Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics

Hidden Collective Factors in Speculative Trading AStudy in Analytical Economics ~ Springer

Hidden Collective Factors in Speculative Trading

Bertrand M. Roehner

Hidden Collective Factors in Speculative Trading A Study in Analytical Economics Second Edition

Professor Bertrand M. Roehner University of Paris 6 LPTHE 4 place Jussieu F-75005 Paris France [email protected]

ISBN 978-3-642-03047-5 e-ISBN 978-3-642-03048-2 DOI 10.1007/978-3-642-03048-2 Springer Dordrecht Heidelberg London New York Library of Congress Control Number: 2009932723 © Springer-Verlag Berlin Heidelberg 2001, 2009 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover design: WMX Design GmbH, Heidelberg Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

History for most beginning economists seems to begin in 1945. How to imbue students with a thirst to learn how the present relates to the past is not obvious. Anna Jacobson Schwartz (1995)

The temptation to form premature theories upon insufficient data is the bane of our profession. Sir Arthur Conan Doyle (1915)

Preface to the second edition

For the present edition four chapters have been added which form the fourth part at the end of the book1 . Entitled “The triumph of neoliberalism”, the new part explains how the implementation worldwide of the neoliberal agenda paved the way for the present crisis. As a matter of fact, the evidence provided in chapter 9 suggests that the present crisis already began to build up in the mid-1970s. It is around 1975 that (real) US wages reached a peak-level they would never regain in following decades. It was also around 1975 that the number of strikes began to fall sharply. The mid-1970s also marked the beginning of a huge inflow of immigrants (in large part of Hispanic origin) into the United States. The inflated supply of labor depressed wages and this had the consequence that consumption could be increased only by an unprecedented development of credit. Perhaps the reader may think that to blame the prevailing economic system for the unfolding depression is a fairly common and all too easy temptation. In fact, the author did not wait until 2009 to question the neoliberal creed and to take it to task2 . Moreover, it can be observed that even at the present time of writing there are but few voices in the media who establish a link between the roots of the crisis and the main tenets of the neoliberal agenda. Indeed, as we will see in chapter 9, the term “neoliberalism” is hardly ever used in major newspapers. As another illustration, one can mention that according to current polls conducted in Britain, the Conservative Party has a lead of 19 percent over the Labour Party. This clearly suggests that most people do not wish to establish any connection between the ideology of Thatcherism which still underpins the political agenda of the Conservative Party and the present woes of the British economy. The most current attitude nowadays is to attribute the crisis to some errors 1

They are numbered 9-12; the “old” chapters (1-8) have been kept unchanged. See for instance chapters 6 and 8 of “Driving Forces of Physical, Biological and SocioEconomic Phenomena” which was published by Cambridge University Press in May 2007. 2

viii

Preface to the second edition

and excesses; this notion is embodied in the expression “subprime crisis”. That this expression is much too narrow is obvious from the fact that, as we will see, the policy of making subprime loans and packaging them into mortgage backed securities was already used during previous real estate booms and can hardly account for the catastrophic consequences seen in 2007– 2008. The policy which results from such a narrow conception is clear: bail out the defaulted companies, make credit flow again, restore confidence and the economic system will be up again for a new period of prosperity. This is to forget that the 1990s were a time of affluence only for those who were able to partake in financial profits. Wages, on the contrary, did hardly increase either in the United States, in Europe or in Japan. To talk about a golden decade is also to ignore that between 1995 and 2007 economic growth in the United States was largely fed by an increasing deficit of the current-account balance (the largest component being the deficit on goods). By this mechanism a substantial part of consumption and growth was achieved through an increase of the outstanding Treasury debt3 . As the deficit of current account was on average of the order of 3% of GDP, it means that once this spurious part of growth is discounted, the real growth was in fact fairly weak, of the order of 1% per year. The same argument also holds for other countries, for instance Britain. In spite of being a major oil producing country, Britain had an average deficit of current account of the order of 2.5% of GDP over the period 2004-2007. If the influenza epidemic which appeared in March 2009 spreads worldwide, it will temporarily hinder international trade. However, the example of the influenza epidemic of October 1918 suggests that this should not have a big impact on economic growth. Indeed, despite the epidemic, the US Gross Domestic Product increased by 9% in 1918. The contraction which occurred in 1920–1921 had nothing to do with the epidemic. The present crisis has been preceded by numerous warnings. From the failure of “Long Term Capital Management” (LTCM) in 1998 to the bankruptcy of Iceland in 2008, there have been many financial and economic crashes. What gave to these crashes the character of specific warnings is the fact that the companies and countries involved embodied the very essence of the freemarket ideology. 3

In this respect the 1970s also marked a turning point. From 1945 to 1975 the ratio of Treasury debt to Gross Domestic Product (GDP) fell from 120% to 35%; after 1980 it began to increase fairly steadily.

Preface to the second edition

ix

• Founded in 1994, LTCM was a US hedge fund whose board comprised two directors, Myron Scholes and Robert C. Merton, who shared the 1997 Nobel prize in economics. Highly leveraged, its operations were based on quantitative stochastic models. At first they were very successful with annual returns of about 40%. The wind turned in mid-1998. After losing $4.6 billion, LTCM was bailed out by a consortium of banks in an operation supervised by the Federal Reserve. The failure lead to an abrupt but short-lived stock market crash to which the Federal Reserve responded by lowering its interest rate. In short, this episode prefigured in several ways the bailout of much bigger companies such as Sears or AIG ten years later. The scenario was very much the same: sudden and unexpected failure, bailout led by the Federal Reserve, stock market crash, lowering of interest rates and massive loans made to the banking system. There was a difference in scale however: in 2008 the losses were (at least) ten times larger than in 1998. What made the failure of LTCM particularly ominous for the future was the fact that the sophisticated hedging procedures which were supposed to provide a shield against heavy losses simply did not work. • Enron Corporation was awarded the title of “Most innovative US corporation” by Fortune Magazine in 2001 just a few months before asking for bankruptcy protection. Two months later, in January 2002, there was the bankruptcy of “Global Crossing” and in July 2002 the failure of WorldCom. What made these failures of particular significance is the fact that they revealed huge accounting scandals. In his book published in 2003, Frank Partnoy shows convincingly that in many other (more technical) ways these collapses prefigured the global financial crisis of 2007-2008. What makes his analysis particularly meaningful is of course the fact that it was published four years before the outbreak of the present crisis. • Argentina used to be the International Monetary Fund’s poster child. During the 1990s the IMF praised Argentina and used it as an example to show the rest of the developing world what sort of economic success could come from following the neo-liberal and free market policies of the IMF. By 1995, 90% of all state enterprises had been privatized. But prosperity lasted only so long as credit was abundant. In December 2001, Argentina plunged into a devastating economic crisis. Privatization policies also lead to lackluster performances in other Latin American countries. • Britain, Hungary, Ireland and Iceland were acclaimed examples of the success of free-market policies. Yet again, in all these cases, the level of indebtedness was staggering and the financial houses of cards collapsed as

Preface to the second edition

x

soon as credit became scarce.

Deflated stock index

The comparative methodology that we advocate and use in this book often leads to testable conclusions and predictions4 . Naturally, predictions are dependent upon the ceteris paribus condition (i.e. “all other things being equal”). This can be illustrated by the prediction regarding the evolution of

50

Prediction

40

Actual course of the NASDAQ

100 90

Buybacks by quarter

80

30

70 60

20

50

40

10 9 8 7 6

30

0

2

4

6

8

10

12

14

16

18

20

Years (peak=10.3)

Fig. 1 NASDAQ Composite index: prediction compared with observation. Year 10 on the horizontal axis corresponds to 2000; the downturn occurred in March 2000 which corresponds to 10.3. The prediction covers the interval 10-16 that is to say 2000-2006. Made in May 2000, it was published in January 2001 in the first edition of this book (Fig. 7.12). The prediction is based on the common pattern of speculative price peaks observed on various stock markets since the mid-19th century. The main purpose of this graph is to illustrate the importance of the ceteris paribus (“all other things being equal”) condition by showing that massive buybacks (which are basically unpredictable) can have a substantial influence on stock prices. The vertical scale on the right-hand side refers to buybacks expressed in billion of dollars; they tripled in 2003-2004. Sources: NASDAQ: http://finance.yahoo.com/; Consumer Price Index: Website of the St Louis Federal Reserve; Buybacks: Press releases of Standard and Poor’s. 4 These predictions rely on observed quantitative regularities rather than on mathematical models. The reasons for this are explained in Roehner (2008 b).

Preface to the second edition

xi

the NASDAQ Composite index in the years after the crash of March 2000 that was proposed in the first edition of this book and is recalled in Fig. 1. It was in agreement with observation for four years, from 2000 to 2003. Then, something unexpected happened. Many companies (e.g. IBM, Intel, Microsoft) started massive buyback programs of their own shares5 . As another illustration, one can mention a prediction about real estate prices in the west of the United States which was made in mid-2005 and published in Roehner (2006). So far it turned out to be correct but it is clear that if in coming months (or years) there is a massive federal program to prop up house prices, the ceteris paribus condition will no longer hold. In this respect, the situation is exactly the same as in experimental physics when an unexpected exogenous factor interferes with the phenomenon that one tries to observe. The previous examples show that one must be very careful in comparing theoretical predictions with observations. However, what makes the discussion of many economic issues quite unsatisfactory is due to a much simpler reason. It is the fact that the lessons of history are completely discarded. This point is underlined in the citation by Anna J. Schwartz that can be read on the first page of this book. As an illustrative example (see chapter 12 for more details), one can mention the fact that in 1893 one third of the American railroad companies had to be bailed out6 . On account of such evidence should it not be clear that the privatization of railroads can hardly be a successful operation? When it was attempted by the government of Margaret Thatcher it turned out to be so successful that the private “Railtrack” company had to be re-nationalized in 2006 in the wake of a series of tragic accidents. Yet, in spite of such converging evidence, the privatization of European railroads is still on the agenda of the Commission of the European Union. Apart from the loss in efficiency due to increased technical and economic segmentation, another adverse effect of privatization was amply demonstrated in the past decade. It can be called the “bonus-triggered doctoring effect”. When the bonuses of executives are tied to specific levels of earnings, it is extremely tempting to do just about anything to meet (or appear to meet) the goal: cooking the books, playing down debt through off-balance sheet ac5

At the same time there was a huge increase in mergers and acquisitions which is also known as a factor which raises stock prices, mainly because to make such operations successful share holders are offered more than the current market price. 6 For one of these companies it was the third bailout in a row. As is well known, British and French railroad companies also faced similar problems.

Preface to the second edition

xii

counting, favoring short-term profit over long-term investments are some of the methods which have been used at Enron, WorldCom, Fannie Mae, Freddie Mac and numerous other companies. For listed companies this temptation is even stronger because many bonuses are paid in stock options (call options) which, to become profitable, require share prices to go up. Thus, shareholders must be cajoled in any possible way. Yet, in current-day mainstream thinking privatization is still regarded as a miracle cure. How can one explain such a paradox? A clue may have been given by former Vice-President Al Gore when, during his lengthy campaign on the issue of carbon dioxide emissions, he observed that: “You cannot make somebody understand something if his (or her) salary depends upon not understanding it.” It seems that the same observation also applies to the agenda of neoliberalism. Tailored for the wealthy, this political philosophy has well served its purpose for many decades. Why then should it be questioned? Bertrand Roehner Paris, 10 May 2009

Preface to the first edition

This book develops a few simple but unconventional concepts. (i) The same basic mechanisms are supposed to be at work in speculative trading whether it concerns property values, stocks, antiquarian books or other speculative items. (ii) We deliberately avoided investigating individual speculative episodes. Instead, for instance, of discussing the “causes” of the crash of October 1929, we address the more general question of why a bull market cannot last forever. The first question belongs to what Harvard sociologist Stanley Lieberson categorizes as “undoable” questions; the second question on the contrary is a prerequisite for any further and more detailed investigation. (iii) A purely economic explanation of speculative trading cannot be satisfactory for, in a very fundamental way, speculation is a social as well as an economic phenomenon. As a matter of fact, the first two ideas are not specific to the question of speculative trading. They delineate a comparative approach which also guided many of our former investigations. What is new at the present time is the fact that the econophysics movement which started around 1996 has given to that approach a new legitimacy and actuality. In very general terms econophysics contributed to a shift from standard anthropocentric modeling where the individual motivation and behavior of economic agents play a key role, to a more impersonal and scientific approach based on patterns of collective behavior. This book relates an inquiry which has been in progress for many years; its writing has been an exhilarating journey in the course of which I came to explore many different facets of speculative trading ranging from diamond or postage stamp markets to property or stock markets. May be in some places I have erred; that is almost inevitable if one considers the diversity of the data that needed to be processed. Needless to say, I welcome in advance notification of possible errors or omissions. Another observation is in order regarding repetitions. Only few readers will probably read that book throughout from first to last chapter; accordingly,

xiv

Preface to the first edition

some useful definitions have been purposely repeated several times in different chapters. It is a pleasure at this point to thank the many people who provided encouragement, support and advice. My first contact with what was to become econophysics was through an email from Didier Sornette on 2nd October 1995; since that time we have had a permanent and stimulating collaboration. Many thanks to him for sharing with me his insight and enthusiasm. Another milestone was my encounter in February 1998 with Gene Stanley by whom the word “econophysics” was coined at about the same time. In the fall of the same year I visited his group at the University of Boston; all my gratitude to him for his warm reception. Yi-Cheng Zhang contacted me in September 1998 while I was in the United States; ever since I have admired his capacity to develop new ideas and to bring people together. At about the same time I came into contact with Dietrich Stauffer and I soon came to appreciate his rigorousness and his invigorating humor. The views expressed in this book have also benefited from discussions with several colleagues to whom I express my appreciation; let me mention in particular Luis Amaral, Doyne Farmer, James Feigenbaum, Peter Freund, Rosario Mantegna and Gilles Zumbach. Moreover, I am indebted to a number of distinguished economists for their unfailing support and interest. Since the time of our first contacts in 1991, my exchanges with Guy Laroque and Edmond Malinvaud have been a permanent source of encouragement. In 1998, at a time when I already was no longer a “conventional” economist (if I ever was one), Jeffrey Williamson invited me to spend the fall term at the Harvard Department of Economics, a stay which was a highly stimulating experience. Many thanks to him for his openmindedness and for having, through the comparative orientation of his own work, set an example that I was glad to follow. Similarly the work of other “comparativists” such as James Foreman-Peck or Gunnar Persson convinced me that I might be on the right track after all. At the present moment there are still too few links between econophysics and economics and Thomas Lux had an instrumental role in bridging the gap between the two fields; many thanks to him for his dedication. Ever since our first contact in 1993 in connection with the publication of “Theory of markets” my exchanges with Werner M¨uller, economics editorial at Springer-Verlag, were a real pleasure. Many thanks to him for his interest and foresight. Finally, I would like to express my thanks to my colleagues at my laboratory and especially to Laurent Baulieu, Bernard Diu, Jean Letessier

Preface to the first edition

xv

and Ahmed Tounsi. The book is dedicated to my wife Brigitte and son Sylvain whose cheerful encouragements and stimulating support have been invaluable. Bertrand Roehner Paris, 22 March 2000

Contents

Preface to the second edition Preface to the first edition PART I

vii xiii

PROLOGUE

Chapter 1

Introduction

1

1 “Practical” versus “solvable” questions 2 A crucial step: finding regularities

1 3 4 4 5 6 7 7 7 8 10

3 4

2.1 Models need precise targets 2.2 In search for regularities 2.3 A parallel with other complex systems 2.4 What can we learn from the parallel with meteorology? 2.5 Devising quasi-experiments Basic features of our approach 3.1 Comparative analysis 3.1 Identification of speculative episodes Presentation of the book

Chapter 2

Overall view of speculative markets

1 Overview of speculative markets 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

What kind of price data do we need? Grain markets Commodities Land and property Postage stamps Antiquarian books Diamonds Other speculative markets

13 14 15 17 20 22 28 31 32 35

Contents

xviii

2

1.9 Stock markets Comparative perspective 2.1 Relative weight of various speculative markets 2.2 Turn over ratio 2.3 Price earnings ratio

PART II Chapter 3

HIDDEN COLLECTIVE DETERMINANTS Rational?

1 Hidden preconceptions: long-term perspective

2

3

1.1 The attitude toward usury 1.2 The attitude toward slave trade 1.3 The attitude toward health care 1.4 Nationalization versus privatization 1.5 A possible clue to shifts in conceptions Hidden preconceptions in economic rationality: medium-term perspective 2.1 The Paris real estate market in 1990 2.2 Shift in accounting rules Application to speculative trading 3.1 The crystal ball 3.2 How a mania spreads 3.3 Short-term phenomena

Chapter 4

Joint crashes

1 Is speculation in diamonds related to speculation in cobalt? 1.1 1.2

2

3

35 43 43 43 45

Other commodity bubbles which burst in 1980 Why it is almost impossible to build econometric models for floating stockpile markets 1.3 A possible clue: a hedge against inflation Are property bubbles in Tokyo and Paris related? 2.1 The expulsions of Jesuits in the 18th century 2.2 Mississippi versus South Sea bubble 2.3 Property crashes Interdependence between different speculative markets 3.1 The link between equity and bond markets

47 48 50 50 52 54 55 56 56 57 58 59 63 64

65 66 67 69 70 72 72 73 74 76 77

4

Contents

xix

3.2 Link between crashes in equity and property markets 3.3 Link between crashes in equity and art markets Conclusion

78 79 79

Chapter 5

Contagion of speculative frenzy

1 Social contagion 1.1 The production of books reflects the spread of bubbles 1.2 How the press participates in the spread of a bubble 1.3 Consumer confidence and stock prices 2 Volume of trade as an indicator of the extension of speculation 2.1 Qualitative evidence 2.2 Relationship between trading volume and price variations 3 Why can a bull market not last forever? 3.1 Is there an upper bound for the price level? 3.2 Is there an upper bound for the volume of transactions? 3.3 Applications 3.4 Empirical evidence about transaction costs on stock markets 3.5 Implications of the formula for the upper bound of trading volume

PART III Chapter 6

3

84 85 89 95 97 98 99 108 108 110 110 112 114

REGULARITIES IN SPECULATIVE EPISODES Peak amplitude: the price multiplier effect

1 A case in point: property speculation in Britain

2

83

1.1 The price multiplier effect 1.2 Spatial propagation of speculative fever 1.3 Examination of other property bubbles Generalization of the price multiplier effect 2.1 Illustrative examples 2.2 Postage stamps 2.3 Antiquarian books 2.4 Summary Implications and significance of the price multiplier effect 3.1 Evolution of the ensemble dispersion during

117 119 120 120 123 125 127 129 132 133 136

Contents

xx

4

a price peak 3.2 Is the price multiplier effect a diffusion phenomenon? Application of the price multiplier effect to stocks

Chapter 7

Peak shape: the sharp peak - flat trough pattern 151

1 Describing speculative peaks

2

1.1 Shape of wheat price peaks 1.2 Mathematical description of price peaks 1.3 Observations 1.4 Unified overview Is there a pattern for stock price peaks? 2.1 Necessity of a taxonomy of stock crashes 2.2 Quantitative analysis 2.3 Is the existence of stock price patterns consistent with market efficiency?

Chapter 8

Stock market bubbles

1 A case in point: 1929

2

1.1 The driving force 1.2 The bubble’s signature 1.3 Role of discount rate increases 1.4 Role of the margin debt Comparative analysis 2.1 The engine 2.2 The bubble’s signature 2.3 Role of discount rate increases 2.4 Role of the margin debt 2.5 What is the bottom line after a stock market crash?

PART IV Chapter 9

152 152 155 157 165 167 168 172 176

179 181 182 185 186 187 188 188 189 192 194 194

THE TRIUMPH OF NEOLIBERALISM Triumph of neoliberalism in economics

1 Neoliberalism: an overview 1.1 1.2

137 143 146

Roots of the neoliberal program Neoliberalism and the financial crisis

197 197 198 201

Contents

2 3 4 5 6

1.3 Neoliberal ideology and social fragmentation 1.4 The promotion of neoliberalism State power and cracks at grass-root level Income inequality: the watershed of 1975 Ownership by “absentee landlords” Weakened national solidarity Appendix: Possible scenario for the economic crisis

Chapter 10 1 2 3 4

203 204 205 207 211 214 215

Triumph of neoliberalism in society

219

Leveling off in education trends Abrupt increase in the trend of incarceration rate Change in the trend of infant mortality Increased segmentation of the American society 4.1 Why should one focus on social segmentation? 4.2 Persistence of the Black-White divide 4.3 “White flight” 4.4 Gated communities 4.5 The self-sufficient world of US Armed forces 4.6 Deepening gap between employers and employees 4.7 Aging and underfunded transportation infrastructures 4.8 Decline in the membership of social organizations

220 221 221 221 221 224 224 225 226 226 227 227

Chapter 11

Triumph of neoliberalism in finance

1 Aggravating factors 2 Greater segmentation 3 How can one follow the buildup of a speculative episode?

4

xxi

3.1 Stocks 3.2 Real estate 3.3 Bonds 3.4 Derivatives Spreading neoliberalism: overlords and vassals 4.1 Dollarization 4.2 Currency board 4.3 The aftermath of the Argentinian collapse 4.4 Historical roots 4.5 Other forms of oversight and control 4.6 Privatization and economic return

229 230 234 236 236 236 238 239 241 241 244 246 247 247 251

Contents

xxii 4.7 4.8 4.9

5 6

Role of the armed forces The future Why does privatization offer inroad opportunities to foreign companies? Fate of absentee ownership in Eastern Europe during the next decade? Previous speculative crises 6.1 Previous “subprime crises” 6.2 Regulation, self-regulation or no regulation at all 6.3 A highly profitable business

Chapter 12 1 2 3 4

5 6 7

8

9

Is there a way out?

Crisis of 1893-1900 1932-1933: Runs on banks and closing of the Stock Exchange 1933-1938: Opposition to New Deal policy Anti New Deal reforms in the aftermath of World War II 4.1 Broken promises 4.2 The turning point of the Cold War 4.3 The Taft-Hartley Law (1947) Will Obama’s presidency mark a turning point? What should be expected in 2010 and 2011? Some tests for the years to come 7.1 Comparison between 1929 and 2008 in terms of synchronicity 7.2 Comparison with the Great Depression in terms of social interactions Supply of labor 8.1 In the 19th century 8.2 After 1945 8.3 Can the trend be reversed? Insight about the future 9.1 Toward a new model? 9.2 Is there a way out?

252 252 253 254 255 255 257 258

261 262 264 267 268 268 270 271 273 282 283 283 287 288 288 288 290 293 293 293

References

295

Index

309

PART I

PROLOGUE

I

Chapter

1

Introduction

Many people plan their summer vacations in March or April and would like to know in advance what weather conditions are going to be. Unfortunately, no reliable weather forecasts can be made three or four months in advance at least in the present state of the meteorological science. Similarly all investors around the world would like to know the date of the next property or stock market crash, all sugar producers would like to know whether and when there will be a downturn for the price of sugar. These are undoubtedly "practical" questions, but they are not yet "solvable". In this introductory chapter we explain why it can be of good strategy to put aside the question of practical usefulness (at least for a while) and to concentrate on those questions one is able to solve, however insignificant they may seem at first.

1

"Practical" versus "solvable" questions

When a block of butter is left in the sun it progressively softens before eventually to tum liquid. On the contrary, in the same conditions an ice cube does not soften, it becomes smaller as more ice smells but the remaining piece of ice remains as hard as it was at the beginning. In his "Discourse on method" the scientist and philosopher Rene Descartes undertook to question the reason for that difference from a scientific point of view (Descartes 1637). This, however, is a very difficult problem especially if one aims at quantitative predictions; even three and a half centuries after Descartes, physics is not really able to provide a satisfactory answer. We mentioned that example in order to emphasize that even in the natural sciences there are phenomena which we can observe in everyday life and for which physics has no clear explanation to offer. Actually the very reason for physics' tremendous success in the past two centuries is precisely that it concentrated on the questions it was able to solve. Countless other problems (like the butter/ice issue) were left aside. B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_1, © Springer-Verlag Berlin Heidelberg 2009

1

2

Chapter 1

For most physicists the acceptance of this methodological option is probably implicit rather than conscious and deliberate; yet some scientists had a very clear understanding of the above distinction as appears in the following citation taken from the introduction of ''Theory of games" by J. Von Neumann and O. Morgenstern (1953): ''The great progress in every science came when, in the study of problems which were modest as compared with ultimate aims, methods were developed which could be extended further and further. The free fall is a very trivial physical phenomenon, but it was the study of this exceedingly simple fact and its comparison with the astronomical material which brought forth mechanics [ ... ] The sound procedure is to obtain first utmost precision and mastery in a limited field, and then to proceed to another, somewhat wider one and so on." That "modest objective" approach was highly successful not only in mechanics, but also in chemistry, genetics or atomic physics. Thus, the fundamental concepts of modern chemistry which lead to Mendeleev's table were laid down by comparing simple chemical elements rather than more useful products such as wood, steel or wine; the laws of genetics were first established for garden peas and their application to more useful questions turned out to be far more difficult; in the same way the hydrogen atom and a few other simple atoms served as testing grounds for the concepts and methods of quantum mechanics while understanding the spectrum of more complex atoms was far more tricky.

When one tries to apply the above paradigm to economics there are two main difficulties: (i) It is not straightforward to identify "simple" systems (ii) Detailed statistical information is available only for a small number of systems which are not necessarily the simplest. Let us examine these two points more closely. Regarding the simplicity/complexity assessment two simple criteria can be used. How many elements does the system comprise? How many different interactions does it involve? Many of the problems physics was able to solve are of the two-body type which means that they involve only two elements and one interaction. Several obvious examples can be mentioned: the SunMars system studied by Kepler and Newton, the proton-electron system (hydrogen atom), the interaction between two heat reservoirs at different temperatures (thermodynamics), the interaction between light and a transparent medium (geometric optics), the gravitational interaction between light and matter (deviation of light by matter in Einstein's general relativity). Similarly Mendel's laws in genetics concern the interaction between only two alleles of a gene (riddled/non riddled peas).

Introduction

3

In economics the two-body problem was also studied extensively: it takes the form of the two-company problem in microeconomics, the two-sector problem in macroeconomics or the two-country problem in international economics. Unfortunately, the predictions made for these simple cases could not be confronted to empirical evidence for the simple reason that in the real economy it was not possible to observe such simple systems separately: exogenous factors due to the rest of the economy turned out to be of the same strength as endogenous effects. Perhaps some simple economies of the kind observed in the mid-19th century in parts of Africa or in some Pacific islands could have provided a close approximation to a two-body problem in economics, but no detailed and reliable statistical data are available for such societies. In the same line of thought, ant colonies could probably provide examples of fairly simple economic systems and in that case it would be possible to generate reliable data by laboratory monitoring of ant colonies. For instance an investigation of the North American harvesting ants (Pogonomyrmex Mayr) could provide a model of a twosector economyl. No doubt that such a track will be explored in the future but to our best knowledge no data of that sort are currently available. On the other hand the problems for which detailed statistical records are available are not necessarily "simple". Financial markets are a case in point. Huge data sets are available for exchange rates or stock prices, but neither the money market nor the stock market can be considered as simple systems. Indeed it is fairly obvious and will be further documented in this book that money-, bond-, stock-, and property markets are strongly interconnected. Therefore it is hardly possible to model one of them without also modeling the others and eventually the whole economy. Is the situation hopeless therefore? Certainly not. The next section proposes some clues.

2

A crucial step: finding regularities

Let us open this section with a personal anecdote. My first research work was in neutrino physics, but at that time, in the late 1960s, only fairly weak neutrino beams could be produced in accelerators. As a result a measurement campaign covering several weeks produced only a few dozens protonneutrino interactions and the experimental results I was trying to model had error bars close to 100 percent. Under such conditions all the models which had been proposed by various groups around the world worked equally well. 1 In an ant-economy there would be no prices (at least for an external observer); thus, one would have to reason in terms of working time and material output.

4

Chapter 1

This was a frustrating situation which discouraged any real progress. The circumstances became more favorable only with the introduction of more adequate neutrino beams a few decades later. This example emphasizes an important point namely that the building of a satisfactory model requires a wellidentified and sharply-defined experimental target. Without Kepler's three laws for the trajectory of a planet around the sun, Newton's theoretical investigation would not have been successful. The situation is to some extent similar in economics as will be seen in the next paragraph.

2.1

Models need precise targets

At the present time only few regularities are known which could provide adequate targets for models. Consider for instance the modeling of stock or commodity prices (e.g. Caldarelli et al. 1997, Farmer 2000, Deaton et al. 1992). The objective of almost all models is to reproduce qualitatively a number of observed features such as the ''fat tail" of the price distribution (which means that the density function of successive price differences decreases as a power law) or the fact that the autocorrelation of price changes falls off very rapidly (random walk behavior). When such general conditions are met the models are found acceptable; this is clearly demonstrated by the phrasing used in most papers: "it describes markets at some level of realism", "[the model] behaves surprisingly realistic". Well, from a scientific point of view the affirmation that a model is "realistic" has no clear meaning. Firstly, as emphasized by M. Friedman (1953), many important theories are not realistic; thus because it does not take into account air resistance, Newton's gravitation law provides only an incomplete model for the fall of an apple (and an even poorer description of the fall of hazel-nut); what makes it nonetheless important is of course the fact that it explains not only the fall of apples but also the fall of the moon. Secondly, there are fairly realistic models which are wrong; for instance Ptolemy's model of the solar system was accurate enough to be used by astrologers until the mid-17th century, in spite of the fact that it was based on inadequate assumptions, moreover it did not explain the fall of apples which means that it was rather a description than a unifying theory. To sum up, before a theory can be built one needs a set of quantitative empirical regularities. Tycho Brahe, Kepler, and Galileo had to precede Newton. The realization of this simple fact was one of the main innovations of econophysics.

a

Introduction

2.2

5

In search for regularities

Almost all early papers in econophysics were based on large data sets and tried to identify empirical regularities and patterns. Let us give two examples chosen from a vast collection (i) "The objective of our study is to uncover empirical scaling regularities about the growth of firms" (Amaral et al. 1997) (ii) [In the Standard and Poor's time series we] identify precursor patterns as well as aftershock signatures and characteristic oscillations of relaxation (Sornette et al. 1996). This book, similarly, has for main objective to disclose new empirical regularities. More specifically, we are looking for what we call "broad spectrum regularities" by which we mean regularities that encompass sets of facts which are as large as possible. We will refrain from proposing a mathematical description until the phenomena under consideration are structured within a network of working regularities. To some readers such an objective may perhaps seem disappointing, but the following analogy explains why in our opinion time is not yet ripe for global theories.

2.3

A parallel with other complex systems

In the previous paragraphs we mentioned Kepler, Galileo and Newton because classical mechanics historically was the first example of a unifying theory in physics. In fact the ideas and mathematical techniques of Newtonian mechanics played a key role in the genesis of mathematical economics. Both Antoine Cournot (1801-1877) and Vilfredo Pareto (1848-1923) were educated as engineers. However, it would be misleading to pursue that parallel with classical mechanics for there is indeed a fundamental difference. In mechanics the interaction of a given system with the rest of the universe are small and can safely be neglected. Consider a pendulum for instance. There is not doubt that its movement is affected by the rotation of the earth or the attraction of the moon, but these effects are small (less than 10- 4 percent) and can therefore be neglected except perhaps in high accuracy measurements. As we have already mentioned the situation is completed different in economics: for most economic systems exogenous and endogenous interactions are of the same order of magnitude. Thus, only an analogy with a strongly interconnected system can be meaningful. C. Granger (1991) has proposed an interesting analogy with a mammalian brain; we consider here an analogy with another complex system namely the meteorological phenomena which take place in the earth's atmosphere (Roehner 1997a). In terms of connectivity, i.e. the number of neurons to which each neuron is directly connected, the world economy probably ranks between the mammalian brain and the

6

Chapter 1

earth's atmosphere. In the brain the connectivity greatly depends on which region one considers but a typical order of magnitude is 105 , a connectivity which is about a thousand times higher than that of the world economy. On the other hand meteorological phenomena are notably less complex than economic phenomena: only a dozen variables are necessary in order to describe the atmosphere at any point in space, while at least three times more are required in order to describe an economic agent at microeconomic level. In addition of being closer to economic systems in terms of connectivity, meteorological phenomena also have the advantage of being well understood. If one excepts a few exceptional phenomena like twisters or cyclones, the basic physical effects are identified and the dynamic equations (Coriolis, NavierStokes equations, etc.) describing movements in the earth's atmosphere are known. The only remaining problem is to implement these equations, that is to say to provide sufficiently detailed initial conditions, and to solve them.

2.4

What can we learn from the parallel with meteorology?

Meteorological phenomena involve a large number of physical phenomena: Archimedes law, relation between pressure and volume (Boyle's law), deviation toward the east due to the rotation of the earth (Coriolis force), phenomena due to air viscosity (Navier-Stokes equation) and so on. Fortunately all these effects could be tested and analyzed in carefully planned laboratory experiments, independently of their role in meteorology. Now, suppose for a moment that we discard what we have learned about meteorological phenomena from laboratory experiments; thus, the situation would be similar to the one with which one is confronted in economics. Under that assumption, meteorology would have to rely solely on its own data that is to say on time series for temperature, wind velocity or pressure recorded at various meteorological stations around the world since the 19th century. The scientific challenge of meteorology then would be from these series to derive the various physical laws that we mentioned above. Well, that would be an almost impossible task. Indeed in a given temperature series, even if it covers only a short time interval (say one month for instance) several different phenomena will be combined such as for instance rain, fog, cold front, warm front, etc. Trying to find a stochastic process which generates a random variable having the same statistical properties as the observed variables, an approach which is commonly used in finance, would certainly not give any real insight in the physical mechanisms that are at work in meteorological phenomena. In economics there are certainly also several specific social or psychological mechanisms at work although we do not know them yet. De-

Introduction

7

riving these laws from an analysis of standard economic time series similarly seems a very difficult if not impossible task. Is the situation hopeless then? Not at all, as we explain now.

2.5

Devising quasi-experiments

Let us for instance imagine how one could derive the Coriolis effect from meteorological observations. First, one would need observations for the speed and direction of wind in several places uniformly spaced over a sufficiently large area. Then, after all other effects (such as for instance the movement ofhighllow pressure centers) have been eliminated, one would be able to observe the way the wind veers, and derive the Coriolis law. Historically the discovery of the Coriolis effect around 1830 was indeed one of the major achievements of the so-called synoptic approach in meteorology (Fierro 1991). In the following chapters we stage several quasi-experiments of that sort in order to unravel various speculative mechanisms. In the next section we explain the general principles on which such quasi-experiments rely.

3

Basic features of our approach

On any speculative market there are several types of agents. For instance on a property market there are (i) Residents who buy and sell apartments or houses for personal use (ii) Speculators and investors who make money by buying and selling property (iii) Insurance companies who buy real estate for longterm investment. For more complex markets such as stock markets there may be more than three types of agents. These different categories of agents do not have the same behavior and the observed evolution of market prices is a weighted average of their respective trading dynamics. One of the main problems that we face is to separate these different contributions. Comparative analysis plays an instrumental role for that purpose.

3.1

Comparative analysis

Because the proportion of the different types of agents mentioned above is not the same in all speculative markets, one should be able to sort them out by way of comparison. For instance, in the postage stamp market there are no type (iii) agents and only a small proportion of type (ii) agents; thus the stamp market allows us to probe the behavior of type (i) agents. In the terminology of comparative analysis the investigation of different speCUlative

8

Chapter 1

markets is referred to as transversal analysis. Longitudinal analysis on the other hand refers to the investigation of different speculative episodes for a given market. In the following chapters we use both transversal and longitudinal analysis, at least so far as data are available. Longitudinal analysis is commonly used by economic historians; see for instance in that respect the interesting contributions of Cohen (1997), Calomiris et al. (1991), Kindleberger (1978). Cohen's work is particularly stimulating because the author emphasizes the continuity between the crashes that occurred before and after World War n, whereas economic historians often restrict their investigation to the period before 1914 (e.g. Calomiris 1991) or 1929 (e.g. Kindleberger 1978). In contrast only few studies use transversal analysis. One reason is probably the difficulty to find relevant data for nonfinancial speculative markets such as the markets for property, antiquarian books or postage stamps. As a matter of fact data collecting absorbed a substantial part of the time and energy that we devoted to the present research. It is precisely in order to facilitate subsequent studies that we included in this book (chapter 10) several of the price series that we used. For short, our approach can be summarized in the following way. The search for regularities requires that a large number of speculative episodes be studied and compared. In particular, in order to distinguish significant factors from those which are merely incidental or spurious one needs a sample of events as large as possible. This lead us to expand our collection of cases in two directions: firstly, we include in our field of investigation all sorts of speculative items for which reliable data are available; secondly, we do not hesitate to include cases which belong to previous centuries whenever good data are available. Before we leave this section we must explain our position on the question of the identification of speculative episodes.

3.2

Identification of speculative episodes

Establishing unambiguously the existence of speculative bubbles is an ongoing controversy which has focused much attention in the last decade. From a logical point of view there is no doubt that it would be desirable to find a satisfactory answer to this question before setting about studying speculative episodes. Unfortunately this is a very tricky problem. The trouble comes from the difficulty of defining a model of fundamental parameters. In order to explain that point on an example we consider the following "Gedanken experiment". Suppose the oil market is in equilibrium at time to when a limited war

Introduction

9

involving oil producing countries breaks out in some part of the world; that will push oil prices up, but after the war is over one expects prices to resume their pre-war equilibrium level. Strictly speaking, this would not be called a speculative bubble since the price peak is triggered by a well-identified shift in the fundamentals, namely a supply-demand imbalance in the countries at war. However, once oil prices begin to climb companies and traders around the world will try to make the most of this new situation. In other words, speculation will develop in parallel with the shift in the fundamentals. In order to make a clear distinction between these effects one would need (i) very detailed production-consumption data which are in practice never available especially in time of war, (ii) a reliable econometric model for the behavior of oil prices in order to derive expected price shifts from the productionconsumption data. As a matter of fact that second condition implies that the problem one tries to elucidate has already been solved. For short it is practically impossible to distinguish between the speculation induced price increase and the part which should be attributed to the shift in fundamentals. In the rest of this book we will usually not try to make that separation. Does this mean that our results should be looked at with suspicion? No, because it is almost always possible to check their consistency fortunately. In a general way if the analysis of several speCUlative episodes displays definite regularities these can be attributed to some common underlying speCUlative mechanism rather than to more or less random shifts in fundamentals. In our previous example we considered a situation in which there is a clearly identified change in fundamentals, but this is in fact an exceptional case. Usually the identification and monitoring of the fundamentals generates a great deal of controversy. Here is an example. Between 1984 and 1993 property prices in the 13th district of Paris rose substantially. Now, this coincided with the construction of a new opera house in that area. Should that element be included into the list of the fundamentals? And if we answer in the affirmative, do we have a reliable econometric model which enables us to predict the price increase to be expected from a new opera house? Clearly these are difficult and in a sense ill-defined questions. To conclude this section let us come back to the parallel with Newton's breakthrough in classical mechanics. He explained the movement of an apple, the Moon or Mars by a force which he called gravitation; but he was not able to define precisely what gravitation was, nor was he able to prove that gravitation did indeed influence these bodies. His assumption was justified a posteriori because his theory could embrace a wide sample of different effects. Similarly we will assume that there is something called "speculation" and try

10

Chapter 1

to identify some of its properties; if significant regularities emerge that will lend weight to our hypothesis. In the next section we summarize some of these regularities.

