models (and even the science of macroeconomics in general). The graphs show gross fixed capital formation as a percentage of gross domestic product. Compared with the gross series of Kuznets they are extended into the past, beyond 1950 and give yearly instead of decadal information which enables more precise conclusions. Kuznets (1955) series include military equipment, just like the newer series for the countries discussed in the following sections and consistent with the latest guidelines (ESA 2010 [2013]). As I was fortunate enough to still have the old Eurostat series, which in true ‘1984’ spirit are erased from the Eurostat databases, the new series compatible with ESA 2010 can be compared with the old data to investigate the influence of these changes in definition.

7.6.1  The USA: exceptionalism The USA graph shows series including and excluding military equipment. This serves to highlight a conceptual problem: ‘how to define investment?’ As there is a clear primary and second hand market in military gear while countries replace equipment lost in wars, tanks and planes do have ‘economic’ value which means that, according to SNA logic, they have to be included in our concept of gross fixed capital, which means that spending on military equipment is an investment. And the USA data surely suggest that military investments did contribute to the end the great depression: note that military spending in the USA increased with leaps and bounds from 1939 onward.4 In a comparative perspective the most remarkable aspect of the US series is, as we will see, the comparatively high rate of investment in the 19th century in combination with, despite some extreme short-term swings, a remarkable long rate stability. For other countries, the high 19th-century-level of US gross fixed capital formation surely was a sign of things to come. The, compared with other countries, relatively high level in the 19th century was caused by a rapid increase of the urban as well as the agricultural population in combination with the extension of the system of railways. This finding is consistent with the finding of Wolff (1991) of a comparatively high as well as stable ratio of fixed depreciable capital to labor in the USA during this period. Part of the high rate of investment might have been caused by the necessity of high investments in the large empty spaces which were opened up for the Western economic system which required initial investments in local roads, bridges and comparable which in the course of centuries already had been carried out in areas like China or Europe (on this also Kuznets 1961). In economic parlance: the high ratio of unimproved land to labor necessitated high investments to improve land. In the short run, US gross fixed capital formation was highly volatile. Net investments (gross investment minus

214  I stands for gross fixed capital formation 40

35

30

25

20

15

10

5

0

1839 1849 1859 1869 1879 1889 1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 2009 'Historical' data, including defense

BEA data, excluding defense

New

Graph 7.2  Nominal gross fixed investment rates, USA (% of GDP) Sources: see Graph 7.1 as well as www.nber.org/databases/macrohistory/rectdata/10/a10037. dat www.nber.org/databases/macrohistory/rectdata/08/a08165.dat; also FRED series A006RE1A156NBEA; A822RE1A156NBEA; W170RC1Q027SBEA; FYGDP

depreciation, not shown here) even became negative during the Great Depression. Especially after World War II, the US investment ratio was rather stable in an arhythmical sense. But was it also stable in an economic sense? As we will also see for other countries – Finland, Spain, Ireland and, to a lesser extent, France – the post-2000 level of investment was driven by a construction bubble and as such unsustainable. Did the fundamental rate of investment change as resource intensive expenditure on houses, roads and harbors included in the rate of fixed investment has become lower in a relative sense once levels of GDP increased, populations aged and expenditure shifted to private cars, recreational vehicles and artificial hips which are not embedded in the statistical framework? Just like the high 19th-century level, the low post-2008 US level of fixed investment (as presently defined) might again be a sign of things to come.

7.6.2  France and the UK: catching up versus falling behind France shows rather low but gently rising levels in the 19th century, followed by increases after 1920 and, again, after 1945 which led to very high investment ratios in the 1960s and 1970s, a period also known as les trentes glorieuses. The

I stands for gross fixed capital formation 215 40 35 30 25 20 15 10 5 0 -5

1815 1825 1835 1845 1855 1865 1875 1885 1895 1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005 2015 France, eurostat+historians

France, Insee

UK

Graph 7.3  Nominal gross fixed investment rate, France and the UK, 1815–2017 Source: see Graph 7.1

UK experience is slightly anomalous: it shows, in a comparative perspective, low rates up to as late as 1920. The relative low rate of investment in the 19th century is, considering the rise of cities like London, the railway system and the like, surprising. The relatively low UK 19th- and 20th-century rates make us distrust the data a bit but in 1979 ‘fixed depreciable capital/labour’ ratios were, contrary to the situation around 1870 and even 1890, way lower than for instance in Germany (Wolff 1991) which corroborates these low levels. The high French rate after 2005 is remarkable, too. Even then both countries large and by do show the same pattern: a gentle increase after about 1920 and a more sudden jump after about 1948 followed by the recent decline. In a European perspective, the fertility rate of both the UK and, especially, France is relatively high while net immigration into the UK has been relatively high compared with other European countries. In combination this lead to relatively high rates of population growth in these countries compared with for instance Germany and Italy. At the moment, this has not yet translated itself into higher rates of gross fixed capital formation.

7.6.3  Germany, the Netherlands and the Nordics: archetypes First, note that the borders of Germany changed quite a lot during the period in question. Despite this the patterns for Germany, the Netherlands and the Nordic countries are consistent: a rather late take-off of the Netherlands in combination

216  I stands for gross fixed capital formation with a rather high level of investment before this period, the high level in Germany during the ‘Gründerjahren’ in the 19th century and a relatively high German level after World War II (graph in next paragraph). The most remarkable aspect of the graph is the sustained post 1973 decline of the German and to a lesser extent Dutch investment rate – even despite German reunification. The comparability with the pattern for the ‘Nordic’ countries is striking. Finland sustained a high rate for longer than the other Nordic countries but as this was related to a construction boom induced by deregulating the financial system this came at the cost of a severe financial crisis around 1990. As such, the 150-year wave-like pattern of Germany, the Netherlands and the Nordics seems to be typical for Western ‘latecomers’ to modern growth. The enduring high level of East German unemployment (20+ percent for almost a decade) in a period when the investment rate declines indicates a consistent pattern of misallocation of resources and shows a striking contrast with the fast decline of West German unemployment after 1950. Eventually the spending shortfall caused by a low rate of investment – also low in a structural way as net government investment was negative for quite a part of this period – was filled by a large external surplus. A strategy which for obvious reasons can’t be used by all countries at the same time. As net government investment in Germany is, at the moment of writing, negative a slight uptick to make up for deteriorating public infrastructure might be expected will have to crowd out the current account surplus. Remarkable are the high levels in Finland and Ireland before respectively 1990 and 2008. These high rates were caused by construction booms. The busts are clear, too. As indicated earlier, the high level of Irish investments at the end of the period was caused by a transfer of ownership of intellectual property products from the USA to Ireland.

7.6.4  Italy, Spain Both Italy and Spain show the same long-term wave, with Spain as a latecomer and, like Finland, at the end of the series showing a financial deregulation connected investment bubble (which eventually resulted in a financial crisis). The gap for Spain between 1958 and 1980 can be filled with the well-known Penn series of the investment rate. The Penn series shows data in constant prices (i.e. nominal data deflated with a price index) which are often are quite different from current price rates. The data suggest that, after a severe decline during the 1959–1960 depression, investments were relatively high throughout the 1960s and especially in the 1970s, declining a little thereafter. The fixed price Penn series shows a much higher rate of investment that the current price series, which calls for an explanation. As the Penn investment data are probably deflated with an investment price index while the Penn GDP data are deflated with the GDP a decline of relative prices of investments may have caused this difference. The Italian peak around 1970 takes place at the same time as the peaks in the Netherlands, Denmark, Norway, Sweden, France and Germany. This is remarkable as Italian productivity was, at the time and like UK productivity, considerably below levels in northern Europe and the USA. It is tempting to assume that as in the

35

30

25

20

15

10

5

0

-5 1807 1817 1827 1837 1847 1857 1867 1877 1887 1897 1907 1917 1927 1937 1947 1957 1967 1977 1987 1997 2007 2017 Germany

Germany revised

Netherlands

Graph 7.4  Nominal gross fixed investment rate, Germany and the Netherlands, 1807–2017 Source: see Graph 7.1 40

35

30

25

20

15

10

5

0 1844 1854 1864 1874 1884 1894 1904 1914 1924 1934 1944 1954 1964 1974 1984 1994 2004 2014 Finland

Denmark

Ireland

Graph 7.5  Nominal gross fixed investment rate, Finland, Denmark and Ireland, 1844–2017 Source: see Graph 7.1

218  I stands for gross fixed capital formation 40

35

30

25

20

15

10

5

0 1850

1865

1880

1895

1910 Spain

1925

1940 Italy

1955

1970

1985

2000

2015

Spain, Penn data

Graph 7.6  Nominal and a ‘real’ gross fixed investment rate Italy and Spain, 1850–2017 Source: see Graph 7.1

other countries a relative lack of investment opportunities led to the synchronized decline of the investment rate, the difference in productivity rates however means that this idea has to be rejected. For the period up to (about) 1955 the results corroborate those of Kuznets (1955) even when there was a surge after about 1955. Except for the USA all countries show, with relatively little difference in timing, a 150-year wave-like pattern of the investment ratio. It is interesting to compare this with the findings of Piketty (which are consistent with the national accounts) who does not look at fixed gross capital formation but at fixed gross capital. Counter intuitively, high investment ratios were characteristic for the period when, according to Piketty and Zucman (2014), the fixed capital/output rate was lowest while the opposite held for the period after 1970. After this year, gross fixed capital formation declined but the fixed capital/output ratio increased. According to Wolff, however, the depreciable fixed capital/output ratio increased during the years of high investment, the difference between this ratio and the total fixed capital ratio mainly being land. It seems that this pattern was also characteristic for the USA but with the onset of this development starting some decades earlier. Clearly, changes in the {total capital/output} ratio were not just influenced

I stands for gross fixed capital formation 219 by investment but also by price developments, most notably the decline of the value of agricultural land after about 1880 (caused by lower prices for agricultural products) and the increase of the value of land underlying houses and structures after around 1980. Most countries show a decline of the gross fixed capital formation rate starting somewhere in the 1970s or 1980s. Countries which managed to keep a stable or even increasing rate of investment after this period seem to have paid for this with severe financial busts. Somehow, the data suggests, the high rates of the 1960s and 1970s were not sustainable anymore after around 1980 which might have been caused by a lower increase of the population, a shift in opportunities and whatever. But just like the value of the stock of capital is influenced by changes in relative prices, so is the rate of investment expressed as a part of GDP: lower relative prices of depreciable fixed assets might have moved the rate of investment down. This view is challenged by a recent report of the ‘Project for strong labor markets and national development’ on business investment (nonfinancial incorporated and unincorporated companies) (N.n.1 2019). It argues that financialization has led to a turnaround in investment behavior: incorporated businesses used to be borrowers and investors in fixed assets but have become, on average, lenders and investors in financial assets. Comparing Figure 7.1 of the report to Graph 7.2 indeed shows that the size and pattern of the decline of the investment rate is almost solely due to a decline in business investment, which shrank from about 10 percent (peaks) of GDP in the 1960s of the 20th century to about 4 percent now. The report points out that according to several measures profitable investment possibilities exist but a combination of the idea of financial shareholder value with excruciating pressure from, especially, institutional stock owners (who do not invest in physical capital but in financial assets) leads to an overly high ratio of dividend payments and share buy backs. They also show that especially around 2000, net borrowing of incorporated companies suddenly changed into net lending, while this did not happen for unincorporated businesses which, up to that point, showed the same level of borrowing as incorporated companies. This calls into question the idea that the structural level of profitable investment has declined even when it does not call into question that investment bubbles caused by real estate booms have become more prominent. It is a sketch of institutionalized financial rent seeking – meaning that compared with the past US companies are not creating markets anymore but leveraging existing and buying their own shares. What to make of all this? Kuznets was a bit at a loss, back in 1955. According to him, it were fixed investments which enabled use of new formal and informal knowledge and which hence led to economic growth. When it came to describing the facts and guessing the historical blanks Kuznets assumed low ‘pre-modern’ rate in every country. This turned out to be true. But he did not foresee the future: the investment and growth wave (in Europe) of the 1960s must have been a surprise to him – tough, considering this surge, he would have been less surprised by the decline after about 1970: swings in the rate of investment were

220  I stands for gross fixed capital formation central to his thinking. Remarkably, these swings are negated by the iconic 1956 Solow growth model (Solow 1956). Kuznets did not succeed in swaying the ahistorical nature of neoclassical macro. Part of this problem might have been caused by the concept of capital in the statistics and the models which focus on respectively the transaction economy or (the models) even the market economy only. Is this right? After 1950 European countries witnessed unprecedented levels of investment. These investments were connected to ‘the rise of the network economy’. Foreshadowed by canals in the Netherlands in the 17th century and railways in the UK and many other countries in the 19th century the decades post-1950 witnessed the growth of large scale water and sewer systems: the electricity grid, highways, radio and television networks, systems of super- and hypermarkets and, recently, 3G, 4G and 5G systems and the Internet. Even suburbs can be understood as part of an ever expanding houses-roads-mallsoffices network. As a consequence and in combination with the mechanization and robotization of agriculture, the ‘decentralized centralization’ of the population in sprawling suburbia ensued, a development mirrored in the overriding importance of houses, other buildings, roads and land underlying structures in our statistics of the stock of capital. Suburbia (including commuting times) is however primary a lifestyle decision, not an economic one. A lifestyle entirely and even dangerously dependent on public investments in roads and airports but also on the private purchase and use of durable consumer goods like cars (and bicycles). Cars (and bicycles) our however left out of the concept and definition of capital of the models as well as the statistics. There are reasons to re-classify this. In the long run, the change of our lifestyle not just depends on car factories (private investment) or roads (government investment) but also by purchases of household durables – and as now the world produces about 100,000,000 passenger cars a year which, valued at EUR€10,000 – each – will be worth about EUR€1 trillion. As is well known, trucks and cars are an important investment item. Trucks and cars owned by companies are supposed to add to fixed capital and hence an investment item. Trucks and cars owned by households and used for personal use, including commuting, are supposed to be durable consumption goods (light or pick-up trucks are classified as trucks, not as cars). Neither the national accounts not the models grasp this. A more integrated approach based on an extended but also more granular concept of investments and capital is needed which also takes account of the differences between commercial use of fixed capital, government use and household use: who spends how much on what, why does this happen and how does this interrelate?