4

Presentation of the book

Although this book contains some guidelines and practical recipes for speculation it does not intend to be a guidebook for investors. Some good books already exist in which interested readers can find detailed and practical hints for making money in the stock market; one should mention in this respect the excellent books by C. Caes (1990, 1993, 1994). The main objective of the present book is rather to find some unifying regularities which could pave the way for an understanding of basic speculative mechanisms. The second chapter provides a panorama of speculative markets. Each of these markets has its peculiarities which one has to keep in mind if one wants to compare them in a meaningful way. In the third chapter we emphasize that the term "rational" which generated so much controversy in connection with speculative markets has a changing meaning in the course of time. For a stockholder the announcement by a company that there will be a fall in expected earnings will undoubtedly be received as a bad news. On the other hand the information that the price earning ratio (PER, the ratio of the stock price to annual past or expected earnings) of the same company has reached a level of 100 will be interpreted differently whether one is in an ascending or descending phase of the business cycle. In a bull market, when profits are expected to increase, a PER of 40 would be considered as acceptable and a level of 100 as exceptional whereas in a bear market one would not be surprised to see a PER as low as 10. (remember that in the long-term the average level of the PER on the NYSE was around 15) The same observation holds for other variables; a level of debt or public deficit which is considered acceptable in one decade will be found excessive in the next. In the fourth chapter we note that market experts, because of their tendency to ignore what is going on in other markets, are always tempted to "explain" a crash by endogenous factors. Such explanations are clearly inadequate when crashes occur simultaneously in different markets, an occurrence which is the rule rather than the exception. In such cases it is probably more productive to posit that a common factor is at work which we do not yet understand. In chapter 5 we show that a speculative bubble is not only an economic but also a social phenomenon. It is not only accompanied but also reinforced by a

Introduction

11

widening public interest. This in tum brings about changes in public opinion about what is to be as regarded sensible or rational. Another manifestation of increased public involvement is the explosion in trading volume which goes together with a speculative episode. Moreover, we argue that whereas there is no "natural" upper bound for stock prices, the increase in trading volume is limited by the balance between transaction costs and the rate of price increase. The next two chapters are devoted to an analysis of price peaks. In chapter 6 we study the peak amplitude while in chapter 7 we analyze the shape of price peaks. The amplitude is ruled by a remarkable regularity that we call the price mUltiplier effect. It states that peak amplitude is an increasing function of the initial price level, a rule which has obvious practical implications for investors. The price multiplier effect is widely observed in speculative markets ranging from property to diamond or postage stamp markets. An alternative formulation of the price multiplier effect is proposed which also applies to stock markets. We also show that in the case of property values a market contagion is at work by which price increase spreads from expensive areas to low-priced regions. Thus, during the 1984-1994 episode in Britain, the speculative wave unfolded from the London area up to Scotland. The shape of a price peak follows what we call the sharp peak - flat trough pattern, which means that the bull to bear market transition is usually quite sudden while the bear to bull transition is slow and progressive. That pattern can be characterized quantitatively by an exponent which is less than 1 for peaks and larger than 1 for troughs. Chapter 8 is devoted to stock market bubbles. We show that stock prices follow what can be called a resilience pattern: the higher a stock climbs during a bull market the better it resists during the subsequent bear market. Finally, chapter 9 summarizes our conclusions and chapter 10 provides a number of statistical series that were used in this book and can be useful for further investigations.

I

Chapter

2

Overall view of speculative markets

"Is the current increase in investment in rare books wholly a good thing? Personally I have reservations. They arise from stories I heard at my father's knee about the crash in the 1930s: books at that time had been driven up in price far beyond the levels that were truly supported by their significance and their rarity". These were in 1994 some of the reflections of A. Rother, past president of the International League of Antiquarian Booksellers. The crash that he refers to began in New York after the spectacular fireworks of the Jerome Kern sales in January 1929 and deepened after the stock market crash of October 1929. Values fell so heavily that in 1933 it was customary to pay fo! Kern copies no more than 10 to 15 percent of what they had fetched at these sales. The stock market crash has been studied in great detail, but to our best knowledge the crash in the antiquarian book market did not attract much attention. And yet it is in a sense even more revealing of the shift from a state of euphoria to one of dejection. As a matter of fact, if the book market fell it was not just because of the drop in revenue which resulted from the stock market crash, for even fairly cheap books in a price range around 10 dollars were no longer bought and their price plunged. As we will see in a subsequent chapter the fall affected all price segments from books costing less than 10 dollars to books costing more than 1,000 dollars. Thus, studying the book market gives an opportunity to learn how the society reacted. Curiously, this sort of information has largely been neglected by economists. From the vast literature about speculative bubbles one would draw the conclusion that speculative trading is confined to financial markets. As an illustration of that predilection one can for instance mention Cohen's book (1997) that we have already cited. Although it considers the question of speculation in a broad historical perspective which makes its reading quite captivating, non-financial markets are' almost completely ignored. Even the property B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_2, © Springer-Verlag Berlin Heidelberg 2009

13

14

Chapter 2

market which has so close connections with financial markets is only briefly mentioned (less than three pages are devoted to it in a total of about four hundreds). The present chapter provides an overview of several speculative markets. Actually, the expression "speculative market" is probably a pleonasm for any market can give rise to speculative episodes. When we use that expression it rather means that we have statistical evidence of speCUlative price peaks. At the end of the chapter we also consider stock markets but only shortly since a special chapter will be devoted to them later on. For each market we restrict ourselves to presenting a number of basic facts; special attention is given to the presentation of statistical sources in the hope of encouraging further studies of these markets. Being more descriptive than analytical, the present chapter is mainly intended for reference purposes; it provides the elements on which subsequent chapters will draw. Accordingly, it can be skipped by well-informed readers or by those who are more interested in analytical results and regularities.

1

Overview of speCUlative markets

In any markets there are at least two kinds of buyers and sellers (i) Those who buy or sell for personal use, subsequently referred to as users. (ii) Investors and speculators who make money by buying and selling with a profit. Such a distinction is particularly clear in property markets: the users are the residents who live in the houses or apartments they have bought while the investors are property developers, real estate agencies, insurance companies and so on. The proportion of investors in a given market is an important parameter which will be referred to as the speculative ratio. The speculative ratio is equal to 1 for stock markets and close to 0 for postage-stamp or coin markets which are dominated by collectors; for property markets it is of the order of 0.2. In the following paragraphs we try to adopt a uniform presentation for each market: first there is a qualitative description of how the market works; then we discuss statistical sources; thirdly we give one or two instances of speculative episodes; finally, we discuss some of the specificities of the market under consideration.

Overall view

1.1

15

What kind of price data do we need?

Physicists give great care and attention to the precision and reliability of experimental results. Let us illustrate that point by two examples. (i) If one wants to measure the lift generated by an airfoil one has to make sure that the wind tunnel does not produce too much turbulence. Failing that, observed results can change by a factor 10. (ii) During the development of the meteorological science in the 18th and 19th century one major problem was to make sure that all thermometers and barometers used in various stations were comparable and it was soon realized that the best guaranty was to impose the use of identical instruments (Fierro 1991). Economists, in contrast, devote great attention to modeling but tend to neglect the measurement process. This is unfortunate for flawed observations will spoil theoretical investigations as well. Depending on the study one is engaged in, the kind of data one needs will be different and in each case one has to make sure that the data are indeed suited to the investigation. In order for instance to study the shape of a price peak annual data are inadequate; one needs higher frequency (i.e. monthly, weekly or even daily) data. If one wants to study the spatial propagation of a price bubble one needs price data for several locations. If the objective is to fit the distribution function of price changes one needs very long records, in order to capture a number of huge but infrequent price peaks. Table 2.1 summarizes information about existent price records for each of the items considered in this chapter. The following explanations are in order. (i) The information given in columns 2-5 is based on sources which will be detailed when we study each market separately (ii) The last column refers to the fact that some items cannot be divided; this is obvious for stamps; it is also true for apartments despite the fact that one can define a price per square meter; the latter indeed will be different for a 5-room apartment or a I-room apartment. In contrast, if one neglects transaction costs the price of 10,000 shares is just ten times the price of 1,000 shares. For spot markets the discount which is usually granted for large quantities can also be attributed to transaction costs. (iii) On line availability of data files will certainly improve in coming years. Incidentally, the fact that early studies in econophysics concentrated on stock markets was certainly in part due to the fact that extensive stock price records were available on line.

Chapter 2

16

Table 2.1 Classification of price records

Item

Price estimate

Price frequency

Prices at several locations

Longest price records

(E)

Price range of items

or real price (R)

month- highest ly frequenor cy data better

over 10

max. number of locations per country

R

Yes

/

cheapest [years]

[days] Grains

most expensive

7

Yes

500

420

No

Commodities Spot

R

Yes

1

Yes

10

200

Futures

R

Yes

<1

No

3

50

No No

Stamps

E

No

365

No

1

130

10,000

Books

R

Yes

15

No

3

200

10,000

Land

R,E

No

365

Yes

20

50

10

Property

R,E

Yes

100

Yes

20

30

20

R

Yes

<1

Yes

50

130

No

Stocks

Notes: The table refers to published price records; more detailed records may possibly be available in archives; such is for instance the case for grain prices: before the 20th century grain markets were held two or three times a week and such primary data can be found in archive records. The table is restricted to those items considered in this book which explains the omission of items such as coins, paintings, collectible cars or phone cards. "Commodities" here refers to other commodities than grains.

Overall view

1.2

17

Grain markets

In Western Europe until the 20th century grains were an essential component in the diet of a majority of the population. In terms of revenue that it generated the production of wheat was the most important single item in national income. At first sight it could seem surprising that such an important commodity was subject to heavy speCUlative trading. Among the factors which favored speCUlation the following probably played a key role: (i) During times of scarcity prices could be multiplied by three or four. For all people whose revenue was high enough a natural reaction was to buy grain reserves before the price was too high; this is well documented by contemporary reports, see for instance Biollay (1885), Modeste (1862), or Martin (1908); for instance a survey conducted by the French government described how people bought up the wheat market at the first signs of a coming wave of high prices and stored large amounts of wheat for future use (Biollay 1885). (ii) For those people who by their function had to buy and collect large quantities of wheat, for instance traders in charge of provisioning the army, it was very tempting to speculate either for the benefit of their administration or their personal profit. Such practices are documented in Bord (1887) and Martin (1908). Moreover the large number of laws and regulations aimed at prohibiting specific forms of speculation attest that such business practices were fairly common, see in this respect Abeille (1764) or Herbert (1755); moreover several authors (e.g. Martin 1908) point out that the existing regulation was not always enforced.

1.2.1

Organization of the market

Lawfully, grains had to be sold only on the market-sales that took place once or twice a week in most cities and towns; in such a way everybody was given the same chance to buy which is the wayan efficient market is supposed to operate; present-day regulation about inside information has the same objective. Nevertheless large quantities of wheat were often sneaked from the producers directly to traders thus bypassing the market-sales. While small traders, the so-called "blatiers", fulfilled a useful role in the distribution of grains, in times of dearth numerous self-proclaimed "blatiers" attracted by the expectation of huge profits entered the trade, a circumstance which amplified the speCUlative frenzy. In the era before 1850 grain prices very much played the same role as stock prices nowadays in shaping the spirits of a population, a parallel which is summarized in Table 2.2a. At first sight it is not completely obvious that high wheat prices should have

18

Chapter 2 Table 2.2a Parallel between wheat and stock markets Yesterday

Nowadays

Wheat prices

Stock prices

Low and declining

High and increasing

.IJ.

.IJ.

High spirits High birth/marriage rate

High spirits High birth/marriage rate

High and increasing

Low and declining

.IJ.

.IJ.

Low spirits Low birth/marriage rate

Low spirits Low birth/marriage rate

Notes: The correlation between population fluctuations (i.e. birth/marriage rates) and wheat or stock prices was pointed out in Roehner (1995, chapter 5).

been associated with a contracting economy for wheat producers and traders may have profited from the price increase, especially if one accepts the standard argument about the so-called King's law l . However, it has been shown that high wheat prices lead to a fall in marriage and birth rates (Roehner 1990, 1995); on the contrary marriages and births increased in times of low prices which reveals a more euphoric social climate. A similar phenomenon occurred in the early 1930s: the depression which followed the stock market crash lead to a sharp decline in the American birth rate; the latter dropped from an average 2.45 percent in the decade 1921-1930 to an average 1.90 percent in the decade 1931-1940. Moreover, it will be shown in a latter chapter that in the late 1990s consumer confidence was strongly correlated with stock price levels.

1.2.2

Price peaks

Fig.2.1a,b show two price peaks which can be considered as fairly typical. One is for Paris in the late 17th century, the other is for Munich in the early 19th century. These peaks share several common features: (i) The total duration of the peak is about 5 years. (ii) The amplitude of the peak that is to say the ratio P2/Pl 1 It states that the price increase is proportionaly larger than the fall in production which means that the revenue received by producers was increased in times of scarcity.

Overall view .~

..

19

50

.S

l

40

i

~

30

20

10 9

~~~~~~~~LLLLLLLLLL~

1691

1692

1693

1694

1695

1696

Fig.2.1 a Monthly wheat price in Paris (1691-1694). Vertical scale (logarithmic): Iivres tournois per setier (1 setier=1.56 hectoliter). Grain markets were held twice a week; the highest figure was reached on 19 May 1694 with a price of 52livres/setier. Source: Baulant and M euvret ( 1960).

8000 7000 6000 5000 4000 3000

2000

1000

L.........I-.........L.........._...L...-'-..-..L_...............I---'-..........

1815

1816

Expl"io. ., TamH,..

1817

1818

1819

1820

Cold ...... r

Fig.2.1 b Wheat price in Munich (1815-1820). Vertical scale (logarithmic): 111 00 gulden per Schaffel (1 Schaffel=2.2 hectoliter). If one excludes the potato famine in Ireland in the 1840s this was the last large-scale dearth in Western Europe. It may be of anecdotal interest to note that the peak was probably triggered by the bad harvest of the year 1816 referred to as the "year without summer"; that meteorological anomaly may have been caused by the explosion of Mount Tambora (5 April 1815), a volcano located in Indonesia which ejected some 40 cubic kilometers of debris in the high atmosphere. Sources: Seuffert (1857), Stommel et at. (1979).

Chapter 2

20

of the peak price to the initial price is about 5. (iii) Although the two curves differ in their details, their overall shape is similar: both the upgoing and the downgoing paths have the shape of a function of the form y = xO! (0: > 1). Since the vertical scale is logarithmic this means that the growth is faster than an exponential. This property is common to most wheat price peaks. It is precisely this similarity which leads to the conclusion that there must be a common mechanism. At the start of the peak there is usually an exogenous factor, but once started the bubble seems to have its own dynamics.

1.2.3

Sources

Thanks to the efforts of several generations of economic historians many long price records have been published in book form. As a result it is easier to find data about wheat prices in the 17th century than about property prices in the 19th (or even 20th) century. Some major sources are listed in Table 2.2b.

1.3

Commodities

The price behavior of commodities is a vast subject. In this paragraph we restrict ourselves to a few basic facts.

1.3.1

Organization of the markets

The price quotations given in financial newspapers are restricted to major exchanges; the Financial Times will report prices from the Metal Exchange and Baltic Exchange both located in London, the Wall Street Journal will give the prices at the New York Mercantile Exchange and the Chicago Board of Trade. This could lead the reader to the conclusion that for each commodity there are only two or three meaningful marketplaces. This would be an oversimplified view however; let us briefly explain why. On commodity exchanges most transactions concern futures. As one knows, a forward contract gives the buyer the right to take delivery of a given, standardized quantity of a given commodity in 3, 6 or 9 months. Actually only a very small proportion (less than five percent) of the transactions on futures markets eventually result in the delivery of physical quantities. Because contracts can be transfered from one market to another (at least if similar contracts exist on both markets) one would expect arbitrage between futures markets to be similar to arbitrage between stock markets, that is to say almost instantaneous and performed with negligible transaction costs. This is indeed confirmed by empirical observation (see Roehner 2000a, p.180). On

Overall view

21

Table 2.2b A selection of wheat price sources Characteristics

References

Long records (over 100 years)

Ebeling et al. (256 years), Baulant et al. (178 years), Drame et al. (428 years)

High frequency (better than weekly)

American cash grain prices (1 day), Baulant et al. (2 days), Seuffert (7 days)

Many locations (intermarket distance < 100 km)

Drame et al. (53 markets), Labrousse et al. (85 locations), Seuffert (12 markets)

High frequency (better than monthly) and several locations

Drame et al. (14 days, 53 markets), American cash grain markets (1 day, 8 markets)

Notes: As in meteorology a major objective is to have good coverage both in time (high frequency data) and in space (many locations). This table reports only market prices. Some series such as the very long Exeter price record (13161820) whose prices come from accounting books of hospitals or monasteries are therefore not included; such series usually have a smaller price volatility than market prices. The daily American cash grain prices though not published in book form can be obtained from the United States Department of Agriculture. Other references for historical price series can be found in Drame et al. (1991) and Grenier (1985).

spot markets, on the contrary, commodities are traded for immediate delivery; thus spatial arbitrage on spot markets requires actual transportation of physical quantities from one point to another. As a result there are usually substantial differences, of the order of 15 percent, between prices of a given commodity on geographically separated spot markets. This is true for international markets (see for instance the prices listed in the UNCTAD Monthly Commodity Price Bulletin) as well as for markets located in the same country, e.g. Middle West versus West Coast wheat markets (Roehner 2000a, p.I77).

1.3.2

Price peaks

Fig.2.2 a,b,c,d show four cases characterized by increasing volatility (i.e. standard deviation of prices). In order to facilitate the comparison the graphs were drawn with the same vertical scale. Note that relative levels of volatility

Chapter 2

22

would be the same in the longer period 1900-1960 (see Roehner 1995, p.1214). Explaining why these markets have these volatilities is probably a difficult problem; for a theoretical attempt based on storage costs see for instance Deaton et al. (1992.). One major difficulty is the fact that the volatility depends not only upon the organization of the market but also upon the intensity of exogenous shocks. Fig.2.3 lists some possible causes of price fluctuations. One simple observation can be made: since the volatility of banana prices is almost zero once seasonal fluctuations are discounted this means that both A and B factors are absent. The second conclusion is of particular interest; as a matter of fact the three specUlative strategies listed in Fig.2.3 (endogenous dynamics) suppose that the commodity under consideration can be stored for a sufficiently long time and at fairly low cost. If, as is the case for bananas, no storage is possible then the B factors disappear as a source of volatility. This analysis is confirmed by the fact that most speculative bubbles that we discuss in subsequent chapters concern goods which can be easily stored: gold, diamonds, stamps, books, and to some extent land and property are some examples. Moreover for all these markets the inventories held by the public exceeds the volume of annual transactions. In short, one is lead to posit that the existence of a floating stock-pile conditions speculation. If that assumption is correct one would for instance not expect a marked speculative dynamics for electricity prices. It will be possible to test that prediction in a few years when the deregulated electricity market which is progressively taking shape in Western Europe will be functioning.

1.3.3

Sources

The two following sources are important because they give monthly prices for all major commodities (i) The Monthly Commodity Price Bulletin published by UNCTAD (United Nations Conference on Trade and Development). (ii) The International Financial Statistics published by the International Monetary Fund. For data concerning a specific commodity the relevant sources can be found in commodity guides (e.g. Radetzki 1990 or Cyclope) or on the computer Web.

1.4

Land and property

Land and property assets represent considerable amounts. The extreme case of 1990 Japan is well known. At that time, Japan's total stock of property was valued at about three times the capitalization of the Tokyo stock market or five

23

Overall view

10 3 f-

1960

I 1964

I 1968

I

I

L

1972

1976

1980

l 1984

1988

Fig.2.2 a Price of bananas. The data are monthly prices for Central America bananas expressed in US cents per pound and deflated by the US consumer price index. Apart from seasonal fluctuations the price remains almost constant. The figures 2.2a,b,c,d are drawn with the same vertical scale in order to facilitate a comparison of price variability. The fact that there are no price peaks for bananas can probably be related to the distinctive fact that they cannot be stored for more than a few weeks. Their production and distribution on the American continent is in the hands of a small number of firms: United Brands, Castle and Cooke, Del Monte. Sources: UNCTAD Monthly Commodity Price Bulletin 1960-1984,1970-1989.

1964

1968

1972

1976

Fig.2.2 b Price of wheat. The data are monthly prices for American wheat (Hard Red Winter) in Gulf of Mexico ports expressed in US dollars per ton and deflated by the US consumer price index. Sources: same as Fig.2.2a.

24

Chapter 2

1975

1980

1985

Fig.2.2 c Price of Douglas-fir sawtimber. The data are annual prices on national forests in Oregon and Washington states; they are expressed in dollar per thousand board feet and are deflated by the US consumer price index. Between 1970 and 1980, according to the industry, there was a bubble which resulted in a multiplication of the real price by a factor 4. This is all the more surprising because the market is largely under the control of the Federal government through the Bureau of Land Management and the U.S. Forest Service. This price peak parallels similar peaks observed at about the same time for diamond, gold, silver, platinum. Source: Mattey (1990).

60

50

........

-

40

-

30

/ \.

.......

... .............. 1960

1964

1968

1972

20

1976

1980

10 1984

1988

Fig.2.2 d Price of sugar. The data are monthly prices observed in Caribbean ports expressed in US cents per pound and deflated by the US consumer price index. The course of sugar prices displays a number of huge peaks of an amplitude comprised between 5 and 10 and a duration comprised between 3 and 6 years. The broken line shows the evolution of world closing stocks. It can be noted that the 1975 peak does not coincide with a sharp drop in closing stock levels. Sources: UNCTAD Monthly Commodity Price Bulletin 1960-1984,1970-1989; McGraw Hill handbook of commodities trend futures (1985).

Overall view

25

"A-FACTOR": EXOGENOUS SHOCKS Meteorological shocks Exogenous increase in production costs Exogenous increase in transportation costs Exogenous Ouctuation in demand (business cycle)

"B-FACTOR" : ENDOGENOUS DYNAMICS Staging a corner (i.e. "buying up the market") Hoarding in wait for an expected price rise Buying spree (e.g. diamonds in 1978-1980)

Fig.2.3 Some factors which may trigger large fiuctuations in the price of a commodity. Exogenous means here that the shock originated outside the market of the commodity under consideration; for instance a recession will reduce the level of demand for a large variety of other goods. On the contrary, the endogenous shocks originate within the market itself either at producer, trader or consumer level.

times the size of Japan's gross national product (Wood 1992). Land speculation also played an important role in the expansion of the United States toward the West (see in this respect Swierenga 1968). Unfortunately economists and statisticians have given little attention to land and property markets which explains the scarcity of statistical information on that matter. 1.4.1

Examples

Fig.2.4a gives an example ofland speculation in the area of Port Said (Egypt). That city is located at the entrance of the Suez canal which was inaugurated in 1869. The Port Said land speculation is typical of a mechanism recurrently observed in the 20th century; for instance in the Tokyo area in the 1980s, or in Hong Kong and Singapore in the 1990s. In all these cases land speculation was accompanied by a rise in stock prices, and both markets crashed simultaneously. Fig.2.4b illustrates the property speculation in Singapore and Hong Kong. In

26

Chapter 2

i

~

l i

~

..~

]

70 60

so 40 30

~

l

20

10 1908

1910

1912

Fig.2.4 a Price of land in Egypt. Vertical scale: average price per square meter expressed in French francs; the data refer to sales made by the company of the Suez canal in the vicinity of Port-Said. The abrupt decline that occurred after 1907 accompanied a stock market crash in Alexandria: between April 1907 and February 1913 stock prices of real estate companies dropped by about 50 percent. In a sense this episode paralleled, though on a much smaller scale, the 1990 stock and property crashes in Japan. Source: Bourgeois ( 1913).

Singapore the government tried to curb real estate speculation by restricting property loans; subsequently, in mid-1997, both in Hong Kong and Singapore the market fell sharply in the wake of the stock market crash.

1.4.2

Sources

Price series are scattered in various publications issued by real estate companies and government agencies. Let us mention some of them. (i) In the UK the Halifax house price series exists since 1983 and trading volumes are available since 1989. Figures for recent years are accessible on line at the following addresses: http://www.houseweb.co.uk and http://www.halifaxp1c.com (ii) For Paris, starting in 1985 price and volume series for old apartments (i.e. apartments which have already been sold at least once) are published by the Chambre des Notaires (lawyers syndicate). Furthermore, data, graphs and comments can be found in the journal "Conseil par des notaires" (Advice from lawyers) (iii) In Hong Kong property price data are published by the Rating and Valuation Department of the Hong Kong government. The series for the past five

Overall view

~

...

"i!

27

400 ,----------------------------------------,

~ ~ 300

Hong Kong Singapore

1 i

.1/.......

~

~200

...

~

~

f/-

. .... ....•.

."

.:................... 100

Fig.2.4 b Average price of residential property in Hong Kong and Singapore. The prices are expressed in (non-deflated) local currency. The index 100 corresponds to a price of 2730 US dollar per square meter for Hong Kong and 2290 US dollar per square meter for Singapore. In May 1996 the government of Singapore took severe measures to limit speculation which is probably why the Singapore market began to plummet before the Hong Kong market. The latter dropped in the wake of the 1997 stock market crash. It should be noted that the overall shape of the curves in Fig.2.4a and b are similar, both are characterized by a downward concavity and a round top. Such a pattern strikingly contrasts with the one for commodities shown in Fig.2.1a,b. Sources: Hong Kong: Financial Times (27 March 1997), Guardian (16 Jan. 1998), Agence France Presse (20 Oct. 1999), on-line site ofthe Rating and Valuation Department of the Hong Kong government. Singapore: Agence France Presse (various dates).

years are available on line at http://www.infogov.hk.rvd (iv) For Japan land prices by prefecture are published in the "Japan statistical yearbook". At this point a word of caution is in order regarding the comparability of real estate series. First, one must distinguish between real price averages and estimates made by experts; the first three series belong to the first category while the fourth belongs to the second. As a rule estimates made by experts tend to lag behind real prices and tend to underestimate the amplitude of price peaks. Furthermore price averages can be performed in a number of different ways. Thus, the index issued by the Hong Kong Rating and Valuation Department is a weighted average based on various size and luxury factors, the details of which are unfortunately not given.

1.5

Postage stamps

Postage stamps are an item which in many ways is of great interest for the study of speculative bubbles.

1.5.1

A useful ''laboratory'' for studying speculative trading

The main advantage of the stamp market is that it is "simpler" than other speculative markets. Let us briefly explain why. (i) The stamp market is relatively isolated from other speculative markets because the proportion of collectors is by far larger than the proportion of investors. Moreover, if one excepts a small number of particularly rare stamps, most of them are fairly inexpensive compared with an apartment or a futures contract and can therefore be bought even in times of recession. (ii) For stamps ''production'' and "consumption" take on particularly simple forms. Production is restricted to a short time span when the stamps are on sale in post offices; moreover production figures, that is to say the total number of stamps sold, are well documented in stamps catalogs. Since most collector's stamps are not used on letters, consumption occurs only by wear and tear or by loss (for instance after the death of a collector). It is therefore not unrealistic to assume that consumption takes place at a slow, and fairly constant rate. (iii) A few decades after they have been issued, stamps can no longer be used on letters. Moreover in contrast to gold or silver coins, stamps cannot be melted. Thus, having no "intrinsic" value, their price is solely determined by the judgment of collectors. (iv) In contrast to other collectibles such a paintings or furniture, stamps can easily be stored and are fairly liquid assets. Any valuable stamp can be sold to a trader at a price indexed on the catalog price, the difference being determined by the state of conservation of the stamp. (v) The stamp market displays huge price bubbles. Multiplication of the current price by a factor of about ten within one decade is not uncommon. (vi) Stamp prices range from a fraction of a dollar to several thousand dollars for very rare stamps. This gives the unique opportunity to observe the speCUlative behavior of collectors when the amount at stake varies over a broad interval. In short, the stamp market constitutes an excellent laboratory for studying speCUlative trading. To be fair, one must recognize that there are also a few drawbacks especially regarding statistical information. For one thing, stamp catalogs are published only every year, and in some cases for instance for foreign stamps only every two or three years. This precludes any investigation of short-term price fluctuations. Secondly, prices given in the catalogs are not "real" prices but price estimates made by experts. Thirdly, since the bulk of

Overall view

29

the transactions take place between private individuals and are therefore not recorded, it is impossible to estimate the trading volume.

1.5.2

Price peaks

Fig.2.5a,b give examples of two huge price peaks. It is worth pointing out that the "Van Gogh" and "Diane au Bain" stamps had the same face value (2 French francs), were put on sale the same day (27 Oct. 1979), with same number of copies (6 millions). Yet, only the first became a speculative target. This provides a clear example of the fact that often there are few objective causes at the origin of a speculative process. One may in fact wonder whether such peaks do not reflect deliberate attempts to comer the market, i.e. to take control of the market by buying up all available stamps. Let us examine that point further. When they were issued the 6 million Van Gogh stamps were worth 2x6 = 12 million French francs, an amount which represented only 4 percent of the annual turnover of the five major French trading houses. In other words, a trader or a rich collector could possibly take control of a sizeable portion of the total stock. Which proportion of the stock (10, 40 or 75 percent?) has in fact to be controlled in order to push the price to such high levels is a question of some interest; let us for the moment assume that the proportion is 75 percent. At the highest point of the peak the Van Gogh stamps were worth 60 francs and 75 percent of the total stock represented about 60 percent of the trade over of the major trading houses. One should keep in mind that there are at least 1,000 commonly traded French stamps and it would be unrealistic to think that the traders would commit 60 percent of their purchasing power to just one stamp. This discussion shows that if this was indeed a comer then it must be possible to take control of the market by buying a proportion of the total stock much smaller than the level of 75 percent that we assumed.

1.5.3

Sources

The basic sources for stamp prices are the stamp catalogs. In most western countries stamp catalogs have been published annually for more than a century. Thus, for 19th century stamps long price records are available. However, as we already mentioned, there are no high frequency data. Another practical difficulty is the fact that stamps catalogs are usually not to be found in university libraries. One has to resort to specialized libraries such as for example in France the Musee de la Poste (Mail museum, 34 Boulevard Vaugirard, Paris). Table 2.3 summarizes some references of stamp catalogs.

30

Chapter 2

.l

/'\\... Pierre Col .•....

'.'.

138 80 19~~~-L~=-LL~~~-L~~LL~~~-L~~~J-~~2~~

Fig.2.S a Speculative peak for two French stamps. The "Van Gogh" and the "Diane au bain" stamps were issued the same day and in very similar conditions (same face value. same number of copies); yet. in contrast to the first the second did not become a target for speculation. Source: Ceres catalogs 1980-1998.

800 700 .: 600 ~ 500

1.

~

l<;

"&400 .~

.... 300

I

200

100 90

80

1955

Fig.2.S b Speculative peak for French stamps during World War II. During the war consumption was drastically curtailed and there was a substantial rate of inflation. it is therefore not surprising that available revenue was spend on such items as stamps or antiquarian books (see Fig.2.6). For the purpose of comparison the horizontal scale is similar in Fig.2.5 a and b. Source: Massacrier (1978).

31

Overall view

Table 2.3 Stamp price sources Sources

Countries covered

American catalog: Scott (Sidney, Ohio, since 1863)

US, Canada, UK, Commonwealth

British catalog: Stanley Gibbon (since 1898)

UK, Commonwealth

French sources: (i) Ceres catalog (since 1942) (ii) Yvert et Tellier catalog (since 1898) (iii) Massacrier (1978)

France World 19th century French stamps

Notes: Massacrier gives price series for all 19th century French stamps from 1904 to 1975.

1.6

Antiquarian books

This paragraph is devoted to the market of antiquarian books; these are also sometimes referred to as rare books but we will avoid that expression because it can be misleading by suggesting that all these books are indeed rare and hence expensive, which is not the case. As a matter of fact, the price range of antiquarian books extends from a few euros to several thousand euros. How is this market organized and why it is of interest for our purpose?

1.6.1

Organization of the market

A distinctive feature of the book market is that auction sales represent a large proportion of total sales. For instance in France in the 1990s it is estimated that auction sales represented one half of the sales, the other half being sales in antiquarian bookstores; there were about 100 auction sales a year, each with about 300 lots of books sold. Because auction sales are a key element of the market, sale prices were recorded and published yearly in book form. That usage has been followed for over a century in countries which had a major book market, e.g. Britain, France, Germany or the United States. While for stamps we had to satisfy ourselves with estimates, for books on the contrary one has access to actual sale prices.

32

Chapter 2

Unfortunately, there is a feature of the book market which raises a serious difficulty. We already mentioned that for stamps the state of conservation is an important element in the determination of the price. For books, that question is even more crucial to the point that each book is almost unique. Let us explain why. As for stamps state of conservation and quality of the binding are of course important factors, but this is only part of the problem. A given title is usually available in different editions and within each edition in successive printings; the number of copies printed in each printing is a key element of the book's rarity and therefore of its price. As an example, a 1831 edition (Gosselin, Paris) of "Notre Dame de Paris" by the French writer Victor Hugo is available in four successive printings; for some reason the name of the author and of the publisher was omitted on the copies of the first printing and as a result these copies are priced three times as much as the three other printings (Clouzot 1996). Similarly a dedication on the front page of a book can inflate its price substantially. For all these reasons one can never be sure that two books are identical. What complicates the matter even more is the fact that the short descriptions of the books which are given in auction catalogs are written by experts who may change in the course of time; thus the very same book appearing in two separate sales can well be described in different terms.

1.6.2

Price peaks

Fig.2.6 gives an example of a price peak for books which parallels the peak of Fig.2.5 observed in the same period for stamps. Unfortunately, since no sale records were published in the years between 1934 and 1942 it is somewhat difficult to identify the beginning of the peak. Other examples of price peaks will be given in subsequent chapters.

1.6.3

Sources

Table 2.4 summarizes the sources in different countries. In recent years all catalogs have become international in the sense that for instance an American catalog will also report sales in Geneva, London or Paris.

1.7

Diamonds

For an economist the diamond market constitutes a very interesting case for it is an example of an almost perfect monopoly; while oligopolies are not uncommon real monopolies are rare. De Beers's, the South African mining

Overall view

33

11

1000 900

~800

~ 700 ~

j600 ~500

==

1400 300

1930

1935

1940

Fig.2.6 Speculation peak for books. Vertical scale: average price of books by Victor Hugo sold at auction sales and deflated by the consumer price index. The factors which lead to the stamp bubble, and particularly the high inflation rate probably also account for the book bubble. Sources: Sales catalogs: Delteil, Brecourt, Grolier (various years).

group, has been in control of the diamond market for over a century; in 1999 it controlled two-thirds of world gem diamond sales.

1.7.1

Organization of the market

It may be tempting to explain the absence (or the existence) of large specu-

lative peaks in a given market by the fact that it is dominated by one company. The case of diamonds clearly shows that such explanations cannot be sufficient. Indeed, although in "normal" times the diamond market is fairly stable, it was marked by two huge peaks which culminated in 1929 and 1980 respectively. The 1980 peak is shown in Fig.2.7. Hence De Beer's claim that through its control of the market it is able to stabilize prices is not entirely true. De Beer's monopoly has other consequences as well: (i) Diamond prices are not made public, a rule which, although seldom stated explicitly, is accepted by most people in the industry. (ii) DeBeer's was indicted in 1994 by the U.S. Justice Department on charge of fixing industrial diamond prices and this was but a new episode in the four-decade long attempt of the American government to break up De Beer's monopoly. De Beer's answer was to withdraw from the U.S. market but the group continues to sell diamonds to American

Chapter 2

34

Table 2.4 Book price sources Country

Reference

Britain

Book-prices current. A record of prices at which books have been sold at auction. Sergeants Press. London.

France

(i) Delteil (L.) 1920-1930: Annuaire des ventes de livres. Delteil. Paris. (ii) Brecourt 1934: Liste alphabetique et prix des livres adjuges a I'hOtel des ventes de Paris d'octobre 1933 ajuin 1934. Brecourt. Paris. (iii) Grolier (E. de) 1945: Le guide du bibliophile et du libraire. Rombaldi. Paris. (iv) L'argus du livre ancien et moderne. Repertoire bibliographique. Editions Promodis.

Germany

lahrbuch der Auktionpreise filr Bucher. Ernst Hauswedell. Stuttgart.

United States

American book-prices current (since 1894): E.P. Dutton. New York; and other publishers subsequently.

dealers through its London trading house (Epstein 1982, Financial Times 31 Jan. 2000).

1.7.2

Diamond prices

From the few price data available it is possible to derive a useful relationship between the weight of a diamond expressed in carats (1 carat=O.2 gram) and its price, namely: price = K mass a

a::=2

where K is a coefficient which depends on the currency unit used. That rule can be derived by regression from the data in table 2.5. The interesting point is the fact that it fits data from the 18th as well as 20th century. It seems therefore to be a very robust regularity in the diamond industry. Since diamond prices are so difficult to obtain the above formula provides a means for deriving prices for diamonds of different sizes from a single figure.

Overall view

-l :

35

D color, one cant -

I:

G color, one carat - -

211

1974

1976

1978

1980

191%

1984

1986

1-

Fig.2.7 Speculation peaks for diamonds in the late 1970s. Vertical scale: thousand of 1980 dollars. In the diamond color classification D denotes the highest quality (white); from D the classification extends down to Z. One should bear in mind that the diamond bubble was not an isolated phenomenon; the factors which affected the price of diamonds also affected the price of cobalt, gold, palladium, platinum and silver. Source: Diamond 1988 (The Economist, Special Report No 1126).

1.8

Other speculative markets

There are many other speculative markets for instance for coins, wines, paintings or furniture. We restrict ourselves to giving some basic references (Table 2.6) where the interested reader will find more details about the organization of the market and price sources.

1.9

Stock markets

Stock markets have come to play an ever increasing role in modem economies. Because they constitute a fascinating example of speculative market they have been studied with much attention by financial analysts and more recently by econophysicists. In this respect one should mention the following important contributions: Bak (1996), Bak et al. (1997), Bouchaud et al. (1997,2000), Caldarelli et a1. (1997), Farmer (2000), Levy et al. (2000), Lux (1995), Lux et a1. (1999), Mandelbrot (1997), Mantegna (1991, 1999), Mantegna et a1. (1995, 1999), Marsili et al. (2000), Shatner et al. (2000), Sornette et a1.(1996), Stauffer et a1. (1999), Zhang (1998). Stock markets however constitute systems of great complexity and a real understanding has not

Chapter 2

36

Table 2.5 Relationship between the weight of a diamond and its price

1720

1997

Weight (m)

Price

Weight (m)

Price (P)

Weight (m)

Price

(P)

[grain]

[livres]

[grain]

[livres]

[carat]

[dollar]

1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 7 8 9 10

13.5 24.5 38 61 68 100 109 150 205 225 315 420 580 800 1,000

11 12 l3 14 15 16 17 18 19 24 30 35 40 45 50 60

1,300 1,550 1,900 2,250 2,750 3,300 3,600 4,000 4,750 6,000 10,250 17,500 22,500 35,000 55,000 67,000

0.5 1 2 3 5

4,000 16,000 52,000 120,000 275,000

(P)

Notes: The 1720 data are for Holland-style cut diamonds; the 1997 prices are for D-IF (D=white, IF=Internally Flawless) diamonds which correspond to the highest quality. These figures are based on expert estimates rather than on actual sale prices. A regression performed on the logarithms leads to a relationship of the form: p = Km U with: <Xl=2.17 ± 0.07 (r=0.99) for the 1720 data, and <X2=1.84 ± 0.07 (r=0.99) for the 1997 data. In other words: p=Km 2 is a reasonable approximation; it means that the price per carat is in proportion to the size of the diamond. A grain corresponds to 0.064 gramme and a carat to 0.2 gramme, which implies that 1 carat=3.13 grains. The largest known diamonds weight about 100 carats. As an illustration the so-called "Grand Sancy" diamond (55 carats) was sold in 1600 for 210,000 livres and in 1604 for 180,000 livres which corresponds to 1220 livres/grain and 1046 livre/grain respectively (Quid p.546), two figures which are consistent with the previous· relationship which gives a price of 1116 livres/grain. Sources: 1720 data: Savary des Bruslons (1723); 1997 data: Rapaport Diamond Report (statistical section) 10 Ja. 1997.

Overall view

37

Table 2.6 Sources for coin, painting and wine prices Item

References

Coins

World Coins (periodical) Montroll et al. (1974, p.200-20S)

Paintings

Buelens et al. (1993), Pommerehne et al. (1989)

Wine

Di Vittorio et al. (1996)

Note: Here is an example of a speculative episode for the coin market. In the late 1970s and early 1980s investors flooded into the coin market pushing coin prices to historic highs and forcing out traditional collectors who felt coins had become overvalued. As soon as the newcomers discovered how difficult it was to sell their coins they fled pulling down prices by around 30 percent (Herald Tribune 28 May 1995).

yet been achieved. One of the major themes of this book is precisely that by studying other, in many respect simpler, speculative markets we are in a better position to understand the basic mechanisms of speculative trading. This does not mean however that stock markets are excluded from our study, rather they are considered in the broader context of speculative behavior. In this paragraph we present a brief historical survey and we explain the organization of the market. For stock markets there are plentiful sources of statistical data, therefore we did not devote a special paragraph to this question; Table 2.7 only lists a number of historical sources and periodicals.

1.9.1

Role of stock markets: an efficient capital allocation machine

The purpose and usefulness of the markets considered in the previous paragraphs were fairly obvious. For instance the stamp market enables collectors to improve and update their collection. What purpose is served by holding stamps is of course another question that we do not consider here. The economic role of a stock market is perhaps less obvious. Standard textbooks tell us that one of its main purpose is to enable companies to raise money. Let us consider that question more closely from a system theory point of view. There are basically four ways for a company to raise money. (i) To borrow

Chapter 2

38

Table 2.7 Sources for stock market data Series beginning before World War II The Commercial and Financial Chronicle (periodical, New York) The Dow Jones averages 1885-1970, M.L. Farrel ed. Dow Jones Books (1972) Common Stock price histories 1910-1987, D.E. Prendeville ed.(1987) and logarithmic supplement (1988). WIT Financial Publishers. Series beginning after World War II Vigreux (P.), Pistre (M.) 1967: Dynamique du marche financier. Editions d'Organisation Paris World Stock Exchange Fact Book (periodical starting in 1995) New York Stock Exchange Fact Book (periodical) London Stock Exchange Fact Book (periodical) Toronto Stock Exchange: annual report and fact book (periodical) Emerging Stock Markets Factbook (periodical starting in 1997) Notes: After 1995 a large number of on line sites came into existence which provide stock market prices; moreover extensive data sets can be purchased in CD-ROM form from various companies.

from a bank (ii) To sell stocks (iii) To sell corporate bonds (iv) To use its own cash-flow. The two last methods can hardly be used by those small new companies which are so important in high-tech industries. Before a bank commits itself it usually asks experts to make a precise technical and financial assessment of the potentialities of the new field. Not only must these experts be renumerated but they may also disagree one with another. Selling stocks provides an easy solution to this thorny assessment problem. When the company is introduced into the stock market it will be scrutinized by all potential buyers;

Overall view

39

probably they will not all agree but that does not matter; those who think that the field has high expectations will buy the stock while others will simply keep away. It is through their collective judgment that the price of the stock will be set. If the project is approved by a majority the price of the stock will go up, otherwise it will fall. In short, the stock market provides the experts, renumerates them, enables them to "vote" and sends the resulting message back to the company in the form of a given stock price level. In other words we have here a machine which collectively is capable of performing a very difficult task. Seen in that way, the buyers and the sellers collectively work in the same way as a neuron network. The shift from bank loans to capital raised by selling stocks is in essence a shift from a visible hand to Adam Smith's invisible and collective hand. It brings firms in direct contact with the public as shematically represented in Fig.2.8. In a sense, it is direct economic democracy at investment level. In 1997 some 470 new companies were introduced in American stock markets and these Initial Public Offerings (IPO) raised 24 billion dollars (The Economist 15 Nov. 1997). If our argument is correct and if this is indeed a more efficient way to raise money, it means that in the future many more companies will could "go public" than is currently the case.

1.9.2

Two safety nets: investment funds and derivatives

Every rose has its thorn. When a mania floods the market careful judgment can be replaced by credulity, over-confidence, or panic. In order to protect stock holders from such pitfalls, stock exchanges have invented several safety nets. The introduction and rapid development of investment funds in the 1920s and of derivatives (options, swaps and so on) in the 1980s and 1990s were two major steps in that direction. By purchasing shares in investment trusts the average investor gets a well managed diversification of stock holdings; and by selecting a given investment trust rather than another he is still able to "vote", although this is a kind of representative democracy rather than the direct democracy we mentioned above. The rapid development of investment funds between 1925 and 1929 (their total asset was mUltiplied by 10) is proof of the fact that they responded to a real need. Of course, when there is a simultaneous crash for all stocks (as was the case in 1929) the protection brought about by diversification becomes ineffective. Subsequently, investment trusts were even blamed for being one of the causes of the crash. May be, but the idea was basically good, as shown by their huge development after World War n. Derivatives provide protection through hedging; they are in fact of form of

40

Chapter 2

Financing and monitoring the growth of a company Company A

Company B

CD-

-CD 2) Raising money by selling stock to share holders

-@

Fig.2.8 Financing and monitoring the growth of a company. The figures schematizes two ways of financing the growth of a company, either through bank loans or by selling stocks to share holders. In the first case the growth of the company is monitored by one agent, in the second it is monitored by the average opinion of a large number of agents.

insurance. For instance options enable a purchaser to buy or sell an asset at a certain price on a given date or allow him to walk away if he wishes. It is a generalization to financial markets of the forward transactions that have existed for decades in commodity markets. The rapid development of the derivatives in the 1990s (total transactions were multiplied by 10) was also proof of the fact that they filled a real need. It is true that there have been some accidents: both the bankruptcy of the Barings bank in 1995 and of LTCM (Long Term Capital Management) in 1998 was the result oftrading in derivatives. But in spite of such incidents the idea is basically sound. After all, airplane crashes do not call into question aviation. As we have seen even the

Overall view

41

disaster of 1929 did not dissuade investors from using investment funds.

1.9.3

The development of stock markets in the late 20th century

Does the previous interpretation account for the development of stock markets in recent decades? It seems it does. Between 1980 and 1997 the number of companies listed on the NYSE increased by 94 percent (from 1570 to 3047) while the number of companies listed on the NASDAQ increased by 90 percent (from 2894 to 5487). Most of the smaller and emerging companies are listed on the less strictly regulated NASDAQ. Even in Europe and in Japan, new markets have appeared in order to help emerging high-tech companies to raise capital. In Europe the EASDAQ (European Association of Securities Dealers Automatic Quotation system) was created in Brussels in the late 1990s, the "Nouveau March6" (new market) was created in Paris in March 1996, the "Neuer Markt" (new market) started in Germany in March 1997 and the MOTHERS (Market of High Growth and Emerging Stocks) started in Tokyo in 1999. All these markets were more or less built on the model of the highly successful American NASDAQ; by 2000 they were still fairly small but rapidly developing. No doubt that evolution will bring about a much closer connection between companies and the general public.

1.9.4

Organization of the market

For the sake of brevity we restrict ourselves to American stock markets. Table 2.8 gives a summary of the main markets. There are basically two kinds of markets: markets with a specific geographicallocation and the so-called over the counter (OTC) markets where shares are traded by telephone or by inter-connected computers. Among the first, the New York Stock Exchange (NYSE) is by far the largest. In fact it is the largest in the world; in terms of capitalization it was surpassed only by the Tokyo stock market when the latter was at its peak in 1989. In the United States since the mid-1980s stock market events are closely monitored by TV networks. In 1989 NBC (News and Business Channel) decided to broadcast twice an hour from the floor of the exchanges. Most of the trade in derivatives takes place on the over the counter markets. These markets have many advantages in terms of flexibility and transaction costs, but they also have some drawbacks. For instance in an OTC market there is a risk of one of the contractors going bust and refusing to pay. From the perspective of speculative trading there is a characteristic of American stock markets that is worth mentioning, namely the so-called special-

42

Chapter 2

Table 2.8 Stock markets in the United States

Number of companies

Capitalization [billion dollars]

1980

1997

1990

1996

1999

1) New York NYSE (New York Stock Exchange)

1,570

3,047

2,820

7,300

11,000

AMEX (American Stock Exchange)

973

102

2,900

5,500 1,500 mainly derivatives

5,000

2) Electronic markets

NASDAQ OTC (Over The Counter) 3) Regional markets

Chicago: Midwest Stock Exchange Philadelphia: Philadelphia Stock Exchange Boston, Cincinnati, Denver, Salt Lake City, San Francisco

less than 5% less than 3% less than 3%

Notes: NASDAQ means National Association of Securities Dealers Automatic Quotation system. Since 1953 the AMEX has replaced the former New York Curb Exchange .. In March 1998 an agreement was settled for a possible future merging of NASDAQ and AMEX. Sources: Quid (p. 2205), Bruchey (1991), Le Monde (14 March 1998).

Overall view

43

ist system. Specialists are members of the Exchange whose obligation is to maintain an orderly market. Acting as buffers between the public and the market, they serve as dealers when there is insufficient public interest to accommodate sellers or buyers at prices reasonably close to the last trade. Thanks to their participation 92 percent of all transactions that occurred in 1990 on the American Exchange were within 0.12 points of the last sale (Bruchey 1991). Their role is particularly important in the event of a panic. For instance during the crash of 14-20 October 1987 they were responsible of about 22 percent of the purchases and 16 percent of the sales whereas on average is the late 1980s they accounted for only 10 percent of the trading volume. In the days following the crash (21-25 October) they were responsible of 17 percent of the purchases and 21 percent of the sales (Bruchey 1991). By permitting to avoid panics the specialist system can have a decisive influence not only on short-term price fluctuations but even on the whole subsequent business situation.

2

Comparative perspective

In the previous section we have examined each market separately; actually, they are more or less closely connected. This will be the theme of a subsequent chapter. In the present section we consider these markets in a global, comparative perspective. To begin with, we try to compare their relative weights.

2.1

Relative weight of various speculative markets

From table 2.9 it can be seen that in terms of capitalization, real estate and stock markets have grossly a similar weight. The weight of the other markets is negligible in comparison with these two giants. Let us emphasize once again that if we study the other markets it is not because they are economically important but because of their simplicity.

2.2 Thrn over ratio Table 2.10 introduces the notion of tum over ratio which is obtained by dividing the value traded (i.e. the turn over) by average market capitalization. For gold or diamonds capitalization is understood as the value of the total stock existing in the world. In terms of turn over ratio there are three groups; (i) the stock markets which have a high turn over ratio (ii) Real estate with a ratio

Chapter 2

44

Table 2.9 Relative weight of speculative markets

Percent Source of GDP A. In terms of total value traded I ) Real estate:

Paris (1997)

0.4

Marche Immobilier Fran
United States (1992)

17

Le Monde (10 March 1992)

Paris (1994)

13

Quid (p.22IO)

NYSE (1996)

60

Statistical Abstract of the US

3) Art: France

0.02

Peyrelevade

4) Postage stamps: France

0.01

Roehner et al. (1999)

5) Antiquarian books: France (1993)

0.01

see notes

2) Stocks (shares)

B. In terms of capitalization I) Real estate

Paris (1997)

13

Marini et al.

United States (1992)

160

Le Monde (10 March 1992)

2) Stocks (shares) Paris (1994)

31

Quid (p.22IO)

NYSE (1996)

90

Statistical Abstract of the US

Tokyo (1990)

150

Wood

Notes: Although the GDP (gross domestic product) is a variable of a different nature than either the total value traded or the capitalization, it seemed convenient to use it as a yardstick of comparison. Note that the French real estate data are restricted to Paris intra muros, i.e. without the suburbs. The value traded for antiquarian books was obtained on the following basis: 100 auction sales a year, 300 lots sold at each sale, average price of a lot 3,000 French francs, sales in bookshops about the same order of magnitude as auction sales.

Overall view

45

between 4 and 10 percent (iii) Gold and diamonds which have an even lower ratio. Table 2.10 Turn over ratio of speculative markets Market

Turn over ratio

Source

I) Real estate Paris (1997)

4%

see Table 2.9

United States (1992)

11 %

see Table 2.9

Paris (1994)

43 %

see Table 2.9

NYSE (1996)

55 %

see Table 2.9

United States (1998)

106%

ESMF (1999)

Japan (1998)

40%

ESMF (1999)

Chile (1998)

7%

ESMF(1999)

3) Gold: world (1990)

5%

Quid (p.2188)

4) Diamonds: world (1980)

2%

Epstein

2) Stocks

Notes: The turn over ratio is calculated by dividing the value traded by average market capitalization. Except for emerging markets such as Chile, stock markets are characterized by a fairly high turn over ratio; in other words, they are liquid markets. ESMF means Emerging Stock Market Factbook.

2.3

Price earnings ratio

The ratio of the current price of a stock to past (or expected) annual earnings is used to estimate the price level of a stock. From 1871 to 1996 the PER of US stocks averaged 14. Before the crash of October 1929 it reached a level of 22. Between January 1986 and the crash of October 1987 the PER (for Standard and Poor's stocks) increased from 12 to 30. Before the crash of November 1989 in Japan the PER was at a level of about 50 (Galletly 1988, Schwert 1996). In January 2000 the PER of the NASDAQ was at a level of 200 while the PER of the 30 Dow Jones Industrial stocks was around 28 (Washington

46

Chapter 2

Post 12 Jan. 2000, Dallas Morning News 26 Feb. 2000). But stocks are not the only values for which one can define a PER. The same concept can be defined for property; in that case the earnings are represented by the monthly rent received by the owner. Before the 1990 property crash in Paris the PER for a 3-room apartment was about 2 million FFI 72,000 FF =28; in the mid-1990s it had dropped to 21. For other speculative items for which there are no earnings one cannot define a PER. However, for rare postage stamps one can define what can be called a generalized PER based on the following argument: observation shows that the deflated price of 19th century French stamps increased fairly steadily between 1920 and 1975 at an annual rate of 5.2 percent (Roehner et al. 1999). If one accepts to interpret that increased valuation as a kind of dividend this leads us to a PER of 1/0.052 = 19.2. The interesting point is that, if we exclude the somewhat exotic value reached on the NASDAQ, most of the above figures are in a rather narrow range between 15 and 30. Each time the PER reached the lower or upper bound of this interval a change in the trend rate occurred.

PART II

HIDDEN COLLECTIVE DETERMINANTS

I

Chapter

3

Rational?

In January 2000 it was customary to receive Internet advertisements for new credit opportunities, such as for instance the following: "Credit cards goodies: free information on cards offering rebates, bonuses and other incentives"; "Increase sales up to 150 percent, accept credit cards over the Internet. Good credit, bad credit or no credit, no problem, it just doesn't matter, everyone gets approved"; "Debt blaster: reduce your credit card and loan debt more effectively". Such advertisements reflected the development of electronic trade, but if we look at them more closely they can also tell us many things about the prevailing mood of the time. Firstly, the fact that such advertisements did arrive at a rate of about two or three a day attests to the Internet's cost effectiveness for advertisement purposes!. Secondly, they very clearly delivered the message that drawing heavily on short term credit and loans was socially acceptable and should in fact be encouraged by any possible means. This idea was in sharp contrast with the attitude which prevailed in the 1930s but is was in line with the rapid development of the private short-term debt in the United States in the 1980s and 1990s; the latter increased from a level of 20 percent of national income in 1970 to about 100 percent in 1999. Undoubtedly the wide acceptance given to the "debt economy" was something new. The prosperity enjoyed by the United States in the two last decades of the 20th century certainly favored acceptance of that concept. What is of special interest in the perspective of this chapter is the fact that within a time period of about twenty years a new set of ideas had won popular as well as academic approval and was included in the set of axioms and assumptions which formed the basis of the economy of that time; similarly deregulation and privatization became widely accepted notions in the late 20th 1 At

the time the rate for bulk email sendings was about 100 dollars for one million emails

sent. B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_3, © Springer-Verlag Berlin Heidelberg 2009

47

48

Chapter 3

century. Yet, it is important to realize that the consensus of opinion on such issues is subject to fluctuations in the course of time. Some notions that seemed "natural" to our ancestors of the 18th century are no longer accepted today and vice versa. Of course it can be argued that profit maximization was the ultimate objective in the past as well as it is today. But the means which were considered socially acceptable in order to realize that objective have been changing a great deal. For instance the attitude toward animal suffering has obvious implications for meat production. The ruthless way cattle and poultry were handled in the intensive production centers of the late 20th century would have seemed unacceptable in previous centuries by a mostly rural popUlation which had stronger feelings for the animals which shared its existence. These are only some examples; in the present chapter others are discussed in more detail which make clear that the expression "economic rationality" has no clear meaning unless one has defined the broad set of political and social assumptions on which it relies. For the sake of clarity we use the terms biased or circumstantial rationality for economic reasonings making implicit use of a set of economic and social presuppositions. The transition from one set of cannons of social acceptability to another can be gradual or sudden; in the six months between July 1929 and February 1930 accepted conceptions probably changed more than during ten years in more normal time. We are not yet able to measure shifts in conceptions; some indicators will be proposed in the next chapter but they remain rudimentary and restricted to the long term evolution. However, I am convinced that once we can measure more precisely and realistically the fluctuations in public opinion it will become clear that their role is at the heart of the question of speculative trading. The present chapter intends to be a first introduction to that crucial question. First, we analyze hidden preconceptions in a long-term perspective. Seeing things from a distance make them stand out more clearly. Then we discuss the same phenomenon in a medium-term perspective (i.e. for time spans of the order of 10 years) which is more relevant to the question of speculative trading. Finally, in the last section we discuss short-term shifts of opinion during speculative bubbles.

1

Hidden preconceptions: long-term perspective

"Despite their reputation for caution and ability, these high profile experts [bankers, brokers, fund managers and company directors] demonstrate far greater levels of speculative fervor and outright irrationality than the mass

Rational?

49