7.7 Overview The information in this chapter leads to the next comparison of investment the models and the accounts:

National Accounts

Neoclassical Macro-Models

Basic method of accounting

Quadruple accounting

Basic method of valuation

Market or unit values or cost prices or estimates of replacement costs

Contains not owned natural capital.

No, meaning that expenditure on for instance catalytic convertors is not an investment as clean air is free. Land itself is not an investment but major improvements of land like clearing woods are considered to be investments. Yes. This also means that research and development is an investment

Basically single accounting. Inclusion of a financial sector in the models by necessity leads to more emphasis on double and even quadruple accounting but this has not yet been fully modelled. Model consistent valuations: a mark-up on the consumer price level. Investment goods do not have an own price level. No. But models could well include investments in catalytic convertors assuming that clean air yields utility.

Investments in ‘unproduced’ natural capital, like land and subsoil stocks of oil Contains owned ‘unproduced’ capital, like production permits and patents Contains government investments, like coastal defenses

Yes

Contains a complete set of sectors

Yes

A distinction between capitalists (‘entrepreneurs’) and labor exists Nature of investment

No

Detailed lists of investments exist, including for instance deconstructing nuclear power facilities. As long as expenditures yield monetary rewards (including non-monetary imputed rent for houses) in the future there are counted as investment.

No

No

In most models, government investments are considered to be a waste of scarce resources. NPISH and monetary financial institutions, i.e. money creating banks, are always or often excluded. There are models who do this, capitalists own the capital and invest. A choice between present and future consumption.

(Continued)

222  I stands for gross fixed capital formation (Continued) National Accounts

Neoclassical Macro-Models Often: mark-up on consumer prices.

Sectoral consistent (sectoral balance sheets match with each other)

Measured amalgam of cost prices, perpetual inventory methods, replacement prices and the market price of items which are sold on the second hand market. Only a limited relation with investment prices. When depreciation is taken into account: yes in the short term. In the long term, there are problems with valuation when technological and other changes cause changes in relative prices. Yes (but measurement problems with Net International Investment position)

Nature of financial market

Money creating banks plus loanable funds.

Measured or derived price of existing capital

Stock-flow consistent with production accounts?

Yes

Theoretically: yes. But sectoral division is incomplete (government, NPISH and monetary financial institutions are excluded). So: no. Loanable funds with international flows of capital.

Notes 1 Coastal societies surviving winters on dried salmon or cod or farming societies with large amounts of animals fed with hay did have quite a stock of capital in the physical sense and, in the case of hay, often also the monetary sense. As our basic unit in this book is the year and not the season investments in such seasonal stocks do not count. 2 On a net basis the accounting identity S = I of course holds when it comes to investment. On a Gross basis, large changes in balance sheets may accompany the events which by necessity make this identity true ex post. In Spain, during the pre2008 building boom, the accounting identity was true. More means were devoted to investment which meant that the net savings rate increased. However, other kinds of spending were financed with foreign loans which, when a ‘sudden refinancing stop’ occurred, lead to banking trouble: the banks had to pay back their short term loans but could not obtain the money from the people borrowing from banks as these loans were long term. Unemployment went from 7 to 25 percent. All the while, S = I was true. 3 Maddison continuously revised and extended his data (Maddison 1992, 1994). 4 The index number of federal fixed assets (1996 = 100) increased from 11 in 1939 to 17 in 1941, 34 in 1942 and 76 in 1944. In 1946 a decrease started which lasted

I stands for gross fixed capital formation 223 till 1951, when an inexorable rise started which accelerated after 1985 and tapered off after 1991. The post 1939, 1951 and 1985 rise were largely due to increased military spending. It is challenging to compare 1960 interstates with 2000 yet fighters, but the pattern of increase and decrease probably shows a genuine development (Nn1 2001, table 2).

Literature Bokan, N., A. Gerali, S. Gomes, P. Jacquinot and M. Pisani (2016). ‘EAGLE-FLI. A macroeconomic model of banking and financial interdependence in the euro area’. European Central Bank working paper series no. 1923. Burriel, P., J. Fernández-Villaverde and J.F. Rubio-Ramirez (2009). ‘Medea: A DSGE model for the Spanish Economy’. Center for Economic Policy research discussion paper no. 7297. Carrere, R. (2013 [2010]). ‘Oil palm in Africa: Past, present and future scenarios’. WRM series on tree plantations no. 15. Central Statistical Office, employment database. www.cso.ie/px/pxeirestat/Statire/ SelectVarVal/saveselections.asp. Accessed 15 May 2019. Christiano, L.J., M. Trabandt and K. Walentin (2011). ‘DSGE models for monetary policy analysis’ in: Handbook of monetary economics, volume 3A 286–364. Amsterdam: Elsevier. Dimsdale, N., S. Hills and R. Thomas (2010). ‘The UK recession in context – what do three centuries of data tell us?’. Bank of England Quarterly Bulletin 2010 Q4 577–591. Domar, R.E. (1946). ‘Capital expansion, rate of growth and employment’. Econometrica 14:2 137–147. Eurostat/European Commission (2013). European system of accounts 2010 (ESA 2010). Luxembourg: Printing Office of the EU. Eurostat (2018). https://ec.europa.eu/eurostat/statistics-explained/index. php?title=Glossary:Gross_capital_formation. Accessed 13 March 2018. Fazarri, S.M., P.E. Ferri, E.C. Greenberg and A.M. Varlato (2013). ‘Aggregate demand, instability and growth’. Review of Keynesian Economics 1:1 1–21. Fitzgerald, J. (2016). ‘Problems with the Irish national accounts and possible solutions’. Internal Memo of the Office of National Statistics. www.cso.ie/en/ media/csoie/newsevents/documents/reportoftheeconomicstatisticsreviewgroup/National_Accounts_-_problems_and_possible_solutions.pdf. Accessed 23 December 2018. Gallman, R.E. (1986). ‘The United States capital stock in the nineteenth century’ in: S.L. Engerman and R.E. Gallman (eds.) Long-term factors in American economic growth 165–214. New York: National Bureau of Economic Research. Goldsmith, R.W. (1985). Comparative national balance sheets. A study of twenty countries, 1688–1978. Chicago: University of Chicago Press. Guiso, L., A.K. Kashyan, F. Panetta and D. Terlizzese (2002). ‘How interest sensitive is investment? Very (when the data are well measured)’. Working paper, University of Chicago. Hamilton, J.D. (2017). ‘Why you should never see the Hodrick-Prescott filter’. NBER working paper no. 23429. Harrod, R.F. (1939). ‘An essay in dynamic theory’. The Economic Journal 49 119–133.

224  I stands for gross fixed capital formation Iwata, Y. (2012). ‘Non-wasteful government spending in an estimated open economy DSGE model. Two fiscal policy puzzles revisited’. ESRI discussion paper series no. 285. Jonung, L. (2009). ‘The Swedish model for resolving the banking crisis of 1991–93. Seven reasons why it was successful’. DG ECFIN economic paper no. 360. Keynes, J.M. (1940). How to pay for the war. A radical plan for the chancellor of the exchequer. London: Macmillan. Khramov, V. (2012). ‘Assessing DSGE models with capital accumulation and indeterminacy’. IMF working paper WP/12/83. Knibbe, M. (2014). ‘The growth of capital: Piketty, Harrod-Domar, Solow and the long run development of the rate of investment’. Real-world Economics Review 69 100–121. Kuznets, S. (1955). ‘International Differences in capital formation and financing’ in: National Bureau of Economic Growth (ed.) Capital formation and economic growth 17–110. Princeton: Princeton University Press. Kuznets, S. (1961). Capital in the American economy. Its formation and financing. London: Oxford University Press. Lee, C. and P.J. Lloyd (2018). ‘A review of the recent literature on institutional economics analysis of the long-run performance of nations’. Journal of Economic Surveys 31:1 1–22. Maddison, A. (1982). Phases of capitalist development. Oxford: Oxford University Press Maddison, A. (1992). ‘A long run perspective on saving’. Scandinavian Journal of Economics 1992:2 181–196. Maddison, A. (1994). ‘Standardised estimates of fixed capital stock: A six country comparison’. Research Memorandum 570 (GD 9). Rijksuniversiteit Groningen. N.n. (2019). American investment in the 21st century. Report of the project for Strong labor markets and national development with a foreword by Senator Marco Rubio. N.n.1. (2001). ‘Fixed assets and durable goods for 1925–2000’. Survey of Current Business September 2001 27–39. Palenzuela, D.R., S. Dees (eds.) and the Saving and Investment Task Force (2016). ‘Savings and investment behaviour in the euro area’. ECB Occasional paper series no. 167. Piketty, T. and G. Zucman (2014). ‘Capital is back: Wealth-income ratios in rich countries 1700–2010’. The Quarterly Journal of Economics 129:3 1255–1310. Rostow, W.W. (1955). ‘The take off into self-sustained growth’. The Economic Journal 66:261 25–48. Schüler, Y. (2018). ‘Detrending and financial cycle facts across G7 countries: Mind a spurious medium term!’ ECB working paper series no. 2138. Sims, E. (2016). ‘Graduate macro theory II: A medium-scale new Keynesian DSGE model’. Teaching paper, University of Notre Dame. https://www3. nd.edu/~esims1/medium_scale_dsge_2016.pdf. Accessed 4 November 2019. Solow, R.M. (1956). ‘A contribution to the theory of economic growth’. The Quarterly Journal of Economics 70:1 65–94. Solow, R.M. (1987). ‘Growth theory and after. Lecture to the memory of Alfred Nobel, December 8, 1987’. www.nobelprize.org/prizes/economic-sciences/1987/ solow/lecture/. Accessed 2 November 2013. Stähler, N. and C. Thomas (2011). ‘Fimod – A DSGE model for fiscal policy simulations’. Banco de España Documentos de Trabajo 1110.

I stands for gross fixed capital formation 225 Wikipedia (2018). https://en.wikipedia.org/wiki/Rail_transport_in_Sweden. Accessed 22 December 2018. Wolff, E.N. (1991). ‘Capital formation and productivity convergence over the long term’. The American Economic Review 81:3 565–579. wrm.org.uy (2019). https:// wrm.org.uy/articles-from-the-wrm-bulletin/section2/oil-palm-in-benin-fromsmall-scale-production-by-women-to-large-scale-corporate-industry/ Accessed 2 April 2019.

8 Unreal production

8.1  Introduction: how real is ‘real’ production Economists often work with ‘real’ variables. These are calculated by deflating nominal data. Nominal series – series based on prices which are actually paid – are divided by a price index which captures how much average prices have increased or decreased. This is supposed to yield a series of production or consumption or investment not influenced by changes in the price level. Intuitively, this makes sense: isn’t the economy not about the money people pay but about the items they purchase or invest in? But how ‘real’ are these calculated variables? Isn’t it the case, in the real world, that people actually pay these money prices? Also, how do economists estimate this price level and changes thereof? And are we using the right price indices? Were price indices developed to calculate real series at all, or where they developed to gauge the purchasing power of nominal income (a wholly different question). The answer to these questions is complicated and confusing. In 2015, Scott Sumner titled a blogpost about deflating nominal income data: ‘The bizarre way economists calculate real income’ (Sumner 2015). He was totally dumbstruck about the price indices used to ‘deflate’ disposable income. There are quite some price indices and it seemed to him as if economists did not care about the choice of the right index, sometimes even leaving prices of investment goods out of the deflation procedure of GDP even when total GDP spending includes investment expenditure. Be aware: his was not a critique of price indices but of the loose way these were used by economists. A year later, Yanis Varoufakis wrote a blogpost about the situation in Greece, where nominal production declined but ‘real’ production increased (Varoufakis 2016). The GDP price level decreased from 101 in 2011 to 94 in 2016 (2019 data from Elstat). A classic case of deflation, which meant that, for one thing, people (and the country) would experience larger difficulties to pay down debts as nominal income declined – a fact not captured by ‘real’ GDP and ‘real’ income as these debts are not an element of the nominal series which are deflated even when paying down a debt is as real as paying for groceries. Varoufakis put forward different arguments than Sumner but came to a comparable conclusion: we have to take great care when we calculate and analyze variables like ‘real production’, ‘real income’ or ‘purchasing power’. And it are not just such rogue economists who

Unreal production 227 take heed. Official statistical institutes do so, too. For instance when they point out that consumer price inflation is influenced in a non-trivial way by value added taxes (Eurostat 2018B). And especially when it comes to value added. Value added is the difference between the nominal value of production minus the nominal value of purchased inputs. Taking deflation logic serious, ‘real’ value added is the deflated value of this difference. Or is it, taking deflation logic serious, the value of deflated production minus deflated purchases? The first deflation method is called: single deflation. The second one: double deflation. According to the IMF ‘double deflation’ is the economists’ method of choice: A volume estimate of GDP is an essential measure of economic activity because it removes the effects of price changes. The System of National Accounts 2008 (2008 SNA) recommends a technique called double deflation. In contrast, single deflation, the deflation with a single price index, is not recommended because it fails to capture important relative price changes. (Alexander et al. 2017) Everybody seems to agree. Great care has to be taken. So, which deflators do we have to use and how should they be applied? Or, to state this more transparant: which set(s) of relative prices do we have to use to compare one period with another? And how does this relate to the neoclassical macro-models? The answers to these questions are somewhat tentative as the tension between deflating short term and long term data has not been fully resolved and some tests on historical production series have been carried out to get a clearer view of the issues involved. Also, the problem of relative prices, and qualities, does not only arise in historical comparisons of this year with last year or many years ago. It also raises its head when we compare different societies using different products and services to cope with different climates and cultures: how to compare the price of rice in Japan in 1925 with the price of potatoes in the Belgium in the same year? Which shifts the question a little. Instead of answering the question which deflators we have to use, it might be better to pose the question how we have to understand the results of using specific deflators or, more transparent, specific sets of relative prices to compare periods of societies. To be able to answer this question, or at least to discuss it, we however first have to investigate single and double deflation.