public". This is how B. Cohen (1997) describes the behavior of economic agents at the height of a bubble. Can it really be described as being irrational? In this section we argue that "rational" is a relative concept, not an absolute one. A given behavior can well be considered as perfectly rational in a given social context and quite fantastic or unreasonable in another. This will be shown by considering some historical examples. By so doing we are in fact trying to (re)introduce a historical dimension into economics. Such a task has already been brilliantly undertaken by G. Snooks (1993, 1998), J. Williamson and his collaborators (Hatton et al. 1998, O'Rourke et al. 1999) and indeed all economists who did not restrict their conceptual horizon to the last fifty years. Once we accept that rationality is conditioned both in a longand short-term perspective by the socio-historical context, the behavior of economic agents during a speculative bubble becomes more understandable. No sound investor would shun a golden opportunity to reap huge profits, but at the height of a speculative episode, profit perspective look so attractive that everybody tends to make the same decision at the same moment. A behavior which is perfectly sound when adopted by 20 percent of the people becomes unsound when 80 percent rush in the same direction. Although after a crash, most traders realize that the climate has changed it is not always obvious to identify precisely the factors which are responsible of that change. Examining long-term shifts provides a clearer perception. In 13th century France it was deemed self-evident that one had to build cathedrals at all costs. Similarly, in the late 20th century nobody questioned the fact that health care had to be improved at all costs. In the 18th century it was obvious for almost everybody that slave trade was a perfectly acceptable business. In the mid-19th century it was admissible to wage war against a country that tried to ban opium from its territory; in the second half of the 20th century it seemed equally admissible to outlaw a country on the ground that it was unwilling to fight drug trade. In the 18th century it was natural to lease tax collection to private companies; in the 1930s it was a well accepted idea that the economy has to be regulated by the state; in the 1980s an opposite view gained acceptance which holds that everything will work better if it is done by private companies; however nobody questioned the fact that tax collection has remained in the hands of the state. While in the 1950s and 1960s a substantial budget deficit was considered acceptable, such was no longer the case in the 1990s; in contrast, in the late 1990s it was regarded as good policy to boost consumption by encouraging people to increase their personal debt, an attitude which would have been looked at with suspicion in previous decades.

50

Chapter 3

The previous list could easily be extended but in order to get a better understanding of such shifts in opinion it is probably better to examine a small number of cases in more detail.

1.1

The attitude toward usury

The word usury is currently restricted to the practice of lending money to be paid back at an unfairly high rate of interest. However, back in the 18th century the expression was used to designate all sorts of loans whatever the interest rate, and it is with that meaning that usury was condemned by Christian and Muslim theologians. As we know this had far reaching social consequences in concentrating many financial activities in the hands of people who where not subject to that kind of interdict: Budhist Hindus in Pakistan and predominantly Muslim parts of India, Confucian Chinese in Indonesia and Malaysia, Jews in Europe and in the Ottoman Empire. In Western Europe opinions began to change on that matter only by the end of the 18th century when the society freed itself from the influence of the Church. The essays published by J. Faiguet de Villeneuve (1770) and by J. Bentham (1796) are representative of that tendency. Yet, even in the early 19th century some writers continued to champion the old conception: see for instance the book by M. Capmas (1829) whose title in substance maintains that interest payments on loans are in contradiction with both natural and divine law. That question is often seen as a scholastic dispute which is no longer of great interest today; however it also provides a typical example of what may be called a social reading of the Bible. By the expression "social reading" we mean the following. Individuals may have divergent interpretations, opinions and beliefs; yet, for a society to keep its cohesion people must share common beliefs regarding important matters. Back in the 18th century the interpretation of the Scriptures certainly was a matter of paramount importance which therefore required a social consensus. The next paragraph provides another and even more dramatic example of the same phenomenon.

1.2

The attitude toward slave trade

In 1783 there was a trial in London which may provide a graphic illustration of the attitude toward slave trade which was prevailing at that time. In 1781 the slave ship Zong, under Captain L. Collingwood, sailed from the coast of Africa to Jamaica with 470 slaves on board. During the crossing 60 Africans and 7 crews died of disease and many of the remaining slaves

Rational?

51

were sick. Collingwood called his officers together and explained to them that if the slaves died of natural causes it would be a total loss for the ship owners, whereas if they were thrown alive into the sea on the grounds of necessity the owners could claim the insurance of 30 pounds a head that applied to goods lost at sea. Accordingly he ordered his crew to throw 130 African slaves overboard. However, the insurers refused to pay on the ground that there was no true necessity to throw the cargo overboard. The case was first heard in March 1783 at the London Guildhall which supported the claim of the ship owners. The insurers then took the owners to court and the case was heard by Lord Chief Justice Mansfield. The legal issue was only about cargo: had the captain had any reasonable alternative to dumping the sick slaves. As a matter of fact throwing sick slaves over board in order to prevent the spread of epidemics was fairly common practice on slave ships. What was unusual in the Zong case was the scale on which the operation was performed (Walvin 1992, Sunday Times 5 July 1992 and 7 Sept. 1997). This episode is revealing of the conception 18th century societies had of slave trade. Of course with the passing of time it is easier to get a clear view of former bias. The ideology of pre-industrial societies rested principally on religious conceptions and the proponents of slavery found their justification in the Bible. It has been argued, e.g. by M. Weber, that from an economic perspective there were significant differences in conceptions between protestants and catholics; regarding slave trade, however, there was a striking similarity: British Anglicans, Dutch Calvinists, Danish Lutherans, French or Spanish Catholics, even American or British Quakers, all took an active role in the slave trade in spite of their different creeds. The attitude toward slave trade once again offers a spectacular illustration of the fundamental difference between what we called individual and social reading. At any time there was a wide range of individual interpretations of the Bible, but the spectrum of its social interpretations was much narrower; and it was the latter, not the former, which was determinant in shaping social attitude. As a matter of fact that social interpretation was maintained until the early 19th century when plantations in the Caribbean became less profitable and slave work less essential. We come back to that point subsequently. In the same line of thought it can me mentioned that two centuries after the Zong trial, apartheid in South Africa was justified and actively supported by the South African Protestant Dutch Church. Other times, other places, other conceptions. The two previous cases could lead us to think that our current conceptions are less prejudiced and have a more "rational" basis. The two following examples are intended to show that this would be a misconception.

52

1.3

Chapter 3

The attitude toward health care

Let us introduce this question with a case which exemplifies the shift in conceptions. In the 1950s the treatment of someone who had fractured his foot consisted in two or three X -ray photographs, a plaster cast for immobilization and a number of physiotherapy sessions. If we denote by E the corresponding expenses expressed in 1999 euros, treatment of the same fractured foot would cost in the late 1990s between 5E and 1OE. Examining the reasons of such a huge increase will help us to understand the shift in current attitude regarding health care. There are mainly two factors but in order to explain them one has to go into some medical technicalities. • In a specific case of the mid-1990s that we were told about, the number of X-ray photographs for a fractured foot reached about 50 and one may well wonder what justified such a huge increase. A fracture in a foot bone may be difficult to identify because X-rays may be screened by surrounding bones. Before taking the X-ray photograph one has to find the right angle and position. In the 1990s time had become so costly that it was more expedient to prescribe a scan rather than to make a careful examination; a scanner is of course a very costly device too, but once it has been bought it has to be used intensively. • Blood tends to accumulate in the veins of an immobilized foot instead of being driven back to the heart. This can lead to the formation of blood clots with a subsequent risk of gangrene or even heart attack. There are basically two ways to solve that problem. The first is to teach the patient how to contract the muscles in his foot and leg in order to boost the blood circulation. The second is to prescribe an anticoagulant. The first solution is almost never used because it would take too much time to teach that technique to patients and time is so costly. On the other hand, anticoagulants are expensive drugs which, in addition, can induce harmful reactions in the organism. Therefore one has to put the patient under constant medical watch which makes this procedure very costly. To sum up, two conclusions can be drawn. Firstly, there is a permanent tendency to trade off time and care against reliance on costly technical devices. Secondly, eliminating the small hazards which subsisted in the treatment used in the 1950s without introducing new ones turned out to be very costly. Fig.3.1 provides a macroeconomic illustration of the above case in terms of diminishing returns of health expenditures. It is not the fact that the benefits in terms of increased life expectations are becoming smaller in the course of time which is surprising for diminishing returns are of course common to many other situ-

53

Rational?

i

10 4

~

:

~

45

\

40

\

....

.

"

35

...

l

iI

50 ••••• Ex......1Ioa or or. I Healtb expenditures per capita II ...~

....••••.

10

30 15

\ •........•. ....

J 1

m 15

i:!:

10 5 1920

2000

Z020

Fig.3.1 Exponential growth of health expenditures and its diminishing returns. The solid line (left-hand scale) shows health expenditures per capita in the United States over a period of more than 60 years; since the vertical scale is logarithmic the growth is seen to be exponential. As a fraction of GNP the share of health expenditures remained stable from 1930 to 1960 at about 5 percent, before beginning to climb; in 1994 it was of the order of 14 percent. The same pattern is observed in other industrialized countries. The dotted curve (right-hand scale) shows an estimate of the health expenditures efficiency; it represents the ratio of life expectation to health expenditures per capita. The additional life expectation brought about by each additional dollar spent on health care is becoming close to zero. Sources: Kurian ( J994), Statistical Abstract of the United States.

ations from agriculture to mining development; what is surprising is that even with such low return levels, expenditures continued to increase exponentially. A parallel in mining development would consist in putting ever increasing amounts of capital in an already depleted mine. No doubt that would be considered as a highly irrational attitude; in the case of health expenditures it is not however, thus reflecting the scale of values of our societies. In 13th century France and Western Germany it was held that the only rewarding undertaking was to celebrate the glory of God by building ever larger cathedrals. About 50 cathedrals were built at that time and no consideration was given to the fact that such a huge immobilization of capital could have calamitous economic consequences. As a matter of fact, the subsequent century was marked by a protracted economic recession. Was it a consequence of the wave of cathedrals building? The question remains open.

54

1.4

Chapter 3

Nationalization versus privatization

In the 1980s, in a move against the over development of the state, many governments embarked into privatization programs. Were such moves really supported by historical evidence or was it rather an ideological option? Here are some elements for an answer. • So far no government has tried to privatize its public universities. The European Union for instance is pushing for privatization in utilities and telephone but there is no plan to create private European universities. Isn't that surprising in view of the fact that in the United States private universities have proved highly successful as shown by their unparalleled record of Nobel prize winners. • In the mid-19th century numerous private railroad companies were set up in all industrialized countries. One century later however, in the very same countries, railroads were run by public companies. What had happened? Have governments really been eager to nationalize railroads companies? Not at all! In fact, they took them over with great reluctance and did so only because one after another they all went bankrupt. The long series of bankruptcies began right from the start. Thus, in Britain the number of companies decreased from 300 in the 1850s to about 120 in 1918 and only four in 1921. In France around 1875, the government was obliged to buy up 2,600 km of line belonging to failing companies, by 1909 it had to buy the western lines which had been operating chronically at a deficit. One of the latest episodes of that kind occurred in the United States in August 1970 and concerned the Penn Central Company. For many years it had been plagued by derailments due to deferred maintenance; in order to survive it needed a loan of 50 million dollars which was at first promised by a consortium of nine banks under the guarantee of the Federal government. Eventually, however, the government withdrew its guarantee and the Penn Central Transportation Company rolled toward bankruptcy with the finality of an express train (Fortune May 1970). Finally after 1971 passenger traffic in the United States came under the supervision of a unique, federally supported corporation, the so-called AMTRAK. We will not analyze here why private companies failed so badly in the railways business whereas they brilliantly succeeded in other economic sectors. One reason is probably that in railroad business they would have had to join forces and work together, something they were not used to, accustomed as they were to compete and fight one another. Whatever the reasons, the fact remains that they failed consistently and everywhere. Nevertheless, in the 1980s when privatization had become politically a popular theme, it was also applied to railroads in a number of countries e.g. Argentina, Brazil, Britain,

Rational?

55

Portugal. After two decades the record of these privatizations was fairly disappointing; not surprisingly, the achievements of private companies were in line with what they had been in the period before nationalization. • As we already mentioned, back in the 18th century taxes were collected by private companies and it was at that time even possible to rent a private army for a summer campaign. So far nobody has ever suggested that privatization should also extend to tax collection or the recruitment of armies. Not so much because that organization proved inefficient but rather for political reasons. To sum up, in cases where privatization would have given excellent results nothing of the sort was attempted, while on the other hand privatization was carried out in economic sectors where private companies had a fairly poor record. It seems therefore fair to conclude that the privatization vogue was driven by political and ideological motives rather than by the "lessons" drawn from history. In the next section we will see that even in a medium-term perspective economic "rationality" is shaped by social preconceptions. But before that, we propose an hypothesis as to why the aforementioned shifts in conceptions occurred.

1.5

A possible clue to shifts in conceptions

Are the previous changes in conceptions completely random or is it possible to find a clue? Let us for instance consider the case of slave trade. E. Williams (1944) argued, quite convincingly in our opinion, that slave trade was abolished by Britain and France once their sugar plantations in the West Indies had lost their capability to produce the huge profits they had generated in the 18th century. The rational behind that argument is that a given ideological framework comes into force to permit and facilitate the exploitation of a profit source of great potential and to give a competitive advantage to the dominant powers of the time. Although this mechanism cannot possibly explain all the ideological changes that took place in recent centuries it seems to fit with the evidence in many cases. Here are a few illustrations. (i) The opium trade imposed to China in order to provide an outlet for the opium production of British India was a case in point. (ii) Ecological requirements about clean air and water became a major international issue only once developed countries had completed their second industrial revolution based on less polluting industries (iii) Until the end of the 19th century both Europe and large parts of the United States have been deforested without much concern for the environment. Some authors even argued that the shortage of wood in

56

Chapter 3

Britain was one of the triggering causes of the first industrial revolution. At that time deforestation was not the ethical issue it would become in the late 20th century. (iv) Free trade usually becomes a dominant conception when the power which is politically dominant is also in the lead economically. Such was the case in the mid-19th century2 when England was the dominant power and after World War II when the United States were the dominant power. (v) The "keynesization" of economic conceptions was a reaction to the devastating impact of the Great Depression in the United States. In the 1970s a generation of managers came to power who had not experienced that depression. Not surprisingly economic conceptions then shifted back to the "less state" creed which was in force before 1929.

2

Hidden preconceptions in economic rationality: medium-term perspective

As in the previous section the role of social preconceptions will be illustrated through a number of specific cases.

2.1

The Paris real estate market in 1990

I remember a discussion I have had with Parisian friends around 1990 about the situation of the real estate market. To me it was obvious that it was in a deadlock. Schematically there are three possible destinations for an apartment (i) To rent it out (ii) To keep it unoccupied before selling it (iii) To use it for oneself . Between 1985 and 1990 current prices of apartments had tripled and this has had several implications. As rents did not follow suit because they were limited by government regulations, the rent revenue drawn by the owner had become an unattractive 3 percent or even less. In the vocabulary of the stock market one would say that the price earnings ratio had strongly increased. To keep the apartment unoccupied was a good choice only so long as prices continued to climb. Finally, at those high price levels most apartments had become too expensive to buy for the average Parisians, as was confirmed by the decrease in the volume of sales. In spite of these omens my friends were convinced that prices would continue 2 The fact that in the late 18th century British manufacturers still favored protectionism can be illustrated by the following observation: in 1784 the city of Manchester sent a petition with 55,000 signatures against any tariff concessions to the Irish; remember in that connection that Irish printed linens entering Britain paid a 65 percent duty (Palmer 1959).

Rational?

57

to climb. The reason of their attitude was simple: between 1945 and 1990, property prices had never substantially fallen in Paris. There may have been some periods during which real (i.e. deflated) prices had fallen but, as is well known, most people are sensitive to nominal rather than real prices. Thus the belief in a continuation of the upward trend had become stronger and stronger to the point of turning into an absolute certitude. When the prices began to fall in 1991 it came as a surprise even to many "high profile experts"; as a matter of fact in the early 1990s several large banks and insurance companies were driven to failure because of their over-exposure in the real estate market.

2.2

Shift in accounting rules

A spacecraft worth several billion euros can be lost because of a faulty electrical wiring. Similarly, even a small change in accounting rules can have devastating consequences for banks and investors. Here are two examples. The first concern Japanese banks. In the 1980s Japanese banks substantially increased their market shares because they were not bound by the requirement of short-term profitability. In contrast, in the United States higher capital adequacy standards were imposed to the banks partly as a safeguard introduced in the aftermath of the Latin American debt crisis. When the matter came under discussion at the BIS (Bank for International Settlements) the American banks argued that those rules regarding capital requirements had placed them at a disadvantage. It is precisely the role of the BIS located in Basel, Switzerland to edict common accounting rules for all internationally active banks of the so-called G-I0 group which comprises most industrialized countries. In July 1988 it was decided that by the end of fiscal year 1991, banks should have a minimum capital adequacy ratio of 8 percent. The precise definition of the capital ratio adopted by the BIS is a matter of considerable importance in practice, but it is also a very technical question, the details of which are unimportant here. For Japanese banks these tighter requirements came at the worst moment. As a result of the Tokyo stock market crash and the collapse of the property market, their capital ratio began to deteriorate which made the 8 percent target even more difficult to comply with. This factor strongly contributed to deepen and prolong the financial crisis (Taniguchi 1993). The second example concerns the relation between financial analysts or auditors and companies headquarters. In 1998 the Association for Investment Management Research, the umbrella group for 34,000 financial analysts explained that "most analysts" who work for the big brokerage houses that sell

58

Chapter 3

stocks to the public are reluctant to write negative reports by fear of black balling. According to a 1998 survey, 80 percent of leading American companies were shown advance copies of analyst reports. What could be merely an accuracy check was also an invitation to censorship. Those fears were expressed in an article of the New York Times (28 Aug. 1998) entitled "Pervasive problems in accounting standards"which was published in the wake of the strong market decline of late August 1998. Subsequently the market resumed its rise and the question of accounting rules was eclipsed by other issues. However, it surfaced now and then. For instance in January 2000, in a rare move that sent shivers through the accounting world, the Securities and Exchange Commission threatened to ban two auditors who failed to detect several accounting tricks while investigating Micro Devices Corp (Wall Street Journal 6 Jan. 2000). Such problems can hardly be separated from the business situation. In a general way, in a bull market a strong social pressure is put to bear on accountants to release reports which support shareholders expectations. In the following paragraphs we explain how such a climate comes to pervade the whole society.

3

Application to speculative trading

In September 1929 the renowned American economist Irwing Fisher noted (the full citation is given below) that, thanks to the rapid development of investment trusts, investing in the stock market had become much safer than it used to be. Should such a statement be considered as reflecting the euphoria that prevailed at the time? Yes and no. Obviously, Fisher's opinion was influenced by the continuous growth that the market had experienced in previous years, although in September 1929 the euphoria was already somewhat eroded. But, even if its timing was ill-fated Fisher's statement was fundamentally true. By the diversification they provide, investment funds objectively reduce the risk for the investor. Except of course in the case of a major crash. In February 1996, Barton Biggs, chairman of Morgan Stanley Asset Management claimed that the US stock market was heavily overvalued; the Dow Jones index had just reached the level 5,600 at that time. He foresaw a sharp decline on Wall Street for 1996 but estimated that emerging markets could return 25 percent (Cohen 1997, p.349). These predictions were spectacularly contradicted by the market: the Dow Jones index reached 7,000 one year later, whereas emerging markets began to falter before plunging in 1997. These two examples illustrate the fact that when we turn from a long-term perspective to a short-term perspective everything becomes more fluctuating,

Rational?

59

more erratic, in one word more complicated to understand. In 1760 there may have been a few people opposed to slave trade but there was a strong consensus to support it. When it comes to judgments about short-term issues the picture becomes more confused. Even in the strong bull market of 1996 there were "high profile experts" who voiced their concern. This means two things. (i) It confirms our thesis about rationality. If all these arguments had a strong rational basis it would be difficult to understand that such strongly divergent opinions can be expressed at the same time. (ii) If economic rationality conceals hidden preconceptions then one must recognize that in a short-term perspective these preconceptions are not identical for everybody. In face of such a bewildering diversity of opinions and judgments trying to find some underlying regularities could seem a hopeless task. Yet, in chapter 5 we will see that it is possible to define social markers enabling us to gauge the global mood of the market. But first of all we want to mention some other examples of individual judgments and predictions just in order to convince the reader that their relevance is almost nil.

3.1

The crystal ball

The crash of 1929 and the subsequent market drop was by far the most spectacular in the history of the New York Stock Exchange, hence it is an interesting "laboratory" for our purpose.

3.1.1

The crash of 1929

First we give in full Fisher's citation that we already mentioned. "A few years ago people were as much afraid of common stocks as they were of a red-hot poker. Why? Mainly because the average investor could invest in only one common stock. Today he obtains wide and well managed diversification of stock holdings by purchasing shares in good investment trusts". There was indeed a tremendous increase in the assets of investment trusts in the late 1920s with a multiplication by ten between January 1927 and October 1929 (Tvede 1990); the irony lies in the fact that it is precisely this rapid growth which was pinpointed as one of the major causes of the over-speculation that lead to the crash. Fisher made another reassuring declaration only one week before the crash: "I do not feel that there will soon, if ever, be a fifty or sixty points break below present levels, such as Babson has predicted. I expect to see the stock market a good deal higher than it is today within a few weeks" (Galletly 1988). In fact, between October 23 and October 29 the Dow Jones index fell from 326 to 230 that is to say 96 points.

3.1.2

The property crash in Hong Kong

Between 1984 and 1996 there was a tenfold increase in the prices of residential property in Hong Kong; a few weeks before the downturn the industry was still seeing clear reasons for further price increases. In an interview given to the Financial Times (24 June 1997) a "high profile expert" mentioned the following reasons: (i) The government failure to release enough land for housing had restricted the supply. (ii) The 1996 new housing stock was one of the lowest in a decade. Accordingly that top Hong Kong banker brushed aside concerns over property exposure in following terms: "The demand is high, the supply at the moment isn't there". That belief was supported by nice charts which showed that the supply was indeed substantially lower than the demand (Financial Times, 27 March 1997). And yet, in July 1997 the market began a downward spiral which resulted in a price decline of about 50 percent in the following three years. 3.1.3

The bull market of the 1990s and early 2000s

During the whole decade 1990-2000, newspapers were full of warnings as to an imminent crash. Here are a few examples taken from a much larger sample. • In November 1992, F. Filloux in an article published in the French newspaper Liberation (23 November) warned against a major drop in the Dow Jones index; and he provided a number of "good" reasons for that. For instance he noted that the price earning ratio (PER) of the 500 stocks listed in the Standard and Poor's index was at the level 24 which marked a historical record. To come in line with a more normal PER the market had to drop by about 20 percent. As we know, in 1993, instead of falling the Dow Jones index gained about 10 percent. • On 14 February 1996, an article entitled "Warning lights start to flash over Wall Street" appeared in the British newspaper "The Independent" The article explained that Wall Street was more highly valued by traditional standards than at any time other than before the last three stock market crashes. • In May 1996, in an article entitled "Downsizing dubious dividends" (Financial Times 14 May) B. Riley warned that the stock market capitalization had risen to almost 90 percent of GDP, beating the previous peak of 82 percent in 1929, the long-run average being estimated to around 48 percent. In spite of this warning the capitalization of the NYSE reached 150 percent of the GDP in July 1998. • In September 1996, P. Yates, the management director ofPDFM (Phi-

Rational?

61

Hips and Drew Fund Management) one of Britain's biggest managers of pension money, declared (Guardian 18 September): "The degree of over-valuation in the UK and US stock markets is unprecedented. If you have done all the analysis which leads you to that view then you have to stick by it". And that is what the firm did by progressively reducing its holding of shares and instead building up a mountain of un-invested cash of about 7 billion pounds. As a result the firm got a poor 3 percent return, while stock values soared by about 30 percent. • On 5 December 1996, Federal Reserve Chairman Alan Greenspan shook financial markets with the question: "How do we know when irrational exuberance has unduly escalated asset values?". From Asia to New York, traders interpreted Greenspan's query to imply that the stock market was overvalued and that the Federal Reserve might be about to increase interest rates to dampen the market. The day after, investors bailed out pushing stock prices down between 2 to 4 percent. But this did not last and soon the market resumed its ascension. • On 16 December 1996, Newsweek's front page portrayed an abysall perspective for stock holders (Fig.3.2). • On 1 October 1998, reacting to the failure of Long-Term Capital Management (LTCM), a large hedge fund, A. Greenspan declared: "Financial markets operate efficiently only when participants can commit to transactions with a reasonable confidence that the risks of nonpayment can be rationally judged and compensated for. Effective and seasoned markets pass this test almost all the time. On rare occasions they do not. Fear, whether irrational or otherwise, grips participants and they unthinkingly disengage from risky assets in favor of those providing safety and liquidity. Assets, good or bad, are dumped indiscriminately in circumstances of high uncertainty and fear that are not conducive to planning and investment". Yet, in spite (or perhaps because) of all these warnings western stock markets continued to climb. What conclusion can we draw from this lengthy enumeration. At any time it is possible to find books or newspaper articles warning against the current speculative fever. Does this not contradict our thesis that a speculative frenzy spreads to the whole society through the communication channels of the media? Certainly one must take into account the fact that predicting a crash is a better copy-selling argument than predicting continuous and unhampered growth. In order to characterize the spread of the stock market bubble it is in fact the increase in the number of articles which matters rather than their content. This is what we examine quantitatively in a subsequent chapter. In the last part of this section we pave the way for such a study

62

Chapter 3

1111.111.11.

Investing in an U.rtain Market Wall Street's Global Impact Fig.3.2 llIustration of an ill-timed warning about a stock price collapse. The Newsweek issue with this cover appeared on 16 December 1996. The title of the article "What goes up must go down" was certainly reasonable, but one never knows when the downturn will occur. In this case the bull market continued for over four years.

by examining some qualitative evidence. However, before closing this paragraph on expectations and predictions, another sort of forecast has to be mentioned.

3.1.4

''Buy back on Derby Day".

"In May go away and buy back on Derby Day [Epsom Derby takes place on first Wednesday in June)". That saying may have been popular among British investors in the 19th century but it is probably no longer used nowadays. This, however, does not mean that similar precepts do not playa role in investors' decisions. At moments when the future looks particularly uncertain it is tempting even for professionals to push aside computer generated data and graphs and to contemplate such rules as the "January barometer" or the "Super bowl predictor". In essence the "January barometer" says "as January goes, so goes the rest of the year" which means that if the stock market rises in January it will be up

Rational?

63

for the entire year. According to the "Super bowl predictor" stock markets generally rise in years when a team from the National Football Conference wins. Since 1950, in the United States, the January barometer has been accurate on 40 out of 50 years; more precisely its accuracy hits 100 percent in odd number years (which may have a connection with the beginning of a new congress) but is only 60 percent correct in even number years. The Super Bowl predictor was discovered in 1979 by R. Stovall, president of "Stoval1!Twenty-First advisers". It has been an accurate predictor of the stock market 85 percent of the time (International Herald Tribune 2 February 2000). Needless to say, these rules have little or no intellectual basis and even investors who use them would probably not admit doing so. Such rules will certainly never be mentioned as possible causes of a crash by economic historians. And yet they may contribute to change the climate of a market. It is in this sense that they can be considered as hidden collective factors.

3.2

How a mania spreads

Enthusiastic titles probably contributed more to the euphoria of the late 1990s than carefully drafted explanations. It may be of interest to give some illustrations in order to suggest the climate that prevailed at that time. The January 2000 title of the French economic magazine "L'Expansion" read: "La Wahoo economie. Prevision 2000: une expansion sans precedent" i.e.: The Wahoo [a word coined by combining the names of two Internet providers: the French "Wanadoo" and the American "Yahoo"] economy. Forecasts for 2000: an unprecedented expansion. At about the same time one could see the following advertisements: "Boursorama: premier site d'information boursiere: avezvous encore une excuse pour ne pas gagner en Bourse?" i.e.: Boursorama: a website that gives free stockmarket advice. There is no longer any reason to stay away from the stock market. At the beginning of this chapter we have already mentioned advertisements received through computer networks; the following is quite revealing: ''We thought you should be interested in our free monthly newsletter called Diamonds in the rough. Each month the Newsletter profiles an undervalued public company that is on the verge of making a big move. My goal is to show that big winners can be found where others fear to look". As one would suspect the mania can also spread through inter-personal exchanges. Unfortunately, such cases of informal communication are only rarely recorded. An instance of that kind concerning the speculative episode in Hong Kong that we already referred to was related by the Los Angeles Times

64

Chapter 3

(27 Sept. 1994) in following terms: "As in L.A. in the 1980s real estate has become a Hong Kong obsession, a recurrent theme of cocktail parties and dinner conversations. Those who made a killing can't resist exulting, while those who stayed out of the market forever lament their misfortune".

3.3

Short-term phenomena

In the sphere of economics and finance panics are he most spectacular manifestation of short-term collective outbursts. However collective frenzy can also manifest itself in other circumstances. For instance, news of the end of the siege of Mafeking on 12 May 1900 (during the Boer War in South Africa) produced a mass hysteria in the streets of London. Similarly, after the victory ofthe French soccer team in the 1998 World Championship the whole country gave itself up to demonstrations of exuberant joy in the streets of all major cities. In the 19th century the runs on the bank were a collective expression of feelings of panic. Spectacular as they were such movements did not always have a lasting influence. For instance the runs on the banks that occurred in New York in 1854 and 1857 had no serious consequences for most of the banks (see in this respect 0' Grada et al. 1999). Similarly, after the panic of October 1929 the Dow Jones index rose again from 230 by the end of October to almost 300 in April 1930; subsequently, however, and this time without spectacular panic, it resumed its downward trend in a fateful spiral which lasted until 1932. In this chapter we have argued that most speculative frenzies are based on beliefs and opinions which, at the time, seem perfectly natural and acceptable because they are shared by most people. Accordingly, it is only when the frenzy has died down, usually after a crash, that one begins to realize that what had been considered as self-evident was in fact the manifestation of a temporary tum of mind; that phenomenon is well described by the term"Zeitgeist", a word borrowed from German which means the spirit of the time. In the next chapter we will see that, somewhat surprisingly, the prevailing Zeitgeist can be the same at a given moment even in distant countries such as Britain and Japan.

I

Chapter

4

Joint crashes

On 9 July 1709 (Gregorian calendar) the Swedish army led by king Charles XII was severely defeated by the Russians at Poltava. Europe was stunned by this defeat for until then the Swedish army was considered invincible. Subsequently scholars "explained" the defeat in various ways, for instance (Swanstrom et al. 1944, Andersson 1973, Scott 1977, Massie 1980) by the fact that having been wounded in the left foot ten days before, Charles could not lead his army in his usual way. Although plausible, such an explanation appears nevertheless surprising when one realizes that the Russian army was far superior in numbers and equipment; as a matter of fact, the Russian army numbered 45,000 and had 72 guns while the Swedish army numbered 26,000 and had only 4 guns (Bodart 1908, p.159). While that obvious reason is in fact given by some historians (e.g. Langer 1968, p.508), the fact that most scholars preferred to mention a circumstantial cause is characteristic of the tendency to discard basic and organic factors. R. Massie (1980), for instance, puts the following words in the mouth of one of the Swedish generals: "Would to God our gracious king had not been wounded for then it had never gone as it did", and he observes that the confusion in the Swedish army stemmed from "the absence of the one commanding figure who rose above jealousies". To show that this is not an isolated case it can be of interest to give two other, perhaps even more striking, examples. The battle of Waterloo (1815) is commonly presented as a watershed in the history of Europe. Historians offered countless speculations as to the causes of Napoleon's defeat: "The ground was moist and it was necessary for it to become firmer so that the artillery might maneuver. Had the earth been dry the action would have began three hours earlier and it would have been won before the arrival of Blucher" or "Had Grouchy been able to join in the battle, history would have been changed". Yet, if one takes a closer look at the situation such speculations appear largely pointless. Opposing forces for the battle of Waterloo were the followB.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_4, © Springer-Verlag Berlin Heidelberg 2009

65

66

Chapter 4

ing: 120,000 Allies of which 24,000 English, 44,000 Dutch-Hanoverians and 52,000 Prussians, against 72,000 French (Bodart 1908). Even if Napoleon had won the Waterloo battle, the outcome of the whole campaign would certainly not have been changed for the strategic situation was overwhelmingly in favor of the Allies. According to Clausewitz (1972) for the whole of the campaign the Allies totaled 645,000 comprising 100,000 Anglo-Dutch-Hanoverians under Wellington, 115,000 Prussians under Blucher, 140,000 Russians and 230,000 Austrians; on the opposite side there were only 195,000 French. A similar and more recent case was the Pacific War. No doubt that each naval battle comprised a great part of uncertainty; yet, from a strategic perspective there was not the slightest doubt about the final outcome of the war for the gross national product of the United States was about 10 times larger than the Japanese GNP. In the long run this implied a massive American superiority in terms of ships, planes and other armaments. Furthermore it should be recalled that Japan was at war not only against the United States but also against Britain and the Netherlands (and even Russia in the last months of the war). Presented in this way, history looks less enthralling, and one can understand the tendency of historians to prefer captivating stories to dull facts and figures. In this chapter that tendency will be observed repeatedly for economic phenomena as well. Our demonstration proceeds in three steps. In the first section we list a number of economic phenomena which are impossible to account for through circumstantial explanations. In the second and third sections we consider several specific examples of inter-connected speculative markets. In many cases simultaneous crashes cannot be explained without invoking some common collective factor.

1

Is speculation in diamonds related to speculation in cobalt?

In January 1979 a one carat, flawless polished diamond cost 20,000 dollars on the Antwerp market. In February 1980, at the peak of the speculative episode, is was worth 60,000 dollars. Subsequently diamond prices plummeted; the same one carat diamond cost 40,000 dollars in January 1981 and 20,000 in January 1982 (Diamonds 1988, The Economist). In order to understand the causes of this price peak it is natural to first turn our attention to the explanations put forward by diamond experts. In 1988 an authoritative economic assessment of the past decade was published by W. Boyajian (1988) the pres-

Joint crashes

67

ident of the Gemological Institute of America. What is his interpretation of the crisis? The story goes as follows. In 1976 Israel was a young but rapidly growing diamond center. Anxious to promote the diamond trade because of its contribution to the country's exports which represented 40 percent of Israel's non agricultural exports, the Israeli government supplied several banks with huge amounts of money at very low interest rates to be passed on to diamond manufacturers so they could build their inventories of rough diamonds. But, because of the rapid increase in diamond prices many diamond-cutters found more profitable to hold their rough diamonds rather than to cut them. Thus, they created an apparent demand for diamonds that really did not exist. Soon the speculative fever spread to Antwerp and New York the two other main diamond centers. In an attempt to bring order to an exploding market the De Beers imposed enormous price increases totaling 140 percent between March 1979 and February 1980. In late 1979 the Israel market began to crumble: for the first time in 30 years employment in diamond manufacturing declined by one third. At first sight such an account appears fairly satisfactory. It sounds plausible, seems to explain the facts and is supported by some quantitative evidence. However if we look at it more closely it becomes clear that there are several loopholes in the argument. Why did De Beers drastic measures not succeed in limiting the speculation? Why did the speCUlative fever spread to Antwerp and New York where no low interest government loans were available? How can we explain in this framework that almost overnight diamonds became a fashionable investment and that so-called diamond investment firms sprang up throughout the U.S. and Europe? Well, owing to the many facets of the diamond market it can hardly be surprising that our previous account was incomplete. However, it would in fact be useless to study that market in greater detail for the same phenomenon also affected several other commodities as shown in the next paragraph.

1.1

Other commodity bubbles which burst in 1980

FigA.1 and Table 4.1 show that there was a huge price peak for several precious metals at the beginning of 1980. In econometric models the price of a commodity is determined by its supply and demand functions; cobalt, diamonds, gold, palladium, platinum or silver being rather dissimilar in terms of production/consumption factors, are characterized by fairly different supply/demand functions. It is true that for some usages gold and platinum are substitutes but this represents only a small part of the platinum market; remember that 35 percent of the platinum production is absorbed by the manu-

68

Chapter 4 ~

600

.--~~~--~~- -~~~~__~~_ _ _---.

]

~

~

500 -

]

Diamonds

e ~

400

~...

300

§

!

Silver Gold Platinnm

I

~.

~

]

~

~

200

l·

100 --

....:::-...---

.~.~~.....

-'---L_L. ~,~~I-LI---L-'-,--<1-"--,-L--L-l ,

1974

1976

1978

1980

1982

[ " 1984

I , 1986

,J

1988

Fig.4.1 Price peak for diamonds, gold, platinum and silver. For gold, silver and platinum the prices are expressed in dollar per troy ounce; for diamonds the price is for a one-carat G (i.e. white) flawless diamond. The collapse occurred almost simultaneously in the four markets. The coincidence is all the more significant because in normal times these prices are only weakly correlated; for instance between 1993 and 1998, gold and silver prices had a completely different behavior. Prices of palladium and cobalt which showed a similar pattern were not represented on the same graph for the sake of clarity. Sources: International Financial Statistics (International Monetary Fund) 1979, 1991; Journal des Finances (26 Oct. 1978); The Economist (5 April 1980), Diamonds 1988 (The Economist); Chalmin (1999).

facturing of catalytic converters for exhaust pipes, an usage for which it cannot be replaced by gold. Cobalt, diamonds and gold are even more dissimilar than gold and platinum. Thus, the coincidence of these price peaks can hardly be explained by standard econometric models. As a matter of fact each of these industries has its own story in order to "explain" each specific peak. For instance the price peak for silver is commonly explained by the failed attempt of Nelson B. Hunt, the Texan billionaire, to comer the market; see for instance in this respect an article in the Financial Times (29 March 1980) entitled ''The silver coup that failed" or a similar article in the Journal des Finances 00 April 1980) entitled: "Echec et mat pour un milliardaire texan (i.e. Checkmate for a Texan billionaire) or the book by S. Fay (1982). Without going here into the details of this story one must emphasize that it is not supported by any precise statistics. For platinum there is another story: in Canada and Russia the extraction of platinum is associated to the extraction of nickel; now, nickel was at that time at a low price which did not encourage production. That reason again sounds plausible, but it is hardly confirmed by actual figures. For instance one may note that Canada

Joint crashes

69

Table 4.1 Simultaneous commodity price peaks Price increase 1978-1980 [$/ounce] Cobalt

213 %

Diamonds

127 %

Gold

133%

Palladium

152 %

Platinum

98 %

Silver

470%

Notes: In contrast the increases in the price of aluminium and copper were much smaller: 47 % and 49 % respectively. Sources: Cobalt, gold, palladium, platinum, silver: La Vie Franraise (14 Jan. 1980); diamonds: Diamond 1988 (Special Report, The Economist).

and Russia together represent only 40 percent of the world production of platinurn, the other 60 percent being produced in South Africa which calls into question the previous explanation. More generally none of the above explanations can explain why there were simultaneous peaks for at least five different commodities. In other words these stories provide just anecdotal interpretations similar to the one about Charles's injury regarding the outcome of the battle ofPoltava. Before going further one must explain in more details why in such a case it is so difficult to build reliable econometric models.

1.2

Why it is almost impossible to build econometric models for floating stockpile markets

It could seem that we jumped a little bit too quickly to the conclusion that supply/demand econometric models are unable to explain the simultaneous price peaks of January 1980. After all the supply/demand framework is one of the main pillars of economics and it is rather worrying if one has to give it up. Two points should be emphasized in this respect.

70

Chapter 4

• Our critic does not concern the supply/demand framework itself but rather the practical possibility to build a satisfactory model in the present case. Basically most of the data that would be required in order to build a model are deemed confidential and are not made public. This is particularly true for diamonds, gold and silver. Remember in this respect that for diamonds even wholesale prices are not made public, not to speak inventory levels, production costs or price elasticities. • The synchronous peaks displayed in Fig.4.1 strongly suggest that these commodities share a common feature. One of their common characteristics is that they are all easy to store and the level of stocks usually exceeds annual production. This is particularly clear for diamonds and for gold for which stocks are respectively 50 and 20 times larger than annual production (Table 4.2). In chapter 2 we referred to such markets as floating stockpile markets. Precise stock data were not available for the other metals but their case is likely to be similar. Remember for instance that the world production of platinum is only 134 tons (1993); with respect to such a small amount stocks whether in the form of strategic or industrial reserves can easily represent several times the annual production. Totalling 20,000 tons, the world production of cobalt is somewhat larger but remains relatively small when compared with for instance the production of copper which is 500 times larger. It is extremely hazardous to build an econometric model for commodities for which stocks are permanently larger than the world production unless one has very precise estimates of all inventories and reliable knowledge concerning stock management strategies; needless to say, such information is hardly available for the markets under consideration.

1.3

A possible clue: a hedge against inflation

The floating stockpile mechanism cannot alone explain why these speculative episodes occurred at the end of the 1970s. A possible clue is that it was a hedge against inflation. Because precious metals and diamonds gave the possibility to stock easily a great amount of wealth in tangibles they were good opportunities for a hedge against inflation. They were not the only ones naturally: paintings, postage-stamps or other collectibles could also be used as a hedge against inflation and several of these items indeed experienced speculative episodes. Fig.4.2 shows that there was a striking correlation between the rise in silver prices and the level of inflation. In that chart we used the consumer price index in the United States, but the inflation phenomenon affected all industrialized countries and in many cases the situation was even worse than in the

71

Joint crashes

Table 4.2 Annual production compared to world inventories

Annual production

World inventories

Annual production as

percentage of inventories Gem-quality diamonds 1980

10 million carat

500 million carat

2%

1990

99 million carat

1,500 million carat

7%

3,000 tons

60,000 tons

5%

Gold 1996

Notes: For the diamond and gold markets annual production represents only a small proportion of existing stocks. Other speculative markets, e.g. land, property, equity markets, share the same characteristic; in real estate for instance annual production of new houses and apartments represents but a small fraction of existing houses. All these markets are particularly sensitive to manias and speculative fever. In the rest of the book they will be referred to as floating stockpile markets. Sources: Diamonds: Epstein (1982), Quid (1997); gold: Quid (1997).

u.s. Thus in the U.K. the inflation reached 24 percent in 1974 and 18 percent in 1979. In a previous paragraph we mentioned the hoarding behavior of diamond cutters in Israel; that phenomenon becomes clearer in the light of the high inflation level experienced by Israel at that time: the inflation rate jumped from 6 percent in 1970 to 40 percent in 1974 and remained at a high level (between 50 and 300 percent) at least until 1985. In the next chapter we will see that the concern about inflation really permeated the whole society. As a matter of fact, it was explicitly mentioned as a persuasive selling point by all diamond investment firms or art merchants. What this thesis fails to explain is why the price peaks culminated between January and March 1980. Would it have been one year later it might have had something to do with the economic program of president Reagan who took office in January 1981. Since science can hardly explain single events (single events can only be described) the only way to get a better understanding of