8.2  Single and double deflation Let’s first look at single deflation. Real purchasing power, like nominal household income deflated with the consumer price index, is a single deflated nominal variable. It’s also possible to deflate total nominal national income (which consists of wages, rents, interest and profit) with the consumer price index. But is this a sensible idea? A part of total national income is spent on investment, which means that it might be a better idea to deflate this with some kind of average of the consumer price index and a price index of investments. But while household consumption

228  Unreal production is relatively stable, investment is quite volatile which causes problems with the weights chosen. Also, poor people and rich people do not buy the same consumption goods and services and one might want to account for this, constructing an Aldi as well as a Gucci consumption price index. On top of this, in a country like Greece in 2010–2012 the consumer price level including changes of VAT and other price increasing taxes declined a full 6.5 percent less than the price level at constant tax rates. To state this otherwise: increases in indirect taxes increased the consumer price level with a full 6.5 percent in three years. Should we in this case use the price level including changes in indirect taxes or the constant tax rate price level to calculate the purchasing power of income? And this are not the only problems with single deflation. We all know that people will change their purchasing habits, to an extent because relative prices have changed but also because products change, or life styles change. An increase in the price level caused by an increase in the price of gasoline might be as large as an increase caused by a rent increase but it might, despite the fact that ‘real’ income is the same in both situations, lead to other behavioral reactions and other consequences for family budgets. This holds for microeconomic situations as well as for the macroeconomy. The development of new products like smartphones might influence behavior, too, just like the greying of populations. These changes are fundamental manifestations of modern economic life. Single deflation in combination with an analysis of changes in the ‘basket’ used to calculate the consumer price index can help us to track such events and to evaluate their influence on purchasing power of households. ­However – total income is of course also used for investments, which means that we should take care to distinguish between household income and income, like profits, which is often used for investment. And use an appropriate deflator to deflate this part of total income. This is the tractable part. Now for the confusion. Value added is not just equal to income. It is also equal to economic production which is equal to the value of outputs minus the value of inputs. Prices of outputs tend to develop in way diverging from the price level of inputs. This means that constructing a ‘real’ measure of value added by subtracting the deflated value of inputs from the deflated value of outputs requires the use of two different deflators, one for inputs and one for outputs, which yields double deflated value added. This makes for some complications. If the following discussion gives you the impression of entering an hall of carnival mirrors do not despair. You are. It will be argued that the weights chosen as well as the aggregation procedures used by statisticians influence the level and growth rates of ‘real’ macroeconomic variables – which distorts our insights unless we compensate for the distortions. As such, this is not disputed among specialists. But it will also be argued that this is inconsistent with the single good approach of the DSGE models. Before doing this it is important to point out what double deflated value added isn’t. It’s not a physical entity like a ton of pig iron, a megaton of CO2 or the per capita amount of calories available for human consumption. It’s also, unlike gross output or gross input, not a price weighted average of such physical entities. Deflated or ‘real’ economic production series, also called the volume of production, are value series of monetary output using fixed prices. And any value

Unreal production 229 based upon a price, fixed or not, is not a physical but a monetary value, even when we call it a ‘volume’.1 This is also true for ‘double deflated’ series of value added. Assuming that relative quantities of outputs and inputs do not change, we might have the idea these series, even nominal ones, still capture some kind of technical relation between inputs and outputs, like the use of feed cakes plus other current inputs like services of the veterinarian per kg. of milk. But relative quantities do change, which means that the series do not capture technical relations. They capture economic relations. But it’s not just the amount of inputs which changes. A large decline in the price of an input might lead to a large increase in its use, changing relative amount of inputs. According to the OECD, which like the IMF and the SNA defends the use of double deflation, this combination of events can lead to bizarre outcomes: ‘Another issue is the occasional occurrence of negative value added figures when double deflation operates with Laspeyres quantity indices. Nothing ensures that the subtraction of constant-price intermediate inputs from constant-price gross output yields a positive number’ (OECD 2001, p. 45). If this happens the OECD is clear on what to do: ‘In these circumstances different accounting methods should be used to estimate an aggregate like value added, such as the methods based on “superlative” index numbers’. When the brakes don’t work – try something else. We will return to superlative index numbers. But first we will explain why a method of choice results in unpalatable outcomes and we will try to give economic meaning not to these outcomes (which indeed are unpalatable) but to the process which engenders them. Why do bizarre outcomes happen? After all: even when double deflated value added is not an unpalatable negative number but a slightly positive it might still be wildly of the mark which means that we do have to know how robust our methods are! The first question is how the double deflation method which tries to disentangle changes in prices from changes in volumes might result in negative ‘real’ value added. An arithmetical example: value added is a nominal variable defined as the nominal value of production minus the nominal value of current inputs. You purchase lemons, lime and sugar for 10 units; you sell the lemonade for 20 units – which means that your value-added is 10 units. Next year, thanks to the eruption of a volcano, prices of inputs will be 50 units – while the lemonade sells for 35. To be able to make a profit, the use of inputs is halved which means that nominal value added still is 10 units. Yes, lemonade quality deteriorates but that doesn’t matter: value added still is 10 units. Double deflation however means that purchase prices are deflated with a price index for purchase prices (lemons, lime, sugar) while selling prices are deflated with a price index for selling prices (Lemonade) which yields ‘real’ value added of 15 units (20 units – 5 units). Production increases with 50%. Using the prices of the second year to calculate real production in the first year yields ‘real’ value added of minus 15 units. A more elaborate example: when one would calculate the value of inputs (purchased feed, fertilizer) used in Dutch agriculture around 1950 using fixed prices the value of purchased inputs as a percentage of output shows a different development than the same in current prices (Graph 8.1).

230  Unreal production 45 40 35 30 25 20 15 10 5 0 1851

1861

1871

1881

1891 Current prices

1901

1911

1921

1931

1941

Fixed prices

Graph 8.1 Intermediate inputs (% of gross output), Dutch agriculture, 1851–1950, current and fixed prices Source: Knibbe (1993)

The outputs and inputs of 1921 weren’t the inputs of 1851 anymore. And the prices of 1921 weren’t the prices of 1851, anymore. Artificial fertilizer had replaced purchased guano and also, quantitatively much more important, had totally replaced self-produced heather sods which, based upon a very labor intensive process, were used as organic fertilizer. Madder disappeared, sugar beet appeared. Massive amounts of corn and feed cakes were imported to feed pigs and cows. Agriculture had changed. As prices of inputs declined relative to outputs, the ratio of inputs to outputs in current prices declined relative to the ratio of inputs to outputs in fixed prices. The current price series captures the change in relative prices, the fixed price series doesn’t (actually, the prices of the fixed price series as shown in this graph are not entirely fixed, as that would have led to an even more implausible high level of inputs, of over one hundred percent of value added). Interestingly the decrease of input prices stimulated the increase of their use and led to fast concomitant institutional changes (Knibbe 1993). At first, farmers didn’t always know how to use artificial fertilizer as they lacked ‘tacit knowledge’ about the mineral content of the soil; government subsidized laboratories took care of this by testing samples and giving advice. Hayek (1945) stresses the importance of tacit knowledge of the individual producer but in this

Unreal production 231 case it were outside organizations which provided producers with this knowledge. For a somewhat comparable example where this tacit knowledge (fat content of milk of individual cows) was produced by a overlapping networks of farms and factories using modern ‘Taylorist’ estimation methods: Knibbe and Molema (2018). Putting it in a Polanyian perspective: the growth of markets for outputs but especially for inputs and the monetization of the input process made farmers more dependent on market-counterparts. Often, these counterparts were much larger than individual farmers while there also where informational and financial asymmetries (economese for: ‘you’re being ripped off’). Cooperative purchases of inputs in combination with cooperative production of outputs like butter, sugar and cheese enabled farmers to reap economies of (also in measurement and breeding) and to change market power in their advantage, breaking monopolies and establishing an entirely new financial infrastructure. Another, more complicated example is discussed by Edquist (2013). According to him the extremely high growth rate of productivity in Swedish production of electronics (+40 percent a year) is due to subtle biases of the double deflation method. In this case and in line with OECD prescriptions, the growth of ‘real’ gross output and ‘real’ inputs were calculated. Subsequently, the growth rate of inputs was weighed according to its share in total nominal gross output and subtracted from the growth rate of gross outputs. These adapted growth rates of gross output still had to be normalized to arrive at a growth rate of value added as value added in some industries is 90 percent of gross turnover while in others it might be 9 percent. The same (output minus weighted input) growth rate of gross output means something different for the change of value added in the first industry as compared with the second industry. The idea is that when this growth rate increases with 1 percent in a 90 percent value added industry this means a {1%/90%}*100% increase in the first industry against a {1%/9%}*100% increase in the second industry, which is why the growth rate is divided by the share of value added in gross production (i.e. multiplied by its inverse). But: Swedish manufacturing of electronics is dominated by one company, Ericsson. And changes in output prices, even when they do not lead to a change in deflated gross turnover (sales are stable in numbers) can still wreak havoc with the share of value added in gross output (turnover declines). This is what happened to Ericsson around 2001: prices of outputs plummeted. The share of value added in turnover changed from around 25 percent to around 4 percent (meaning: hefty operational losses), which of course affects the estimates of productivity, even when physical productivity changes are pretty stable. The inverse of the share of value added which was used to multiply the adapted growth rate of gross output increased from 4 to 25 percent, leading to an incredible fast increase of estimated productivity, even when physical productivity did not change too much. Looking at double deflated data conflates the situation instead of illuminating it. In Edquist’s words: ‘when productivity is analyzed for these types of industries it is important to base the analysis on both value added and gross output to test the robustness of the results’ (Edquist 2013, p. 9).

232  Unreal production Peter Lindert recently took a related question head on to answer one of the large questions of economic history: why and when did ‘the West’ surge ahead of the rest of the world? He rephrased it and discussed if we should use present day prices to evaluate the past, which led to the question if and when the West surged ahead of the rest: was this already in the 17th and 18th century or only in the 19th century (Lindert 2017)? Instead of using anachronistic 1990 prices of inputs and outputs to compare the relative level of prosperity in centuries bygone (the equivalent of the double deflation method with fixed prices) a more direct approach may be advisable: ‘It is far easier, and more appropriate to historical contexts, to stick with direct current-price comparisons of the countries’ nominal incomes per capita back then, deflated by prices people paid back then’. From a production oriented double deflation perspective based upon 20th century prices he moved to a consumption oriented single deflation based on contemporary prices. Doing this made a difference. For one thing, the Western surge was not always a question of Western growth but also of non-Western stagnation and decline, exemplified (according to the data of Lindert) by the long run development of the Dutch colony of Java which, looking at per capita consumption, fell back during Dutch rule, while India was already very poor to begin with (1595). He shows that the choice of prices and methods to measure such processes matters. On the most basic level, this is caused by the fact that we are living in a monetary world – and the spatial and historical nature of relative money prices influence our estimates. Value added is a fundamental monetary variable. Changes of relative prices are as important for the development of value added as changes in the amount of products produced, which means that we should take great care to use relative prices which from different periods of times. We can’t understand the development of these prices as the outcome of some kind of eternal efficient market. Technologies, companies, market power, coordination and exploitation systems, markets and even people change. Also: prices may be crucial in a process of monetization and de-monetization of production and changing the production boundary as happened in the case of artificial fertilizer in the Netherlands when monetized inputs replaced non-monetized inputs. This importance of relative prices and dynamic changes is also central to the discussion of purchasing power parities and consumption (Allen 2017). To put it simple: how do we compare the standard of living of a 19th century peasant living in Ireland dependent on potatoes with the standard of living of a Javanese peasant dependent on rice? Allen states that estimates which take due account of circumstances and bodily requirements should be used instead of measurements based on today’s market prices. To state this in economese: ‘a process of production which is efficient at one set of prices, may not be very efficient at another set of relative prices. If the other set of prices is very different, the inefficiency of the process may reveal itself in a very conspicuous form, namely negative value added’ (SNA 1993, quoted in OECD 2001). We have to add that the process itself can change, too. The frontier of the production possibilities might change. This goes for a comparison of Dutch agriculture in 1880 with Dutch agriculture in 1950 as well as for a comparison of India in 1595 with Indonesia in 1800 or for a historical and comparative comparison of productivity

Unreal production 233 in Swedish electronics around 2001 and for a comparison of poverty in workers in Saint Petersburg around 1910 with workers in Bombay around 1920 (Allen 2017). ‘Real’ production and consumption are monetary variables based on prices which are contingent for a specific situation – even when it’s sometimes tricky to recognize this. DSGE models do not know such complexities. They know only one, unchanging, eternal good. Changes in relative prices and quantities combined with changes in technologies are defined away. Sometimes, this one good is called a ‘complex good’ which consists of other goods. As the volumes and relative prices of the complex good do not change this however does not change the picture: it is like producing the same radio receiver again and again and again and relative prices and quantities do not change. And as there is only one good there is also only one price. This disables sound, deep, non-trivial analysis of the process of economic growth and development. Modern economy is almost defined by changing relative prices and quantities – the very things ruled out by the models. Economic growth is not ‘more of the same’ but ‘more, better, new and different’ (more formal: Dietzenbacher and Hoen 1999). In a way DSGE models are designed to analyze an economy bereft of everything which makes our economies exiting and interesting.

8.3 Putting formulas to the test: superlative index numbers and Dutch agricultural production and prices, 1851–2016 In an arithmetical sense, problems like those mentioned in the previous section arise because relative prices and quantities change, nullifying the assumption of un unchanging production and production. In the case of value added, the use of two indices, price and volume, compounds this problem. In an economic sense, the problem with double deflated value added arises because production and productivity, as economists define these, are not something physical like production of wheat or production of wheat per hectare, phone calls per tele call employee, spectators watching a particular football game or the occupancy rate of hotels or planes. Economic production is a monetary phenomenon influenced by prices. We’ve seen that the OECD advises to use ‘superlative’ indices to solve such problems and to capture the dynamic nature of modern economies. Diewert (1976), after a long exposé pointing out that stable neoclassical production functions can be consistent with superlative price indices, states in his note 16: ‘However, as a matter of general principle, it would seem that the chain method of calculating index numbers would be preferable, since over longer periods of time, the underlying functional form for the aggregator function may gradually change’. Production functions aren’t stable, which necessitates a continuing rebasing of the indices. In a useful overview Diewert (1992) lists 21 ‘reasonable’ tests indices have to satisfy – many frequently used indices do not satisfy a lot of these and speaks out in favor of the Fisher index. But he refrains from stating how superlative indices can be used in dynamic economic analysis. Present day national accounts invariably use

234  Unreal production such indices. Using these is not new. The ‘Landbouwcijfers’ (Agricultural data) series dating to the 1950s of the Dutch Landbouwkundig Economisch Instituut (LEI) already contain chained Fisher indices of prices and production. Superlative indices enable one to estimate a price or volume series which uses a new base year every year, which prevents the problem of using anachronistic prices of a fixed base year which becomes less relevant with every year that passes, which leads to the question: are superlative indices indeed an alternative? As I could not find serious long-term tests of this indices, I decided to put them to a preliminary test myself, extending some of the price and volume series of Knibbe (1993) and recalculating these using a number of indices. As is well known, the choice of a deflator does make a difference. As I did not know any systematic collection of agricultural prices for the 1950–2015 period I thought it a good idea to assemble such a collection and to extent the series. It turned out that there was a valid reason why such a collection did not exist – especially potato and milk prices were ‘complicated’. To stick to milk: dairy prices contain loads of mark ups and mark downs for bacteriological quality, fat content, protein content as well as some fixed costs while these fixed costs are calculated after subtracting part of the proceedings of products like whey. Also and for a time, the Dutch government paid part of the milk price but also subtracted some levies to finance practical and fundamental research. In case of cooperatives, they also contain an element of return on investment as well as (as part of prices were paid ex post) interest. The details of how milk and potato prices were, eventually, constructed will be published elsewhere, together with institutional detail (Knibbe and Molema, in progress). But in the end, a farm gate price connected to selling a product is a farm gate price even when two parties (the factory and the government) both pay a part of it and Excel sheets are pliable which enabled the calculation of two superlative indices: a Fisher index and a Törnquist index which behaved remarkably similar. Graph 8.2 compares two superlative price indices: a Törnquist and a chained Fisher index. The main conceptual difference: the Törnquist index uses an average of this and yester years quantities, the Fisher index averages an index totally based on this year’s quantities with one based on yesteryears quantities, which means that we expect the Törnquist index to be slightly smoother, as quantities are basically a two year running average. This seems to work out. But the by far most impressing aspect of the graph: there does not seem to be any kind of systematic bias between the two indices over a period of 160 years. We should take care. Wheat today is not the same wheat as one hundred years (Omstead and Rhodes 2002) even when it is still recognizable as wheat, this contrary to the difference between a tractor and a horse. Even then, products like madder disappeared while new products like sugar beet appeared and while the use of inputs like artificial fertilizer, insecticides, herbicides and fungicides became increasingly monetized. So, changes took place. This puts in question if using fixed products for such a long period is a viable procedure. And we should take note that even when the Törnquist and the Fisher index show amazing consistency, the Paasche and Laspeyres index underlying the Fisher index (not shown in the graph) do not, the Paasche index almost increasing twice as much as the

Unreal production 235 1200

1000

800

600

400

200

0 1851

1861

1871

1881

1891

1901

1911

1921

1931

Fisher

1941

1951

1961

1971

1981

1991

2001

2011

Törnquist

Graph 8.2 A Törnquist and a chained Fisher Price Index, Dutch gross arable output, 1851–2016 Sources: Up-to-1950 prices from Knibbe (1993). Prices 1951–1995: Landbouweconomisch Instituut: Landbouwcijfers and Land en Tuinbouwcijfers. Prices post-1995: Wageningen University and Research: Agrimatie.

Laspeyres index. The consistency is also no law of nature (even when it might be a tendency of economics), we will see that vehement price and quantity changes related to predictable sales of grocery products can throw even superlative indices of kilter. So, care has to be taken, however. The basket of arable products in 2015 was not the same one, anymore, as the basket of 1851. Also, around 1890 rye was mainly used as feed. The crop however disappeared and was replaced by imported corn – a product not included in the graph (but which has to be included in a price index of inputs of livestock farming). Aside – whatever all the caveats the graph shows that there has been no price inflation with regard to farm prices of arable products since 1972. When we compare the superlative indices with the chained unsurprisingly it turns out that the Paasche and Laspeyres indices (no graph) show large differences, the Törnquist and, of course, the Fisher index being in the middle of these two. Using unchained Paasche of Laspeyres indices makes the differences even larger, especially for the Laspeyres index, and Diewert (1976) is right to state that these just should not be used. But does this mean that we can use the superlative indices? Using one of the famous Irving Fisher tests for price indices (the increase of the index should lie in between the price increases of the goods and/or services with the lowest and the highest price increase) shows that even the Törnquist index only barely survives this test for these data. So: is there a right one? We can apply some plausibility checks. Graphs 8.3 and 8.4 show price

1200

1000

800

600

400

Livestock producon

2011

2001

1991

1981

1971

1961

1951

1941

1931

1921

1911

1901

1891

1881

1871

1861

0

1851

200

Arable producon

Graph 8.3  Producer prices of arable and livestock products, the Netherlands, 1851–2016 Sources: see graph 8.2

1,400,000,000

1,200,000,000

1,000,000,000

800,000,000

600,000,000

400,000,000

Gross livestock producon

2011

2001

1991

1981

1971

1961

1951

1941

1931

1921

1911

1901

1891

1881

1871

1861

0

1851

200,000,000

Gross arable producon

Graph 8.4 Gross arable and livestock output, Törnquist deflated volume, the Netherlands, 1851–2016 Source: see Graph 8.2

Unreal production 237 and ‘volume’ indices for gross arable and livestock production (gross meaning that rye used to feed pigs or buttermilk used to feed cows is not subtracted). The results are interesting and also in line with historiography, even when the ‘sudden stop’ of livestock production growth turns out to be even more sudden than expected. Fitting exponential growth curves describes the arable data quite well until 2001 and is followed by stagnation. Livestock data until 1989 shows high but unstable growth. Post-1950 growth was too high compared to the period before 1950 and three curves have to be estimated: 1851–1950, 1950– 1989 and post-1989. If we cut out the Great Depression and the World War II and 1945–1955 data, a fitting exponential growth curve however surfaces – it is interesting to speculate that livestock growth was tied to German agricultural imports (which only started to recover after about 1955, also in the case of East Germany). Interesting as this may be: besides these general patterns, and even these have to be looked at with caution, we should not attach too much quantitative precision to such estimates. Even when – a new finding – it is remarkable that the level of livestock relative to arable prices is back to its 1851 value, again (Graphs 8.3 and 8.4). It is also possible to calculate the share of arable and livestock production using current prices gross output and deflated gross output (Graph 8.5). It all seems plausible. The most interesting data point maybe is 1929. This year knew a bumper crop but also witnessed a very considerable fall in arable prices which was followed only with a lag by livestock prices. The series neatly show this

90 80 70 60 50 40 30 20 10 0 1851

1861

1871

1881

1891

1901

1911

1921

Current prices

1931

1941

1951

1961

1971

1981

1991

2001

2011

Törnquist deflated volume

Graph 8.5 Gross arable and livestock production (% of total gross agricultural production), current and fixed prices, 1851–2016 Source: see Graph 8.2

238  Unreal production (the share of arable production in constant prices goes up, the share in current prices goes down). A comparable effect is visible after World War II when arable production initially stayed high but when relative arable prices declined. Interesting is also the rather stable level of the volume share of arable output after 1880, indicating that the increasing importance of livestock in this period (even accounting for the use of part of arable production as an input for livestock production) was to quite an extent a price development, while it was only between 1950 and 1985 when the volume of livestock output increased faster than the volume arable output. Interestingly, this is also the period when a long-term decline of relative aggregate livestock prices started (the idea that EU policies increased the agricultural price level is, at least for the period after 1985 – not true at all). Such results all seem plausible. But they are just that: plausible. They might enhance our understanding of developments in this period (and in my opinion they do). Even then, economists have to take care. Looking at an extreme period, World War I, we do see that the Törnquist index for artificial fertilizer, which as stated uses weights based on the current as well as the preceding year, goes rogue and is higher than price indices for individual fertilizers (Graph 8.6). What happened? Especially during World War I, price and physical volume changes of amounts purchased were as extreme as it gets (Table 8.1). It’s arithmetically reminiscent of ‘bouncing’ behavior of weekly supermarket prices and sales: whenever a product is on sale, the price of course drops while sales increase sharply, especially when this is about what marketeers call convenience goods like diapers, peanuts, detergents and toilet paper (De Haan en Van der Grient (2011)). That means that the

600

500

400

300

200

K2O

1950

P2O5

1945

1930 nitrogen

1940

Törnquist

1935

Fisher

1925

1920

1915

1910

0

1905

100

Graph 8.6 Prices of different kinds of artificial fertilizer, plus Törnquist and chained Fisher Price Index Source: Knibbe (1993)

Unreal production 239 weight of the good is influenced by this sales pattern, leading when weights of the running period are used to too much emphasis on price decreases and too little on price increases the next week, when hardly anything is sold with as a consequence a downward drift of the price index much less visible when monthly prices and sales are used. To state this otherwise: even when in week 3 (after the sale) the prices and sales are equal to those in week 1, the price index will be lower than in week 1 – no matter what kind of index is used (the authors even use a truncated Jevons index which gives every item above a certain threshold a standard weight of 1). Somehow, the extreme changes of fertilizer prices and volumes in World War I led to a somewhat comparable development. But the situation is even more complicated. Lots of organic fertilizer – manure – was still used, which also contained nitrogen and potassium. There sure was a manure market, but by far the larger amount stayed on farms and did not enter this market. Should these amounts be included in the calculation of the price indexes? We won’t answer that question here, but it does show some limitations or at least complexities of price and volume indexes. The Fisher index behaved more ‘Plausible’ than the Törnquist index. But even when we use the chained Fisher index for output as well as input (which is as good as it gets) it shows that the double deflated series of inputs, expressed as a percentage of output, behaves quite a bit different than the current prices percentage. Double deflation value added would not become negative but it would be lower, relatively, than current prices output. Economic statisticians are of course well aware of this and other problems and in reality use, next to the double deflation method, a plethora of methods, maybe not all of them theoretically sound, to produce volume estimates of value added (Eurostat 2015, especially figure 5), which reminds us of the suggestion of Edquist, cited earlier. Using just one method only to calculate production and productivity might not be the

Table 8.1  Use and prices of artificial fertilizer in the Netherlands, 1914–1923

1914 1915 1916 1917 1918 1919 1920 1921 1922 1923

Vol. of Minerals in Artificial Fertilizer (tons*1,100)

Price per Unit of Mineral

Nitrogen

P2O5

Potassium

Nitrogen

P2O5

Potassium

24 28 17 13 0.3 33 33 31 25 28

53 17 7 12 1.0 25 49 42 74 74

42 12 18 16 24 21 41 41 60 52

690 1120 2000 1710 2720 1300 1920 1010 990 830

160 360 570 650 860 510 550 330 250 190

150 150 180 200 460 280 270 180 100 100

Source: Knibbe (1993)

240  Unreal production answer. A useful addition to the discussion is Oulton et al (2018). The show that the equality of aggregate expenditure and aggregate output is not just a characteristic of nominal accounts but also, at the GDP level, of double deflated ‘real’ accounts, at least when ‘consistent’ sets of prices are used. The intuition behind this: when real agricultural production increases but relative prices of agricultural products decline, purchasing power of farm incomes might still deteriorate. However, the negative development of the terms of trade for agriculture is a positive development for other sectors, according to Oulton e.a. these effects can be made consistent in multi sectoral input output tables using consistent prices. But the confusion continues: there are different consistent sets of estimated prices possible which yield different estimates of real GDP and, more important, quite different growth rates. Their estimate of GDP growth for the UK for the 1997– 2015 period is more than 1% per year lower than the official estimate (25% for the entire period) which is just too much of a difference for me. They do not discuss the economic side of this. But as the official growth rate is based on single deflation it means that use of inputs has grown much, much faster than output. They also do not discuss the fact that even when a consistent set of prices does lead to real expenditure being equal to real production (contrary to the situation when single deflation is used) there is no guarantee that these prices are indeed the prices relevant to the budgets and decisions of companies and households. This is even more important when we move from the accounts to comparisons of different regions in different periods: consistent sets of price scan be used to estimate consumption of GDP in several regions but we do not know if the cattle or milk or silk price index used was relevant in an economic way. It is, however, a step ahead. Summarizing: we do seem to have reasonably stable long term deflation methods as well as methods to calculate real aggregate expenditure and output which match, just like nominal output and expenditure (and income, not treated by Oulton e.a.) match. So, what to do? Estimating ‘real’ economic growth is a worthwhile pursuit. But economists should take heed that value added is, fundamentally, a nominal and therewith a historical variable. Deflating will, among other problems, lead to all the well-known technical problems related to the use of deflators for constructing time series but also to problems of interpretation. There is no ‘right’ way to disentangle price increases from changes in relative prices and historical knowledge is necessary to understand the results. But knowledge of technical aspects of deflators is necessary, too, to do this. Deflating nominal agricultural value added with the consumer price index yields a measure of purchasing power – but in the case of gross value added, this is of limited use as part of gross (as well as net) value added is used to invest and not to consume, even when this part might fluctuate quite a bit. Deflating nominal agricultural value added with a price index based on prices paid to farmers uses an average selling price for nominal production. Comparing this price with for instance prices on inputs in one way or another is interesting but doing this by using a double deflation procedure might, for the reasons mentioned earlier, yield nonsense results. Such procedures are beset with troubles. Products purchased change and

Unreal production 241 so do relative prices, which means that we run the risk of calculating purchasing power using an outdated Sears catalogue. Double deflation doubles that risk and adds the possibility that unpalatable magnitudes of ‘real’ value added are calculated, which obfuscate instead of enlighten. Instead of doing this – better: next to this – economists should focus on intermediate inputs which are subtracted from output and are hence in a way discarded from discussions, while it might be exactly these items, like energy, feed cakes and the like, which are interesting: the famous ‘residual of Abramovitz’ which might not be a residual but the true object of economic investigation. In the same way, changes in distribution of income over factors of production and purposes (consumption and investment) themselves are the interesting developments, sometimes more so than changes in purchasing power. Fortunately, we do have existing estimates of real ‘real’ use of inputs – information about tons and gigawatts or in agriculture feed and phosphorous – which are largely based upon the same national accounts as the value added series. Eurostat (2018a) calculates use of material resources and production of CO2 related to final use (consumption, investment, net exports), which shows that post 2008 housing bust led to a sharp decrease of the use of materials. A recent flagship UN report written by the International resource panel and titled ‘resource efficiency: potential and economic implications’ (International Resource Panel 2017) contains extensive data on use of a whole array of resources, like different materials, land, water and energy and how these are used. Inputs should not be subtracted, they should be investigated! The nominal national accounts are a great resource to enable this – we need really real, i.e. nominal, accounts as well as estimates of real use of flows of resources and outputs, like oil or cars or CO2.