Chapter 4

72

2000 -

Prlceor.u_ Inn.1Ion over 5.4.. •••••.•.•.

1000 900 800 700 600

SOD 400 300

1995

Fig.4.2 Deflated price of silver in New York and inflation rate in the U.S. Vertical scale: 1980 US cents per troy ounce. The prices considered here are annual prices; in terms of daily prices the maximum would be at a level of 5,000 cents/ounce. The inflation rate refers to the US consumer price index. The broken line represents a dichotomic function: 0 for an inflation rate under 5.4 percent, 1 when it is over. The graph shows that the price peak was largely the result of a flight from inflation and that a 5 percent rate of inflation seems to be a critical level in that respect. Sources: International Financial Statistics (International Monetary Fund) 1979, 1991; Historical Statistics of the United States (1975).

that phenomenon is to find similar episodes of hedges against inflation.

2

Are property bubbles in Tokyo and Paris related?

Before we examine the relationship between property bubbles on spatially separated markets it can be enlightening to consider an historical episode marked by spatially correlated events.

2.1

The expUlsions of Jesuits in the 18th century

Because they had to pledge total obedience to their general in Rome, Jesuits were an ideal target for nationalistic agitation; because they managed to win the favor of kings and princes they were subject to the jealousy of the nobility and high clergy; because their colleges were granted special privileges Jesuits were seen as unfair competitors by the universities; in short, their expUlsions can be seen as the critical culminating point of a kind of nationalistic bubble. From the creation of the order in 1540 to the end of the 20th century at least

Joint crashes

73

thirty expulsion episodes occurred worldwide which constitute an ideal "laboratory" for studying social mobilization and collective outbursts (for more details see Roehner 1997b). Here we restrict our attention to four expulsion episodes which occurred in a row between 1758 and 1767. In 1758 Jesuits were banished from Portugal on the charge of a conspiracy to assassinate the king. In 1764 following the fraudulent bankruptcy of father La Valette, they were expelled from France. In April 1767 , in the wake of the Sombrero riot they were banished from Spain and in November of the same year from the kingdom of the Two Sicilies (Italy) on the charge of a plot to murder the king's fiancee. Finally, in July 1773 the order was (temporarily) suppressed by pope Clement XlV. Seen from a distance two centuries later, it is fairly obvious that these successive expulsions were different manifestations of the same wave of anti-Jesuit feelings. But in most historical accounts that common factor is completely ignored. Instead, the circumstantial facts that we briefly alluded to, are described as the real "causes" of the events. In analyzing property crashes we face a similar situation. For each crash it is possible to identify and enumerate a number of circumstantial causes. But in a comparative perspective it becomes clear that these causes were nothing more than triggering factors. Before coming to property crashes let us mention another historical example.

2.2

Mississippi versus South Sea bubble

The Mississippi bubble in France and the South Sea bubble in Britain burst within a few months, the first in May 1720 and the second in September 1720. Both were brought about by plans about the development of overseas trade and in particular slave trade. The fact that both Britain and France had at that time a huge public debt makes the parallel even closer. Certainly, one could think, economic historians did not fail to notice the strong connection between these events and studied them in parallel. That is hardly the case however. While numerous studies were devoted separately to each of these events only few mention a possible link between them and an even smaller number analyze it in detail. Not surprisingly, the South Sea bubble was primarily studied by English-speaking economic historians, e.g. Carswell (1993), Cowles (1960), while the Mississippi scheme was mainly studied by French historians, e.g. Cellard (1996), Faure (1977), and by a few English-speaking historians mainly with a focus on John Law himself. The language divide thus helps to understand why there are but few studies, e.g. Klaveren (1961) or Cohen (1997), in which these crashes are considered as two manifestations

74

Chapter 4

of the same wave of speculative frenzy.

2.3

Property crashes

In 1990 there was a pivotal downturn in the Japanese land and property markets. How is it explained by specialists of the Japanese economy (Wood 1992, Aveline 1995)? Land has a special status in Japan; whether this is due to its scarcity or to the fact that the Japanese society treasures its rural roots, the fact remains that land and property are considered as privileged forms of wealth; it supports for instance only fairly slight taxes. In the 1980s land came to represent a substantial fraction of the assets of many companies. In 1990 Japan's total stock of property was valued at four times the value of the total stock of property in the United States. The circumstances which led to such an extreme situation could be described in even greater detail: one could mention the role of the jiage-ya who were in charge of piecing together dispersed parcels of land or the role of the eight jusen, the major housing-loan companies. However it would be counterproductive to study the Japanese property crash in greater detail for that crash was not restricted to Japan, it also affected other property markets such as California, London, Paris or Sweden (Fig.4.3). In each of these cases the consequences were very similar: failure of a number of major banking institutions, weakness of the stock markets. In the previous section we described synchronous crashes for different commodity markets; here is something similar but for spatially separated markets. How can we explain these simultaneous crashes? An honest answer is that we do not yet have a satisfactory explanation. Undoubtedly, it would not be satisfactory to pretend that it was merely a coincidence. It is also difficult to explain them by a parallel evolution of underlying fundamentals; as a matter of fact the countries under consideration differed notably in terms of interest rate, taxes, demographic situation and so on. Of course it is always easy in economics to imagine possible mechanisms which would explain the connection between these markets, but it is quite another matter to prove that these mechanisms can indeed account for the observed link. For instance, can overseas capital movement provide a satisfactory explanation? Between 1985 and 1993 the Japanese invested 77 billion dollars in American property, a third of which went to California. But this amount represents less than 1 percent of American property transactions; even for California Japanese purchases represented only 2.7 percent (The Economist 23 July 1994, Le Monde 10 March 1992). It is doubtful that such small percentages have had more than a marginal impact on American property markets.

Joint crashes

75

Japan CaliforuJa Paris

Sweden

.................... 120

'

.......

.....

Fig.4.3 Simultaneous price peaks for residential property markets. In California the price peak occurred somewhat earlier than in the other countries. In this respect it is of interest to note that property markets in different parts of the United States are usually not synchronized; in Texas for instance the peak took place in 1984, while the north east coast experienced a speculative episode which paralleled the one on the west coast; thus in Boston the price of land doubled between 1985 and 1990. Sources: Japan, Sweden: Ramses (1994, Dunod, Paris); California: The Economist (23 June 1990, 23 July 1994); Paris: Conseil par des notaires; Australia: Financial TImes (12 Oct. 1992); Texas: The Economist (15 June 1991); Boston: Taniguchi (1993).

It can be observed that the land and property markets belong to the category of what in the previous section we called floating stockpile markets. This, of course, does not provide an explanation for the simultaneous crashes but it can explain why these markets were more sensitive than others to the influence of a possible common factor, the precise definition of which will be left open. Our above observation as to the similarity of the consequences of property crashes in Japan and other countries can at first sight seem surprising. In the 1990s Japan went through a severe financial recession while this was not the case in any of the other countries listed in Fig.4.3. Japan's recession was in line with the scale of its property bubble; elsewhere the effects of the crash were similar albeit less devastating. For instance in California the bankruptcy of the savings banks resulted in the loss of at least 200 billion dollars; in 1992 about 105 financial institutions had to be closed (Le Monde, 10 Nov. 1992). In July 1992 unemployment in California reached 9.5 percent while the national average was 7.8 percent. Fortunately the crisis remained confined to the West Coast; therefore the recession did not spread to the whole country;

76

Chapter 4

in particular it had only a limited impact on the stock market. In France the property crash in Paris forced several banks into bankruptcy, among them the Credit Foncier and the Credit Lyonnais. The failure of the latter (though it was eventually bailed out by the state) resulted in the loss of over 20 billion dollars. Fortunately, the crisis was less severe in provincial cities which explains that it did not take the proportion of a national disaster. Nevertheless, as in Japan, it brought about a stock market contraction: in the three years from 1990 to the end of 1992 the SBF index at the Paris stock market lost 13 percent, while in the same time interval the Dow Jones index at the New York Stock Exchange climbed 42 percent. In the first section we have seen that there are connections between price peaks for different commodities; in the second we have shown that speculative episodes of spatially separated property markets are in fact related. It should be emphasized that these connections only materialize during speculative episodes. In "normal times" the connection between the prices of cobalt and diamonds is very loose. Even for commodities which are to some extent substitutes of one another such as for instance platinum and gold or platinum and palladium the price correlation is very low in the absence of strong fluctuations. Similarly, in "normal" times the property markets in Tokyo and Paris are almost uncorrelated. A similar phenomenon is known in statistical physics: the interaction range between distant elements becomes larger as a system undergoes a phase transition; see in this respect Roehner (2000a, p.182).

3

Interdependence between different speculative markets

It is a well known rule of thumb that a rise in interest rates usually leads to a fall in stock prices. This rule, we will see, is far from being always true, but if we accept it, it provides a first example of a connection between two different markets, namely the equity and bond markets. Such a connection is hardly surprising for stocks and bonds are both traded in stock exchanges and to a large extent by the same people, a circumstance which facilitates arbitrage. But we will see that at times of major crashes such a connection extends to other markets as well, such as for instance the property, art, book or stamp markets. But first of all we discuss the link between equity and bond markets on a specific case.

Joint crashes

3.1

77

The link between equity and bond markets

Our illustration is taken from the Paris stock market in the period 1955-1965. At that time the capitalization and trading volume were limited; in 1960 the capitalization was only 26 percent of the gross national product but in fact this very circumstance will facilitate our analysis. Our objective is to examine how an increased issue of new bonds can affect the level of share prices; for a given volume of bonds such an impact will be more apparent for a small equity market than for a large one. Fig.4.4 shows the evolution of share prices together with the volume of public bond issues (the volume of private bonds was at that time four times smaller). We see that large bond issues attracted flows of money which in turn pro-

-(BOO

,-----------------------------------~9

i.. 700

8

I.

7 !

500 400

300

I

:

i

/

......1

6

5 4

\ ••••••••••••••••••••• ;::;... Issues

3

Fig.4.4 Relationship between equity and bond markets in Paris. Solid curve: deflated share prices. Broken curve: yearly bond issues, vertical right-hand scale in billions of French francs of 1960. When large amounts of bonds are issued, funds are channeled from the equity into the bond market, which in turn provokes a fall in stock prices, hence the negative correlation (-0.35) which is observed on the graph. It is an example of a communicating vessel mechanism. That relationship between the bond and equity markets if the basis of the Levy-Levy-Solomon model described in Oliveira et al. (1999). Source: Vigreux (1967).

voked a drop in share prices. In other words, the two markets seem to operate like communicating vessels: what is lost by the first is gained by the second and vice versa. By generating a greater money demand, large bond issues can push up interest rates; however, such an outcome can be thwarted by many other factors; this is precisely what happened in our case: between 1955 and 1965 interest rates, measured by the central bank discount rate, fluctuated fairly randomly between 3.5 and 4.25 percent. This helps to explain why the

78

Chapter 4

correlation between interest rates and the level of stock prices is rather loose.

3.2

Link between crashes in equity and property markets

Observation shows that crashes in equity and property markets often, but not always, occur simultaneously. A spectacular example was provided by the Tokyo stock and property market crashes in 1990 (Fig.4.5a). As a rule one 240

500 Japan

United States

450

220

...,:'

200

.!

180

..... ....

160 140 120

400

if\ ,

f

(!

(;!

~ \\

350

\\

\~

300

\,

250

\.....

.....

100 80

200 150 100

1988

1990

1992

1994

1925

1930

Fig.4.5 a,b Relationship between stock and property crashes in Japan and the United States. (a) Solid line: Nikkei index of share prices; broken line: residential land price; dotted line: residential property price. (b) Solid line: Standard and Poor's index of common stocks (deflated); broken line: land price in Chicago (deflated). The land bubble was of much larger amplitude in Japan than in Chicago: between 1985 and 1990 land prices in metropolitan Tokyo more than tripled. Sources: Japan, stocks: Taniguchi ( 1993); Japan, land: Financial Times (17 Oct. 1997); Japan, residential property: Ramses (1994, Dunod, Paris), US. , stocks: Historical Statistics o/the US.; US., land: Sakolvski (1932, p.4lO).

observes that in such cases the fall in shares'prices is faster than the fall in property prices, but this does not necessarily mean that the second is a consequence of the first. The fact that fluctuations in the property market are less abrupt is indeed a general characteristic of that market which can be attributed to its long transaction delays; three or four years after the downturn the magnitude of the fall was almost the same in both markets. During the 1997 crisis in Thailand, Malaysia, Singapore, and Hong Kong there was also a strong parallelism in the evolution of stock and property prices. On the other hand, Fig.4.5b shows a case where the connection is much weaker: American prop-

Joint crashes

79

erty prices already reached a plateau in 1926 and their fall after 1931 was probably caused by the economic depression rather than by the stock market crash.

3.3

Link between crashes in equity and art markets

In Fig.4.6a we examine the link between the stock prices on the NYSE and the prices of antiquarian books and paintings. The art market is seen to react with a delay of about two years. On the other hand changes in book prices paralleled stock price changes. Fig.4.6b provides evidence for the period 19801995 of a correlation between property (and stock) prices in Tokyo and prices on the art market. At first sight it could seem that there is an obvious and so to say mechanical explanation to such connections. The causal chain would be as follows: the collapse in stock prices results in a disappearance of wealth, therefore one would expect all superfluous, lUxury expenses to be reduced with immediate consequences for the art and rare book markets. But such an explanation is unable to account for several specific facts. • The fall in the book market already began after February 1929 that is to say six months before the crash at Wall Street. • The decline in book prices did not only affect expensive books but also books in a 10-20 dollars range. For such low prices the decrease in wealth can hardly be a satisfactory explanation. This leads us to an alternative interpretation. Both the stock and the book market collapse were a consequence of a change in the social climate. For such an explanation to be of some value one must of course be able to probe and gauge what we called the social climate. In the next chapter we propose some measurement means.

4

Conclusion

For some readers the present chapter may perhaps prove disappointing. Its main message was to show that standard explanations of market collapses are inadequate. Confronted with a phenomenon as spectacular as the burst of a speCUlative bubble the natural tendency of the human mind is to search for "causes" which are specific to the market under consideration; that bias is of course accentuated by the fact that "high profile experts" usually know only their own market. We have shown that such explanations most often miss completely the point because market collapses occur simultaneously on

Chapter 4

80 i

3OO

~ -(-

r--------------------,

~

Sloduo

Palntl... CoI...ted ........

I

1\

~ ~

\

\\

100 !IO

80 70

1931

1932

1933

Fig.4.6 a Stocks, paintings and collected books. Solid line: Standard and Poor's index of common stocks: price index for paintings constructed by N. Buelens and V. Ginsburgh; dotted line: average price of a random sample of books sold at auction sales in the United States. For authors such as H. Fielding or P.B. Shelley who had the favor of collectors the amplitude of the price peak would be at least five times larger. As to the timing, it can be observed that whereas the peak for stocks and books occurred the same year, the peak for paintings was two years latter. In the same line of thought it can also be noted that the early 1930s were marked by a collapse in diamond prices. Sources: Stocks: Historical Statistics of the United States (1975); paintings: Buelens et al. (1993); books: American book-prices current: several years.

=

1!188

....... prlce ... Tokyo-Art 100 Index

600

500

400 300

1995

Fig.4.6 b Land and art bubble. Solid line: price of commercial land in greater Tokyo; broken line: Daily Telegraph Art 100 index. Source: The Economist (30 November 1996,22 March 1997).

Joint crashes

81

several markets. It is true that we were not always able to identify the factor which triggered such series of collapses. However, if one of the objectives of science is to discover regularities and links between seemingly unrelated phenomena, then the present chapter was a step in the right direction. It seems to us that before market collapses can be fruitfully investigated one has to get rid of ad hoc explanations and to recognize that we do not yet fully understand the collective phenomena which are at work.

I

Chapter

5

Contagion of speculative frenzy

In August 1857 the failure of the Ohio Life and Trust Company started a banking panic which ended the boom decade which followed the Mexican War. In New York City, between 28 September and 13 October over five hundred depositors, i.e. 9 percent of customers, of the Emigrant Industrial Saving Bank (EISB) closed their accounts. The EISB was a mutual saving bank which predominantly attracted Irish immigrants to the point that in the late 1850s they held almost 90 percent of accounts. In a pioneering study, M. Kelly and C. O'Grada (1999) studied the contagion of panic among the depositors. Their results are most striking for it turns out that contagion channels of panic were largely dependent upon the emigrant's place of origin in Ireland (the latter determined their living place in New York). For instance the proportion of depositors who closed their accounts is about three times larger for emigrants from the province of Munster, in the south-west of Ireland, than for those coming from Ulster in the north-east; the impact of this factor can be traced even to county level. This example graphically shows the role of social proximity in the process of panic contagion. It is the objective of this chapter to examine these mechanisms more closely. The above example referred to a panic, however the mechanism is probably similar for good as well as for bad news: individuals hear some bad/good news and communicate it to their acquaintances who pass it on in tum. Although certainly of importance for the formation of speculative bubbles that mechanism has practically not been studied by economists, Kelly and O'Grada's study being a notable exception. This is of course understandable in so far as the study of social contagion belongs to sociology rather than to economics. Throughout this book one of our main methodological messages is precisely that such a separation between connected fields is a major obstacle to a better understanding of speculation. That situation is hardly uncommon in science; for instance to explain how a fuel-cell works one has to draw both B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_5, © Springer-Verlag Berlin Heidelberg 2009

83

Chapter 5

84

on physics (electricity) and on chemistry. The chapter proceeds as follows. In the first section we examine the role of two major communication channels namely books and the press; then in the second section we examine the relationship between the expansion of a bubble and the increase in trading volume. We argue that there is an upper bound for the trading volume which helps to explain why a bubble cannot expand indefinitely.

1

Social contagion

For the following discussion it is convenient to distinguish between three kinds of communication channels. • Direct inter-personal communication between individuals either by word of mouth or by mail, phone, email, etc. • Commercial communication means, such as newspapers, magazines, books, radio, television, etc. • Institutional channels such as companies, churches, political parties, unions, armed forces, etc. In order to assess the role played by these channels the crucial point is to find sources from which one can draw quantitative evidence. Inter-personal communication is of course the most difficult to study for it does not leave any records in archives. In some cases that obstacle can be overcome, however, as for instance in the investigation ofEISB depositors; in that case it was possible to know depositors' addresses in New York City as well as their place of origin in Ireland and to derive from these data their personal links. Similarly, by relying on surveys it was possible in some cases to trace how a technological innovation spread through personal contacts. Classical examples are the studies by Ryan et al. (1943), Griliches (1957), Coleman (1964); additional examples are mentioned in Hamblin et al. (1973). In the present chapter we limit ourselves to the study of two commercial channels, namely the publishing business and the press. As a preliminary piece of evidence one can mention the progression in the sales of specialized magazines during bull markets; thus between 1997 and 1999 sales of the French magazine "Investir" have almost doubled from 100,000 to 190,000; another magazine entitled "La Vie Financiere" has increased its sales from 90,000 in 1998 to 145,000 by the end of 1999 (Le Monde, 10 April 2000).

Contagion of speculative frenzy

85

1.1 The production of books reflects the spread of bubbles That there is a loose connection between the publication of books and the spread of a bubble would not be surprising; in this paragraph we intend to show a much stronger result, namely that there is a close quantitative correlation between the pUblication of books about a given speculative item and the price of that item. The main difficulty in this study is to devise adequate means of observation. In the next paragraphs we set up, test and put to use two procedures (for short we call them sociological microscopes) in order to scrutinize the book and newspaper business.

1.1.1 Preliminary test: assessing inflation from the number of published books The first "microscope" relies on the catalogs of major university libraries; that tool has become particularly convenient to use in recent times because these catalogs are now available on computer networks. Before using that tool for our purpose it is appropriate to test it by first performing a control experiment. For that purpose we listed the number of yearly published books on the topic of inflation in two university libraries, namely (i) The Harvard library which is a major general library (ii) The library of the National Foundation for Political Science (Fondation Nationale des Sciences Politiques, FNSP, Paris) which is a library specialized in economics and political sciences. Furthermore, in order to test whether the results depend on the counting procedure we used two different criteria. At Harvard we counted all books whose title contains the word "inflation"; at the FNSP we counted the books listed under the "inflation" heading of the subject classification. In spite of the fact that this second criterion to some extent depends on the judgment of the librarian in charge of the classification, both criteria lead to similar results, as shown in Fig.5.l. The numbers of yearly published books are represented by the thick (Harvard) and thin (FNSP) solid lines while the rate of inflation is represented by the dotted line; there is a close connection between the three curves: all three cross-correlations are over 0.82.

Remarks

We also tried another methodological option by normalizing the number of newly published books on inflation with respect to the total number of economic books entering the library in the same year. But this did not lead to better results; on the contrary, at Harvard it biased the results for the following reason. Periodically, microfilmed copies of 18th or 19th century books are bought by the library and introduced in the catalog; in some years

86

Chapter 5

1100

12

1

10

..

f i

'"

J

80

8 80 6 40

4 20

2

0

0

Fig.5.1 Inflation rate versus number of books published about inflation. Dotted line: yearly inflation rate in the United States (right-hand scale); thick line: number of books published yearly whose title contained the word "inflation"as listed in the catalog of Harvard Library; thin line: number of books published yearly which are listed under the "inflation" heading of the subject classification at the FNSP library (Paris). The two last curves were shifted one year backward which is approximately the time it takes to publish a book. The cross-correlation between any pair of these three curves is over 0.82. Source: adapted from Roehner (1995).

the number of new microfilmed books was substantial enough to bias the normalized variable. As far as the accuracy of the measurement is concerned another observation is in order. In the last two decades in addition to its use in economics, the word "inflation" has acquired a technical meaning in cosmology in relation with theories of the big-bang. Fortunately, this represented only a small number of books and did not substantially affect our results. It is precisely in order to control for such problems that we complemented the procedure based on titles with the one based on subject headings. The previous observations lead to the following conclusions. • When inflation increases the concern about price escalation seems to pervade large sections of the society: it makes publishers more prone to publish books on that topic, authors more willing to write them, librarians more ready to buy them and the public more desirous to read them. In fact the correlation is so good that the level of inflation could be measured fairly accurately (albeit with a time lag of one or two years) through the number of books published on that topic.

Contagion of speculative frenzy

87

• Estimates of the number of books published on an issue of worldwide interest are fairly robust with respect to the country where the library is located, the type of the library (general versus specialized for instance) or the method used, e.g. whether based on title or subject search. In order to better assess the significance of the previous result, the following comparison may be helpful. When a new experiment in the field of particle physics is carried out, one would certainly expect an increased production of papers on that topic. What we have observed in the case of inflation is fairly different however for at least two reasons. (i) The book/inflation relationship is not restricted to major inflation peaks; even modest upsurges in times of low inflation (under 5 percent) brought about an increased book production. (ii) The study of the inflation phenomenon is by no means facilitated in times of high inflation. On the contrary, from a scientific point of view, one would be in a better position 5 or 10 years after an inflation peak for one could then study both the upward and downward phase of the inflation episode. In short, Fig.5.l shows that the society reacted with great sensitivity (so to say as a resonance chamber) to fluctuations of the price index. It is important to note that, in this case, one cannot argue that the increased book production stimulated inflation. On the contrary, in the case of speculative trading to which we turn in the next paragraph, an augmented book production has a positive feedback effect: it feeds speCUlation by sustaining the public's interest.

1.1.2

Connection between speculative trading and production of books

In this paragraph, we propose three examples which point to a close connection between a speCUlative bubble in a given sector and the production of new books on that topic. All these experiments have been conducted on the electronic catalogs of the Harvard Library which can be accessed through the instruction: telnet hollis.harvard.edu. The solid line curve in Fig.5.2a shows the number of books published yearly whose titles contain one of the words: "stocks", "stock market" or "speculation". The broken line curve gives the level of stock prices. The connection between both phenomena is apparent especially during the peak of 19251932. The book production lags behind the price index, the time lag being of the order of 1.5 years, which is approximately the time it takes to write and publish a book; with this time lag taken into account, the correlation is 0.57;

Chapter 5

88

t f

1

f

f •'J"

30

25

20

15

10

5

Fig.5.2 a Stock prices versus number of books published on the topic of stock markets (1910-1940). Broken line: Standard and Poor's index of common stocks (NYSE), right-hand scale; solid line: number of books published yearly whose title contains one or several of the words "stocks", "stock market", "speculation" (as listed in the catalog of Harvard library). As in the previous figure there is approximately a one-year delay between the two curves but here the book curve was not shifted; the same remark applies to following figures. Source for stock prices: Historical statistics of the United States.

t

I

50

45

8000 7000

......71

4000

1

f

f

•J'"

6000 5000

40

3000 25

2000

20

15 10

5

...........

.•

\

\/Stock prkes

0 ~~~-L~~~~-L~~~~-L~~~~~J-~

1000 !IOO 800 700 600

1970

Fig.5.2 b Stock prices versus number of books published about stock markets (1970-1996). Broken line: Dow Jones (Industrial) index (NYSE), right-hand scale; solid line: number of books published yearly whose title contains one or several of the words "stocks", "stock market", "speculation" (as listed in the catalog of Harvard library). Source for stock prices: Statistical Abstract of the United States.

Fig.S.2 c Diamond price versus number of books published about diamonds. Broken line: diamond prices on the Antwerp market. (right-hand scale); solid line: number of books published yearly whose title contains one (or both) of the words "diamond" and "diamonds" (as listed in the catalog of Harvard library). Source for diamond prices: see Fig. 2. 7.

of course it would be substantially higher if the series were restricted to the 1925-1932 peak. Fig.5.2b tells a similar story for a more recent time period, namely the 19801996 bull market. The correlation is 0.70 (a zero- or a one year-lag leads to almost the same correlation). The last example concerns the diamond bubble of 1975-1984 (see Fig.5.2c). There is a close connection between the number of books whose titles contain the words "diamond" or "diamonds" and the actual price of diamonds. The correlation (without lag) is 0.66. The augmented book production is of course but one of the manifestations of the society's increased interest for a speculative item. By its very nature, because of the time it takes to write and publish a book, this indicator displays considerable inertia. Needless to say, it would be of interest to find an indicator responding more quickly to a change in the society's climate. We consider such indicators in the next paragraph.

1.2

How the press participates in the spread of a bubble

Our first task is to design an adequate "microscope".

1.2.1

Designing the microscope

Lexis-Nexis (Elsevier) is a huge computer data base which contains millions of articles published in newspapers, magazines or issued by news services. It covers the period from 1979 to present; not surprisingly, in the course of time as more and more newspapers became available in computerized form there was a massive increase in the number of articles. On Lexis-Nexis it is possible to select all articles published in a given time interval which contain specific words. The main problem is to select these words adequately; that problem already existed in our previous investigation but it is more serious in the present case because here it is the whole text of the article which is searched. Not surprisingly, in the course of time as more and more newspapers became available in computerized form, there was a massive increase in the annual number of articles; as a consequence the number of articles in the data base was much larger in the late 1990s than in 1978 which implies that the observed numbers of articles containing specific words must be normalized. In the next paragraph we explain the specific procedure used.

1.2.2

Property bubbles in the press

Suppose we want to follow the growth of a property bubble in Hong Kong. We first determine the number of articles published in a given year which contain the words "Hong Kong" and "property prices"; then, in order to normalize these numbers, we divide them by the number of articles which contain only "Hong Kong". From a practical point of view one must in this case restrict the time interval to a single day for otherwise the number of retrieved articles would exceed the capacity of the system, namely 1,000 articles. Fig.5.3a and b show the results for Hong Kong and Singapore. The normalized numbers of articles follow reasonably well the price level; the correlation is 0.51 and 0.69 respectively. In those two cases we were in a favorable situation in the sense that both in Hong Kong and Singapore several major newspapers are published in English. One is in a less favorable position when trying to apply the same procedure to Japan for in this case an overwhelming majority of newspapers are written in Japanese. As a result, most of the articles retrieved by searching for the words "Tokyo" and "property prices" correspond to non-Japanese newspapers. The results shown in Fig.5.3c follow reasonably well (correlation is 0.77) the price bubble until 1993; after that date there are a number of outliers.

91

Contagion of speculative frenzy ~

~

=' •:l!

1•

t

4QO

200

Number or artlcles--

300

Property prk:e 1118 II 70

" 51

ZOO

1

40

3G

f

:zo

l

18

8

~

•,

100

7

5 1!1116

I_

-

!III

1998

1994

1991

1996

1998

Fig.5.3 a Property price in Hong Kong versus number of articles published yearly on property. Broken line: property price in Hong Kong (right-hand scale); solid line: number of articles published yearly whose text contains the two expressions "property prices" and "Hong Kong"; that number has been estimated through the Lexis-Nexis newspaper data base. The correlation is 0.51. Source for property prices: see Fig.2.4b.

r-----------------------------------------~300

Number or artIcles--

Property price

200

....

;1 ..........

1986

1988

1990

1992

.~

100 90 1994

1996

1998

2000

Fig.5.3 b Property price in Singapore versus number of articles published yearlyon property. Broken line: property price in Singapore (right-hand scale); solid line: number of articles published yearly whose text contains the two expressions "property prices" and "Singapore"; that number has been estimated through the LexisNexis newspaper data base. The correlation is 0.69. Source for property prices: see Fig.2.4b.

1.2.3

The coin bubble of 1965

This example is taken from a very stimulating study by Montroll and Badger (1974, p.200). In the early 1960s, there was a speculative bubble for a number of American coins. For instance, a roll of fifty 1960 Denver-minted pennies of the small date variety which sold for 4 dollars in 1961 fetched 21 dollars in 1964; similarly a roll of 1955 half dollars went from 20 dollars to 190 dollars in the same period. One of the main journals in the field of coin collecting is "Coin World" which has appeared weekly since 1959. Undoubtedly a weekly publication can reflect the speCUlative mood of the market more swiftly and accurately than newly published books. The solid line curve in Fig.5.3d shows the number of paid copies of "Coin World" that have been distributed. The curve began to level off by October 1964 and the maximum was reached on March 1, 1965. On the same figure, we plot the price history ofthe 1960 D small date penny and ofthe 1955 half dollar; these curves are smoothed least square fits to the weekly price fluctuations given in "Coin World". Once again there is a conspicuous parallelism and this time the circulation curve does not lag behind the price series; its leveling off even precedes the turning point of the half dollar coin. In this case, the distribution data could have been used as a warning signal. Of course, for the penny price, the leveling off occurred even earlier, but the price of one coin can hardly be used to predict the price of another coin; in contrast the fluctuations in "Coin World" distribution provides a global estimate of the number of people who have an interest in the coin market. Incidentally, it is also of interest to consider the price increases themselves. The price of the 1960 Dpenny increased by a factor 5.2 (from 4 to 21 dollars), while the price of the 1955 half dollar was multiplied by 9.5 (from 20 to 190 dollars). This provides an illustration of the price multiplier effect to be examined in the next chapter; it says that for more expensive items speculation is stronger and leads to larger price increases. What is the rationale of the effect described in this section or, in other words, why does the interest shown by the public for a given item reflect actual price changes during a speculative bubble? Basically, each new person showing an interest in the market is a potential customer. When the number of interested people levels off the flow of new customers runs dry. Observation shows that with a fixed pool of customers the high price level attained in a market can be sustained only for a short time (during which "old customers" buy from and sell to each other). Is is true that these customers could decide to de-

93

Contagion of speculative frenzy

~~

80 70

60

• 1•

SO

Number of artides--

Property price

40

It

30

/

10

/..•.•.•......\\

I \

...~..

J

.'

.•.•..

.~

1 !

18

8 7 6

5 4 100

3 1986

1988

1990

1991

1994

1996

lOOO

1998

Fig.5.3 c Property price in Tokyo versus number of articles published yearly on property. Broken line: property price in Tokyo (right-hand scale); solid line: number of articles published yearly whose text contains the two expressions "property prices" and "Tokyo"; that number has been estimated through the Lexis-Nexis newspaper data base. The correlation is 0.77. Source for property prices: see Fig.4.3.

1II 180 170 ~

.s

Half dollar

160

~ ~ 150

.s

a~ 140

...............................

.

'

.1 130

1

120 110

100 90 8O~Lw~~~~~~~~~~~~~~~~~

Fig.5.3 d Comparison between the distribution of the magazine "Coin World" and the price of two American coins (1962-1970). "Coin World" (solid line) is one of the major magazines destined to coin collectors; the curve labeled "Penny" refers to the price of a roll of 1960 D small date pennies; the curve labeled "Half dollar" refers to the price of a roll of 1955 half dollars. Source: adapted from Montroll et at. (1974).

94

Chapter 5

vote a larger share of their revenue to buying coins; in this case, the bubble would continue to grow even with a fixed pool of customers. However, observation shows that this does not happen, at least in a short-term perspective. The example of the 1960 D penny is particularly revealing in that respect: for such an item which cost only 20 dollars even at its peak price, it would be fairly easy for many collectors to buy more; and yet, they do not. As soon as the number of new customers entering the market began to level off, the bubble began to falter. This very same mechanism has been documented by Galbraith (1997) in his analysis of the speculative bubble preceding the 1929 crash. More recently, Krugman (1995, 1998) has shown that the same mechanism has been at work in the speculative bubble preceding the Asian collapse in 1997. Is it possible to find suitable indicators for the mood of stock markets? The most serious difficulty is to find a market journal which is representative of the level of public interest. The "Wall Street Journal" has been the most important newspaper in the securities field for many years; however, it has also achieved recognition as a national newspaper so that it is read for general information as well as for investment news. As a matter of fact, its average yearly circulation does not closely follow the fluctuation of stock prices. For instance, during the bear market of 1968-1974, its circulation increased from 1.1 million copies to 1.3 million, a 18% increase. That increase was however much smaller than the one that occurred during the 1960-1968 bull market, namely 72%. In short, the circulation of the Wall Street Journal to some extent exhibits the phenomenon described above but because of the journal's broad spectrum it cannot serve as an accurate thermometer of speculative fren-

zy!. 1.2.4

''The hidden persuaders"

It would be naive to think that all articles published during a speculative episode had for sole purpose the information of the reader. There is good evidence that a substantial number of them were deliberate attempts to shape the public's tastes and interests. In a book from which we borrowed the title of this paragraph Vance Packard made that point very convincingly. Although that issue is only of anecdotal interest from the perspective of this chapter (what matters is that, whether deliberately of not, these forces contribute to the spread of the bubble), it can be of interest to illustrate that aspect by some 1 I am grateful to Ms Claire de Buttet from the Paris bureau of the Wall Street Journal for kindly sending me the journal's circulation data.

Contagion of speculative frenzy

95

examples. As a case in point we consider the diamond market. In the 1970s De Beers, the company which controls the diamond market, established a "Diamond Information Bureau" in the United States which released a flood of historical data and news. Among its successes the agency fostered the publication of an article exalting diamonds in "National Geographic" and helped to prepare it. In the same line, after a long series of conferences with Paramount officials it succeeded in obtaining that the title of a film be changed from ''Diamonds are dangerous" to "Adventures in diamonds". In another film called "Skylark" intense lobbying resulted in the insertion of a long scene dealing with the selection of a diamond clip and bracelet for the actress Claudette Colbert (Epstein 1982). In the press it is not uncommon in the days before Christmas to see articles about De Beers which along with fairly flimsy and trivial information contain beautiful color pictures of diamonds and jewelry. One example was an article entitled "Diamonds, production and supply. De Beer's control" which appeared in "The Economist" on 20 December 1997, another was a series of two articles about De Beers which appeared in the 1935 May/June issues of "Fortune". In conclusion we have shown that in a number of cases the excitement which accompanies a sustained bull market or a property boom is mirrored in the activity of communication channels. It was possible to check that connection not only qualitatively but also quantitatively. The results are summarized in Table 5.1. One should also keep in mind that the media shape the mind of the public in the sense that we have described in chapter 3. Moreover, we argued that the ability to attract new customers is an essential condition for the continuation of a speculative peak. In principle the same effect could also be achieved by convincing customers to allocate a larger share of their revenue to the specUlative item. However, observation suggests that the percentage of hislher revenue that an investor allocates to a given speculative item is a structural factor which varies only slowly in the course of time.

1.3

Consumer confidence and stock prices

Is there a relationship between the level of consumer confidence as it can be estimated through surveys and the direction in which stock prices move? Fig.5.3e provides an answer for the period 1997-2000. Two different indexes have been used: the consumer sentiment index (dotted line) published by the Survey Research Center of the University of Michigan and the consumer con-

Chapter 5

96

Table 5.1 Role of the media in the diffusion of a speculation mania

Item

1.

2. 3.

4. 5. 6.

Stocks (NYSE) Stocks (NYSE) Diamonds Property (Hong Kong) Property (Singapore) Property (Tokyo)

Time interval

1920-1929 1980-1998 1978-1980 1986-1998 1990-1997 1986-1990

Medium

books books books press press press

Number of publications: annual increase rate

Price: annual increase rate

11%

10%

9%

10%

65%

58%

33%

31%

36%

26%

88%

26%

Notes: In this table we consider the upgoing phase of a speculative peak. For cases 1 and 2, the increase rate in the number of publications refers to the number of books published annualy whose title contains the words "stocks", "stock market" or "speculation"; for case 3 the title must contain "diamond" or "diamonds". For cases 4, 5 and 6 the procedure is similar but technically somewhat more complicated: the methodology is explained in the text. It can be seen that, except for property prices in Tokyo, the figures in the two last columns are surprisingly close.

fidence index (broken line) published by the Conference Board (NFO Research, Greenwich, Connecticut). The first is based on phone interviews of a sample of 500 people, while the second is based on answers to questionnaires sent by mail to 5,000 people. As can be seen the monthly changes of the two indexes are fairly parallel and they seem to follow the changes in the Dow Jones index (solid line) with a time lag comprised between one and two months. In other words consumer confidence as measured by these indexes is affected by the level of stock prices rather than the other way round. On a longer time scale it can be noted that between 1991 and 1999 that is to say during the stock market boom the consumer sentiment index has grown fairly steadily from 65 to 105. That impact of stock markets on public confidence is something new however. Thus, during the 1955-1968 stock market boom the consumer sentiment has not shown any significant increase, fluctuating rather randomly between 80 and 100. One should also note that consumer sentiment may be strongly affected by events which have no impact whatsoever on stock markets. Thus, in 1991 the Gulf War lead to a fall in the

Contagion of speculative frenzy

97

~ 1~r-------------------------------~

il48

1130 11'

f J

120

110

100

Im.5

1998.5

1999

1999.5

2000

2000.5

Fig.5.3 e Relationship between consumer confidence and stock price changes. Solid line: Dow Jones index (monthly data); broken line: consumer confidence index (Conference Board); dotted line: consumer sentiment (University of Michigan). The consumer indexes appear to follow the changes in stock prices with a lag of about two months. In other words the situation of stock markets have a determining influence on consumer confidence; this constitutes an important self-reinforcing effect. Sources: consumer sentiment index: Survey Research Center of the University of Michigan (Web site); consumer confidence index: various Internet press releases.

index from 93 to 63 while on the other hand stock prices continued to climb. The fact that consumer confidence is more and more influenced by the behavior of the stock market has important implications. As one knows consumer's propensity to buy is closely correlated with the consumer confidence; thus, an euphoric stock market is doubly beneficial to the economy: not only is it easier for companies to raise capital on the stock market, but their sales are also boosted. While the first effect is well-known, the second is often overlooked.

2

Volume of trade as an indicator of the extension of speculation

For stock markets there is a vast literature on possible relations between trading volume and prices; see for instance Crouch (1970), Karpoff (1987), Rogalski (1978), Schneller (1978), Tauchen et al. (1983), Ying (1966). Most of these papers are concerned with short-term (Le. hourly, daily or weekly) changes; since price and volume are two fairly erratic variables this turns out to be a difficult statistical investigation. Crouch (1970) makes two interesting

Chapter 5

98

points in that respect. • Hourly correlation between price and volume is fairly larger than daily correlation (0.13 and 0.29 respectively) • Total trading volume has a higher correlation with the NYSE index (i.e. the index which includes all stocks quoted on the New York Stock Exchange) than with the Dow Jones index: the correlation increases from 0.29 to 0.69. In this section we concentrate our attention on long-term changes. Moreover, we consider the volume of sales as a convenient indicator of the public's interest for a given speculative item. In that sense this section is a logical continuation of our analysis in the previous section. With the exception of stock markets, statistics about trading volume are not easy to find. For the postage stamp market for instance no such statistics are available because a large part of the transactions take place between private collectors and are not recorded. In this section we consider stock markets and some property markets for which we were able to get adequate data. From the evidence available it appears that in the upgoing phase of a speculative episode there is a strong correlation between price and trading volume. Before giving the results derived from correlation and regression analysis, we examine some qualitative evidence.

2.1 2.1.1

Qualitative evidence Stock markets

Wall Street has several adages about trading volume, for instance: "It takes volumes to make prices move" (cited in Schneller 1978) or ''Transaction volume [i.e. the number of shares traded] tends to be high in bull markets and low in bear markets" (cited in Karpoff 1987, p.112). More precisely it will be seen that transaction volumes are high and growing in bull markets and low and decreasing in bear markets. This conclusion holds for stock markets as well as for other speculative markets. The volume of shares traded on the NYSE between 1895 and 1940 is plotted in Fig.5.4a (solid line) along with the (deflated) Standard and Poor's index. Between 1895 ad 1914, there is a clear parallelism between volume and price (the correlation is 0.60). During the bull market of the 1920s, the yearly volume of shares traded increased from 0.2 billion to one billion; this five fold increase matched a multiplication by three of the price index (the correlation is 0.76). Similarly, Fig.5.4b shows that, during the bull market of the 1980s and 1990s, there was a graphic connection between trading volume and share

Contagion of speculative frenzy

99

prices: the volume increased from about 10 billions to more than 100 billions, a ten fold increase which matched an eightfold increase of the price index (the correlation is 0.98). The huge increase in the volume of transactions is direct proof of the fact that there has been a permanent inflow of new customers into the market; that inflow is itself a manifestation of a widening public interest for stock investment. One may wonder if this effect was accompanied by an increase of the funds each customer was prepared to invest in the stock market. The data in Table 5.2a,b give some indirect information in that respect. Thus it appears that the number of transactions of more than 5,000 shares (i.e. more than 250,000 dollars if one takes the price of a share to be of the order of 50 dollars) has markedly increased between 1975 and 1985. This period saw the creation of huge investment funds and the entrance in stockmarkets of pension funds and insurance companies. It would be of great interest to know if the progression has continued after 1992, unfortunately no such data are available after this date.

2.1.2

Property markets

Fig.5.5a concerns sales of land in Paris. The solid line shows the number of transactions while the broken line gives the price per square meter. During the 1982-1991 bull market the two curves were almost parallel (correlation is 0.95). The second example (Fig.5.5b) concerns the property market in Paris intra-muros. Again the curves are almost parallel in the upgoing phase. Then, during the 1991-1997 bear market the number of transactions fell more abruptly than the price. In this case price and transaction statistics are also available at the more detailed level of the 20 districts (the so-called "arrondissements"). This gives the possibility of a more accurate test as will be seen in the next paragraph.

2.2 2.2.1