Note 1 A ‘line estimate’ is a better description of an index of real output or purchases or income than a ‘volume estimate’.

Literature Allen, R. (2017). ‘Absolute poverty. When necessity displaces desire’. American Economic Review 107:12 3690–3721. Agrawal, R., T. Büttgenbach, J. Findley, A. Jeddy, M. Petry, J. Kondo, B. Lewis, G. Subramanian, A. Borsch-Supan, K. Huang and S. Greene (1996). Capital productivity. Washington: McKinsey Global Institute. Alexander, T., C. Dziobek, M. Marini, E. Metreau and M. Stanger (2017). ‘Measure up: A better way to calculate GDP’. IMF Staff discussion note 17/02. Dietzenbacher, E and A.R. Hoen (1999). ‘Double deflation and aggregation’. Environment and Planning A 31 1695–1704. Diewert, E.W. (1976). ‘Exact and superlative index numbers’. Journal of Econometrics 4:2 115–145. Diewert, E.W. (1992). ‘Fisher ideal output, input and productivity indexes revisited’. Journal of Productivity Analysis 3:3 211–248.

242  Unreal production Edquist, H. (2013). ‘Did double deflation create the Swedish manufacturing growth miracle? Is there a lesson for other western European countries and the US?’. Economics Letters 121:2 303–305. Eurostat (2015). ‘Building the system of national accounts – volume measures’. https://ec.europa.eu/eurostat/statistics-explained/index.php/Building_the_Sys tem_of_National_Accounts_-_volume_measures. Accessed 19 April 2017. Eurostat (2018a). ‘Statistics explained. Greenhouse gas emission statistics’. https:// ec.europa.eu/eurostat/statistics-explained/pdfscache/1180.pdf. Accessed 17 May 2019. Eurostat (2018b). ‘Statistics explained. HICP at constant tax rates’. https:// ec.europa.eu/eurostat/statistics-explained/index.php/HICP_at_constant_tax_ rates. Accessed 31 January 2019. Haan, J. de and H.A. van der Grient (2011). ‘Eliminating chain drift in price indexes based on scanner data’. Journal of Econometrics 161:1 36–46. Hayami, Y. and V.W. Ruttan (1984). ‘Toward a theory of induced institutional innovation’. Discussion paper 200, Center for Economic Research. Department of Economics, University of Minnesota. Hayek, F.A. von (1945). ‘The use of knowledge in society’. The American Economic Review 35:4 519–530 International Resource Panel (2017). Resource efficiency: Potential and economic implications. New York: United Nations Environment Program. Knibbe, M. (1993). Agriculture in the Netherlands 1851–1950. Production and institutional change. Amsterdam: Nederlands Economisch Historisch Archief. Knibbe, M. and M. Molema (2018). ‘Institutionalization of knowledge-based growth illustrated by the Dutch-Friesian dairy sector, 1895–1950’. Rural History: Economy, Society, Culture 29:2 217–235. Lindert, P. (17 April 2017). ‘European and Asian incomes in 1914: New take on the Great Divergence’. Voxeu. https://voxeu.org/article/european-and-asianincomes-1914-new-take-great-divergence OECD (2001). Measuring productivity. OECD manual measurement of aggregate and industry-level productivity growth. Paris: OECD. Omstead, A.L. and P. Rhodes (2002). ‘The Red Queen and the Hard Reds: Productivity growth in American wheat, 1800–1940’. The Journal of Economic history 62:4 929–966. Oulton, N., A. Rincon-Aznar, L. Samek and S. Srinivasan (2018). ‘Double deflation: Theory and Practice’. Paper prepared for the 35th IARIW General Conference Copenhagen, Denmark, August 20–25, 2018 Session 2E: Pushing the Boundaries of the SNA. Sumner, S. (2015). ‘The bizarre way economists calculate real income’. Blogpost on Econlog. www.econlib.org/archives/2015/03/the_bizarre_way.html. Accessed 8 January 2017. Varoufakis, Y (2016). ‘Real versus money incomes. The one thing we need to understand during deflationary times (with an illustration from Greece and Cyprus)’. Blogpost on ‘Thoughts for the Post 2008 World,’ 13 March 2016. www.yanisvarou fakis.eu/2016/03/13/real-vs-money-incomes-the-one-thing-we-need-to-under stand-during-deflationary-times-with-an-illustration-from-greece-and-cyprus/

9 Macroeconomic unit labor costs as we measure them are no indicator of competitivity

The fallacy of composition is the false assumption that what is true for a part will also be true for the whole. Keynes carefully distinguished the adjustment process for a single market from the adjustment process for the aggregate economy, arguing that the effects differ significantly between the two. (https://quizlet.com/44764462/macro-econ-ch6flash-cards, accessed 20 August 2019)

9.1 Introduction Neoclassical macroeconomists model production as a single good which, with some twists like branding, can be used for exports, consumption, investments or even saving. It might be called a ‘composite’ good just like a cake is a composite food but the recipe does not change. The ‘single good’ is a pure market good – or, well, it isn’t always but when the government produces or uses it this is considered to be wasting the good or wasting resources. Again, there are models which amend this idea – but in the canonical models either used for teaching or used by central banks, and these are the models which really matter, no regrets about this situation are voiced. What happens when this attitude spills over into the national political realm? What happens when abstracting from non-market situations (government but also household production), treating GDP as a single unchanging good, abstracting from sharply diverging sectoral cost structures, output gaps and global production chains enters the policy room (European Commission 2019)?

9.2  NULC, RULC and ULC An example are Nominal Unit Labor Costs (NULC), defined as ‘Unit labour costs (ULC) measure the average cost of labour per unit of output and are calculated as the ratio of total labour costs to real output’ (OECD 2019). The definition of the OECD is however about NULC which are not the same as labor costs per unit of nominal output (ULC). Also, We have RULC, real labor costs per unit of output. For the time being, we’ll stick with NULC. NULC are a staple of applied macroeconomic statistics and are defined as nominal labor costs per

244  Macroeconomic labor costs & competitivity unit of real GDP. As the OECD glossary states: ‘They play a large role in policy advise not just with regard to competitivity but also with regard to the institutional set up of countries especially in the eurozone’ (Lebrun and Pérez 2011). They are included in the EZ macroeconomic imbalance procedure next to variables like credit, the current account, house prices and unemployment. Specifically, NULC should not increase too much. The upper threshold is a: ‘3-year percentage change in nominal unit labour cost, with thresholds of +9% for euro area countries and +12% for non-euro area countries’ (Internet 8.1). We will argue that in the real world the abstractions mentioned earlier wreak havoc with these policy proposals. Crucial is the idea that GDP can be approximated as a single good. Labor costs per unit of this approximated single good are calculated on a regular basis by the OESO, Eurostat and the US Bureau of Labor Statistics and they figure, though less prominently than a few years back, on the websites of organizations like the ECB and the European Commission. They are also one of the core variables of the macroeconomic imbalance procedure (MIP) of the European Union, which indicates their political significance. After the introduction of the EU, NULC, originally developed to gauge the influence of wage increases on the rate of inflation, gained prominence as an aggregate indicator of competitivity. The introduction of the euro bereft the eurozone countries of the possibility to use exchange rates to try to adjust the price competitivity of their export sector and enhance the competitive position of producers for the domestic market vis-à-vis imported goods. Originally, it was assumed that the abandonment of exchange rate flexibility would not lead to any kind of problems. Market forces would force countries to keep prices in check. As expected, a process of price convergence took place within the eurozone. Prices in periphery countries increased faster than in the core which meant that differences in price levels declined. Asset prices increased, too. This indeed made the market kick in: capital flows to countries with faster increasing price levels greatly increased. Being able to invest, without any kind of exchange rate risk, in countries were assets were almost guaranteed to increase in price was a free lunch. Current account deficits in the eurozone of countries with relatively low but increasing price levels exploded. When in 2008 disaster struck and sudden stops of capital flows disabled the financing of the large current account deficits of Southern but also of Eastern European countries the economy came to a halt and unemployment exploded. These eurozone countries were put under strong pressure to decrease their NULC relative to other countries. A relative decline of periphery NULC compared with other eurozone countries was understood to be the policy of choice to increase international competitivity, to increase exports to compensate for the sudden stop of inflowing capital and to lower unemployment (Horn and Wolff 2017). Relatively high NULC were understood to be a sign of low competitiveness (Draghi 2013). The preferred way to decrease NULC was to moderate or halt wage increases or even to slash wages. Graph 9.1 shows that after 2008 some countries indeed managed to lower NULC. But was this caused by keeping wage increases low or by lowering wages? Graph 9.2 shows that Greece and Ireland, which until 2013 witnessed a comparable total decrease in NULC of about 15 percent, had

120 110 100 90 80 70 60 50 40

Germany

Finland

Portugal

Spain

Italy

Greece

Ireland

Graph 9.1  Nominal unit labor costs, selected countries in the eurozone, 2000–2018 Source: Eurostat, labour productivity and unit labour costs data, nama_10_lp_ulc

120

100

80

60

40

20

0

Ireland

Greece

Spain

Portugal 2007

Finland

Italy

2017

Graph 9.2 Compensation of employees per hour worked (% of German compensation), selected eurozone countries Source: Eurostat, labour productivity and unit labour costs, nama_10_lp_ulc

246  Macroeconomic labor costs & competitivity a radically different development of wages. Greek wages showed a large and, for Europe post-World War II, unprecedented decline while Irish wages continued to increase. After 2013 the difference even exacerbated and Irish NULC fell through the floor – even despite continuing wage increases. This suggests that other factors were important, too. NULC just did not react as expected. It just might be that we can’t look at GDP as an unchanging good which, in the case of the EU, doesn’t even show national differences. And if this is the case, can NULC still be understood as a valid indicator of competitiveness? Or should we understand the economic aggregates as what they are: aggregates of changing underlying sectoral developments? These factors will be investigated and it will be argued that NULC should not be used as an indicator of macro-productivity and competitiveness. On the micro-level, wage costs per unit of product – such as wage costs for installing a solar panel – are an important indicator of productivity. This might even spill over to the sectoral scale: solar electricity in low wage countries might have a price edge compared to solar energy in high wage countries. On the macro-level this is not the case. GDP (the denominator in the formula) is an aggregate based upon the aggregation of sectors with wildly different labor costs (graph 9.3). These sectors are aggregated using weights which, because of Schumpeterian dynamics or booms and busts, can change quite fast. This can lead to changes in NULC which are sometimes even in the opposite direction to changes in wages levels. Also, many of these sectors have next to no connection with the competitiveness of an economy.

130 110 90 70 50 30 10 –10 Mining and quarrying

Energy

Informaon, communicaon Lowest

Manufacturing

Recreaon, hospitality

Construcon

Highest

Graph 9.3 Labor share, highest and lowest share per sector, sectoral data, the Netherlands, 2008–2013 Source: Central bureau voor de Statistiek, Tijdreeksen Nationale rekeningen 1995–2015, available at www.cbs.nl/nl-nl/maatwerk/2017/24/tijdreeksen-nationale-rekeningen-1969-1995, accessed 17 May 2019

Macroeconomic labor costs & competitivity 247 Importantly, competitiveness is only to a limited extent dependent on the wage level (see the ECB competitiveness research group (2013) or Felipe and Kumar (2011). Consequently, while product or even sectoral ULC might be useful to compare individual companies and maybe even sectors, they are not fit to compare countries. Felipe and Kumar do a good job debunking the use of this metric. However, their list of reasons is far from exhaustive and I’ll have to complement it. To do this, I first have to explain how ULC are calculated. Eurostat provides us with the following definition: The unit labour cost (ULC) is defined as the ratio of labour costs to labour productivity. Nominal ULC (NULC) = (D1/EEM)/(B1GM/ETO) with D1 = Compensation of employees, all industries, current prices EEM = Employees, all industries, in persons (domestic concept) B1GM = Gross domestic product at market prices in millions, chain-linked volumes reference year 2010 ETO = Total employment, all industries, in persons (domestic concept) In other words, NULC are nominal labor costs per employee divided by real average value added (GDP) per worker. Mind the difference between ‘employee’ and ‘worker’. Eurostat continues: ‘Please note that the variables used in the numerator (compensation, employees) relate to employed labour only while those in the denominator (GDP, employment) refer to all labor, including self-employed’. It is a crude approximation for the share of GDP going to workers. This definition leaves us with the following conceptual problems: A It is an indicator which contrary to many statements should increase, considering stated policy goals. Imagine a country with (like the eurozone) an inflation target of 1.8 percent but which (unlike the eurozone) does not target consumer price inflation but the GDP deflator. Such a level of inflation cannot be sustained when, in the medium run, wages do not increase by at least the same percentage (in fact: a slightly higher percentage, assuming some increase of labor productivity). The idea that NULC have to be stable is not consistent with the inflation target of the central bank. The low increase of German NULC before 2010 should have been a cause for concern and sorrow for the ECB. B The Eurostat caveat is significant. The use of employees in the numerator and all workers in the denominator means that countries which had a large share of self-employed, which left this status to start to work for wages which are about as high or slightly higher than their previous ‘mixed income’ as self-employed (e.g. a shift out of peasant farming and into tourism in Greece) might see an increase in NULC, because of structural modernization. (In extreme but conceivable cases this won’t happen when, thanks to this shift, GDP (the denominator) rises relatively faster than total

248  Macroeconomic labor costs & competitivity wage income (the nominator) because of a very rapid increase in net capital income – a case would be a shift of labor from peasant production to the extraction of natural gas). C A related problem exists because of differences between economic sectors. Ireland is an example. The bust of the building boom led to the demise of lots of industries with relatively high ULC’s while industries with a low ULC and a high share of capital, like the pharmaceutical industry, were much less affected by this bust. As a result, the average ULC of Irish industry declined (O’Brien and Scally 2012) while the ULC’s of subsectors of Irish industry barely changed. Talk about a fallacy of composition. One should not underestimate the magnitude of these differences. Graph 9.3 shows the labor share in value added for economic sectors in the Netherlands (which I took as I had the data at hand and also because of the exceptional low labor share in the natural gas sector). Differences are clearly very large (the 4 percent for mining and quarrying, aka natural gas, is real) and show large changes while for some sectors the labor share can even be more than 100 percent, especially in sectors with many self-employed, who, according to the formula, are assumed to earn the average wage. Often, however, they earn less. This is due to the accounting assumption that the wage income of the self-employed is ‘average’. It means that their high labor share is compensated by negative profits (i.e. losses). Applying such ideas: the decrease in NULC in Greece was probably caused by lower wages but the Irish decrease was to a large extent caused by the bust, which caused construction (which has high NULC) to decline. Two totally different events show up in the data in the same way. The post2014 decline in Ireland was due to the shift of the legal Microsoft headquarters from the USA to Ireland, which meant that Microsoft profits (generated worldwide) were allocated to the tiny Irish economy, resulting in a decline of NULC – while wages continued to increase. D Non-tradables. Between 2000 and 2011, average German wages did not rise much. Consequently there was a low increase of the German macroULC. But to quite some extent this was caused by stable nominal wages (and falling purchasing power) of teachers. Wages in industry rose a little more than average and Germany still is one of the few countries where industrial wages are higher than economy wide average wages (even despite very high wages in the financial sector). But do decreasing real wages of teachers really increase the international competitiveness and exports of a country? It might affect the current account as German teachers will have had to restrain consumption of, among other things, imported products. But at least for me the relationship with gross exports is not readily apparent. E Felipe and Kumar (2011) mention the Kaldor-paradox: the empirical evidence about the ULC and competitiveness in fact suggests that high increases in ULC do not cause a decline in competitiveness but are, on the contrary, a sign of successful export performance.

Macroeconomic labor costs & competitivity 249 F Another fallacy of composition. An individual firm can increase its competitive position by cutting the wage level as the wages it pays (almost) do not affect the demand for its products. But when every company decreases wages total demand will suffer. Greece (where nominal wages have decreased by about 20 percent) is an extreme example of this. Low rates of capacity utilization and the accompanying decrease in productivity may prevent the decline in NULC and improved competitiveness. G Global supply chains. The share of ‘domestic’ labor in the cost price of tradeable products shows sustained declines. Giordano and Zollino (2013) mentions that the ‘domestic labor share’ of Germany declined from 27 percent to 21 percent of gross output (not the same as value added, the concept used to calculate GDP) while the Italian share declined from 21 to 18 percent. This means that lowering domestic wages has a more limited effect on total production costs than it used to have. H Owner occupied houses are, with good reason, supposed to add to total GDP. Household labor is however not counted in the GDP accounts, which means that, according to the OESO (McKenzie and Brackfield 2008, p. 14) ‘in the case of Ownership of Dwellings there are no employees, and so this component of value added has nothing to do with the relationship between output and labour costs. Consequently, it should ideally be removed from calculations of ULC indexes. . . . If included it has the potential to distort the comparability of ULC indexes across countries, in particular where there are large differences in the level or, more importantly, changes over time across countries in the contribution of Ownership of Dwellings to value added’. I Aside from NULC we have RULC, real unit labor costs. According to the IMF (Huemer, Scheubel and Walch 2013) these are defined as either: ‘the ratios of real wages to productivity, labor compensation to nominal GDP or nominal unit labor costs to the GDP deflator’. In all these cases it is (something like) the labor share in the economy. Though interesting in its own right this metric is bound to have limited variability as the labor share is part of total production while it is, as an indicator of competitiveness, prone to the same problems as the NULC (an increase of the production of natural gas in the Netherlands will lower the RULC, an increase in construction will increase RULC). Summarizing: NULC are a seriously flawed macro-indicator of competitiveness, much better ones are available, such as the composite one proposed in the ECB study by Huemer, Scheubel and Walch (2013). GDP is a composite, transaction based aggregate. So, returning to the question of the introduction, what happens if these are used for policy purposes? It leads to flawed advice. We can even extrapolate this to the micro-level. Mercedes-Benz makes technologically advanced, competitive cars. That’s not because they keep cutting wages – even when we take account of global production chains.

250  Macroeconomic labor costs & competitivity

Literature Internet 8.1. European Commission (2019). https://ec.europa.eu/info/businesseconomy-euro/economic-and-fiscal-policy-coordination/eu-economic-govern ance-monitoring-prevention-correction/macroeconomic-imbalance-procedure/ scoreboard_en. Accessed 17 May 2019. Draghi, M. (2013). ‘Euro area economic situation and the foundations for growth. Presentation by Mario Draghi President of the European Central Bank at the Euro Summit Brussels, 14 March 2013’. www.ecb.europa.eu/press/key/date/2013/ html/sp130315.en.pdf?8fdd86d374a7fb3eb880870eb6f8b41b. Accessed 25 May 2018. European Central Bank (2013). ‘Competitiveness research network: First year results. June 2013’. www.ecb.europa.eu/home/pdf/research/compnet/CompNet_First_ Year_Results.pdf European Commission (2019). https://ec.europa.eu/info/business-economy-euro/ economic-and-fiscal-policy-coordination/eu-economic-governance-monitoringprevention-correction/macroeconomic-imbalance-procedure/scoreboard_en. Accessed 17 May 2019. Felipe, J. and U. Kumar (2011). ‘Unit labor costs in the Eurozone: The competitiveness debate again’. Levi Institute working paper no. 651. Giordano, C. and F. Zollino (2013). ‘Going beyond the mystery of Italy’s pricecompetitiveness indicators’. Voxeu, 13 July 2013. https://voxeu.org/article/goingbeyond-mystery-italy-s-price-competitiveness-indicators. Accessed 17 May 2019. Horn, G.A. and A. Wolff (2017). ‘Wages and nominal and real unit labour cost’. European Commission fellowship initiative papers. Discussion paper no. 059. Huemer, S., B. Scheubel and F. Walch (2013). ‘Measuring institutional competitiveness in Europe’. Working paper series no. 1556. Lebrun, I. and E. Pérez (2011). ‘Real unit labor costs differentials in EMU: How big, how benign and how reversible?’. IMF working paper WP/11/109. McKenzie, R. and D. Brackfield (2008). ‘The OECD system of unit labour cost and related indicators’. OECD statistics working papers, 2008/04. O’Brien, D. and J. Scally (2012). ‘Cost competitiveness and export performance of the Irish economy’. Quarterly Bulletin 3 (July 2012) 86–102. OECD (2019). ‘Glossary of statistical terms’. https://stats.oecd.org/glossary/ detail.asp?ID=2809. Accessed 17 May 2019.

Epilogue

Epilogue: where to go The road we have to travel to mend the rift is long and windy but also clear and is two-pronged. The modelers have to become aware of the classical roots of their models as well as the extent to which these have been discarded (‘class’, ‘land’) and of the extent to which neoclassical economics has failed to develop genuine macroeconomic statistics consistent with their models. Any model calling itself a macro-model will have to encompass, as a matter of routine (most of this is already happening but only for individual models, not on the level of the general methodology): • Class; • Land and other unproduced capital, natural and legal; • Government production of goods and services; • Durable consumer goods; • Non-rational expectations; • Unemployment as a negative instead of as a positive and labor as positive instead of a negative; • The concept of the fallacy of composition has to be introduced – what’s true at the micro-level is not necessarily true at the macro-level which means that the models will have to be reconstructed to enable aggregation of micro-data; • Much more precise and explicit definitions of variables like the intertemporal substitution of labor which take the phase of the business and the financial cycle into account; • A relation between utility and transactions (as present, the key variable of the models, utility, is not measured); • A full set of balance sheets (partial models without balance sheet are not genuine macro-models, also balance sheets which are ‘inert’ today may be important tomorrow) – one of the items in these balance sheets will be money; • A full set of sectors including banks and non-profits;

252  Epilogue • The idea that ‘deep’ or ‘fundamental’ parameters are not time invariant but influenced by the business as well as the financial cycle while they are also historically embedded; • Fixed capital as an intertemporal heterogenous entity often without a clear market price or value. Relative prices as well as technologies and interest rates and depreciation rates change all the time which makes it rather complicated to attach a specific transaction value to many items. While the models have to get rid of: •

The idea of labor as a negative (this means fundamentally changing the Euler equation with regard to the minus before ‘labor’ as well as the assumption that the relation between higher wages and labor supply is stable over the business and the financial cycle as well as positive); • The idea of unique equilibria; • The idea that market prices govern non-market behavior (they don’t – money relations are guided by other processes than mon-money relations); • The assumption that economies will return to a unique equilibrium; • Perfect forward-looking behavior; • Utility as an undefined, unmeasurable variable (for all other items in this list one can point to old or new authors which have published promising ideas, but not for this one – the core variable of the models seems to escape measurement). When it comes to the statistics these do already account for many of the factors mentioned earlier. But improvements can be made: • Class in the classical sense has to be added. Marxists still use this idea, but making a difference between owners of capitals and sellers of labor was a classical, not a Marxist edifice. • A clearer distinction between and coupling of the realm of monetary transactions and the world of non-monetary behavior (mainly the family), especially when it comes to labor and to ownership and use of durable consumer goods like cars and solar cells has to be made. • Attention to how this relates to ‘time poverty, as well as ‘capabilities and functioning’ is important. • It is possible to be more explicit about composition effects on prices levels, wage levels and other macro-variables. • Explaining over and over and again and again that GDP is not (repeat: not) the same thing as utility (not at all) while macro-statistics are about the accounts, including labor accounts which includes the ILO estimates of slave labor, and not just about GDP. • The same for the concept of ownership of capital as a factor of distribution instead of production.

Epilogue 253 At the same time the macro-statisticians have to stick to the world of monetary transactions, minimizing imputations, while keeping acknowledging that this world is much larger than just the world of market transactions as Samuelson (1954) forcefully Make Veblen proud of you!

Index

Page numbers in italic indicate a figure and page numbers in bold indicate a table on the corres-ponding page. 8-hour working day 56 ABP see Algemeen Burgerlijk Pensioenfonds ‘Accounting for Princes’ (book) 15 accounting identity 222n2 account of money, single accounting concepts of: choice theoretical framework 90; Divisia indices 91–94; Friedman and Schwartz work on 88–91; inflation targeting policies 89; microfounded models 90, 91 ‘actual individual consumption’ 18, 20, 21; concept development 180; concept of 8; of households 177, 178 ‘administered prices’ 49 aggregate demand 205–206 aggregate expenditure and aggregate output, equality of 240 aggregate investment 210 aggregate transactions 11 aggregate wealth and income series 170 aggregation problems 9, 11 AIC see ‘actual individual consumption’ Algemeen Burgerlijk Pensioenfonds 88 Arrow, Kenneth 11, 192 artificial fertilizer: Törnquist index of 238, 238–239; use and prices of 239 assets: economic 146; financial and non-financial 152; fixed see fixed assets; liquid 45; non-financial see non-financial assets; non-produced see non-produced assets; subsoil 153 ‘augmented product’ 53

balance sheets 146, 150 ‘battle of the Cambridges,’ fall out of 169 Bavel, Bas van 48, 54, 57, 58 Bears Ears’, law suits to stop sale of 178 ‘Beter Leven’ meat 53 bond purchase from MFI banks 87 Borio, Claudio 13 borrowers 76, 77, 85, 86 borrowing 77, 86, 93, 162, 165, 166, 187, 188, 219 bottom up aggregation 9–10 bovine tuberculosis, eradication of 69n6 broad unemployment 109, 115, 131, 134; headline and 110, 111; ILO definition 114 bubbles, measurement of 13 Buiter, Willem 7 business cycle analysis 28 business cycle indicators, Mitchell-style 28–30; coinciding with declines in GDP 28, 29; correlation with national accounts data 30; in DSGE models 28–29 business cycle measurements 14 capital 145; concepts and definitions 150–154; definition 146; dual nature of 150; formula for 32; historical character of 146; measurements 154–156; natural 162–164; slaves as 147; stock/flow consistent estimates of 208; valuations 156–159; value of 146; volumes 159–162