Relationship between trading volume and price variations Methodology

Our main concern in this part is to obtain regression results which can validly be compared for various speCUlative markets. For instance, a regression of the form p = av + b where v stands for volume and p for price would lead to very different results for stock and property markets for the simple reason that the orders of magnitude of the variables v and p are different. A remedy is to work with scale invariant variables. Possible regressions which satisfy this condition are:

Chapter 5

100

I

I9811

i.. III

I ... '7GI

1GB to

•

J is

JaG

7t

II

j

~

.

loe M

40

70

60

51

11195

Fig.5.4 a Stock prices versus trading volume on the NYSE (1895-1940). Broken line: annual average of the Standard and Poor's index of common stocks (deflated), right-hand scale; solid line: yearly trading volume. The NYSE was closed from 1 August to 10 December 1914 which is why the 1914 volume figure was omitted. The correlation is 0.60 for the period 1897-1913 and 0.76 for 1915-1940. Source: Historical Statistics of the United States (1975).

I ~

f !:l ~

I. " II 70

" 51

.a • II

I

I

.'.' ..•..' .'.'

.'

......

--

.................................."

31

21

,..

1000

I-

I- I,.. I'"

1m

I'"

-

III

I'"

Fig.5.4 b Stock prices versus trading volume on the NYSE (1980·1998). Broken line: annual average of the Dow Jones index (deflated), right-hand scale; solid line: yearly trading volume. The correlation is 0.94. Source: Statistical Abstract of the United States (various years).

101

Contagion of speculative frenzy

Table 5.2a Distribution (by size) of transactions on the New York Stock Exchange, 1975-1985 (percent)

75

76

77

Under 100 shares

78

3.0 2.9 2.7 2.3 100-900 shares 42 38 33 30 1,000-4,900 shares 31 32 32 34 5,000-9,900 shares 8.7 9.7 10 10 Over 10,000 shares 18 20 24 25

80

81

82

83

84

2.2 1.8 27 25 34 31

1.6 21 31 12 36

1.3

1.2 15 27

1.0 0.9 11 11 25 24 14 14 50 51

79

11

27

11

32

16 28 13 42

13

46

85

Notes: These data are not available before 1975; moreover, the data for the period after 1985 involve only four size classes. Source: Statistical Abstract of the United States (various years).

Table 5.2b Distribution (by size) of transactions on the New York Stock Exchange, 1920-1992 (percent)

1920

1970

1975

1980

1985

1990

1992

Under 100 shares

25

5.9

100-4,900 shares Over 5,000 shares

(NA)

(NA)

(NA)

(NA)

3.0 73 27

1.8 57 43

0.95 35 65

0.71 35 65

0.87 33 66

Notes: There seems to be a general trend for larger transactions; this movement was particularly strong between 1975 and 1985. The average price of a share is comprised between 50 and 100 dollars which means that 5,000 shares represent an amount comprised between 250,000 and 500,000 dollars. Source: Statistical Abstract of the United States (various years); for 1920: Meeker (1922).

102

Chapter 5

I 311

~

I]

31

..................

Salol PrIce

/1

311

20

,

10

8 7

III

41

"II

5

.

"

4

Fig.S.S a Price ofland in Paris versus number of transactions. Broken line: annual price of land, right-hand scale; solid line: number of transactions. The correlation is 0.94. Source: Le Marche Immobilier Franfais (1993).

............

.....• .-

Salol

•...•..

Price

.

40

....::..

. .'.' '

......

20

......... ...•...•

....

......

...........

...................... 30

10 9

8 7

1998

Fig.S.S b Price of apartments in Paris versus number of transactions. Broken line: average annual price of apartments per square meter (deflated), right-hand scale; solid line: number of transactions. The correlation for the upgoing phase is 0.70. Source: Chambre des notaires de Paris.

Contagion of speculative frenzy

lnp

t::.p/p t::.p/p

= =

=

a1lnv + b1 a2 lnv + b2 a3 t::.v/v + b3

103

(1) (2) (3)

Case (3) must be excluded for it does not lead to robust regression results. This is not surprising for we know that the volume of sales is a highly erratic variable and taking its first difference generates a large amount of noise which precludes any significant regression. For case (2) one must examine separately the upgoing and downgoing trajectory. The reason is simple; suppose that the price peak: under consideration is very sharp, then t::.p/p is positive (and large) just before the peak: and it becomes negative (and large) just after the downturn; as a result those two points can hardly belong to the same regression line. The main results of the investigation are summarized in Table 5.3a,b and discussed in the following paragraphs. 2.2.2

Relationship between logarithms of volume and of price

From Table 5.3a we see that during a speculative episode volume and price increase (or decrease) together. For stocks, volume changes are somewhat stronger that price changes while for property the two variables change almost at the same rate. On the whole, one can keep in mind that during a price peak: the changes in the two variables are roughly parallel. For the property market in Paris detailed data are available for the 20 districts and this enables us to perform an interesting robustness test of the above price/volume relationship. Fig.5.5c provides an overview of volume changes for a sample of 9 districts. We see that the behavior of the number of transactions is markedly different from one district to another in contrast to the price curves which are more or less similar. There is nevertheless a common feature: after 1990 all curves show a sharp decline. This simultaneous fall is quite remarkable especially if one remembers that in contrast for the prices the downturn did not occur everywhere at the same moment (the turning points are distributed over a period of about 2 years from early 1990 to late 1991). The relationship between price and volume in each district is given in Table 5.3b. Does the connection between volume and price level also apply to individual stocks? Yes and no. Yes, because if one compares shares that are only rarely traded and actively traded shares the latter clearly demonstrate a greater price

104

Chapter 5

Table S.3a Relationship between logarithm of volume (v) and of price (p): lop ajlnv + hI

=

Regression coefficient

Correlation

(aj)

1) Upgoing phase (bull market) Stock market, NYSE

1920-1929

0.67±O.1

0.98

Stock market, NYSE

1980-1996

0.61±O.l

0.96

Stock market, NASDAQ

1990-1999

0.75±O.1

0.99

Land market, Paris

1980-1991

1.07±O.23

0.95

Property market, Paris

1981-1988

1.l2±O.2

0.69

Stock market, NYSE

1930-1932

0.80±O.3

0.99

Property market, Paris

1990-1996

2) Downgoing phase (bear market) no significant relationship

Notes: All prices have been deflated. The results in this table show that volume and price increase (or decrease) together and at almost the same rate: the average value of aj is 0.84 while the coefficient of variation (cr/m) is 0.2110.84=25 %. The parameter b j is not given because its value depends on the currency unit used.

increase potential than the former. As an illustration one can consider the extreme case of a rapid growth company like Amazon.com: between June 1998 and May 1999 its price was multiplied by 10 and in January 2000 it had a tum overratio of 710 percent, that is to say seven times larger than the average tum over ratio of US stock markets (Washington Post 12 January 2000). On the other hand, once a company has become actively traded a high tum over does not necessarily guarantee the continuation of the price increase; thus between June 1999 and January 2000 the price of Amazon.com's share has remained stationary while its tum over ratio nevertheless remained very large. What are the implications of the connection between volume and price revealed by Tables 5.3a,b? It means that if for some reason further increase of one of the variable is barred then the progression of the other is likely to be broken too. In other words, in order to explain a price downturn it can be worthwhile to examine the factors which restrain the growth in trading vol-

Contagion of speculative frenzy

105

1.75

1.75

1.75

1.5

1.5

1.5

1.25

1.25

1.25

1

1

1

0.75

0.75

0.75

1980

1990

1980

1990

1980

1.75

1.75

1.75

1.5

1.5

1.5

1.25

1.25

1.25

1

1

1

0.75

0.75

0.75

1980

1990

1980

1990

6

1980

1.75

1.75

1.75

1.5

1.5

1.5

1.25

1.25

1.25

1

1

1

0.75

0.75

0.75 1980

1990

1980

1990

1990

1980

1990

1990

Fig.5.5 c Annual number of transactions in 9 of the 20 districts composing Paris intra-muros. The number of the district is indicated in the upper-right comer of each panel. Although the changes in the different districts are fairly diverse. there is one common feature namely the sharp drop after 1990. Sources: Chambre des notaires de Paris. I would like to express my gratitude to Muriel Chavret of the Chambre des notaires for her kind assistance.

ume.

2.2.3

Relationship between logarithm of volume and price variation rate

The main results are summarized in Table 5.3c. The dispersion in the results for the regression coefficient is larger (and the correlation lower) than for the results in Table 5.3a. This is hardly surprising because the replacement of the price by its first difference increases the level of noise. Note however

Chapter 5

106

Table S.3b Relationship between logarithm of volume (v) and of price (p): lup =ajlnv + hI for the Paris property market 1981-1988 District number

2 3 4 5 6 7 8 9 10

aj

Correlation

District number

a]

Correlation

1.34 1.14 2.04 0.86 1.69 0.76 1.51 0.24 1.29 1.16

0.78 0.74 0.94 0.27 0.67 0.34 0.33 0.22 0.89 0.96

11

1.14 1.24 0.82 1.79 1.17 0.38 1.58 0.90 0.62 0.79

0.93 0.74 0.93 0.96 0.39 0.22 0.93 0.89 0.74 0.94

12 13 14 15 16 17 18 19 20

Notes: For the sake of clarity we did not indicate the error interval for a] • The average value of a] is 1.12 while the coefficient of variation (aIm) is 0.47/1.12 = 42 %. As a rule the correlation is lower in wealthy districts (4,5,7,8,16) than in less costly districts (3,9,10,13,18,20). Sources: see Fig.5.5c.

that, contrary to Table 5.3a, the downgoing phase of the property bubble in Paris leads to a significant regression. The results in this table provides more detailed information about the shape of the price curve than Table 5.3a: a positive regression coefficient indicates that the price increase accelerates as the bubble develops.

2.2.4

Corners

The connection established in the previous paragraphs between volume and trade is only valid in the long-run. Both in the short- and long-run the price of an item is of course determined by the demand-supply equilibrium condition, but in the short-run there can be some spurious distortions. If for some reason nobody wants to sell, the price can go up in thin trading volume. This occurs for instance in the case of a comer, i.e. when somebody tries to get the control of the market by buying up all available supply and by holding

107

Contagion of speculative frenzy Table 5.3c Relationship between logarithm of volume and price variation rate: !J.p/p = a21nv + b2 Regression coefficient

Correlation

(a2)

I) Upgoing phase (bull market)

Stock market, NYSE

1920-1929

0.12±O.16

0.49

Stock market, NYSE

1980-1996

0.06±O.lO

0.29

Stock market, NASDAQ

1990-1999

0.20±0.lS

0.62

Land market, Paris

1980-1991

Property market, Paris

1981-1988

0.35±O.lO

0.47

Stock market, NYSE

1930-1932

0.40±O.02

0.99

Property market, Paris

1990-1996

0.17±O.46

0.52

no significant relationship

2) Downgoing phase (bear market)

Notes: The average value of a2 is 0.21 while the coefficient of variation «(J/m) is: 0.13/0.21=63%. All prices have been deflated. IJ.p/p denotes the annual price variation; this variable is of significance for in speculative markets it is the price increase which drives and fuels the purchases. In contrast in a flat-price market there is no incentive to trade. These results give us information about the shape of the price increase (or decrease) curve. For instance if we admit that the logarithm of volume increases steadily in the course of time, then a positive value of a2 tells us that IJ.p/p is larger at the end of the upgoing phase than at the beginning; in other words it means that the price increase accelerates. The fact that the correlations are lower in this table than in Table 5.3a was expected in so far as the differentiation of lnp generates additional noise.

up all new sales. Comers were fairly common in American stock markets in the 19th century. A more recent illustration is provided by the silver market between June 1979 and January 1980. At the Chicago Board of Trade the price of silver rose from 10 to 40 dollars per ounce while at the same time the trading volume, instead of increasing, dropped from 0.8 million contracts to 0.2 million2 . When the magazine "Fortune" (15 May 1980) reported this 2However at the New York Commodity Exchange the silver market was very active to the point that in January 1980 the Comex had to limit trading at 50 contracts per month for each

108

Chapter 5

strange phenomenon the journalist marked its astonishment by entitling the article "Where was everybody when prices set records?". A comer of that kind cannot last very long however, because holding huge inventories over several years would be very costly.

3

Why can a bull market not last forever?

Why can a bull market not last forever? Far from being a naive question this is in fact a fundamental problem in the study of speculative trading. In the following paragraphs we will see that, at least for stock markets, there is no obvious answer.

3.1

Is there an upper bound for the price level?

For property markets one can reason in the following way. In the mid-1990s the "normal" price of an apartment in Paris was about 2,500 euro per square meter; now assume that during a speculative episode this price is multiplied by five; at 12,500 euros per square meter even small apartments will become too expensive to buy for someone with an average revenue. One could therefore be tempted to reach the conclusion that prices cannot reach that level. There is a flaw in the reasoning however, because it does not take into account investors, that is to say people who buy not for their personal use but in order to sell with a profit some time later. For an investor the price level is basically irrelevant, only the profit matters3 . In other words for a market dominated by investors there is no "natural" upper price bound. In Paris around 1995 only about 20 percent of the transactions were attributed to investors (La Vie Fran9aise, 18 April 1998) which means that price increases were indeed restrained by the average income of non-investors. Observation indeed confirms that price increases during property bubbles never exceed 300 percent; the fivefold price increase that we assumed above had never occurred. Would it be the same in a market where investors represent 70 or 80 percent? Only observation can provide an answer to that question. Stock markets provide the closest approximation to a "pure investor" market and one would therefore expect that there is no ''natural'' upper price limit on such markets. Is this confirmed by observation? Before one can answer that question one must discuss how stock prices should be gauged. The price of investor (Figaro. 2 February 1980). 3 However from a global point of view there can be macroeconomic constraints: more loans means a higher demand for money which in turn leads to a rise in interest rates.

Contagion of speculative frenzy

109

a stock has no meaning in itself; when it becomes larger than (say) 150 dollars a split is usually performed which means that every shareholder gets two 75 dollars shares for each 150 dollar share. In that way the price of shares is more or less maintained in a range between 50 and 100 dollars. As an illustration, for Microsoft between September 1987 and April 1999 there were 8 splits. If one takes into account the splits one arrives to the result that the price of a Microsoft share has been multiplied by a factor 800 in that time interval. This could seem to be a huge increase but in fact it has no obvious meaning. Remember that a share represents a fraction of the asset of a company; for a company like Microsoft which experienced a very rapid growth in the 1980s and 1990s a 1999 share represented something fairly different than a 1987 share. The same phenomenon can occur in the property market for a rapidly developing city; a tenfold popUlation increase from 10,000 to 100,000 transforms a rural town into an important urban center; as a result one square meter of an apartment no longer represents the same good. Needless to say, a large city such as Paris cannot undergo such a drastic transformation even in several decades. If the price of a stock cannot be used in order to characterize the growth of a speculative bubble, is there another variable which can be used for that purpose? The price earning ratio (PER) which we have already mentioned in previous chapters comes to the mind as an obvious candidate. From 1871 to 1995 the PER of all US stocks averaged about 14. At their peak in December 1972 the hot growth companies of the 1970s including IBM, Xerox, McDonald's or Polaroid had an average PER of 41; the priciest company, namely Polaroid, had a PER of 95. Despite the astonishing price increase of Microsoft stock its PER was only around 51 in late 1999. In contrast, the average PER of the companies listed on the NASDAQ neared 100 in late 1998 and about 200 at year-end 1999, a figure 14 times larger than the long-term average (Washington Post, 12 January 2000). Some companies had an even larger PER; thus in early April 2000 the PER of Yahoo Inc. was equal to 1601. While the average PER ratio of the 30 companies composing the Dow Jones Industrial index was around 28 in September 1999, there were also some smaller companies on the NYSE with a high PER; thus the PER of SK Telekom was 173. As a matter of fact, one can wonder if the PER is really an adequate measure for rapidly growing companies. Consider for instance the company Amazon.com (retail Internet commerce); with its capitalization of 24 billion dollars this is by no means a small company; yet, by September 1999, instead of earnings it had cumulative losses of almost 600 million dollars. If it had posted nil or negligible earnings its PER would have been infinite; with negative earnings the

Chapter 5

110

situation is even worse but one is unable to compute a significant price earnings ratio. A possible remedy would be to consider expected earnings instead of past earnings, but the PER would then become a fairly subjective variable. To sum up that discussion one can conclude that: • By current standards (e.g. stock price or PER) there does not seem to be a "natural" bound for stock price increases: a multiplication by a factor 10 or more is not uncommon. In some cases prices or even PERs are multiplied by a factor larger than 100. • It is not easy to define a variable which would reliably and unequivocally represent the price level of a stock. If there is no upper bound for stock prices is there an upper bound for trading volume? The answer is "yes" as is shown in the next paragraph.

3.2

Is there an upper bound for the volume of transactions?

Consider a market where the average price increases at a constant (annual) rate a, assume further that there are N investors (and no users) which buy/sell an average number of L items at an (annual) average frequency f. Under such conditions the trading volume will be: V = NfL. Thus for V to increase, either N, for L must increase. But f is limited by transaction costs; indeed, if we assume them to represent t percent, in order to make a profit the buyer must keep the item at least during", years where", is determined by the inequality (", can be smaller than 1):

In other words:

f

= 1/",

:s a/to Thus: a

V
(3.1)

In the following paragraphs we examine a number of illustrations and applications of that equation.

3.3

Applications

We can interpret formula (3.1) in two different ways. • One can assume that at any moment investors indeed trade as often as permitted by the constraint", ~ t/a that is to say at a frequency fmax = a/t; in that case V is all the time equal to Vm and formula (3.1) gives a estimate for actual trading volume.

Contagion of speculative frenzy

111

• On the other hand one can also suppose that in "normal" times investors take a substantial margin by keeping the items they have bought longer than the required minimum time span. Thus, in the first years of a speculative episode they may trade at a frequency (1/2)fmax or (1/3)fmax and come to trade at frequency fmax only in the last phase of the speculative episode when the growth potential due to N or L is exhausted. Under such an assumption formula (3.1) gives an estimate of actual trading volume only for the last phase of a bull market. Only observation can decide which of these interpretations is correct. In the following paragraphs we apply formula (3.1) to property and stock markets.

3.3.1

The Paris property market

In a general way, because of fixed costs, transaction costs are a decreasing percentage of the price; this is true for property as well as for stocks. For the property market at Paris the transaction cost is of the order of t = 10 percent for a 2- or 3-room apartment. Furthermore, during the 1983-1990 price peak the average annual price increase was: a = 7.8 percent. If one assumes that equality holds in 1}a ~ t one gets 1} = 10/7.8 = 1.3 year, which means that an investor will wait a little more than one year before reselling a purchased apartment; this is a reasonable order of magnitude. Equation (3.1) becomes: 7.8

Vm = lONL = 0.78NL

(3.2)

Fig.5.5b tells us that the number of sales reached a maximum of 40,000 per year, but our reasoning only applies to transactions made by investors which we know to represent 20 percent of the total. Thus, equation (3.2) gives: N L = 40,000 x 0.2/0.78 = 10,260. Let us see if this order of magnitude is plausible. Around 2000 there were in Paris some 2,400 real estate agencies 4 • However, one cannot identify the number of agencies with the number of investors for agencies often play only a middle man role: they bring buyers and sellers into contact and are renumerated by a fixed percentage of the selling price. If we nevertheless accept N = 2,400 as a crude approximation, one gets L = 4.2 which means that each investor sells 4 to 5 apartments a year, which is a fairly plausible order of magnitude. 41 am grateful to Mr. Sylvain Marion of the Federation Nationale des Agents Immobiliers for pointing that figure to my attention.

Chapter 5

112

3.3.2

Stock markets

As a second application of equation (3.1) we compare the bull markets in the 1920s and in the 1990s. In 1929, Y29 = 109 shares/year while in 1997: V97 = 130 109 shares/year. How can one explain that huge increase in trading volume? In the two cases the annual (deflated) price increase was of the same order of magnitude, namely: a ~ 10 percent. If we assume that the trading volume is given by Vm , equation (3.1) leads to:

(LN) t

= 130 97

(LN) t

29

The Tables 5.2a,b give some information about the evolution of L, the average number of shares per transaction. Unfortunately, because the range of the last interval, i.e. over 5,000 or 10,000, remains unbounded it is impossible to compute an average from these data. Moreover, for the period 1929-1975 we do not have any information at all. Thus one is reduced to making an educated guess; perhaps: L97 / L29 '" 5 would not be too far off the mark. This leads to: (

~)

97 -

26 (

~)

29

Leaving for a moment the question of the transaction cost t aside (it is treated in the next paragraph), we see that the crucial variable in the above relation is the number, N of investors. It is precisely in order to estimate the increase rate of this variable that we analyzed (in the first section of this chapter) the way a speCUlative bubble spreads. Tables 5.4a,b show that between the 1920s and the 1990s the diffusion of stocks into the American society has been multiplied at least by a factor 4. This sti1lleaves a factor 26/4 =6.5 which remains unexplained. In the next paragraph we will see that the reduction in transaction costs can account for that factor.

3.4

Empirical evidence about transaction costs on stock markets

One of the main problems when one tries to confront equation (3.1) with observation is the fact that transaction costs are known only with great uncertainty. In this paragraph we have collected the few empirical data that we could finds. 51 express my sincere gratitude to Professor Richard Sylla for communicating me some relevant data and references.

113

Contagion of speculative frenzy Table 5.4a Number of American people in possession of shares or participations in investment funds 1952

1970

1975

1980

1995

6.5 4%

30.8

25.3 12%

29.8

56 22%

Number (millions) In percent of population

15%

13%

Notes: In 1970 the American stock market was at its peak; the year 1975, in contrast, marked a trough after which the market began slowly to recover; not surprisingly, the diffusion of stocks in the population paralleled the evolution in stock prices. A percentage of 22 percent with respect to total population means that well over fifty percent of American households were share owners around 1995, either directly or indirectly. That evolution began almost a century ago; thus, between 1910 and 1920 the numbers of shareholders of United Steel and American Telephone and Telegraph (ATT) stocks have been multiplied by 4.7 and 3.4 respectively. In Germany at the end of 1999 about 13 percent of the population was in possession of shares or participations in investment funds; in former East Germany the proportion was only 3.3 percent but it had been multiplied by five since 1992. Sources: Quid (p.2206), Statistical Abstract of the United States 1999, Meeker (1922), Le Monde (10 April 2000).

Table 5.4b Number of stock and bond salesmen

Number (million)

1900

1910

1920

1930

1940

1950

1960

1970

4

6

11

22

18

11

29

99

Note: There was a peak in 1930 at the height of the bubble. Source: Historical Statistics of the United States (1975, p.142).

114

Chapter 5

From 1792 to 1975 the New York Stock Exchange had rules fixing minimum commissions; for instance not less than 12.5 cents per share for shares selling in the 10-15 dollars range, which means that the commission represented about 0.7 percent; and not less than 30 cents per share for stocks selling in the 200-250 dollars range which represented a transaction cost of about 0.1 percent (Meeker 1930). Actual transaction costs were substantially higher however. A reasonable order of magnitude is 0.2 percent; for so-called odd lots, that is to say transactions of less than 100 shares, the cost could reach 1 percent. As a result of the deregulation that was introduced in the early 1980s there are no longer any fixed minima; commissions are subject to all kinds of competitive forces which means that the investor must "shop around" to identify the best deals. The general introduction of computer on line transactions in the 1990s has reduced transaction costs even further. The evolution is schematically represented in Fig.5.6. One can note that on European markets the transaction costs were notably higher than in the United States. For instance in the 1970s at the Paris exchange they were around 1.5 percent per transaction; even after deregulation and the shift to electronic transactions they remained fairly higher than in the United States. Basically one would expect that, everything else being equal, if transaction costs are lower trading volume will tend to increase at a faster rate. The NASDAQ market being an automated market one can expect it to have lower transaction costs than the NYSE which is not. Hence one would expect a larger increase in trading volume on the NASDAQ. This is indeed confirmed for the period 1980-1990 (see Table 5.5). The period 1990-1998 has seen the generalization of on line stock trading, an evolution which has implied larger transaction cost reductions on the NYSE than on the already automated NASDAQ. Thus the results in Table 5.5 are in fairly good qualitative agreement with the predictions of formula (3.1).

3.5 Implications of the formula for the upper bound of trading volume In the previous paragraphs we have illustrated formula (3.1) through various numerical applications but we did not yet really explain how it can be used to forecast stock price downturns. First let us remember that its derivation implied the following steps and assumptions (i) It relies on the empirical observation that during a bull market there is a strong correlation between trading volume and stock prices. (ii) Whereas there is no ''natural'' upper bound

Contagion of speculative frenzy

Paris New York

10

10

·1

-

t . ~~. .-.-.. .- -

115

+. . . . . . . . . . . .+.1-

•••

too,

..

'

I

.2

1900

1920

1940

1960

1980

2000

Fig.5.6 Transaction cost for buying or selling shares. There was usually a substantial difference between official transaction costs as fixed by the exchange and the rates actually applied. Before the deregulation there were (at least officially) no quantity discounts. The solid line gives an estimate of the cost in New York for an order of about 5,000 shares. For odd lots that is to say orders of less than 100 shares the cost can be substantially higher because the trader must lump together several odd lots before being able to put them on the market. For orders over 5,000 shares the commission in the late 1990s was of the order of 1 cent per share. In this connection let us recall that around 1992 orders of more than 5,000 shares represented 66 percent of transactions. The dotted line shows the transaction cost on the Paris exchange. Sources: Meeker (1922), Blume et al. (1993), Annuaire des valeurs admises a la cote officielle de la Bourse de Paris (1924), Congres international des valeurs mobilieres (Paris 1900), Internet sources for present day figures. I would like to express my gratitude to Etienne Kah and Richard Syllafor their explanations regarding the Paris and American figures respectively.

for the price level, increase in trading volume has to be fueled by profit expectation which imposes a limit on sale frequency. (iii) Apart from the sale frequency, the other parameters in formula (3.1) are of a structural nature (at least under "ordinary" circumstances) which prevents them from changing rapidly. (iv) The shift from bull to bear market has to be brought about by endogenous factors; crashes triggered by a domino effect as for instance in the South Asian stock market crisis in 1997 are to be excluded for they are obviously triggered by exogenous factors. In order to eliminate that difficulty one can concentrate on the NYSE or the NASDAQ; being the world's largest they influence other markets more than they are influenced by them. Observation shows that during a bull market the price increase rate a is relatively constant. In order to use formula (3.1) for forecasting purposes the

Chapter 5

116

Table 5.5 Incidence of the level of transaction costs on trading volume

NASDAQ NYSE

1980-1990

1990-1998

2.7 1.2

1.0 1.2

Notes: The table gives the ratio of the increase-rate of trading volume to the increase-rate of the price. According to formula (3.1) this ratio should be equal to the product NUt where N is the number of investors and L the average number of shares traded in one transaction. If we assume Nand L to be similar on the NASDAQ and on the NYSE, the determining factor is the transaction cost t. In the period 1980-1990 one can reasonably assume that transaction costs were smaller on the automated NASDAQ than on the non-automated NYSE; this is reflected in a higher ratio on the NASDAQ. Then, in the 1990s, with the introduction of on line trading, transaction costs on the NYSE and on the NASDAQ become more and more similar.

main problem is to get adequate estimates for the increase potential of N and L and for the reduction potential of transaction costs. The number of investors (N) is determined by sociological constraints and by law; for instance a crucial question is whether or not life insurance and pension funds are allowed to invest their assets (or part of it) in stocks. The average number of shares traded by each investor (L) is determined by revenue levels and the availability of loans. Once the growth potential of these factors is exhausted the increase in trading volume is stopped and is soon paralleled by a leveling off of the price level. The subsequent drop can follow closely or after a period of seesaw oscillations.

PART III

REGULARITIES IN SPECULATIVE EPISODES

Chapter

6

Peak amplitude: the price multiplier effect

From a microeconomic perspective the crucial question regarding the amplitude of speculative peaks can be formulated in following terms. Suppose you have bought diamonds worth one million euros two years ago, assume that in the meanwhile their price has risen by 360 percent, but that in recent weeks there was a sudden 25 percent fall. Will you sell, or rather keep (or even augment) your holding in the hope that the price will soon go up again? In the second case and provided many investors react in the same way, speculation will be fueled and prices will be pushed to even higher levels. In the first case, on the contrary, prices will continue to fall eventually leading to the bubble's implosion. In short, the attitude toward risk plays a crucial role in the dynamic of a speculative episode; conversely, comparing various price peaks provides a means for exploring the attitude of investors when confronted to crucial investment decisions. This effect belongs to the category of what in previous chapters we called hidden collective factors for the investor's behavior will be heavily influenced by the attitude of investors around him. Probably he/she does not himself know in advance how he/she will react in such critical situations where high stakes are at risk. In this and the next chapter we explore price peaks in a attempt to unravel some key factors in the behavior of speculators. In the present chapter we restrict ourselves to the amplitude of the peaks whereas in the next chapter we analyze their shape. The first study will lead us to what will be called the price multiplier effect, while the second will disclose what will be called the sharp peak - flat trough pattern. Considering the number of papers that have been devoted to that question in recent decades it could seem that attitude toward risk has been thoroughly inB.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_6, © Springer-Verlag Berlin Heidelberg 2009

117

118

Chapter 6

vestigated by economists. This question is related to the more general framework of expected utility theory and to the controversy about the so-called Allais paradox, after the name of Nobel prize-winner (1988) Maurice Allais. However, most of the pUblications on these topics are either purely theoretical or based on "experimental" data by which one means data provided through answers to questionnaires, a methodology pioneered by Allais in 1952. A subject will for instance be asked to choose between the following options (Conlisk 1989): (i) Certainty of receiving one million dollars (ii) 11100 chance of receiving nothing, 89/100 of receiving one million dollars, 10/100 of receiving 5 million dollars. Do such experiments provide a reliable model for speculative behavior? Several objections can be raised such as for instance the following. • It is not easy to perceive intuitively a probability as low as 11100; personally I would be very embarrassed to choose between the previous options and my choice would certainly not be very "robust" anyway, in the sense that I might well give another answer one week, one month or one year later.

• The results of such experiments may be biased by the fact that they are often conducted with students who may know the various theories of choice under uncertainty which are supposed to be tested. Remember in this respect that in the experiment performed by Allais in 1952 the questionnaires were addressed to the participants of a Symposium organized in Paris on the question of choice under uncertainty, all of whom where brilliant economists or mathematicians (e.g. K. Arrow, M. Frechet, M. Friedman, E. Malinvaud, P. Samuelson just to mention of few of them) who knew very well the various theories in competition. • The situation of the person who answers a questionnaire can not even remotely be compared with the actual situation of an investor. It is true that in order to make the game more realistic, money awards are sometimes granted to participants. The fact that these amounts are small compared to the million dollars mentioned in the question certainly makes a difference, but the most crucial aspect is the fact that the participant knows that he/she cannot lose a substantial fraction of his/her personal wealth. For an investor there is no safety net and that is probably the crux of the matter. • In real life the probabilities of a return or loss are never specified with such precision as in the above options. In fact, even the issues themselves may not be clearly defined, a circumstance which greatly contributes to making decisions difficult and hazardous. In short, even with able and unbiased subjects it is not clear how the results of such experiments are connected with actual investors' decisions. Readers

Price multiplier effect

119

interested in this approach will find more details, along with additional references in Harless et al. (1994), Hey et al. (1994), Machina (1987), Shiller (1990)1 . The paper by M. Machina provides a survey of major theoretical approaches. Shiller's paper is based on questionnaires sent to traders and investors after the stock market crash of 1987 in order to find out what motivated their decision. This is certainly a more realistic investigation than the aforementioned procedure, yet it assumes that individual traders are perfectly aware of the whys and wherefores of their decisions; in short, it tends to underestimate collective factors. The chapter proceeds as follows. The property speculation in Britain between 1984-1994 provides a case in point on which we explain the main ideas introduced in this chapter, namely the contagion of speculation from high-growth to low-growth sectors and the resulting price multiplier effect. In the second section the price multiplier effect is generalized to other markets, and in particular to the markets of postage stamps or antiquarian books. In the third section we discuss an implication of the price multiplier rule for the behavior of ensemble dispersion. Finally, we show how the price multiplier effect can be reformulated in an alternative form which can also be applied to stock prices.

1

A case in point: property speculation in Britain

The property speculation episode in Britain in the period 1984-1994 will serve as an introductory example for the presentation of the price mUltiplier mechanism. Fig.6.1a depicts the evolution of house prices in the United Kingdom. Each curve corresponds to one of the 12 regions. Two facts are visible. (i) The higher the price at the beginning of the bubble, the stronger the price peak. (ii) Regions where speculation was stronger peaked first; in contrast, for regions where speculation was weak the maximum of the curve occurs with a time-lag which is of the order of two years for the lowest peaks; note that the lowest curve which corresponds to Northern Ireland has no real peak whatsoever and should therefore not be included in this discussion. In what follows we discuss these two points in greater detail. 1 I am indebted to Professor E. Malinvaud for pointing to my attention many of these references.

120

1.1

Chapter 6

The price multiplier effect

We denote by PI the initial price at the beginning of the speculative episode and by P2 the price at the peak; the ratio A = P2/PI will be called the peak amplitude. Fig.6.1b analyzes the relationship between A (vertical scale) and the logarithm of PI (the horizontal scale is logarithmic). The coefficient of linear correlation between A and In PI is equal to 0.59 (the confidence interval at probability level 0.95 is 0.02 to 0.87). A log-linear regression leads to the following adjustment: A

= alnpI + b

a

= 1.13 ± 0.96,

b = -3.12 ± 0.16

For reasons to be explained later on the price PI is expressed in euros of 1st January 1999. The speculative mechanism which leads to such a relationship is what we call the price multiplier effect. It will be seen subsequently that a relationship of the same kind holds for other speculative items; as a rule speculation is stronger (and the peak amplitude higher) for those items which have a high initial price. The fact that speculation is stronger when the stakes are higher can be seen as a risk affinity effect. In that sense it has to some extent a connection with a rule which has been observed in lotteries and according to which gamblers are more attracted by the prospect of winning a large prize with small likelihood than by a substantial chance of winning a modest prize (Rubner 1966). This observation may have something to do with the fact mentioned above that the human mind can hardly make a difference between a probability of 0.1,0.01 or 10- 6 . In the previous example since the range of the interval for PI is fairly narrow it is not obvious that a log-linear fit is markedly better than a linear fit. That point will become more obvious subsequently when we examine cases for which the range for PI is much larger. Before that, however, we analyze the spatial mechanism of the British speCUlative episode.

1.2

Spatial propagation of speCUlative fever

The British property bubble can be interpreted as a speculation which began in the London area and then spread to the rest of the country. Fig.6.2a,b support that interpretation. In Fig.6.2a the peak amplitude is represented as a function of the lag between peak times for different regions. Property prices first peaked in the area of London: the Greater London, East Anglia and South East regions can be considered as the source, center and engine of the speculative process. From there, speculation moved to more peripheral regions while at the same time losing some of its momentum. It reached Scotland

121

Price multiplier effect 90 r - - - - - - - - - - - - - - - - - - - - - - - - - ,

Ii

80

:: .j!

70

~

...

j

60

~ ...

50

f

40

0---=--

1992

Fig.6.1 a House prices in Britain during the speculative episode of 1984-1994. Each curve represents the price in one of the 12 regions composing the United Kingdom. The two highest curves correspond to "Greater London" and "South East". The circles indicate the positions of the maxima; the lower the maximum. the longer the time lag with respect to the leading regions. Note that the maximum of the lowest curve which corresponds to Northern Ireland is obviously spurious for this curve has no maximum. Source: British property data are collected and published by the Halifax group (West Yorkshire, England). I am most grateful to the people at Halifax/or their kind assistance.

i1

2.4 EAST ANGLIA

2.2 2 1.8 YORKS.HVM

1.6 1.4 SCOTLANO

1.2 N.IRELAND

1 60

70

80

90

100

Price allhe slarl of Ihe bubble

Fig.6.1 b The price multiplier effect. Horizontal scale: prices expressed in thousandeuros of 1 January 1999; the vertical scale represents the ratio P2/PI of the peak price P2 to the price PI at the start of the bubble. The correlation is 0.59 (confidence interval at probability 0.95: 0.02 to 0.87). In following graphs the spurious peak for Northern Ireland will be left aside. Source: see Fig.6.1a.

Chapter 6

122

tl

2.4 EASTANUUA

2.2 2

GHA............ 1OIJt1I ......

SOIR'IIWIlS1'

EMIDIAN. . W.MIDLANDS

1.8

YCJaKS.lIUM

_nr......

w.....

1.6

-..

1.4Peak prtJJIGIGtio"

)

l.2

I

1-0.5

0

0.5

1

1.5

2

Fig.6.2 a Relation between the amplitude of the peaks and the moment in time when they occur. Horizontal scale: time interval expressed in years between peaks in different regions. The figure confirms the mechanism identified in Fig.6.1a: the price buoyancy starts in "East Anglia", "Greater London" and "South East" and from there it spreads to the other regions while at the same time losing some of its momentum. The correlation is -0.91. Source: see Fig.6.1a.

Fig.6.2 b Map of Great Britain showing the 11 regions.

Price mUltiplier effect

123

with a lag of 2.5 years and an amplitude which was only half the amplitude in the source region. Fig.6.2c illustrates the propagation aspect of the process: the horizontal and vertical scales respectively represent time and geographical distance. The slope of the regression line represents the speed of the speculative wave; it is equal to 16 ± 4 kmlmonth (correlation is 0.93). _700 ,---------------------------------------,

:!. i 600

.s

~~

SCOTLAND

"

: 500 --

~

i!l

400 f-

NORTH

WALES SOUTHWEST

NORTH WF.5T

3001YORKS,HUM

200 I-

W.MIDLANDS SOUTH EAST E.MIDLANDS

GREATER I..oNDON

100 I-

o 1-

EAST ANGUA

1

1

o

0.5

,

1

1.5

I

2

2.5 Peak delay (year)

Fig.6.2 c Relations between distances to London and time of occurrence of the peaks. The horizontal scale is the same as inFig.6.2a but the vertical scale represents geographical distance to London. The more a region is distant from the London area where the bubble started, the slower and weaker is its response to the price increase. The correlation is 0.93. The slope of the regression line, which is equal to 16 ± 4 km/month, gives the velocity of the price wave. Source: see Fig. 6. 1a.

Since the amplitude of the price peak is higher in East Anglia than in Greater London would it not be logical to consider East Anglia as the real center of the speculative process? If one draws a graph similar to Fig.6.2c under that assumption one gets a coefficient of correlation of 0.82, that is to say slightly lower than before; the speed is also somewhat lower: 12 kmlmonth instead of 16. This suggest that the real center of the speCUlation process is indeed Greater London and that the higher amplitude in East Anglia is merely a fortuitous circumstance.

1.3

Examination of other property bubbles

The 1983-1993 property bubble in Britain stands out as a particularly clear and suggestive example for a number or reasons.

Chapter 6

124

• There was only one speculative center; in the United States there would in general be at least three centers for property speculation, namely: the NorthEast, Texas and California. Similarly in Spain there would be two main centers, namely Madrid and Barcelona. When there are several speculation centers the resulting pattern is likely to be more complicated. • Because Britain's north-south extension is much larger than its eastwest extension we are confronted with a one- rather than two-dimensional problem as would be the case for a square-shaped country like France. • In property speculation there is a spatial interpretation which helps to better understand the propagation mechanism. In other cases (postage stamps, books, etc.) that we examine subsequently such an interpretation is not possible. Are there are other examples of property bubbles similar to the one in Britain? To answer this question we examine the cases of France and Japan. In these two countries there was a property bubble at about the same time as in Britain. Unfortunately no regional price series seem to be available in France; such series are published by the FNAIM (Federation Nationale des Agents Immobiliers, i.e. national association of real estate agents) since 1997 but before that date only estimated price ranges seem to be available. There is one exception, namely Paris for which detailed spatial price statistics are published by the Chambre des Notaires. These data will permit us to show that the same mechanisms are basically at work as in Britain. Fig.6.3 shows the price evolution in four of the twenty districts which compose Paris intra-muros. Prices at the beginning of the speculative episode are listed in the upper-left comer; the amplitude of the peak turns out to increase with the initial price, in other words the price multiplier effect again holds in this case. This is confirmed by studying the whole sample of the 20 districts; the regression reads:

A = InpI

+b

a

= 0.6 ± 0.5,

b = -1.7 ± 0.1

PI is expressed in January 1999 euros and the correlation is equal to 0.48. Do we also have the same spatial contagion mechanism as in Britain? We are here in a less favorable situation to observe that phenomenon firstly because the area is much smaller and secondly because only prices by semester (instead of quarterly prices) are available. Nevertheless Fig.6.4a discloses a similar phenomenon of spatial contagion even if the correlation is here markedly smaller (r = -0.21). For Japan Fig.6.4b shows a similar contagion process. The source of the spec-

Price multiplier effect

125

16th district (110 rr.)17th district (79 fr.) - 11th district (64 fr.) - 18th district (59 fr.) - -

Fig.6.3 Price of apartments in Paris by district during the speculative episode 1984-1994. Vertical scale: deflated price per square meter. The price multiplier effect is clearly visible in the fact that more expensive apartments are characterized by greater peak amplitudes. Source: Chambre des Notaires de Paris.

ulation seemed to be the commercial property market in the center of Tokyo. The bubble then spread to the whole city and to the residential property market in Tokyo itself and in Chiba prefecture which is located to the south east of Tokyo. In conclusion we can say that the contagion mechanism identified in the British case was also at work in the Paris and Tokyo property markets. In the subsequent section we examine whether the price multiplier effect also holds for other (non-spatial) speculative markets.

2

Generalization of the price multiplier effect

The price multiplier rule will be seen to apply to a great variety of speculative markets. The first paragraph contains a number of illustrative examples; in the second we examine the case of postage stamps; in the third we consider the market of antiquarian books; finally we summarize the evidence supporting the price multiplier effect and we discuss a number of apparent counterexamples.

Chapter 6

126

j

I !

3(!) 2.8

r-

®

2 r-

Fig.6.4 a Relationship between the amplitudes of the peaks and the moments in time when they occur. Horizontal scale: delay between peak times expressed in years. This figure parallels Fig.6.2a but due to much shorter distances the correlation (-0.23) is poorer. Source: see Fig.6.3a.

;:- 120 , - - - - - - - - - - - - - - - - - - - - - - ,

I1 i

.S

Tokyo center (comm.) Tokyo (comm.)

100

Tokyo South (resld.) Chlllll prefecture (resId.)

80

1 40

20

1984

1985

1986

1987

1989

1990

Fig.6.4 b Increase rate of land price in the Tokyo area. Chiba prefecture is located in the south-east of Tokyo. The figure suggests that the bubble started for commercial property in the center of Tokyo and then spread to neighboring places. Source: Aveline (1995).

Price multiplier effect

2.1

127

Illustrative examples

Fig.6.5a-d present the price multiplier effect for four different speculative markets. In each graph the two curves refer to items whose prices are markedly different; the upper curve always corresponds to the most costly item. The first figure concerns the price of one-carat diamonds; the upper curve is for a diamond of class D (the highest quality) while the lower one corresponds to a class G diamond. Before (and also after) the bubble the price ratio was about 2, but it became 2.4 at the peak. The graph in panel (b) describes the price peaks for two American coins. The upper and lower curves respectively correspond to a roll of 1955 half dollars and to a roll of 1960 small data pennies. E. Montroll (1974) from which we borrow that example notes that both coins were "great speculative items"; once again speculation was stronger for the most expensive item. The graphs in panels (c) and (d) concern postage stamps. For our purpose postage stamps are of particular interest because their value can range from a fraction of euro to several thousand euros. As we have already seen in chapter 2, during World War II in France there was a strong speculative bubble for stamps. Again we verify that the amplitude of the peak was largest for the stamp with the highest initial price: the 5,000 francs stamp displays an amplitude of 5.9 while the 9-franc stamp has an amplitude of 1.9. Panel (d) tells a similar story for British stamps: the amplitude is 4.9 for the 5,OOO-franc stamp against 2.2 for the 275-franc stamp. We also included in panel (d) a stamp for which there seems to be no speculati ve peak at all. As a matter of fact, stamp catalogs tell us that speculation only concerned stamps with a fairly large face value (close to one pound and over). One may wonder why. The price multiplier effect provides a simple explanation. Speculation in fact affected all stamps but (according to the price multiplier rule) in proportion to initial prices. For a stamp as inexpensive as 1.5 franc the expected peak amplitude is small to the point that the peak cannot be observed. In the next paragraph we examine the case of stamps in more detail. But before we provide a non-spatial example for apartments. Fig.6.6 shows the evolution of apartment prices in Paris during the property bubble 1985-1995; from bottom to top the curves correspond to bigger, and hence more expensive apartments. The regression reads:

A = lnpi

+b

a = 0.13

± 0.08, b =

0.62

± 0.07

PI is expressed in euros of 1st January 1999 and the correlation is equal to 0.83.

Chapter 6

128 Diamonds

American coins

1000 900

10,000 doUan

10doUan 4doUIII'S

800

700 600

500

:

.....

·j · ..~••. .. ·· ...•. ·:· · . .:.. •

400

300

200

+.

•

+

+

+

.. 1976

1978

1980

1982

1984

1962

700r-=---~---------.

600

French stamps

400

5,000 francs

500

1966

1964

British stamps

'francs

300

400

300 200 200

". ···..... ....· .

\

100 90

80 70

.

70 60 1940

1945

60

1950

1980

1990

Fig.6.S Speculative peaks for low- versus high-price items. In each panel the two figures under the title give the prices of the two items at the beginning of the bubble; the upper price always corresponds to the solid line. (a): Polished diamonds. Solid line: one carat, D clarity; broken line: one carat, G clarity; the price data are deflated prices on the Antwerp market. (b) Prices of American coins. Solid line: price of a roll of 1960 small date pennies; broken line: price of a roll of 1955 half dollars. (c) Prices of French postage stamps. Solid line: One of the most expensive French stamps (Ceres No 2, not postmarked); broken line: price of a fairly cheap stamp (Ceres No 16, postmarked). (d) Prices of British postage stamps. Solid line: One of the most eXQ~nsive British stamps (Yvert et Tellier No 90); broken line: medium priced stamp (Yvert et Tellier No 286); dotted line: price of a fairly cheap stamp (Yvert et Tellier No 106). Sources: Diamonds: Diamonds 1988, The Economist, Special Report No1126; Coins: Montroll et al. (1974); French stamps: Massacrier (1978); British stamps: Yvert et Tellier catalogs, various years.

Price multiplier effect

i

129

6+

i:::. 275 300

6-room - S-room - -

~ 250

4-room - 3-room - 2-room - -

.;

I

225