Index  255 capital and neoclassicals: concept of 164–167; work of John Bates Clark 167–168 capital goods and households 209 Capital in the Twenty-First Century (book) 148 ‘Caring Dairy’ initiative 53 cash-in-advance constraint 98 cashless economy 99–100 Central Bank of Ireland 150; financial statistics 45 central banks 46, 76, 85, 89, 95, 100, 247; estimates of money creation 77; inflation targeting policies 89, 90; monetary aggregates estimation 84 see also European Central Bank; money; US Federal Reserve central government, monetary liabilities of 84 Chocolonely ‘no slavery’ statement 53 Christiano, Lawrence J. 122, 124 Churchill, Winston 108 Clark, Bowley 18 Clark, Colin 18 Clark, John Bates 167–168 class, economic definition of 14 Clower, Robert W. 98 Clower constraint see cash-in-advance constraint coastal societies 222n1 coconuts-concept of investment 207 ‘collective government expenditure’ 183 commercial credit 76 commodification of labor 54–56 conceptualization phase 30 construction bubble 13, 22, 214 construction projects 50 consumer credit, transactions based on 38 consumer durables 4 consumer price index 22, 31, 42, 160, 161, 169, 227 consumer price level and indirect taxes, link between 228 consumption 182; concept of 3, 177– 178; definition of 178–179; in DSGE models 185–194; and expenditure 179; historical development of components of 179, 180; in national accounts 178–185; nonmarket monetary 178; omissions in concept of 3–4; price deflator 101 cooperative factories 69n7

Copeland, Morris 11, 23, 47, 82, 90 cost of living index 22 cotton production 148–149 credit 91; advanced to Irish resident private-sector enterprises 6; boom 201; commercial 76; and money, sectoral data on 96; trade 77 credit cards 38, 75 credit growth rates 92–93 credit to construction sector in Ireland 13 Crusoe, Robinson 40, 91, 107, 121 cycle oriented statistics 13 dairy prices 234 Day, Doris 134 debt based transactions 40 ‘decentralized centralization’ of population 220 deflation 226 see also double deflation; single deflation deflationary UK policies 108 Denmark: gross fixed capital formation in 205; gross fixed investment rate 217 deposits: domestic 86; liability of MFI bank 99; long-term 84, 85; originated by MFIs 46 Derksen, Jan 20, 60 distributional accounts of USA 82, 83 Divisia indices 91–94, 102n2; vs. ECB weighted monetary press release aggregates 92; and M3, divergence between 92; of money-aggregate 92; statistically significant responses to 92 Divisia money 92, 93 Domar, Evsey 205 domestic deposit creation 86 domestic labor 59; decline in 62, 63; definition of 62; ILO report on 61–62; NBER publication about 61; paid see paid domestic labor; share and product price 249; statistics 62, 62; unpaid 61 domestic services 60; and its perils, ILO on 65; produced by households 60; unpaid 61 double deflated value added 228, 229, 233, 239–240 double deflation 229; conflating the situation 231; in Dutch agriculture 229–230, 232; IMF on 227; negative ‘real’ value added by 229; procedure

256 Index 240–241; Swedish production of electronics 231, 233 DSGE macro-models 7; DSGE ‘EAGLE’ model 76; and macromeasurements, differences between 4, 25–27; privatization of postal services 8 DSGE model, labor concepts of: anomaly 129–135; assumption of ‘complete markets’ 125; Frisch elasticity 123; high unemployment 125–127; labor market 124–125; labor services 122, 124; leisure 121, 122; long-term unemployment 123; model of ‘representative household’ behavior 121–124; neoclassical story of Great Depression 135–139; search theory 127–128 DSGE models 2, 91, 99, 105, 228, 233; banks 13; business cycle indicators role in 28–29; denouncing government production 20; government expenditure 2; ‘investment’ as variable in see ‘investment’ as variable in DSGE models; money role in 97–98; and national accounts, variables of 29; omissions in consumption concept 3–4; treatment of unemployment 112; unemployment mistaken for ‘leisure’ 115–116; water analogy to 107; workhorse model 97 Dutch agricultural production: double deflation in 229–230, 232; superlative index numbers and 233–239 Dutch dairy chain 48 Dutch-UK unemployment spread 108, 109 Dutch unemployment and wages 108 dynamic stochastic general equilibrium models see DSGE models Eastern Germany, unemployment rate in 108 ECB see European Central Bank ecclesiastical organizations, expropriation of 149 economic asset 146 see also assets economic class 13, 168, 170, 209 economic cycles 90, 91, 95 economic downturn of 2008 244; borrowing and lending 162; Danish investments after 205; DSGE

research agenda and 29; rise in monetary savings rate during 184; theoretical macroeconomics and 2; unemployment after 124–128, 131 economic owner 146 economic policy, analytical pillars of 88–89 economic production 139, 228, 233 ELA see emergency liquidity assistance emergency liquidity assistance 46 employment and unemployment 39 ‘entrepreneurs’ 13, 170 equation of exchange 92 ‘equivalent market rental prices’ 179– 181 Ericsson 231 ESA 2010 30–31; guidelines 60; pristine forest 153 ESA 2013 202–203 European Central Bank: credit to government 87; ELA to banks 46; eurozone monetary aggregates 84–85; manual 76; M3-money aggregate 75; model 100; monetary analysis 89; monetary press releases of 94; monthly monetary press release of 84–86, 90; phasing out the €500 note 92; weighted monetary press release aggregates vs. Divisia index 92 European Central Bank, operationalization of money: bond purchase from MFI banks 87; changes in 84; M3 money 85–87; quadruple accounting 85, 88 European Union 125, 129, 131; introduction of NULC 244; MIP of 244; monetary poverty risk in 122 Eurostat national accounts database 45 eurozone countries 46, 86; compensation of employees per hour worked 244, 245, 246; M3 money and Divisia money 92, 93; nominal unit labor costs 244, 245 eurozone monetary aggregates 84; definition 84; long-term deposits 84, 85 ‘evolving power law’ pattern 57 ‘exchange’ money 92 expropriations 149 Fair Labor Standards Act of 1938 65 Farmer, Roger 126–127

Index  257 farm gate milk prices, the Netherlands 52 federal fixed assets, index number of 222n3, 223n3 female participation rate 126, 134 fiat money system 46, 75, 96–97 final consumption, concepts of 30–31 financial assets and liabilities 150 financial crises and business cycle, distinction between 12–13 financial cycles 13, 90 financial investments and fixed investments, difference between 58 financialization 219 financial wealth, preponderance of 58 Finland: compensation of employees per hour worked 245; gross fixed investment rate 216, 217; nominal unit labor costs in 245; unemployment rates in 108 Fisher, Irving 92 Fisher indices 233, 234; Dutch gross arable output 235, 235; of prices and production 233, 234 fixed assets: Dutch ownership 154, 154; gross 150; value of 150 fixed capital 210; characteristics of 146; estimation of consumption of 156–157; evolutionary change of ownership of 149; fixed capital/ output rate 218; items listed in SNA 2010 152; ownership rights 148; transfer of ownership of 205 flow of funds 10, 11; agenda of 23; based data 13, 90; contemporary description of 77; distributional accounts of USA 82, 83; estimating 24; flows of credit and spending 82; graphs 40; long-term loans 77, 82; modern national accounts and 24–25; for money measurement 82; national accounts and DSGE models 25–27; origin of 24; as overarching framework 24; as overarching model 77–84; prime goals of 24; project initiated by NBER 82; quadruple accounting 85; trade credits 77; US 77, 78–81 flows of labor 54, 128; changes in 65; statistics 118; UK, 2001–2018 118, 119

Friedman, Milton 89; farm prices for milk in 50–51 Frisch elasticity 123 GDP see gross domestic product General Theory, The 107 Germany: average number of hours worked per person 124; domestic labor share of 249; gross fixed investment rate 215–216, 217; household consumption components 181; nominal unit labor costs in 245; unemployment after the war 131; unemployment during 1910–2018 133 gift exchange 42 global competition 56 gold standard and unemployment, link between 107–108, 131 Google, quarterly balance sheet of 76 Gordon, Robert A. 115 government consumption 7, 8, 20, 29, 32, 182, 183, 194; collective 179; incongruous concepts of 2, 20; individual 179, 183; USA 29 government expenditure 2, 14, 16, 18, 193; analysis of 8; collective 183; DSGE models of 2, 7; household and, border between 180; 1954 Samuelson article about 8 government guarantees 43, 75, 96 government production 14, 20; failure to introduce 192; neglect of 8; positive 20 government spending 7, 210; wasteful nature of 194, 210 Great Depression: neoclassical explanation of labor during and after 135–139; unemployment during 132–139 Great Transformation, The (book) 47 Greece: compensation of employees per hour worked 245; consumer price level in 2010–2012 228; household consumption components 181; labor unemployment in 107; nominal unit labor costs in 245, 248 gross arable and livestock production 238; current and fixed prices 237, 237; Törnquist deflated volume indices 236, 237 gross domestic product 22, 28, 45, 246, 247, 249; decline of 63, 134, 137;

258 Index definitions 19; deflation procedure of 226; deflator 249; volume estimate of 227, 240 gross domestic product, percentage of: decline in level of investment 158; household consumption components 181; nominal gross fixed investment rates as 199–200, 214, 219 gross fixed capital 198 gross fixed capital formation: changing nature and definition of 199–202; credit boom 201; decline of 200–201; definition of 198; ESA 2013 and 202–203; investment rate of Western world 212; Kuznets’ views on 212– 213; long run development of 199, 199; newly industrialized countries 199; patents and software in 204; process 202; statistical definitions 202–205; Swedish economy 199–201 gross fixed investment rate: comparison with findings of Piketty 218–219; decline of 219; Denmark and Ireland 205; expenditure/ acquisition components 202–204; Finland, Denmark and Ireland 217; France 214, 215; Germany and the Netherlands 215–216, 217; importance of 212; Italy and Spain 216, 218; Sweden 199, 200; UK 215, 215; USA 213, 214, 218 Haavelmo, T. 2, 21 Haitian dollar 47 Hayek, Friedrich von 47, 48, 54–55, 56 ‘headline’ unemployment 110, 114 ‘Heterodox’ 76 hoboes 132, 135 homo economicus 4 household accounts 39 household consumption 3, 227–228; components of 181; definition of 178; historical development of components of 180 household expenditure 177, 179; and government, border between 180 household labor 249 household production and technological developments 65–66 household purchases 32, 180, 182, 189, 198 houses-roads-malls-offices network 220 human capital 153

ILO see International Labor Organization IMF: on double deflation 227; on real unit labor costs 249 individual consumption 3, 180; and public government, national accounts definition of 182 see also ‘actual individual consumption’ individual government consumption 179, 183, 194 International Chamber of Commerce 56 International Labor Organization 54, 65, 135; classifications of people in labor force framework 116, 116–117; domestic labor statistics 62; first annual report of 55; functions of 113; history of 113; report on domestic labor 61–62; unemployment definitions 114–115, 117, 131; work definition 113–114 intersectoral debt, historical changes in 40 ‘investment’: as expenditures or acquisition 202–203; household durables and 203 ‘investment’ as variable in DSGE models 205–211; aggregate demand 206; capital goods 209; changing environment 207; coconuts-concept of investment 207; government spending 210; households 209; long-term estimates of capital 208; modern growth theory and 208; neoclassical general equilibrium 206–207; smoothed series 211; Solow retrogression 206; stock/ flow consistent estimates of capital 208–209; stock of capital 206–207, 211; subsistence accounting system 207; time consistency 211; time inconsistency 207; ‘workhorse’ concept of capital and 205–206 investment bubbles 216, 219 investments see gross fixed capital formation investment spending 205, 206 Involuntary Unemployment (book) 122 ‘involuntary unemployment,’ concept of 105, 106; after 1940 108; evidences supporting 107–108; lower wages and unemployment 106–107; Lucas’ arguments against 109–112

Index  259 Ireland: accounts receivable and payable, flows 45, 46; compensation of employees per hour worked 245; construction bubble 13, 22, 205, 214; debts of construction sector 150; employment in 205; explosion of household debt 93; gross fixed capital formation in 205; gross fixed investment rate 217; nominal unit labor costs in 245; other accounts receivable and payable, flows 45; status as tax heaven 46 Irish households, stock and flow of ‘long-term loans’ of 87–88, 88 Irish industry, ULC of 248 Irish resident private-sector enterprises, credit advanced to 6 Italy: compensation of employees per hour worked 245; domestic labor share of 249; gross fixed investment rate 216, 218; nominal unit labor costs in 245 Jevons, William Stanley 10 job creation/destruction statistics 119– 120, 127–128 joint liability rule 44 Kaldor-paradox 248 Keynes, John Maynard 10, 15, 20, 95, 105, 115, 137, 206; concept of ‘involuntary unemployment’ 106– 107; on models of Colin Clark 19; on national economy 18; national income innovation 18; ‘profit inflation’ policy in India 18 Klamer, Arjo 7 Koopmans, Tjalling 28 Kuznets, Simon 11, 12, 15, 19, 60, 206, 208, 212, 219–220 labor: and capital, productive combination of 57; class, rise of 54; commodification of 54–56; flows see flows of labor; underutilization 117 labor force 55, 118, 125–126, 136, 138; domestic servants 63; framework, ILO classification of people in 116, 116– 117; participation 111; statistics 115; unemployment see unemployment labor market 64, 105, 112; after 2008 financial crisis 125–126; DSGE models 112, 124–125; dynamic

nature of 118; neoclassical ideas about working of 121–129; rules 53 labor share per sector of Netherlands 246, 248 labor supply 123; depressed 136; elasticity of 123 land 153–154, 155, 158 Landbouwkundig Economisch Instituut 234 Laspeyres quantity indices 229 legal owner 146 LEI see Landbouwkundig Economisch Instituut ‘leisure,’ unemployment equated with 5, 6, 97, 111–112, 115–116, 121–123, 136–137, 193 les trentes glorieuses 214 Lidl 53 lifestyle changes and household durables 220 Lijempf dairy company, 1919 annual report of 69n8 Lindert, Peter 232 linguistic confusion 2 liquid assets 45 long-term loans 82; Irish non-financial companies 150, 151; stock and flow of 88 long-term unemployment 112, 123, 129, 133–134 Low Countries 57 lower wages and unemployment, link between 105–106, 112, 123 Lucas, Robert 108–111; business cycle framework 109; on cyclical patterns of unemployment 111; equated unemployment with ‘leisure’ 111 macro-economic data measurement 2 ‘macroeconomic formula of everything’ 84 macroeconomic imbalance procedure 244 macroeconomic monetary statistics 82 macroeconomics 2, 15 macroeconomic statistics: macroeconomic events outside framework of 12–14; political nature of 115 macroeconomic theory and macroeconomy measurement craft, link between 2 macro-economy 15