I-room - -

~200

l

t

~

175 150 125

100 I

1984

1986

1988

1990

1992

,

1994

1996

Fig_6.6 Price of apartments in Paris by size. Vertical scale: deflated price per square meter averaged over all districts. The amplitude of the peak is larger for bigger apartments. Source: 1984-1990: Conseils par des Notaires (No 117, 23 Dec. 1991); 1990-1996: Chambre des Notaires de Paris. 1 am most grateful to Delphine Slanoski and Christine Vauthier of the Chambre des Notaires for their kind assistance.

2.2

Postage stamps

Stamp catalogs provide the prices of all existing stamps. In countries such as Britain, France, Germany orthe United States such catalogs have been published annually for over a century and they constitute a unique source of information for anyone who wants to study either long term trends or mediumterm price fluctuations. However, since the prices given in catalogs are estimates made by experts and traders rather than real market prices, the question arises of whether these prices really reflect prices in actual transactions. This question which is of crucial importance for our purpose requires some explanations. Prices listed in catalogs are for stamps which are supposed to be in a state of perfect conservation. It is obvious, however, that a stamp bought several decades ago can hardly be in a perfect state. In other words its price will always be less than the catalog price. Nevertheless, there is a close connection between transaction prices and catalog prices. This was shown by Feuilloley (1996) on a sample of 300 stamps; the correlation was 0.90, while the regression coefficient was about 0.5 which means that on average actual prices were only half the prices listed in the "Yvert and Tellier" catalog. Since we are interest in the evolution of relative prices rather than in their absolute magnitude, catalogs indeed provide satisfactory price estimates.

Chapter 6

130

In 1978 A. Massacrier has published a book which proved very valuable in our perspective. It provides price series from 1904 to 1975 for all nineteenth century French stamps. Using Massacrier's work allows a considerable economy of work. The present study concentrates on the speculative episode that occurred in France during World War II. Although the "causes" of this bubble are only of anecdotal interest from our perspective one can mention the following. • During the war savings accumulated as a result of consumption restrictions; moreover between 1940 and 1945 the average annual inflation rate was 23 percent which provided a strong incentive to buy whatever valuables were available. • As a more circumstantial cause one can also mention the demand generated by a number of passionate collectors belonging to the German occupation troops or the need to sweep black market profit under the carpet. Using Massacrier's price source one can investigate the behavior of stamp prices during the war. For postmarked stamps the results are summarized in Fig.6.7. Each dot represents one the 57 nineteenth century stamps listed in Massacrier's work. The triangles represent average peak amplitudes for adjacent subsets of 10 stamps. As can be seen from the positions of the triangles there is no non-linear effect which means that a linear regression indeed provides an adequate fit; it reads:

A = lnpi

+b

a

= 0.26 ± 0.12,

b = 1.7 ± 0.3

PI is expressed in January 1999 euros and the correlation is equal to 0.49 (confidence interval at probability level 0.95: 0.26 to 0.66). A similar graph drawn for non-postmarked stamps shows a larger dispersion and, as a result, the linear fit is poorer. Detailed figures are given in Table 6.1. A possible explanation may be connected with the fact that for a given stamp the non-postmarked variety is usually substantially more expensive (often 5 to 10 times more expensive). Being rarer, these stamps are not traded very often which makes their price estimates more uncertain. In Britain between 1970 and 1980 the annual inflation rate was on average equal to 13 percent. As in France during World War II that double digit inflation rate strongly boosted stamp trade and lead to a speculative price peak (see Fig.6.5d). That bubble provides an other opportunity for testing the price multiplier law; the figures given in Table 6.1 show that the agreement is fairly good. We now turn to testing the price multiplier effect on the antiquarian book market.

131

Price multiplier effect Table 6.1 Price multiplier effect: A = alnp1 + b Item

Numb. of classes

a

b

Coef. of correlation

12

1.13 ±O.96

-3.12 ±O.16

0.59

I-room apartments

4

0.28 ±O.6

0.09 ±1.0

0.55

2-room apartments

4

0.27 ±O.24

-0.08 ±O.O5

0.84

3-room apartments

4

0.28 ±O.47

-0.l7 ±O.12

0.63

4-room apartments

4

0.27 ±O.12

-0.27 ±O.O4

0.95

I. Property A.Various locations 1) Houses in Britain (84-90) 2) Apartments in Paris area (84-90)

5-room apartments

4

0.43 ±O.30

-1.13 ±O.14

0.89

+6-room apartments

4

0.13 ±O.35

0.45

20

0.60 ±O.50

0.42 ±O.22 -1.7 ±O.l

6

0.13 ±O.O8

0.62 ±O.D7

0.83

France, WWII, postmarked

57

0.26 ±O.12

1.7 ±O.3

0.49

France, WWII, non postmarked

57

0.14 ±O.24

2.47 ±O.4

0.16

Britain, 1975-1995

13

0.39 ±O.6

2.3 ±O.6

0.80

US, 1927-1933, sample 1

10

0.62 ±O.O9

-1.6 ±O.l1

0.98

US, 1927-1933, sample 2

10

0.33 ±O.O5

-0.48 ±O.O5

0.96

0.38 ±O.l

-0.07 ±O.O6

0.69

Apartments (all sizes) in Paris

0.48

B. Various sizes Apartments in Paris II. Postage stamps

III. Books

Average

Notes: A is the amplitude of a peak, Pl is the price at the start of the bubble; all prices are expressed in euros of 1st January 1999. In (A-I) the various classes correspond to the 12 regions composing Britain. For the first 6 lines of (A2) the price classes correspond to the four departments (Paris, Hauts-de-Seine, SeineSaint-Denis, Val-de-Marne) composing the "lIe-de-France" region. For the "apartments (all sizes)" case, the classes correspond to the 20 arrondissements in Paris. In (B) the classes are apartment sizes. For postage stamps each stamp represents a price-class. For books the 10 classes are the price-deciJes.

132

2.3

Chapter 6

Antiquarian books

As a laboratory for the investigation of price bubbles the book market has a number of advantages but also an important drawback. That question was already discussed in chapter 2; let us just recall the main points. On the advantage side is the fact that in contrast to stamp prices, the prices recorded in book catalogs are not estimates but actual sale prices. On the drawback side is the fact that each book is so to say unique; even two books printed on the same day by the same printer may no longer be identical when proposed on an auction sale one century later. For instance one of them may contain an autograph which makes it much more valuable than the other. Even the simple fact that a book had previously belonged to some illustrious person is enough to give it an additional value. That circumstance compels us to devise a new methodology for the testing of the price mUltiplier effect. We explain that procedure on the example of the American book market in the 1920s.

2.3.1

The methodology

Each annual volume of "Annual book-prices current" published in the late 1920s contains approximately 9,000 price records. The first step consists in selecting a sample; as this can be done in a number of different ways, we must check whether our results depend upon the way the sample was selected. We used two kinds of samples: one based on selected authors, and another based on a random selection of books. For the first set we used the following authors: G. Byron, J. Galsworthy , H. Fielding. E.A. Poe and P. Shelley. In the late 1920s these were very popular authors; collectors were fond of books such as "History of Tom Jones", "A man of Devon" or "Prometheus unbound", just to mention a few of the most demanded titles. We supplemented this author sample by a "random" sample corresponding to the first five pages of each volume and which comprised approximately 70 prices. Thereafter, the expression sample 1 refers to the author sample supplemented by the random sample; it comprises about 220 prices. We have also used a second sample, thereafter referred to as sample 2, which comprised about 600 randomly selected prices. In a second step, for each of the years of the period under consideration, we computed the average price for the different sets in order to check if there was indeed a price bubble. For the period 1927-1932 the answer was positive: for each of the authors as well as for the random samples the average price went through a maximum in 1929.

Price multiplier effect

133

In a third step one had to distinguish different price intervals in order to test the price multiplier rule. To this aim we ordered each sample into a sequence of increasing prices and we formed sample deciles. Thus, if the sample comprised for instance 600 prices, the first decile contained the 60 lowest prices. At this point it was of interest to check whether within each decile the average price displayed a price peak. This was indeed the case: in each of the deciles the price went through a maximum in 1929. This means that even books which cost as little as about 10 dollars have experienced a price bubble; for instance the prices in the second decile jumped from an average of 11 dollars in 1927 to 14.5 dollars in 1929 and fell to 9.5 dollars in 1932. For the prices in the 10th decile the jump was of much larger amplitude: from 600 dollars in 1927 to 2,250 dollars in 1929 and 250 dollars in 1932. By plotting all the amplitudes as a function of the prices in 1927 one is lead to the graph in Fig.6.8. The figures for the correlation and log-linear regression are indicated in Table 6.1 for sample 1 and 2. The price multiplier effect is confirmed in both cases: the regression coefficient a turns out to be higher for sample 1, an observation which can be explained by the fact that the speculation fever particularly affected the most popular authors. For Byron and Fielding the price increase is particularly impressive: the amplitude is equal to 7 and 8 respectively as compared to 3.2 for Shelley, 2.5 for Galsworthy, 2 for Poe and 2 also for the random sample. It is interesting to observe that along with the price increase there was also an increase in the number of sales; the sales amplitudes are the following: Byron (1.5), Fielding (1.5), Galsworthy (3.8), Poe (1.1), Shelley (1.2).

2.4

Summary

In this paragraph we summarize the evidence brought to light in our previous discussion. But first of all we must consider the question of the unit of currency used in the regression equation. We have already indicated that we used "January 1999 euros" as unit of currency but we did not yet explain the argument on which such a choice is based,

2.4.1

Currency scale

The regression equation reads: A = a In PI + b; in order to make our results independent of the inflation rate all prices that we used were real (i.e. deflated) prices; but this does not settle the matter completely for, in contrast to A and a which do not depend upon the unit of currency used, the coefficient b is affected by that choice. Changing PI to K PI changes b to b + a In K. For

Chapter 6

134

6 f5 f-

.. . .

4 3

.. ...........A.············· . ............ ...... • jl... 1::1. It.

I--

-

•

2f-

A. .•...•...•...~. ••••••••••••.,

....A·c.

...~•••• L.l

:

1 -

~•••••

•

-.

••

•• •

n. .:

,I

,I'

1

10

••••••••

•

.-

Pric. at Ih. rtIuf o/Ib. bubblt

Fig.6.7 Price multiplier effect for French stamps during the speculative epi. sode of 1936·1946. The graph refers to 57 nineteenth century postmarked stamps, Each dot corresponds to a different stamp, Horizontal scale: price at the start of the bubble expressed in 1999 euros; vertical scale: peak amplitude for deflated prices, The triangles are average values for consecutive intervals of initial prices; they provide a control test for possible non linear effects. The correlation is 0.49 (confidence interval at 0.95 probability level: 0.26 to 0.66). Source: Massacrier (1978).

t 3: ~

•

3 2.52

1.5 -

Pric. allh••'art oJ Ih. bubblt (l9!19 .uros)

Fig.6.S Price multiplier effect for antiquarian books during the speculative ep· isode 1927·1933. The graph refers to auction sales in the United States, A sample of 220 books has been divided into sample deciles; each square corresponds to one decile, High price deciles give rise to price peaks of greater amplitude than low price deciles, The correlation is 0.98, Source: American book-prices current (19271933).

Price multiplier effect

135

the sake of comparability one must therefore express all prices in the same currency. Why did we select the euro of 1st January 1999 (J99-euro). In that matter one of the most important conditions is that the exchange rates with other currencies be defined with good precision. The J99-euro meets that condition fairly well. Floating exchange rates fluctuate continuously in the course of time; from one month to the next the fluctuation of the dollar-yen or dollar-euro exchange rates can reach as much as 10 percent; even daily changes can reach one or two percent. One of the advantages of the J99-euro is that since January 1, 1999 the exchange rates for the countries of the eurozone are fixed once and for all and with great precision; thus the franc-euro exchange rate is defined with a precision of 10- 5 . Selecting the euro instead of the dollar which would have been in some sense an obvious choice due to its importance in the world economy guarantees outstanding precision for all the countries in the eurozone and standard precision for other currencies.

2.4.2

Evidence

What conclusions can be drawn from the evidence summarized in Table 6.1 ? First it can be seen that the evidence is in good agreement with the price multiplier rule. A second important point concerns the order of magnitude of a. For stamps and books, a seems to be of the order of 0.4. For property prices considered in a spatial perspective the range goes from a = 0.28 in the Paris area to a = 1.13 for Britain. That last case is somewhat special however for it involves the propagation of a bubble over large distances; one would definitely need one or several cases of that kind in order to confirm that value of a. In conclusion the results contained in Table 6.1 can be summarized in the following statement:

Price multiplier effect

Consider a good (e.g. apartments) which is available in different quality (or size) varieties. During a speculative episode the price of each item jumps from an initial level PI to a peak-level P2 before approximately returning to level Pl. For each item the ratio A = P2/PI, referred to as the peak amplitude, is related to the initial price PI by a relationship of the form: A = alnpI + b The average value of the coefficient a is about 0.4.

Chapter 6

136

2.4.3 Possible mechanisms In economics it is fairly easy to imagine plausible mechanisms capable of "explaining" a given observation, especially if one contents oneself with qualitative arguments. For the price multiplier effect one can propose at least two mechanisms. • The first is the propagation mechanism already mentioned in the first section: speculation starts in an expensive area and then spreads to other, least costly, areas or items. • The second mechanism is based on a distinction between two types of operators (i) users who buy the item for personal use (ii) investors who buy for the only purpose of selling with a profit later on. As a rule users are discouraged by too high prices; on the contrary, for an investor the price makes little difference only profit matters. Consequently one can expect that for costly goods most transactions will be done by investors. The first mechanism will be studied more closely subsequently. The soundness of the second mechanism can be checked to some extent in the following way. In a separate paper (Roehner 1999) we estimated the percentage of purchases due to investors in different districts of Paris. On average it is known to be of the order of 20 percent (La Vie Fran~aise 18 April 1998) and it varies in a range from 10 percent in the 2nd district to 36 percent in the 15th district. For the 20 districts the regression between the peak amplitude A and the fraction of investors s reads: A = 1.0 s + 2, the correlation being equal to 0.28. Thus, the peak amplitudes are equal (though with a low correlation) to the proportions of investors. It would of course be of great interest to be able to perform similar tests for other situations of the same kind in order to confirm that result.

3

Implications and significance of the price multiplier effect

It is obvious that the price multiplier rule provides a useful guideline for speculation purposes. This is illustrated by the following example. Suppose that at the beginning of a price peak a real estate agent in Paris has obtained from hislher banker a loan of 2 million euros. In order to maximize hislher profits and assuming of course that he/she buys at the start of the bubble and sells at its peak, he/she should buy three 5-room apartments in the most expensive district rather than 20 one-room apartments in the eastern (i.e. less expensive) districts. In the first case the profit would be at least 2 millions euros

Price mUltiplier effect

137

(see Fig.6.6) while in the second it would be only 0.75 million, which makes a substantial difference. May be it is precisely because real estate agents are, consciously or intuitively, aware of that rule that the price amplitude of expensive items is larger. The price multiplier effect clearly is a self-reinforcing phenomenon. However, in this section it is not our purpose to examine the implications of the price multiplier effect from the perspective of profit maximization. Rather we want to examine if it can help us to understand other observations made during speculative bubbles. The first paragraph examines the behavior of ensemble dispersion during a price peak; in the second, we discuss the relation between the price multiplier effect and the phenomenon of diffusion.

3.1

Evolution of the ensemble dispersion during a price peak

First let us see how the question should be formulated.

3.1.1

The question

Consider the public auctions for antiquarian books in New York during the 1920s. Such sales took place about twice a week. In the previous section we have already used the ensemble average m for the prices of books. For the period 1927-1932 the evolution of this ensemble average is schematized by the curve in the upper panel of Fig.6.9a. Similarly we can define the standard deviation (J characterizing the ensemble dispersion of book prices. In what follows we consider the following question: what is the behavior of the coefficient of variation (JIm in the course of time? Three possible answers are schematized in Fig.6.9a. Option 1 corresponds to a dispersion which, in the upgoing phase, increases faster than the average price, and similarly decreases faster in the downgoing phase; in option 2, (J changes at the same rate as m; finally in option 3, (J changes less than m. Which of these eventualities is actually observed?

3.1.2

Observations for book and property markets

The evidence given in Table 6.2 and Fig. 6.9b,c shows that for book and property markets it is the first scenario which applies; in other words the dispersion increases more than the average price. For the book market (Table 6.2) we can test several samples, but since the duration of the peak was short the precision with respect to time is poor. We can complement that case by analyzing the 1984-1994 property bubble in Paris; the result is shown in Fig.6.9b.

Chapter 6

138

aIm, option 1

aIm, option 2

Fig.6.9 a What is the behavior of the ensemble coefficient of variation during a price peak? The curve in the upper panel shows the evolution of the average price (m) for a sample of books. The coefficient of variation (0" 1m) is defined as the ratio of the standard deviation 0" to the mean m; both 0" and m are ensemble characteristics; they may for instance refer to prices at book auction sales. There are basically three possibilities: (i) In scenario 1 the dispersion increases (and after the peak decreases) faster than the average price. (ii) 0" and m change at the same rate with the result that the coefficient of variation is stationary. (iii) Option 3 is the opposite of option 1. Observation shows and theory confirms that whenever the price multiplier effect applies the behavior of the coefficient of variation follows the first pattern.

139

Price multiplier effect Table 6.2 Evolution of ensemble dispersion during a speculative episode Sample

(cr/m) initial

(cr/m) peak

(cr/m) trough

1927

1929

1931

1.3

1.7

2.19 (1928)

3.0

0.9 2.95 (1930)

Byron

2.0

3.0

l.5

Fielding

2.8

4.4

1.4

Galsworthy Poe

0.5 l.9

2.1 3.7 (1928)

1.2 2.0

Shelley

2.3

4.3

2.2

First random sample (70 books) Second random sample (600 books)

Notes: The table shows that during a speculative episode the standard deviation of a sample of items varies (increases or decreases) faster than the average price. That result is a direct consequence of the price multiplier rule.

Due to a longer duration of the price peak, we can observe that the ensemble coefficient of variation already peaked in 1987, that is to say about three years before the price peak. A similar result is shown in Fig.6.9c for the property bubble in Britain; in this case the coefficient of variation peaked 9 months before prices. Before we explain how the behavior of the ensemble dispersion can be derived from the price multiplier rule, let us put the previous observations into perspective by contrasting them with a quite different case.

3.1.3

Observations for wheat markets

In the 19th century there were several hundreds grain markets in Europe and this provides an excellent opportunity to observe dispersion results for sets of markets. Will the coefficient of variation also increase during a price peak? In order to answer that question we consider two observations, one in Bavaria and the other in France. Between 1815 and 1820 there was a huge peak for wheat prices in Bavaria (see Fig.2.1b). As a matter of fact, there was a similar (though less sharp) peak in other regions of continental Europe as well. If we compute the coef-

Chapter 6

140 coer. or ••rI.1Ion

t

!

--

A ••rage price

O.3Z

.....•.......•...•..•........... .....

~

11 0.3

........ .....

.a

t

1°'ZS ;

101 0.16

0.24

O.1Z

1984

1986

1988

1990

1994

199Z

Fig.6.9 b Behavior of the ensemble coefficient of variation for apartments in Paris. The ensemble comprises the 20 districts of Paris intra-muros. Since the price multiplier rule applies to this case one expects the coefficient of variation to be larger at the peak than at the beginning of the bubble which is indeed confirmed by observation. In the present case the lag of 3.5 years observed between the two peaks is particularly wide. The coefficient of variation begins to decrease when the increase rate of the highest prices slows down. Source: see Fig.6.3a

!0.45

i

._--Coer. or ..rlolloo

-

Average price

.•.•............

.... 0.4 0

'~'\..

.j

to

.35

1 0.3

'.'

a

.... ..............

o.:zs O.Z

1984

1986

1988

1990

199Z

1994

Fig.6.9 c Behavior of the ensemble coefficient of variation for house prices in Britain. The ensemble comprises the 12 regions of Great Britain. Once again, as the price multiplier rule applies, one expects the coefficient of variation to be larger at the height of the price peak. The lag between the two peaks is about 9 months. Source: see Fig.6.1a

141

Price multiplier effect

ficient of variation for 12 markets located in Bavaria we observe that in the course of time it shows a trough (Fig.6.9d) which corresponds to scenario 3. Intuitively, this suggest that as the bubble develops correlation between the _

!

O.IB . - - - - - - - - - - - - - - - - - - _ - ,

coer. of variation

:1

-

SSOO

AveraIII' priee

II

0.16

SOOII

';. iI 0.14

4SOO

~§ O.ll

4000

=i

!

0.1

..'.'.'

.'.'

-

3500

-

lOOII

ZSOII

.•.......

2000

0.08

1500 0.06 , I

1815.5

lB17

lB17.5

. I·

IBI8

100II

"I

IBI8.5

1819

1819.5

Fig.6.9 d Behavior of the ensemble coefficient of variation for wheat prices in Bavaria (19th century). The ensemble comprises 12 geographically distant markets. As the price multiplier effect does not apply one does not expect the coefficient of variation to increase; observation indeed shows that it decreased which is scenario 3 of Fig.6.9a. Source: Seuffert (1857).

different markets increases more than it does in the cases when the price multiplier pattern applies. A more detailed analysis shows that if PI, P2 are random variables representing the initial and peak prices for the various markets one can, at least as a first approximation, admit that P2 = PI + K where K is a fixed number. This means that for each market the difference between peak price and initial price is almost the same. This model implies that: • E (P2) = E (PI) + K (where E (.) represents the mathematical expectation) and a(p2) = a(pd which leads to:

in agreement with Fig.6.9a. • A = P2 / PI = 1+K / PI, a relationship which is very different from the price multiplier pattern; in particular, the function A (PI) is decreasing instead of being increasing. Let us test this simple model on the case of French wheat markets.

Chapter 6

142

Between 1809 and 1814 there was a major price peak which culminated in April-May 1812. Computing the coefficient of variation for a set of 40 markets at the beginning ofthe price increase (January 1809) and at the peak, one gets:

= 0.24,

c(pd

C(P2)

= 0.13

Thus, we see that once again scenario 3 applies. Since we have here a larger number of markets than in the previous case we can directly test the relationship between A = P2/PI and PI; this is done in Fig.6.ge. The dotted curve

j

4.5 -

1 'l!

~

3S 22

4

37

27

3.5 -

20

3 29

2.5 I40

2

I 10

I

I

I

I

I

12

14

16

18

20

,

22

24 Initial price

Fig.6.9 e Relationship between peak amplitude and initial price for wheat markets. We observe a set of 40 French wheat markets during the 1809-1812 price peak; each label corresponds to a different market; the hyperbola A = 1 + K I PI, K = 40 turns out to provide an acceptable fit. Source: Labrousse et al. (1970).

corresponds to the hyperbolic fit: A model.

1

+ K / PI

predicted by the simple

What can be concluded from these examples? It seems that whenever the price multiplier rule applies the ensemble coefficient of variation has a peak that parallels (or even precedes) the price peak itself. Moreover, the examples taken from wheat market show that whenever the price multiplier does not apply then the coefficient of variation does not display a peak. In short, there seems to be a direct connection between the behavior of the coefficient of variation and the price multiplier effect. This is confirmed by the derivation in the next paragraph.

Price multiplier effect

3.1.4

143

Connection with the price multiplier effect

First we propose the following conjecture. Dispersion conjecture X I is a random variable admitting a second order moment and a variable X 2 is defined as being a function of Xl: X2 = f(Xd where f is supposed to be strictly convex, i.e. for any Xl

t= x2:

f((XI

+ x2)/2) < [f(xt) + f(X2)]/2.

Then, if c(X) = O"(X) / E(X) denotes the coefficient of variation of X, one has: c(X2) > c(XI ); if f is strictly concave then C(X2) <

C(XI). Various numerical tests were performed which supported the conjecture. For instance if X I is distributed according to a bounded Pareto distribution as is indeed the case for book prices in auction sales, then:

f(x) = x1.1 f(x) = xO. 9

(convex function) (concave function)

==> c(X2 )/C(XI ) = 1.05 ==> c(X2)/c(Xd = 0.88

If this conjecture is accepted it can in particular be applied to the functions f(x) = xk, k> lor f(x) = X Inx which are known to be convex (Polya et al. 1972, p.66). Let us now write the price multiplier rule in the form: P2 = PI (a In PI + b). If f(PI) denotes the function on the right-hand side one has: !"(PI) = a/PI which shows that: sgn(f"(Pt)) = sgn(a). From our previous estimates we know that a is always positive; thus f is convex and the above conjecture leads to: C(P2) > c(pt), a result which can be summarized in the following statement: Behavior of the coefficient of variation For a good which follows the price mUltiplier rule, the coefficient of variation for a set of items increases in the upgoing phase of a speculative episode and decreases in the downgoing phase. In the next paragraph we examine the connection between the price mUltiplier effect and the phenomenon of diffusion.

3.2

Is the price multiplier effect a diffusion phenomenon?

In the case of British house prices we have seen that the speculation started in the London area and then spread to the rest of the country, a picture which might suggest that the price multiplier effect should be interpreted as a diffusion phenomenon of a price bubble initially restricted to the London area. In the present paragraph we test that interpretation quantitatively.

144

Chapter 6

First we recall the expression of the Green's function of the standard diffusion equation (Tychonov et al. 1964, p.193). The function: (2.1)

represents the temperature at point x and time t when an amount of heat Q = 1 is set free at the origin x = 0 at time t = O. The function G satisfies the heat (or diffusion) equation: (2.2)

where: a2 = klcp, with k: diffusion constant, c: specific heat capacity, p: density. This equation also describes the diffusion of a gas or liquid from places of higher concentration to places of lower concentration. In that case G (x, t) denotes the concentration of the gas or liquid an c is an index of porosity. Now, for a domain in which a diffusion mechanism is assumed to take place let us observe the evolution of the concentration at various points Xl, ... ,X n for time t > tl. The diffusion phenomenon being symmetrical with respect to x = 0, we can take the Xi positive. The quantities G(Xi' td will be denoted PI(i). Providing that tl is small enough, at each point Xi we will see the concentration first increase, go through a maximum and then decrease as 1/0 (that is to say very slowly). Let us denote the maximum by P2:

We have thus constructed a mechanism which, at least qualitatively, is similar to the propagation of a price bubble. We can for instance compute the amplitudes A( i) = P2 (i) IpI (i) and investigate the relationship between the A( i) and the PI (i) in order to see whether or not the price multiplier rule is satisfied. The answer is given in Fig.6.10a,b. As can be seen the amplitudes A(i) decrease with PI; the decrease is approximately of the form: A(i) = Cexp(-0.6PI(i)). Such a behavior is not very different from what can be observed for wheat prices, but it is definitely at variance with the price multiplier rule. In other words, the mechanism behind the price multiplier effect is not a simple diffusion phenomenon. In the last section of this chapter we examine how the price multiplier effect can be applied to stock markets.

145

Price multiplier effect

.j

8

:II

.~

~ I!i

7 (j

5 4

3 2

Fig.6.10a Diffusion process: behavior of concentration (or temperature) at three locations. Horizontal scale: time; vertical scale: concentration (or temperature) as given by the diffusion ~quation. At time t = 0 ~nd loc~tion x = 0, a concentration peak was generated WhICh for t > 0 moves to IncreaSIng values of x. Each curve refers to what can be observed at a different location; labels give di~tances to the origin; thus, the upper curve (0.030) corresponds to the observatIOn POInt closest to the origin . ... 3.5

,.----~------------------_.

!1::

&:

2.5 I-

2

I-

1.5 I-

I

I-

1

2

3

4

5

(j

7

8 9 PI

Fig.6.10 b Test of the price multiplier effect for the peaks generated by a diflusion phenomenon. As in the previous figure a concentration peak was generated at the origin at t = 0, which subsequently moved to increasing values of x. At t = tl (here tl = 10- 4 ) the concentration at the different observation points is noted PI; P2 denotes the peak concentration. The curve shows the relationship between the peak amplitudes P2/PI and the initial concentrations Pl. The labels on the curve give the x-values of the observation points. Clearly, the price multiplier rule does not apRly here: the correlation is negative instead of being positive. However, this result fits the pattern that we observed for wheat prices.

146

4

Chapter 6

Application of the price multiplier effect to stocks

The fact that the price multiplier effect does not apply to wheat prices is hardly surprising for in this case there is no real element of choice in the buying decision. The wheat sold on each market is basically the same, and the fact that a trader selects one market rather than another is determined mainly by transportation constraints and inventory availability. On the contrary, for stock markets the investor can make hislher selection from a huge number of stocks. Therefore one can expect that bolder investors (remember that there are no "users" on stock markets) will select high growth companies, while more careful investors will concentrate on blue chips. In other words one would not be surprised to see the price multiplier effect playa role. Yet, it can clearly not be tested in the form in which it was stated above for the price of a stock is in itself without significance; it is not because a stock costs 150 dollars instead of 20 that it is more likely to see its price increase rapidly. It should be recalled in this connection that stock prices are periodically readjusted through stock splits. Is there an alternative formulation of the price mUltiplier effect which can apply to stock prices? The answer is yes, and the reasoning goes as follows. We have seen that between the start of a speculative episode and the time when it reaches its peak the price of an expensive item jumps to higher levels; this implies that the average price increase-rate is larger for expensive items. Thus, in addition to the correlation between initial price levels and peak amplitudes there should also be a correlation between price increase-rates (k) and peak amplitudes. This is confirmed in Table 6.3 for several cases. Naturally, the correlation depends on the time interval, I, over which one computes the increase-rate. For an interval I extending right to the peak itself, the correlation would be one, but in that case the statement is nothing but a tautology. For smaller intervals the correlation is reduced, but for a one- or two-year interval there is usually a significant correlation. In this form the price multiplier effect is more general and can be applied to stock prices as well. It is true that this form is somewhat less satisfactory because of the introduction of an additional parameter, namely the length I of the time interval. The application of this alternative formulation to the New York Stock Exchange in the years preceding 1929 is illustrated in Fig.6.11. When the interval I covers one year namely 1925 the relationship reads:

A

= ak +b

a

= 4.7 ±

3,

b = 3.1 ± 1

The correlation is 0.72; if k is expressed in percent as is the case in the figure a has to be divided by 100. According to this rule, high growth companies

147

Price multiplier effect Table 6.3 Correlation between peak amplitude (A) and initial price increase rate (k)

Time interval for measuring the initial price increase rate

Correlation between A andk

[years] 1) Price of houses in Britain, by region for the 1984-1990 peak (12 cases) 0.5

0.31

1.5

0.20 0.17

2

0.33

2.5

0.55

3

0.61

4

0.73

2) Price of apartments in Paris, by district for the 1984-1990 peak (20 cases) 1

0.01

2

0.69

3

0.70

4

0.84

3) Price of apartments in Paris, by size for the 1984-1990 peak (5 cases) 1

0.84

1.5 2

0.90 0.93

2.5

0.95

3

0.98

Notes: With a time interval going right to the peak, the correlation between the amplitude and the average increase rate would be equal to 1; for smaller intervals the correlation is of course smaller than one; the interesting point is that the correlation is fairly high even for an interval of the order of only one or two years. This permits a relatively early prediction of the amplitude of the peak.

Chapter 6

148 8

0-

~

IBM

.;,

...s:

7 -

j t

1 0;

6

f-

5

f-

4 3 -

2 _ BETlfASTM ATI I

o

I

20

I

40

60

80

Price increase rate in 1925 (90)

Fig.6.11 Price multiplier effect for American stocks 1925-1929. For stocks one cannot use the initial price level PI as independent variable; instead we use the initial price increase rate. The correlation is 0.72. Meaning of labels: ALLIS: AllisChalmers; AMCAN: American Can; ATT: American Telephone and Telegraph; BETHL: Bethlehem Steel Corp.; COCAC: Coca-Cola Company; EASTM: Eastman Kodak; GENEL: General Electric; GENMO: General Motors; IBM: International Business Machines. Source: Commercial and Financial Chronicle (various years).

such as "Coca-Cola" or "General Motors" will have a larger peak amplitude than companies such as "Bethlehem Steel". Thus, the price multiplier effect can be stated in an alternative form as follows: Price multiplier rule for stock markets Consider the succession of a bull and bear market and the corresponding price peaks for different stocks. For each stock the price jumps from an initial price PI to a peak-level P2 before returning more or less to level Pl. The ratio A = P2 / PI, referred to as the peak amplitude, is related to the initial price increase-rate k by a relationship of the form:

A = ak+ b,

a>O

The coefficients a and b depend upon the length of the interval I used for estimating the initial price increase-rate. In this form the price multiplier effect appears somewhat similar to Newton's law of inertia. A ball thrown up from an altitude Zl will reach an altitude Z2

149

Price multiplier effect

which is related to the initial velocity

A = -Z2 = ZI

VI

by the relationship:

(1) -2gzI

where 9 denotes the acceleration of gravity.

VI2

+1

Chapter

7

Peak shape: the sharp peak - flat trough pattern

In the autumn of 1998 the American Federal Reserve set up a consortium of banks and financial institutions in order to bailout a large hedge fund called Long-Term Capital Management. The clientele of this mutual fund comprised a small number of wealthy individuals for whom it tried to secure high rates of return through intricate transactions mainly in the derivatives market for interest rates. In a statement delivered on October 1, 1998 before a committee of the House of Representatives, A. Greenspan, the chairman of the Federal Reserve justified his policy by pointing out that LTCM's portfolio was so entangled that it was virtually impossible to avoid what in the financial jargon is called a fire sale i.e. a sale at a considerable discount. Given the fragility of the markets, he added, such a liquidation could have triggered other failures and eventually lead to a severe drying up of market liquidity. At the same time the Federal Reserve cut interest rates three times to prop up stock markets. Subsequently, in December 1999, the Fed flooded the banking system with money to deal with the so-called Y2K (Le. transition to year 2000) problem, an action that has been credited with fueling the December and January 2000 spurt of the NASDAQ index (Commercial Appeal, 23 January 2000). No doubt that by taking such moves the Fed has supported and prolonged the bull market. Later on in this chapter we will see that there are indeed good reasons to think that without such measures the end of 1998 would have marked a turning point. We mentioned that episode in order to emphasize once again that predicting the downturn of a bull market is an almost impossible task in the sense that it heavily depends upon fortuitous circumstances. Yet, the precise date of the turning point, whether it occurs two years later or earlier, would not have changed the overall trajectory of the market; once it began and gathered moB.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_7, © Springer-Verlag Berlin Heidelberg 2009

151

152

Chapter 7

mentum in the early 1990s the scenario of the unfolding of the bull market was fairly predictable. That is the point that will be emphasized in this chapter. It will indeed be shown that the shape of speculative price peaks follows some definite patterns. In the first section we consider speculative price peaks for commodity and property markets. In the first case one can rely on numerous long price series which contain a large sample of peaks. We will see that, at least as a first approximation, one can admit that price peaks are symmetrical with respect to a vertical line drawn trough their summit. As a closer approximation, it will be seen that the rising phase is about 10 to 20 percent longer than the falling phase. If one insists on even greater precision, which corresponds to shifting from a medium-term perspective covering several years to a shortterm perspective, one can in many cases detect oscillations whose frequency becomes higher as one approaches the turning point. This is referred to as the log-periodic oscillation pattern and was introduced in 1996 in a famous paper (Sornette et al.). These results hold for commodity markets as well as for various other speculative markets and in particular stock markets. For the latter, however, the sample of available cases is fairly limited which in tum implies larger confidence intervals for the results.

1

Describing speculative peaks

First, in order to introduce the topic and main ideas of our approach we consider in some detail the specific case of wheat markets.

1.1

Shape of wheat price peaks

Several peaks for wheat prices are shown in chapter 2. Moreover, we have already pointed out in that chapter that although they may have been triggered by meteorological factors, such peaks also reflected a good deal of speculative behavior. In the study of price peaks the first problem one has to face is how they should be identified and selected. Fig.7.1a and b provide an overall view by showing typical wheat price series in two periods of 50 years that are three centuries apart. Three important observations can be made. • The price paths in the two panels look very similar. This means that in spite of the transformations that occurred in production and transportation techniques wheat markets retained most of their characteristics from the 16th century to the early 19th century. Needless to say, a similar graph drawn for

Sharp peak - flat trough pattern

153

400 300 200

1500

1525

1550

Fig.7.1 a Monthly wheat price in Toulouse (south-west of France): 1500-1550. Vertical scale: French francs per hectoliter. If the figure is rotated upside down its aspect does not remain identical which reveals that peaks and troughs are shaped differently: peaks are sharp while the bottom of the troughs is rather flat. Note that the inflation rate was particularly high in that period; the average annual inflation rate is 2 percent against 0.77 percent over the period 1500-1900. Source: Drame et al. (1991.

4000

3000

1800

1825

1850

Fig.7.1 b Monthly wheat price in Toulouse (south-west of France): 1800-1850. Vertical scale: French francs per hectoliter. The overall aspect of the price series appears qualitatively similar to figure (a); even the amplitude of the largest peaks is of the same order of magnitude. Source: Drame et al. (1991.

154

Chapter 7

the period 1900-1950 would reveal drastic changes; in particular the standard deviation of prices would be reduced by at least a factor 4. Regarding price levels expressed in French francs per hectoliter one can notice that in the course of three centuries they were mUltiplied by a factor 10 which corresponds to an average annual inflation rate of 0.77 percent. That estimate is confirmed by inflation figures based on purely monetary considerations; thus, between 1520 and 1720 the price of 244 grams of gold (i.e. what is called a "marc" of gold) increased from 156livres to 740 livres (Marion 1976, p.384) which corresponds to an average annual inflation rate of 0.78 percent. • If Fig.7.1 a (or b) is rotated 180 degrees we see that its aspect changes completely. This test reveals that peaks and troughs are of different shape. The peaks are rather sharp while the troughs are rather round with fairly flat bottom sections. We will refer to that regularity as the sharp peak -flat trough (SP-FT) pattern. A large part of that chapter is devoted to a quantitative analysis of that pattern. In the next paragraph we examine how the peaks and troughs can be described mathematically. • In order to better perceive the specificity of our previous observations one can take a look at Fig.7 .lc which refers to monthly precipitation in Berlin over a 50-year period. In that case no pattern can be detected in particular the 200 100

50

10 5

1850

1875

1900

Fig.7.1 c Monthly precipitation at Berlin. When rotated upside down the aspect of the graphic remains the same. As a matter of fact this series has spikes rather than peaks; it looks very much like white noise, i.e. noise with an autocorrelation which is zero everywhere except for a zero time lag. Source: Clayton ( 1944).

aspect of the graphic does not change when the figure is turned upside down. In the language of stochastic processes one would say that the data in panel (c) are pure white noise.

155

Sharp peak - flat trough pattern

1.2

Mathematical description of price peaks

Fig.7.2 shows a typical price peak. The level PI from which it starts, the level

T(u )

T(down)

Fig.7.2 Some parameters which define a typical price peak. The prices PI, P2, P3 and the time intervals T(up), T(down) define what can be called the envelope of the peak. In addition the shape of the price path is defined by the shape parameters 0: and T (see text).

P2 reached at the peak and the bottom level P3 to which the price drops before beginning to rise again are obviously three important parameters. Thus, the ratio P2/PI defines the magnitude of the peak; thereafter it will be called the peak amplitude. The durations T (up) and T (down) of the rising and falling paths characterize the unfolding of the peak in the course of time. As we have already observed neither P2 nor T(up) can be predicted; but once the price has begun to plummet it is a simpler problem to predict how low it will fall. In other words, knowing PI and P2 one would like to predict P3, or knowing T(up) one would like to predict T(down). That is why we will be particularly interested in the so-called asymmetry ratios: P3/PI and T(up)/T(down). The previous parameters define limits within which the price path will be contained, but they do not specify its shape. For that purpose we introduce the following mathematical description of a peak:

p(t) = P2 exp [-

I-t -7t210<]

(1.1)

Chapter 7

156

In this expression (t2' P2) are the coordinates of the peak and the duration 7 defines the time scale of the process. The exponent a quantifies the abruptness of the peak. When a = 1 , the price increase (or decrease) is exponential, when a < 1 the increase is steeper than an exponential; as a matter of fact in the rising phase: In(p(t)/p2) = -[(t2 - t)/7jO!; thus when a < 1 the curve is concave on its upperside and the tangent to the curve at t = t2 is vertical; this explains why the faster than exponential character of the rising path is especially obvious at its end. On the other hand, when a > 1 the curve is concave on its underside and the tangent to the curve at t = t2 is horizontal; it is what we will call a flat peak. The description by equation (1.1) can also be applied to troughs by changing the sign of the argument in the exponential function:

t -7t210!] p(t) = P2 exp [I -

(1.2)

Both (1.1) and (1.2) can be summarized by the following expression:

p(t)

=

[

t-t210!] P2 exp -Sgn(7) -7-

I

(1.3)

where 7 is allowed to be either positive (peak) or negative (trough). Thus (1.3) describes four different shapes as summarized in Fig.7.3a. To sum up we can say that (1.3) is a parsimonious and fairly flexible description of peaks and troughs which contains only two shape-parameters a and 7. As will tum out subsequently a is the most important as it is independent of the time unit used. Note that for 7 > 0 and a < 1 expression (1.3) represents the zero-order term in the expansion of the function describing a log-periodic behavior (Sornette et a1. 1996, Johansen et a1. 2000). By combining two by two the peaks and troughs in Fig.7.3a one can build four types of price paths; they can be labeled as SP-FT, FP-FT, SP-ST, FPST. The aspect of these price paths is schematically illustrated in Fig.7.3b. Note that in actual price series peaks and troughs are usually not contiguous but separated by some transition intervals (see in this respect Fig.7.4). Some of the patterns shown in Fig.7.3b are more frequent than others; the SP-FT pattern is probably the most frequent, while the FP-FT is almost non-existent. The FP-FT pattern occurs especially in markets in which prices respond with considerable time lag to supply/demand fluctuations. Such is for instance the case in real estate markets, in which a transaction can take from one to several

Sharp peak - flat trough pattem

ex< 1 't>O

ex> 1 't
ex> 1 't>O

ex< 1 't
157

Fig.7.3 a Different shapes of peaks and troughs The four panels correspond to different values of the parameters in the generalized exponential representation (1.3). The vertical scale is logarithmic; the first and fourth graphs correspond to a = 0.4, the second and third correspond to a = 1.5.

months; in a sense the FP-FT pattern can be considered as a smoothed (for instance through a moving average process) version of the SP-FT pattern. We now turn to estimating the parameters that define the magnitude and shape of price peaks.

1.3 1.3.1

Observations Procedure

First of all one has to define peaks and troughs in an unambiguous way. Everybody would certainly agree that a peak is a price path which first goes up, reaches a maximum and then falls. However, real price paths present in addition many short-term fluctuations as illustrated in Fig.7.1a,b. This means that the definition of the starting points and end points of a given peak is not straightforward. We propose the following procedure which for definiteness we explain here for the case of peaks. • We subject the price series p(ti) to a centered moving average (comprising binomial weighting coefficients) with a window width W. Thus, local fluctuations below the time scale W basically will be smoothed out. Let us

158

Chapter 7

0.=1.6

o

10

Fig.7.3 b Four different peak and trough patterns. Upper-left: sharp peak - fiat trough; many commodity price follow that pattern, e.g. coffee, copper, jute, lead, sugar, tea, wheat, zinc. Upper-right: fiat peak - fiat trough; one finds that pattern for property prices or exotic wood. Lower-left: sharp peak - sharp trough; that pattern is found only rarely, two examples are cotton and copra. Lower-right: fiat peak - sharp trough; that pattern does not seem to occur in actual observations.

denote by S(ti) the smoothed series. • The first differences D(td = S(td - S(ti-l) are computed and all successive values of i for which D(td is positive are considered as belonging to a given rising path. Similarly a falling path corresponds to successive values of i for which D(ti) is negative. The reunion of a rising and falling path defines a peak. In practice we found that for monthly series a 41-month window was adequate to smooth out local irregularities without affecting the overall shape of the peaks. To some extent the procedure is of course dependent upon the choice of the window width, but by varying its size in a ±20 percent range we found out that the results are almost not modified. The identification procedure of peaks and troughs is illustrated in Fig.7.4 . • The amplitude of the peak A = P2/Pl is then computed from the original series. As we are only interested in large peaks all peaks for which the amplitude is less than a given threshold are discarded. In our study we have considered two different thresholds A > 1.5 (i.e. a 50 percent increase) and A > 2 (i.e. a 100 percent increase). As the results are fairly similar in both cases (see in this respect Roehner et al. 1998), only one of these thresholds

Sharp peak - flat trough pattern

159

will be considered in that follows. • For all selected peaks the parameters are computed from the original series. The shape parameters a and T are computed by linearization; this enables us to use standard regression techniques instead of non-linear least squares minimization. Fig.7.5 illustrates a typical fit.

1.3.2 Results for wheat markets The wheat price series that we used are listed in Table 7.1. As they concern different countries they can to a large extent be considered as independent realizations. Their lengths total about 1,200 years and they contain 44 peaks and 22 troughs of amplitude A > 2. With such a large sample one is in a good position to get reliable results. Table 7.2a summarizes the statistical results for the first set of parameters. It appears that the rising path is slightly longer in duration and has also a somewhat larger amplitude than the falling path. Nevertheless one can roughly say that after a speculative bubble the price more or less falls back to its initial level. Table 7.2b summarizes the results of the fit for the shape parameters a and T. One can note the remarkably small dispersion in the values of a; only the Vienna market is somewhat out of range. That small dispersion even holds for the fits of individual peaks as shown by the results in Table 7.2c; the coefficient of variation of the a is only 21 percent. Moreover, no clear trend can be detected in the course of time; neither is there any significant correlation between a and the amplitude of the peaks.

1.3.3 The turning point In (1.3), the turning point t2 is a singular point for a > 1 in the sense thatp(t) is not differentiable at t2' Is this singularity solely a mathematical artefact or does it correspond to some real economic features? We believe that the latter holds and that the singularity reflects a genuine cooperative behavior of the market not unlike to those studied in critical phenomena in physics. Indeed, during a major price peak, the correlation length of the markets tends to become very large. As in statistical mechanics the correlation length characterizes the range of the interaction between the markets. In previous papers (Roehner 1989,2000a) it has been shown that during the price boost of 18161819 the correlation length jumped to a level about 20 times above its "normal" level. This could be interpreted in two different ways. First, one could think that it is the occurrence of an exogenous perturbation that forced prices upward on all markets, thus producing almost simultaneous price jumps

Chapter 7

160

80 GO 40

30

10

5~

__

mJO

~

____

1800

~

____

1810

~

____-L____

1820

~

____

~~

1830

Fig.'.4 Example of selected fluctuations of amplitude larger than 100 percent. Vertical scale: wheat price in Munich expressed in gUlden per Schaffel (1 gulden = 2.1 French francs, 1 Schaffel =2.2 hectoliters). From left to right the selected fluctuations are (heavy line): a peak, a rising path, a peak, a trough, a fall, a peak. Source: Seuffert (1857).