260 Index macro-statistical variable, phases of development of: conceptualization 30; definitions 30–31; measurement 31; operationalization 31 market contracts 49 Marshall, Alfred 4, 10–12, 15–17, 19, 183, 192; on aggregation problems 11; Principles of Economics 16; ‘usevalue’ related variable 11–12 mass production processes: farm prices for milk 50–51; pricing systems in 50 ‘measured unemployment’ 111, 115, 122, 127, 136–138 ‘measurement without theory’ (article) 28 Measuring Business Cycles (book) 28 Mesopotamia 57 MFIs see monetary financial institutions microfounded models 9–10, 90 Microsoft, balance sheet of 68n1 Microsoft Ireland, payables and receivables of 44–45, 45 micro-utility concept 11 military investments 213 milk, farm pricing for 50–53 Mill, John Stuart 10, 94 MIP see macroeconomic imbalance procedure Mises, Ludwig von 56 Mitchell, Wesley 10–11, 14, 20, 23, 28, 39, 47, 89; FOF agenda 23–24; on money economy 17; as statistical hero 12 Mitchellian business cycle 94; analysis 109; methodology 89 Mitra-Kahn, B.H. 15, 19 M3 money 75, 77, 85–87, 95; and Divisia money year on year growth rates 93; from MFI credit to 87 M2 money aggregate 90 models and measurements, differences between 4, 12 modern economic life, fundamental manifestations of 228 modern economies 13, 54, 57, 58, 233 modern growth theory 208 monetary economy 40, 48, 56, 87, 105, 146, 178 monetary financial institutions 24; credit 76; credit to M3 money 86–87; deposits originated by 46; license to create money 75, 76; monetary liabilities of 84

monetary market societies 57 monetary models without money 97 monetary prices 21, 41, 47, 48, 50, 159 monetary societies: books about nature of 47; evolutionary change of 47; monetary transactions in 39; prices and pricing in see prices and pricing monetary statistics 45 monetary transactional activities 9, 10–11 monetary transactions see transactions, monetary monetary value 146 monetary variables 13 monetary world 95, 97, 101, 191, 232 money 40, 42, 99; accounting 87; as bills of exchange 44; creating transactions 42; creation of 75–77; definition 43, 99; ‘Divisia’ definition of 86, 87; joint liability rule and 44; kinds of 45–46; manuals 76; as means of exchange 43, 47; ‘M3’ money measurement 75, 77; and models 97–101; origins and uses 92; payables and receivables as 44–46, 45; private 43; role in DSGE models 97–99; stamps 43, 43; as unit of account 45–47; velocity 94; weighted aggregate of amounts of 91 money creating transactions, models abstract from 99 money creation 85; as function of sectoral growth of credit 24; process 99 money economy 10, 17 ‘money in utility function’ approach 98 moral elements in production chains 53 mortgage contracts 77 mortgage debt in Ireland 87 NA see national accounts NA/FOF data 13, 22 national accounting 15–16, 18; development of 19–20; institutionalization 20; literature on 15–16 national accounts 10, 11, 12, 44; ‘chief points’ of 60; development of 21; division of non-financial assets 152– 153; and DSGE models, comparison of variables of 29; financial corollary to 23; as instrument of control 68; integration with flows of credit

Index  261 and lending 82; maintenance costs and 177; nature of logic of 20–21; vs. neoclassical macro-models 170–172, 194–195, 221–222; as political accounts 22–23; production boundary 62; production of state into 15; statistics, categories in 32 national accounts data 13; publishing 20; volatility of investment 28 National Bureau of Economic Research 28; FOF project initiated by 82; study on production boundary 17 ‘national dividend’ 16, 17 national economies 15, 17, 18, 22 national income: concept of 19, 60; statistics 68 National Income and Its Composition, 1919–1938, Volume I (book) 19 national prosperity 16 national resources, depletion of 69n9 natural capital 54, 162–164; and national accounts, relation between 164 see also capital natural non-produced assets 147 ‘Natuurmonumenten’ (charity) 177 NBER see National Bureau of Economic Research neoclassical ‘DSGE’ macromodels: design of 7; introduction of 169 see also DSGE macro-models neoclassical economics 3 neoclassical general equilibrium 206– 207 neoclassical macroeconomic models: macroeconomic events outside framework of 12–14; micro-founded models 9; vs. national accounts 170– 172; without credit 94 neoclassical macroeconomists 3 neoclassical macro theory and statistical macro-measurement, conceptual differences between 5–9, 11, 14; concept of consumption 7–8; consumer goods 8; government expenditure 8; health 9; history of 10–12; unemployment 5–7 neoclassical models of government 20 Netherlands: farm gate milk prices 52; gross fixed investment rate 215–216, 217; job creation/ destruction statistics 119; labor share per sector of 246, 248; producer prices of arable and livestock products

235, 236; and UK, unemployment of 108, 109 ‘New Deal’ rules 135 New Deal ‘workfare’ jobs 133 New Keynesian complete markets macroeconomic theories 7 NHS see UK National Health Service nominal agricultural value added, deflating 240 nominal and real income in USA, time series of 60 nominal unit labor costs: conceptual problems with 247–249; definition 247; definition of 243–244; of eurozone countries 244; upper threshold 244; way to decrease 244– 245, 245 ‘no money no buy’ approach 98 non-financial assets, division of 152–153 non-MFI financial institutions 87 non-monetary labor 39 non-monetary world, interest rate in 97 non-produced assets 147–148, 152–153 Non Profit Institutions Serving Households 30 non-tradables 248 North Italy 57 NPISHs see Non Profit Institutions Serving Households NULC see nominal unit labor costs Office for National Statistics 15, 23 O’Hara, Scarlett 58 ONS see Office for National Statistics Paasche and Laspeyres index 234 paid domestic labor 61, 62; black women 64; immigrant women 64; measurement 67; minimum wage for 65; shifts between 63 peer to peer consumer credit system 38 pension fund 75 people and jobs, difference between 118–120 Perkins, Frances 65 ‘perpetual inventory method’ 206 Petty, William 16 Piketty, Thomas 10 Polanyi, Karl 47, 48, 54, 64 ‘Polanyi problem’ 56 political accounts, national accounts as 22–23 Political Arithmetick (book) 16

262 Index political economy 8, 14, 16 Portugal: compensation of employees per hour worked 245; nominal unit labor costs in 245 postal companies 68n2 postal services 46 potato prices 234 price indices 226 prices, as elements of contracts 58 prices and pricing 41–42, 47–59; adapting production and purchases to 49–50; ‘administered prices’ 49, 54; farm prices for milk 50–53; fixed prices 50; inputs and outputs 48; labor market rules 53; mass production processes 50; moral elements in production chains 53; multi stage production processes 49; overproduction and underproduction 48–49; price contracts 49, 50; role in market coordination 48–49 Principles of Economics (book) 16 pristine forest 153 private banks 100 private money 43, 43, 46 producers and consumers 49 production, formula for 32 production boundary 147; government 17; government and 15; SNA 117 production processes, multi stage 49 productive economy 16 productive government actions 16 products and relative prices change 40–41 profitable investment, structural level of 219 ‘profit inflation’ policy in India 18 ‘Project for strong labor markets and national development’ 219 public goods 9, 180, 183, 193, 194, 203; in DSGE models 178; Samuelson’s definition of 169 purchasing power: calculation 240–241; care needed for calculating 226–227; parities 232 QE effect 92 quadruple accounting 85, 88, 92; aggregates 94–96; flow of funds 90 ‘real’ economic growth 240 ‘real’ economic production series 228

‘real’ income 60, 228; care needed for calculating 226–227 ‘real’ production: care needed for calculating 226–227; and consumption 233 real purchasing power 227 real unit labor costs 243, 249 ‘real’ value added 227 real wages 125–126; declines in 123; and labor sold, relation between 122; and unemployment, relation between 106–108 reciprocal exchange 42 regulated prices 42 relative money prices, spatial and historical nature of 232 relative prices, importance of 232 rent incomes 58 representative consumer 4, 39, 121; vs. NA/FOF 10; and real persons, utility of 11 ‘representative household’ 121 representative patient household 39, 100, 121, 186 Road to Serfdom, The (book) 47 RULC see real unit labor costs Samuelson, P. 8 Schularick, Moritz 10 Schwartz, Anna 89 Scrooge McDuck approach 98 search behavior 117 search theory 127–128 second hand market 156 sectoral consumption flows 181 shadow prices: calculation 42; concept of 41 single deflation 227, 228, 232 slavery, abolishment of 149 slave states of USA, pre-Civil War fixed capital in 148 slave wealth 58 Smith, Adam 15, 16 SNA see System of National Accounts social monetary network 40 social realm of models and measurement 12 Solow, growth theory of 168–169 Solow growth model 220 Solow retrogression 206 Spain: during 2008 building boom 222n2; compensation of employees per hour worked 245; female

Index  263 participation rate in 126; flows of labor in 118; gross fixed investment rate 216, 218; nominal unit labor costs in 245; unemployment and year on year change of earnings of 107 Staatsbosbeheer 177 standard of living 232 statistical heroes 12 statistical production boundary, nature of 59–68 statisticians and theoreticians, social and cultural world of 12 Stevin, Simon 15 stock of capital 206–207 subsistence accounting system 207 subsoil assets 153 Sumner, Scott 226 superlative price indices 233, 234 Sweden: economic history 199–201; entry into modern era 200; nominal gross fixed investment rate in 199, 199–201; unemployment rates in 108 Swedish production of electronics, double deflation in 231, 233 system banks 46 System of National Accounts 147, 195, 229; productive activities beyond 117; sectoral consumption flows according 181; SNA 2008 227; SNA 2010 150, 152 taxes 21 TC see traditional contracting technological developments and household production 65–66 theoretical models and statistical models, difficulty in comparing 3 time poverty 39, 252 Tinbergen, Jan 20 Törnquist index 234; artificial fertilizer 238, 238–239; Dutch gross arable output 235, 235; gross arable and livestock production 236, 237 total consumption 8, 187, 188, 190 trade credits 77 traditional contracting 50 transactional macroeconomy 20 transactional monetary economy 15, 40 transaction cost economics 50 transaction economy 220 transactions, economic 100 transactions, monetary 21, 42, 59, 75; based on consumer credit 38;

debt based 40; definition 40; ex ante dowries 42; involving money 38, 40; and market 41; modeling 40; monetary income and expenditure 41; in monetary society 39; shaping relations 38; social nature of 67–68; workhorse model in 97 TSRPiESiMoAN winners 12 UK: 2001–2018 flows of labor in 118, 119; GDP growth for 1997–2015 period 240; leaving the gold standard 108, 131; and the Netherlands, unemployment of 108, 109; unemployment and year on year change of earnings of labor 106–107 UK National Health Service 9 ULC see unit labour costs unemployment 5, 106; ‘classical’ economists’ views on 106; Congressional hearings on measuring 114–115; cyclical patterns 109–111; data for prewar years 131; definitions 105, 114–115; equated with ‘leisure’ 111–112; during the Great Depression 132–134, 135–139; ILO definitions 114, 117; involuntary 105; long-term 123; long-term vs. short-term 112; and low wages, link between 105, 112, 123; post-2008 131; pre-war data 131; U-3 114; UK and the Netherlands 108, 109; US definitions 115–115; as wage problem 105; and year on year change of earnings of UK ‘labor’ 106–107 see also US unemployment Unilever 53 unit labour costs: calculation 247; definition 247; definition of 243; German 248; product or sectoral 247 unpaid household labor 61 unweighted money aggregates: construction of 92; M1, M2 and M3 92 USA: ‘Beveridge curve’ 133, 134; distributional accounts of 82, 83; economy as overheated economy 134; gross fixed investment rate 213, 214; gross job gains and losses quarterly data 120; real government consumption plus investment 29; real private consumption 29; real private investment 29

264 Index US Federal Reserve 23, 82, 90; US flow of funds 25, 77, 78–81 US unemployment: during 1910–2018 133; after 1929, rapid rise in, 129–130, 131; after World War II 134–135; broad and normal 112, 132; declines in hours worked 129– 130, 130, 134–135; during the Great Depression 132–134; headline and broad 110, 131; post vs. pre-war 134; wage deflation and 130, 132–133 see also unemployment utility curve for homo economicus 9 U-3 unemployment 115 valuation of capital 156–159 value added 227, 229, 233; constructing ‘real’ measure of 228; fundamental monetary variable 232 Veblen, Thorstein 3, 7, 10, 11, 20 vehicle registrations, worldwide 167 volume of capital 159–162

volume of production see ‘real’ economic production series wage costs per unit of product 246 wage employment 54, 56 wage labor 54, 97, 107, 147; Lucas’ views on 111; rise of 160 wage level and competitive position, link between 249 wages and hierarchy position, link between 57 wartime inflation, distributional effects of 18 Wealth of Nations 16, 183 Western Christian wedding vows 42 Western surge, analysis of 232 ‘work,’ concept of 113–114 workers of the world 55 ‘workhorse’ concept of capital 205–206 Zuid-Kennemerland National Park 177, 178