1815

1816

1817

1818

1819

1820

Fig.'.S An example of adjustment of the parameters 0: and T. The graph refers to monthly wheat prices in Munich expressed in gUlden per Schaffel. Stars: observed prices; solid line: theoretical fit (0: = 0.68) through a generalized exponential (expression (1.3». Source: Seuffert (1857).

161

Sharp peak - flat trough pattern

Table 7.1 Wheat price series used in the analysis of price peaks Market

Country

Period

Cologne

Germany

1532-1796

265

2

Grabow

Germany

1785-1870

86

Beitriige (1873)

3

Munich

Germany

1790-1855

66

Seuffert (1857)

4

Paris

France

1521-1698

178

Baulant et al. (1960)

5

Toulouse

France

1486-1913

428

Drame et al (1991)

6

Vienna

Austria

1692-1913

222

Pribram (1938)

InSource terval [year]

Total:

Ebeling et al. (1976)

1,248

Table 7.2a Shape characteristics for peaks and troughs of wheat prices Peaks 44

22

23.2

49.4

P2 / PI

3.2 ±O.3

3.1 ±O.3

T(up)+T(down)

44±5

53 ±9

Number of fluctuations A verage interval between fluctuations [years] A verage amplitude Average duration [months]

Troughs

Asymmetry parameters Left price / right price Left durat. / right durat.

P3 / PI

1.0 ±O.J

1.1 ±O.2

T(up)/T(down)

1.4 ±O.3

1.3 ±O.4

Notes: Only peaks and troughs of an amplitude larger than 100% are considered in this table. The asymmetry parameters are (almost) consistent with the simple assumption that the peaks (and troughs) are symmetrical with respect to peak (or trough) value.

Chapter 7

162

Table 7.2b Shape characteristics for peaks and troughs of wheat prices: parameters of the generalized exponential (1.3) Market

Number of fluctuations

Cologne

13

Grabow

't

Goodness of fit

0.64

27.3

10%

5

0.60

23.8

Munich

7

0.63

38.9

15% 11%

Paris Toulouse

26

0.67

36

0.62

16.9 26.7

14% 13%

Vienna

16

0.80

23.5

10%

0.66

26.2

1) Peaks

Average

2) Troughs Cologne

4

1.14

-17.0

12% 10%

Grabow

5

1.17

-15.4

Munich

7

1.30

-7.7

12%

Paris

1.23

-9.8

Toulouse

16 31

1.05

Vienna

16

1.08

-17.2 -20.0

17% 15%

1.16

-14.5

Average

11%

Notes: In this table we considered peaks and troughs of an amplitude larger than 50% (for peaks of amplitude> 100 % the number of fluctuations would be too small for some of the markets). For the sake of clarity the error margins have been omitted. It is interesting to observe that there is a fairly small dispersion for the parameter a. ; this supports the view that, although the peaks are triggered by supply/demand shocks, their dynamics is largely governed by speculation behavior; otherwise the form of the peaks would be more dependent upon exogenous shocks. The goodness of fit is defined by the ratio: e=L (ti-oj)2 / L oi 2 , where the ti are the theoretical values and 0i the observations; subsequently the coefficient of linear correlation for the linearized data will be used as an alternative measure of adjustment quality.

163

Sharp peak - flat trough pattern

Table 7.2c Estimates of the parameter ex for individual wheat price peaks

1498 3. I 0.453

1508 3.2 0.455

1516 2.1 0.482

1529 4.0 0.509

1539 3.3 0.512

1546 4.6 0.773

1563 3.3 0.784

Amplitude ex

1573 4.0 0.644

1580 2.2 0.775

1595 2.1 0.500

1614 2.1 0.750

1631 3.6 0.581

1644 3.8 0.597

1694 2.5 0.495

Year

1710

1720

1812

Amplitude ex

3.7 0.894

1713 2.2 0.606

3.5 0.613

3.5 0.625

1817 2.6 0.595

Year Amplitude ex Year

Notes: The results are for wheat prices in Toulouse. The estimates of ex do not show any definite trend in the course of time; neither is there any correlation between ex and the amplitude of the peak.

throughout the country. Alternatively one may interpret the jump in the correlation length as reflecting a genuine endogenous increase of the strength and range of the interaction between different markets. To distinguish between the two interpretations, the correlation length of meteorological factors was calculated (Roehner 1989) and it was found that there is no significant increase during or even before the occurrence of the peak. In further support to the second interpretation, we note that it has been emphasized by many historians (e.g. Meuvret 1956,1971) that wheat trade expanded in periods of high prices both in volume and in range. Locally, many self-appointed retailers emerged; regionally, traders were combing the countryside in order to buy every available bushel of wheat for the supply of the cities (even at the expense of the countryside supply); internationally, the government often encouraged (and even subsidized) the importation of additional quantities of wheat from distant markets.

1.3.4

Twentieth century commodity markets

Whereas for wheat prices it was possible to carry out a systematic analysis with a relatively large data set, the situation is much less favorable for twenti-

Chapter 7

164

eth century markets. For most commodities, there have been only a few major peaks during this century. What makes things even worse is that during the first half of the century, the world economy has been disrupted by three major perturbations: World War I, the Great Depression and World War II. For this reason, we restrict our attention to the second half of the century. Our objective in this paragraph is to show by a few examples that the SP-FT pattern also applies to various twentieth century commodity markets. Fig.7.6 provides an illustration for sugar price peaks.

1000

900 800 700 600

500 400 300 200

o

10

20

30

40

50

Fig.7.6 Price peaks for sugar. The data refer to monthly prices in Caribbean ports. For the parameters a andT the estimates areas follows: 1962-1965: a = 0.97, T = 11.3; 1972-1976: a = 0.63, T = 7.3; 1978-1981: a = 1.03, T = 9.5. Source: Commodity price Bulletin, UNCTAD (1985,1990).

In Table 7.3a we have listed some results. In order to better interpret them one would need to know a number of technical characteristics regarding each commodity such as for example price elasticities of demand and supply, storage and transportation costs, amplitude of meteorological shocks, etc.; in that regard a comprehensive commodity handbook giving these empirical data would be very valuable.

1.3.5

Real estate markets

The main results are summarized in Table 7.3b. As already mentioned all the values of a are larger than one which means that property prices follow a FP pattern. Note that in this case there is a positive correlation between a and

A.

Sbarp peak - flat trougb pattern

165

Table 7.3a Estimates of the parameter a. for commodity price peaks in the period 1960-1988 Commodity

Sugar

Gold

Silver

Coffe

Zinc

Tea

Amplitude a.

10

8.0

6.6

4.5

4.3

2.7

0.63

0.36

0.51

0.88

0.72

0.90

Notes: These markets are not directly comparable: the amplitude of exogenous shocks and the dynamics of the market may be different. If for a moment one forgets these differences one can be tempted to observe that there is a negative correlation between the amplitude A and a. which means that the higher the peak, the smaller is a.; the regression reads: A=-ao.+b, a=-9.1 ±9, b=I2±2, r=-0.71.

1.3.6

Postage stamp markets

The main results are summarized in Table 7.3c. In this case a seems to be smaller than one in some cases and larger than one in others. But one can wonder if the a > 1 cases are not a statistical artifact. When discussing with stamp experts one becomes aware of the fact that stamp catalogs want to please collectors and therefore are more reluctant to post price decreases than price increases. As a result there is often a considerable lag of time between the actual occurrence of a downturn and the moment when it is eventually recorded in stamp catalogs.

1.4

Unified overview

The amplitude A = P2/Pl characterizes the magnitude of a peak; the power a characterizes its shape; together these two parameters provide a condensed description of a speculative peak. It is therefore of interest to plot the various results that we have obtained on a chart where the two axis correspond to a and A. This was done in Fig.7.7. The broken line a = 1 separates the plot into two regions; at the left of this line is the sharp peak (SP) region; at its right is the flat trough (FT) region. While that distinction is certainly important from a time series perspective, it remains to understand its economic significance. There are two distinct regression lines between a and A. The fact that small and large values of a can both give rise to high amplitudes is intuitively fairly obvious. For instance both a = 0.5 and a = 2 correspond to a behavior char-

166

Chapter 7 Table 7.3b Shape characteristics of real estate price peaks

Country

Period

A

T(up)/ T(down)

P3/PI

al

a2

a

Port Said London

1898-1912

5.1 2.0

1.6 1.13

0.53 1.24

2.08 1.00

1.31

1983-1993

1.70 1.10

1.02 1.28 not yet finished

0.96

1.22

1.09

1.38

1.27

1.32

1.20

1.43

1.31

Paris Singapore

1984-1997

2.1

1992-1999

2.6

Average

1.26

1.28

1.12

Notes: Port Said is located at the entrance of the Suez canal. a 1 and a2 respectively correspond to the up- and down-going phase; a denotes the average of a I and a2. The regression between a and A reads: A=5.4a-4 (r=0.92).

Table 7.3c Estimates ofthe parameter a for postage-stamp price peaks Country

Period

Britain

1965-1981

90, 155

1.32

France France

1930-1944 1941-1944

16 2

0.73

France

1980-1987

2060,2391

1.13

Stamp numbers

a

0.46

Notes: In the first three lines the stamp numbers refer to the "Yvert and Tellier" catalogs while in the last line they refer to the Ceres catalogs.

Sharp peak - flat trough pattern ~

....

10 r----------------------,r---------------------~

I

•

NASDAQ

t

!

167

8 7 (;

s

• .GoId • lOver •

I I I

•

Copn l

I

•

4

3

•

Tofyo

Wheat

.-.-"

- .....

•

•

Stamps •

•

•

•

2

Peak expo".'" (a)

~ig. ~.7 Representation of speculative markets in the (a, A) plane. London, Pans, Smgapore and Tokyo refer to the real estate markets in these cities. There are two regressions lines, one for a < 1, namely A = -6.6a + 9.4 (r = :-0.63), and the other for a > 1, namely A = 7.6a - 6.4 (r = 0.94)

acterized by a steep increase but the detailed dynamics is completely different; for Q = 0.5 the rate of increase is largest at the end of the rising path, while for Q = 2 the increase rate is already large at the beginning of the peak and remains fairly constant throughout the rising phase. On Fig.7.7 one can see that the largest amplitude corresponds to the NASDAQ stock market; that result invites us to turn now to the study of stock markets.

2

Is there a pattern for stock price peaks?

For the markets studied in the previous section, i.e. commodities, postage stamps, real estate, the statistical investigation could rely on the analysis of several dozens price peaks. From a scientific perspective one was therefore in a good position to reach significant and reliable conclusions. The situation is completely different for stock markets. As will be seen subsequently their statistical investigation has to rely on less than a dozen cases. Furthermore, many of these cases are mixed rather than homogeneous price bubbles (in a sense that will be explained below), a fact which is a direct consequence of the

168

Chapter?

mixed character of the stock markets themselves. Thus, scientifically speaking, one is not in the best position to detect patterns. Nevertheless it can be of interest to review the available evidence if only to examine whether or not the SP pattern conclusions can be generalized to stock markets. For instance the symmetry property with respect to the summit has obvious predictive implications and it is therefore important to look at it more closely. The present section proceeds as follows. First, we emphasize that different "species" have to be distinguished among stock price peaks before valid comparisons can be made. That classification work is an essential prerequisite in any comparative study and it is often its most tricky part. Then, once we know what we want to compare, we proceed to give some quantitative measures of the shape of stock price peaks. It will be seen that the quantitative estimates (i) confirm and support the classification of price peaks (ii) lead to orders of magnitude quite similar to those obtained in the previous section for other speculative markets. The very idea that there can be definite patterns for stock prices has been denied or at least questioned on the basis of the socalled market efficiency and random walk assumptions. In the last section we discuss these objections in the light of our findings.

2.1

Necessity of a taxonomy of stock crashes

The necessity of a meaningful taxonomy for stocks has been emphasized in a different context by R. Mantegna (1999). That requirement is no less essential in the present context of stock crashes. Some stock crashes are due to the implosion of speculative bubbles while others are due to different factors. How can one differentiate the former from the latter? This is a preliminary question that has to be considered. We argue that a good guide is provided by overall economic circumstances, an aspect that we illustrate in the present paragraph by a qualitative discussion of several case studies. Then, in the next paragraph, we show that the parameter a introduced in section one can provide an estimate of the amount of speculation in a given bull market (the smaller a, the more speculation). Each of the three following examples serves to illustrate a different class of stock market crashes. • Between mid-1972 and late 1974 the London stock market suffered a crash which was in fact worse than the one which occurred between 1929 and 1933. By the end of 1974 the Financial Times Index which comprises only blue chip companies had fallen by 74 percent from its peak (Fig.7.8). The All Share Index made up of 714 individual companies, when adjusted

Sharp peak - flat trough pattern

169

for inflation, was in late 1974 standing at just 1/6 of its peak value two and a half years earlier. The average price earning ratio (ratio of share prices to the company's annual earnings, thereafter denoted PER) plunged to about 4. As a matter of comparison in the crash of Black October 1987 the PER fell only to about 11 (Beckman 1988, Galletly 1988). Was that crash due to the burst of a speculative bubble? Well, there was undoubtedly a certain amount of speculation in particular in the property market with nominal property prices climbing by 52 percent in 1973, but at the level of the stock market speculation was fairly modest both in magnitude and in duration (see Fig.7.8) and can certainly not explain the severity of the crash. By all accounts the decline was a downward spiral occurring in a bleak financial context. First there was an intractable problem of inflation: the inflation rate was 16 percent in 1974 and 24 percent in 1975; secondly the "bears were back" in the sense that in New York, between 1970 and 1975 the Dow Jones index lost about 40 percent. The crash of the property market led to a banking panic: 26 secondary banks had to be bailed out by a consortium conducted by the Bank of England (Sampson 1982); that is the reason why this crisis is often referred to as the "Secondary banking crisis". Finally let us mention that for business circles the political context was as grim as the financial perspectives with the coming to power of a labor government in February 1974. • In late 1998 and early 1999 the bull market of the 1990s apparently came to an end for American banks and airline companies. As can be seen by looking at the stocks of individual companies stock price plummeted (or at best remained stationary) between January 1999 and January 2000. An example is provided in Fig.7. 9: the stock of the First Union bank posted a loss of 30 percent in that period. And yet, the Dow Jones index (as well as the broader Standard and Poor's 500) continued to climb from about 9,000 in January 1999 to 11,500 one year later. The reason is well known. The growth was driven by Internet and telecommunication stocks as reflected in the strong increase experienced by the NASDAQ during the same period of time. That is why we call the bull market of the NYSE a mixed bubble. Schematically the scenario was the following: first from 1982 to about 1990 there was a recovery period from the 1971-1980 bear market, then from 1990 to mid-1998 there was a broad bull market. By the end of 1998 stocks prices began to dwindle in many sectors with the result that the bull market lost most of its momentum. If one likes analogies one can think of the NYSE in the 1990s as being similar to the space shuttle (or to the Ariane 5 launcher which has the same structure); from 1990 to 1998 it climbed and accelerated on the thrust of both its main engines and its solid rocket boosters, then in 1998 the two boost-

170

Chapter 7

1

London (1970-1975)

]100 ~

]

90

~ 80 ~

70 60 50 40

30

1970

1974

1975

1976

Fig.7.S Crash of the London stock market in 1972. The crash that occurred in 1972 eventually lead to a price decrease of about 75 percent, a greater fall than during the Great Depression. That crash, however, was not primarily the burst of a speculative bubble. This can be seen from two facts (i) The increase during the upgoing phase was a modest 30 percent (ii) The bottom level in late 1974 was much lower than the initial price level in early 1971. This points to a downward spiral which was the result of a financial crisis in a world in turmoil (increase in oil prices). Sources: Main economic indicators. Historical Statistics. OEeD 1989; Liesner (1989).

FIRST UNION CORP. (1995·lOOO)

100

1999

2000

2001

Fig.7.9 Price of First Union Corp. stock. With a capitalization of over 31 billion dollars (April 2000), the First Union is a major American bank. Between January 1990 and July 1998 the price of its stock climbed from 6 to 58 dollars that is to say 2.7 times faster than the Standard and Poor's 500 index, which means that this company (as indeed all major banks) played the role of an engine in the bull market. Sources: http://finance.yahoo.com, http://minneapolisfed.org/economy.

Sharp peak - flat trough pattern

171

ers were shut down and the climb continued driven by the main engines alone. Thus, at the present moment, the model appears to be a two-stage rocket, but no basic principle can rule out the possibility of a three- or four-stage rocket. Both the Internet and the telecommunication boom were brought about by major technical breakthroughs: enhanced computer compatibility in the first case and numerical compression techniques permitting a cost reduction by a factor 10 in the second. Another technical revolution could probably ignite a third stage and boost stock markets higher. Whether this will happen or not is not, strictly speaking, a scientific question in the sense that its answer depends upon fortuitous circumstances which are exogenous to stock markets. It is clear for instance that the discovery of cheap and convenient supraconductors at room temperature would trigger a new revolution in many sectors related to electrical engineering. The two previous examples concerned major stock markets. The first crash was not due to the burst of a speculative bubble, while in the second case there was indeed a speculative peak, but of mixed rather than homogeneous type. One may wonder whether by turning to smaller markets it is possible to substantially increase the pool of stock market crashes. It is to this question that we turn now • Crashes on smaller market are only rarely due to the burst of an endogenous speculative bubble for (at least) two reasons. (i) Because of the so-called domino effect (Vandewalle 2000), a crash on a major stock market often triggers crashes on smaller markets. For instance the crash of October 1929 in New York was followed by a similar crash in Canada, Sweden, the Netherlands and the United Kingdom!; thus, these crashes can hardly be considered as providing separate instances. (ii) Crashes on small markets are often caused by exogenous financial factors. The wave of collapses in South West Asia and South Korea in the fall of 1997 is a case in point. In none of these countries (with the possible exception of Hong Kong) was the crash preceded by a phase of rapid increase, as would be expected for genuine speculative peaks. In Thailand the decrease had already begun in early 1996, in Malaysia and the Philippines in February 1997 and in Singapore stock prices have been almost stationary since early 1994. In short, the stock market collapses that occurred between August 1997 and March 1998 were brought about by the financial crisis, somewhat in the same way as for the case of London that we discussed above. 1 The latter was of course by no means a small market; note that in France and Germany the fall had already begun earlier, namely in January 1929 and June 1927 respectively (White 1996, p.348).

Chapter 7

172

• A last remark is in order. At first sight one could think that by considering individual stocks instead of indexes one may increase the number of cases. But the diversity in the price path of individual stocks is just too great. One has to resort to some kind of averaging process; a possible solution would be to define stock price indexes for each economic sector. The previous examples suggest a rudimentary classification of stock market crashes into (at least) three classes as summarized in Table 7.4. That classiTable 7.4 Embryonic classification of stock market crashes Category

Features

Examples

Speculative price peaks

The increase phase lasts

NYSE 1857,

several years and

1929

accelerates at the end

Tokyo 1990

Crashes brought about by

A crash which occurs on a

Canada 1929

domino effect

small market and closely

Sweden 1929

follows a crash on a dominant market Crashes brought about by a

The crash is not preceded

London 1994

financial/economic crisis

by a period of rapid

Thailand 1997

increase; the financial

Indonesia 1997

crisis unfolds before the stock market crash.

fication is confirmed by the quantitative analysis to which we turn now.

2.2

Quantitative analysis

In Fig.7.10 three major speculative price peaks are drawn side by side and their comparison reveals a similitude in their shape. In the present section we try to assess that similitude quantitatively. Both a and r can be estimated by linearization plus standard regression. These parameters will be estimated separately for the rising and falling phases. An example of fit is shown in Fig.7.11. The results for a number of major peaks are summarized in Tables

173

Sharp peak - flat trough pattern ~ ~ '-~NA~S~D-AQ~(-I~--2-00G-)----~~------------------------~

1·5 800 700

i 600

NYSE (\921·\932/ Tokyo (\987·1992)

oiSOO ~

400 300 200

100 90 80

·10

·8

2

4

Fig.7.10 Comparison of three speculative episodes. Horizontal scale: time in years (origin at the peak). Solid line: NASDAQ Composite index; broken line: Dow Jones index (Industrial); dotted line: Nikkei index. Sources: NASDAQ: http://ji· nance.yahoo.com, http://minneapolisJed.org/economy; NYSE: Farrel (1972), Liesner (1989); Tokyo: Main economic indicators. Historical Statistics. OEeD 1989, http://ea.grolier.com, http://www.stat.go.jp/english .

.~

t. 500

1 450 1:;

l ~

400

3SO 300

2SO 200

ISO 100

Fig.7.11 Fit of the 1921-1932 price peak on the NYSE. Each triangle corresponds to one month. The fit is performed after linearization which permits to use standard regression techniques.

Chapter 7

174

7.Sa,b,c. The cases considered are mostly what in the previous paragraph we Table 7.Sa Shape characteristics of stock price peaks Asymmetry Dates

NYSE

2 3 4 5 6 7 8 9

NYSE NYSE London Tokyo NYSE First Union National City NASDAQ

1903-1907 1921-1932 1950-1982 1970-1975 1985-1993 1986-1987 1995-2000 1995-2000 1990-?

Overall average (1-8) Average (1,2,3,5,6)

Duration

Amplitude

Duration

[years]

P2/Pl

T(up) / T(down)

3.1 10.7 28.2 2.7 4.7 2.0

2.3 5.5 4.0 1.5 2.5 1.6

>3.8 >3.5 ?

>2.4 >1.9 >9

Price level P3/Pl

0.78 3.0 0.96 0.77 1.45 2.37 1.44 1.57 ?

1.26 0.83 1.85 0.31 1.25 1.13

1.53 1.71

1.20 1.26

1.37 1.58 ?

Notes: All dat.a are mon.~ly prices except for case (3) where annual prices were used. At the tIme of wrItmg the peaks number 7,8,9 were still in progress. The second average is restricted to speculative peaks (which excludes London) for which the end of the story is known.

called genuine speculative price peaks; however for the purpose of comparison we also included a few other cases. Thus, from our former discussion we strongly suspect the London crash to be merely the effect of a financial crisis. As a matter of fact it is not a sharp peak (a > 1 in the upgoing phase) and its asymmetry ratio is less than one for the duration as well as for the price. For the 1906 crash on the NYSE we had no prior knowledge but the a parameter suggest that it might be a true speCUlative peak, a conclusion which is comforted by the fact that the financial crisis broke out in October 1907 that is to say about 10 months after the stock market crash. For the NASDAQ only the upgoing part could be estimated; it has the smallest al value in the whole series which makes it the most extreme example of speculative peak in the list. Two factors probably contributed to this. (i) One

175

Sharp peak - flat trough pattern Table 7.5b Shape characteristics of stock price peaks:

(X

adjustment

Dates

(XI

(X2

(X

0.81 0.64

0.77

0.79

0.70

0.67 1.16

I

NYSE

2

NYSE

1903-1907 1921-1932

3

NYSE

1950-1982

1.33

4

London

1970-1975

1.17

0.99 1.22

5

Tokyo

0.89

0.93

0.91

6

NYSE

1985-1993 1986-1987

0.71

1.69

1.20

7

First Union

1995-2000

1.11

2.3

1.71

8

National City

1995-2000

0.92

1.15

1.54

9

NASDAQ

1990-?

0.51

?

?

Overall average (1-8)

0.95

1.21

1.15

Average (1,2,3,5,6)

0.88

1.02

0.95

1.20

Notes: (X I and (X2 correspond to the rising and falling path respectively. The average coefficient of linear correlation is 0.92 for the rising paths and 0.88 for the falling paths.

Table 7.5c Comparison between the characteristics of speculative price peaks for stock markets and other markets

(X

P3 / PI

T(up) / T(down)

Stocks

0.95

1.26

1.71

Wheat

0.66

1

1.3

Gold, silver

0.44

Postage stamps

0.60

1.6

Notes: The T(up) / T(down) ratio is difficult to estimate for postage stamps for there is often a considerable time lag between actual prices and prices reported in stamp catalogs.

Chapter 7

176

should remember that the NASDAQ comprises a high proportion of high-tech stocks which explains that it is a much more homogeneous market than the NYSE; equivalenty, using our previous analogy, one could say that until 2000 it was one-stage rocket. (ii) The fact that it is an automated market (remember that NASDAQ means: National Association of Securities Dealers Automatic Quotation system) makes it a less strictly regulated market than the NYSE, a circumstance which probably favored the unfolding of speCUlation. Fig.7.12 shows the shape of a typical price peak which is based on the previous regularities and regressions. In the last paragraph we discuss the question

20

,

10

8 7 6

Y..". (Jmdr=1/J)

Fig.7.12 Theoretical shape of a typical speculative stock price peak. For definiteness the scale and parameters of the upgoing part were modeled on the rise path of the NASDAQ composite index. The parameters of the downgoing part were obtained from the empirical relationships derived from Table 7.Sb, namely: 02 = 0.5501 + 0.43, P3/P1 = 1.20, T(up)/T(down) = 1.71

of stock price patterns from the perspective of the efficient-market hypothesis.

2.3

Is the existence of stock price patterns consistent with market efficiency?

According to the standard belief about market efficiency any well-defined pattern will be quickly detected and used by skillful traders in order to make extra profit; these innovators will soon be imitated and followed by other traders with the result that the identified pattern will be short-lived. In other words,

Sharp peak - flat trough pattern

177

any breach in randomness will be temporary and self-suppressing. This argument was stated very clearly by E. Fama (1965) and it has ever since been accepted as a strong justification of the random behavior of stock prices. Can we accept that argument without restrictions? This is clearly a question of fundamental importance in connection with the study of stock price peaks. In some situations the case of randomness is fairly strong as for instance at the level of individual stocks. Suppose for instance that two companies A and B belonging to the same economic sector, are similar in terms of basic fundamentals such as earnings and growth perspectives but that A is priced twice as high as B. This would clearly be an anomaly and one can indeed expect that after being detected it will be corrected within a few weeks. There is a weakness in the argument however. Indeed, if earnings can be estimated unambiguously, such is not the case for growth perspectives. That 'analysts may diverge in that respect by more than a few percent is clearly suggested by the following example. In April 2000 the PER of major American airline companies (Continental, Delta, North West) was on average equal to 6.8, but the PER of US Airways was equal to 84. One may wonder how such a huge difference could be justified in terms of growth perspectives. When shifting from individual companies to global market behavior the randomness argument becomes even less compelling. The main reason is that the PER criterion provides a much more reliable guide at global level. As one knows the long term average of the PER on the NYSE between 1871 and 1995 was about 14. Moreover (as we have already noted in a previous chapter), every time it approached 25 this was followed by a market correction. In 1929 just before the crash the PER was around 22; in early 1936 before a 40 percent correction is was near 25; in 1971, at the peak of the bull market of the 1950s and 1960s, the PER was approximately equal to 20 (the average PER of hot growth companies such as IBM, Xerox, or McDonald's was about 40); in October 1987, just before the crash the PER reached for a short time a level of 30 (White 1996). In short, before the 1990s, the PER seemed to provide a fairly reliable indicator of the overall state of the market. Then in 1991 the PER of the NASDAQ reached 40, in early 1999 it reached 100 and in January 2000 it rose to 200 (Washington Post, 12 January 2000)2. PERs of that order of magnitude were obviously at variance with a century of empirical observations. How can one explain that a guideline which had proved so reliable in the course of time was suddenly discarded? The answer of traders was very 2 At about the same moment the PER of the NYSE was equal to 23 (Dallas Morning News 26 February 2000).

178

Chapter 7

simple: "We are making history" and that sentence indeed encapsulated the crux of the mattey3. If a rule as firmly established as the PER rule can be discarded in that way, it means that similarly any other accepted pattern will be swept away if it stands in the way of a vigorous bull market. In short, collective auto-persuasion and the expectation of high gains seem to be stronger than the most firmly established empirical rules. In a sense such an attitude is coherent with the knowledge that in the social sciences empirical rules are indeed subject to historical changes. The real question is not the change itself but rather its rate. The fact for a variable that for more than a century has remained confined in a 10-30 interval, to jump to a level of 100 and remain there for several years certainly implied quite revolutionary changes.

3 It is true that not all traders accepted that interpretation; for instance on 31 March 2000 Julian Robertson the manager of Tiger Management, a huge hedge fund, preferred to withdraw from what he called an irrational market which "overshadowed time-tested investment strategies",

I

Chapter

8

Stock market bubbles

In a Reuter's headline of 29 April 2000 one could read: "The Fed chairman is scratching his head, wondering why the volley of interest-rate increases hasn't done the trick [i.e. cooling the speculative fever]. One reason is that the new economy has changed the way business operates; companies are not as dependent on short-term debt as they used to be". This illustrates how easily people can forget the lessons of history. Between March 1999 and May 2000 the so-called Fed funds rate (i.e. the rate charged for overnight bank lending) has indeed risen from 4.75 to 6.5 percent (Fig.8.1) that is to say by 1.75 percent. But in similar historical situations larger increases were re-

..~

i

.s. ~

i..!!,

'"~

7

6.5 6 5.5 5 4.5 1999

2000

Fig.S.1 Rise in federal funds rate. This rate which is used for overnight bank lending is fixed by the Federal Reserve; from 1999 to mid-2000, in order to cool down an ebullient economy, it was raised in steps from 4.75 to 6 percent. Sources: Far Eastern Economic Review, 2 Sept. 1999, Le Monde (Web site). B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_8, © Springer-Verlag Berlin Heidelberg 2009

179

180

Chapter 8

quired in order to cool speculation. In 1928 the discount rate was increased by steps from 3.75 to 5 percent without any effect on speculation; the downturn was only brought about by a further one percent increase in August 1929. Here is another example which also emphasizes that former historical situations must be used in a proper quantitative way. In late 1999 and early 2000 a number of articles pointed out the potential risk of an increasing margin debt (i.e. the amount of money borrowed by investors in order to buy stocks): "We have had an expansion of margin debt the likes of which haven't been seen since the 1920s. It's a domino effect. When prices start to fall, the value of margined portfolios falls; lenders issue margin calls; the margin calls force the sale of stock; the mass sales of stock pressure prices even further. That happened with a vengeance in 1929" (Time.com, August 1999; Money.com, April 2000). The parallel with 1929 was farfetched however. As we will see, in relative terms, i.e. with respect to market capitalization, the margin debt was about 6 times larger in 1929 than in 1999, 10 percent against 1.5 to 2 percent. Of course this does not mean that the previous domino effect cannot playa role in crashes, but a shock wave affecting 1.5 percent of the market is obviously less dangerous than one affecting 10 percent. The present chapter is not about stock market crashes. On the contrary it is likely that the fascination exercised by these dramatic events (e.g. see Arbel 1989, Granville 1985, Preston 1973) can only confuse our understanding of the mechanisms at work. In that connection it should be remembered that after having lost 37 percent in October 1929 the Dow Jones index regained 24 percent in the first months of 1930. Then, after April 1930, without the kind of panic that occurred in October, began a long slide that halted only in 1932 after stock prices had lost 5/6 of their value. In other words, the determinant factor was not the October crash but the subsequent downward spiral. Similarly, between November 1989 and mid-1992 the Nikkei index at the Tokyo stock exchange lost more than 50 percent without any spectacular crash or panic. Once again the fatal factor was a downward spiral that affected not only financial markets but also the property market as well as the rest of the economy. A stock market bubble has many facets. In the present chapter we restrict our attention to a few specific questions. How does a market gather momentum at the beginning of a bubble? Is it possible to define some variables which can permit to identify stock bubbles? What is the role of interest rates? How can one gauge market uncertainty? First, we consider these questions for the case of the 1921-1933 bubble; then, in the second section we examine if the

Stock market bubbles

181

observations made in this specific case-do also apply to other stock bubbles.

1

A case in point: 1929

The crash of October 1929 and the Great Depression have been studied extensively. In this section our purpose is different however. We will not stress the idiosyncrasies of the crisis but rather use it as an introduction to the comparative analysis that we develop in the next section. For instance, the fact that the 1929-1933 depression was particularly severe in the United States was probably due to domestic factors; however, even if one discards such circumstances it can hardly be disputed that the strong stock market rise of the 1920s could not last indefinitely. For short, our main objective in this section is to identify factors that are common to different stock bubbles. What triggered the bull market? How did it develop? Why did the bubble last so long? In trying to answer these questions one can roughly distinguish three levels of explanations. • One can propose one or several plausible qualitative mechanisms which may possibly account for the effect under consideration. One writer will be tempted to stress micro-economic causes, while another will point to monetary or more generally macro-economic factors. In economics it is not difficult to identify plausible causes, the real problem is rather to decide which ones are most relevant. • If it is possible to describe the previous mechanisms quantitatively one is in a better position to decide whether they really played a role. Our previous example regarding the role of margin debt was a good illustration: at a 10 percent level it may indeed playa role, but at a 1.5 percent level it is dubious whether it can be of more than marginal importance. • In order to prove conclusively that a given mechanism can account for the effect under consideration a mathematical model has to be developed and tested on several similar circumstances l . lOne faces the same three levels of explanation in other scientific questions as well; as a matter of illustration let us consider them for the much debated question of the disappearance of the dinosaurs. The first level would correspond to enumerating plausible causes: change of climate, lack of food, fall of a big meteorite, etc. The second level would consist in assessing quantitatively one (or several) ofthese factors; for instance number and size of meteorites that hit the earth. Even if large meteorites fell at the time of the extinction of the dinosaurs, that does not prove that meteorites were indeed the determinant factor; after all, numerous other species survived. The third level would necessitate the construction of a mathematical model that can account not only for the extinction of the dinosaurs but also for the extinction of other groups of species in other time periods.

182

1.1

Chapter 8

The driving force

Based on a scientific or technical breakthrough a new sector emerges and enters a period of exponential development. In the 1850s it was the railroad; in the 1920s, automobile and electric utilities; in the 1950s and 1960s office equipment, television and rubber industries; in the 1990s, electronic banking, telecommunication and the Internet. Merely listing these sectors would amount to what we called a type (i) explanation. In the present chapter we try to go somewhat further by assessing to what extent the growth of these sectors was indeed faster than the development of the rest of the economy. However, even if we can prove that these sectors grew say twice as fast that does not prove that the bull market was indeed triggered by these sectors. To prove that point one would need to develop a model accounting for the spread of the fever from the high-growth sectors to the rest of the economy. This is a fairly difficult task that will not be taken up here. Armed with a better understanding of the goals as well as limitations of the present study let us come back to the 1921-1929 bull market. Fig.8.2a shows the evolution of stock prices for three groups of companies (i) Utilities (ii) The Dow Jones Industrial average which was based on 20 companies until October 1928 and then on 30 companies as is still the case nowadays. (iii) Railroads. From that chart it is obvious that utilities indeed played the role of a driving force. Their rapid price rise certainly contributed to attract more money into the stock market. In contrast, railroads which had been the highgrowth companies of the 19th century lagged behind the rest of the market. Particularly striking is the fact that, although railroads rose a modest 100 percent, in the aftermath of the crash they lost much more ground than the utility sector or the market average (represented either by the Dow Jones or the Standard and Poor's 500). Though the picture emerging from Fig.8.2a could appear fairly clear a number of precisions and qualifications are in order. • As one can expect the utilities were not the only companies which did markedly better than the market average. At the more detailed level of individual companies one can mention: Coca Cola (multiplication by 9, Fig.8.2b), Dow Chemical (mUltiplication by 12), General Motors (multiplication by 10), IBM (multiplication by 13, Fig.8.2b). Here again one can observe that there is a definite correlation between P2 / PI and P3 / PI, where PI denotes the initial price, P2 the peak price and P3 the price at the end of the fall (Table 8.1a). This is confirmed by the results given in Table 8.1b for another sample of stocks. The regressions in Table 8.1 a,b can be stated in the form of the following rule.

183

Stock market bubbles ~

Utilities

.S 1000 .,. 900 .~ 800 ~ 700 600

DJ Industrials - - Railroads

500

400 300

200 .__

._

..... •••••••••••••••

.........-

........

........... -.

...... . -.-. . ..-.e•

...

a. ._

100 90

1922

1924

1928

1926

1930

1932

Fig.8.2 a Differentiated rises in stock prices. All prices are yearly highs. The Dow Jones Utility index (thin broken line) was created only in 1929; for the period 1921-1928 we have considered an average of five major utilities namely: American Electric Power, Columbia Gas System, Consolidated Edison, Baltimore Gas and Electricity, Pacific Gas and Electricity. The railroad index is an average of Burlington, Norfolk and Pacific Union. Source: Common stock (1988), Farrel (1972).

i1

.....................................................

1000

if!

600 500

..' ..'

.'

.:It ./.~_ - 1 '

400 300

200

133 80 70 60

Fig.8.2 b Price of two stocks with respect to the Dow Jones index. Heavy solid line: Dow Jones Industrial index; thin solid line: Coca Cola; broken line: IBM. For these high-growth companies the crash of October 1929 was nothing more than a slight disturbance; their prices decreased by less than 15 percent. Sources: Common stock (1988), Farrel (1972).

Chapter 8

184

Table 8.1a The higher a stock climbs during a bull market the better it resists in the subsequent bear market (1921-1932)

P2/PI

P3/PI

IBM

13.0

8.0

Dow Chemical

12.5

6.1

Utilities (5 companies)

11.0

3.5

General Motors

10.5

3.8

Coca Cola

9.0

6.5

Dow Jones Industrial Index

5.0

1.4

Firestone

3.1

1.4

Railroads (3 companies)

2.2

0.95

Notes: The results in this table are based on annual highs. The 5 utility companies and the 3 railroad companies are the same as in Fig.8.2a. PI denotes the price in 1921, P2 the peak price (1929) and P3 the bottom price (it took place in 1932 except in a few cases where it was in 1933 or even 1934). All prices are deflated prices. For the 13 companies considered, there is a close correlation between P2/Pl and P3/PI; the regression reads: P3/PI 0.29 (PVPI) - 0.82 (r=0.86) Source: Common stock (1988)

=

Resilience pattern The higher the price of a stock climbs during a bull market, the less it falls during the subsequent bear market. Note that the above rule does not give any guarantee about the subsequent evolution. "Columbia Gas System" constitutes a case in point. It is an utility company whose capitalization was 5.2 billion dollars in April 2000 (under the name Columbia Energy, NYSE: CG). Between 1921 and 1929 its price rose (in annual highs) from 8 to 150 dollars, i.e. P2/Pl = 18.7; in 1932 its price was still 20 dollars (i.e. P3/Pl = 2.5) and it remained at that level until 1937. After 1937 a downward spiral brought it as low as 2.5 dollars in 1941. Then a steady (but slow) progression in the second half of the 20th century brought the price to 60 dollars by early 2000, that is to say half as much as in 1929 in nominal terms; after inflation is taken into account this corresponds to a real price about 20 times smaller than in 1929.

185

Stock market bubbles

Table 8.1b The higher a stock climbs during a bull market the better it resists in the subsequent bear market (1925-1932) n/pl

P3 / Pl

1. Allis-Chalmers

3.80

0.23

2. American Can

6.33

1.28

3. American Telephone and Telegraph

2.23

0.67

4. Bethlehem Steel

2.35

0.16

5. Coca Cola

15.1

2.9

6. Eastman Kodak

2.25

0.35

7. General Electric

6.38

0.49

8. General Motors

5.66

0.60

9. IBM

7.41

2.02

Notes: PI is the price in 1925, PZ the peak price and P3 the bottom price after the fall. The results in this table are based on monthly prices (weekly during the peak year); therefore they cannot be directly compared to the results in the previous table. However, there is again a close connection between PZ/Pl and PJ/Pl; for the 9 companies, the regression reads: P3/Pl = 0.21 (pZ/pl) - 0.22 (r=0.90) Source: Commercial and Financial Chronicle (various issues)

1.2

The bubble's signature

The word "signature" is used here in the same sense as in "sonar signature" or "radar signature"; in other words we try to define variables which permit the identification and characterization of a bubble. Our attention will particularly be focused on four variables: trading volume, rise in price earning ratio (PER), role of discount rate and interest rate spread. The strong rise (1921-1929) followed by a sharp (1929-1932) drop has already been documented in chapter 5 (Fig.5.4a). More precisely: V2/VI = 5.5 and V3/VI = 2.2, where VI denotes the trading volume in 1921, V2 in 1929 and V3 in 1932. After a short stabilization in 1933, the decline of the trading volume continued. Not surprisingly, the lowest point was reached during World War II, in 1942, with a trading volume of 126 million shares, about ten times less than in 1929 (Historical Statistics of the United States 1974, p.1007). Both the price earning ratio and the price dividend are spiky (and fairly paral-

Chapter 8

186

leI) curves; this is hardly surprising because the level of earnings (and hence of dividends) can change drastically from one quarter to another. For that reason the PER signature is less reliable than the trading volume signature. Nevertheless the rule, already mentioned in previous chapters, according to which the PER usually remains under the 20 mark is fairly well respected at least until 1995. Between 1926 and October 1930 the PER increased steadily from 12 to 21; then it dropped to around 15 until jumping shortly to about 25 in 1934 when, as a result of the Great Depression, the earnings were at their lowest.

1.3

Role of discount rate increases

Fig.8.3 shows that the stock market crash occurred shortly after the one percent discount rate increase of 9 August 1929. The question of whether the

r

.. 550 , - - - - - - - - - - - - - - - - - - - - - - - - - , ~ 500

1450

Discount rate Dow Jones

-

6

-

5

i

"'400

I

i

~

4

350 3

300

250

2

200 I

ISO 100 1'-9.L28-'--'--"--'19--.J28'-•.L5-'--'-....l.1--'92......J9L.L..-'-1-'-9....l.2'--'.-5L.L..-'-I'-'-3--'O-"-'-I-'-'3-'-0....l..s--'--J'--I'-'-'-3-'-1....l...J.--.J 1' --3.LI 5 0 .J.....J.

Fig.8.3 Discount rate versus Dow Jones index. The increases that occurred during the first semester of 1928 did not seem to have much effect; yet, the one percent increase in August 1929 might have been the last straw that broke the camel's back. Note that the repeated lowerings that occurred after the crash did not stop the market collapse. Sources: Wyckoff(1972}, Farrel (1972).

crash was indeed caused by that increase, although of great practical interest, is fairly ill defined scientifically since it is impossible to repeat the "experiment" under different circumstances. Furthermore, even in the absence of any discount rate increase the boom must have come to an end anyway.

Stock market bubbles

187

In support to this view we will see in the next section that the 1989 crash in Tokyo occurred at the very beginning of a series of discount rate increases. Whatever the role of the last increase, Fig.8.3 suggests that (i) The three increments that occurred in the first semester of 1928 had little influence on the bull market. As a matter of fact the period of October 1928 to April 1929 was marked by a particularly rapid increase of stock prices. (ii) The seven lowerings that occurred between 1st November 1929 and mid-1930 did not stop the price fall. The same observation can be made in the aftermath of the Tokyo collapse.

1.4

Role of the margin debt

Buying on margin is the expression used by stock traders for buying stocks by employing credit. In the 1920s, an investor could buy stocks worth 10,000 dollars with an amount in cash of only 2,500 dollars; the other 7,500 dollars were obtained by a loan arranged by the broker. Since 1974 the margin requirement is 50 percent which means that with 2,500 dollars one can buy stocks for only 5,000 dollars. In times of decreasing stock prices the margin debt can give rise to a selfreinforcing spiral: suppose that with an initial amount of 2,500 dollars Mr Jones has bought 100 shares of US Steel at 100 dollars/share. Should the value of these shares suddenly decline from 100 to 70 Mr Jones' holding would shrink to 7,000 dollars. As a result the broker would found himself liable for 7,500 dollars with only 7,000 dollars worth in stock to secure it. To protect himself against possible losses he will therefore demand that his customer put up more margin (i.e. cash); and he will do the same after each substantial stock decline. To put up more margin Mr Jones may be obliged to sell other stocks and preferentially stocks which did not decline so far (otherwise he will face an additional loss). It is clear that if a substantial number of investors are in the same situation as Mr Jones their sales can generate a downward price spiral. The above mechanism: price decrease --+ margin calls --+ more sales --+ faster price decrease was invoked as one of the causes of the crash of October 1929. The increase in the debt margin (which was called broker's loans at that time) is shown in Fig.8.4. Particularly impressive is the sharp decrease after the crash. The level of the margin debt relative to the market's capitalization is a major variable from which one can determine whether a self-reinforcing spiral can start. In 1929 the capitalization of the NYSE was around 80 billion dollars (80 percent of

Chapter 8

188

i

11 7

~

3

II

~ e.

.~ :!!

6 5 4

2

d

0.7

G.6 0.5 0.4

0.3 fI.2

1930

1931

1932

Fig.8.4 Margin debt. At its highest point in October 1929, the margin debt represented about 11 percent of the market capitalization. In the 1920s, with one dollar cash one could buy for 4 dollar stocks; since 1974,50 percent in cash is required. Source: Commercial and Financial Chronicle, April 18, 1933.

the GDP); at its maximum the margin debt was 8.55 billion thus representing 11 percent of market capitalization. As we will see in the next section that was a very high level when compared with similar bull market situations.

2

Comparative analysis

In this section we examine the role the aforementioned factors played in other bull markets, such as for instance the 1857 crisis, the bull market of the 1950s and 1960s, the Tokyo price bubble (1990), the NASDAQ price peak of the late 1990s.

2.1

The engine

Table 8.2 gives an overall view of the role played by leading sectors in several peaks. As we already pointed out, for the explanation to be really compelling one would have to show how the fever in a small number of sectors spread to other sectors and eventually lead to the "take off" of the whole market (or at least a substantial part of it). In a sense this problem is similar to the description of the take off of an economy, and that parallelism is hardly surprising

189

Stock market bubbles Table 8.2 Bull market engines Price peak

Engines

NYSE

1857

Railroads

NYSE

1929

Utilities

NYSE

1969

Office equipment Television Rubber

5.1 4.2 3.7

Tokyo

1990

Banks Communication

1.7 1.4

NASDAQ

2000

Internet Telecommunications

4.3

Ratio: engine / average

2

Notes: The ratio engine / average denotes the ratio of the amplitude of the peak for the drivng sector to the amplitude of the peak for the market average. The different results are based on different sources and can therefore not be compared one with another. The interesting point however is the fact that all the ratios are larger than one which confirms the role of the engine sectors as driving forces of the market. Sources: The NYSE (1929) estimate is based on the discussion in the previous section; NYSE (1969): based on the data given in the Dow Jones Investor's Handbook (1972); Tokyo (1990): Japan Statistical Yearbook; Internet estimate: based on the AMEX Internet index (the AMEX has an association with the NASDAQ).

since the stock market is a financial representation of the whole economy. Remembering how intricate and elusive the problem of economic growth turned out to be and how laborious it was to find clear-cut rules, permits to gauge the difficulty of the present challenge.

2.2 2.2.1

The bubble's signature Price versus trading volume

The relationship between price and level of trading volume has already been discussed in chapter 5. It was shown that for a given price increase rate the increase in trading volume is limited by the number of traders, the share of

ChapterS

190

their wealth they are disposed to invest in stocks and by transaction costs. Yet, the question cannot be considered as completely solved. In particular it is not clear why the trading volume should drop so drastically after the price peak. After all, for major individual stocks trading volume often remains high even during the phase of falling prices.

2.2.2

PER

Table 8.3 summarizes PER data for several markets. During the bull market Table 8.3 Price earnings ratio at the peak

NYSE NYSE

Tokyo NASDAQ

Price peak

PER

1929 1969 1988 2000

21 18 50 200

Sources: White (1996), Galletly (1988), Washington Post (12 Jan. 2000).

of 1921-1929 the earnings remained almost constant with the result that the increase in the PER paralleled the price increase. In the Tokyo and NASDAQ cases a different mechanism must have been at work for the price increase cannot alone account for such high values as 50 or 200. A very simple model which has been first proposed by D. Sornette (2000) can help us to understand that effect. We present it here in a somewhat modified form. Consider a company paying an annual dividend d per share. Let r be the real interest rate supposed to remain constant in the course of time ("real" means nominal interest rate minus inflation rate). The time discount factor per year is thus 1/ (1 +r ) because one euro in year n + 1 is equivalent to 1/ (1 + r) euro in year n. In standard value theory the price p of a share is the present value of all dividends that will be payed in the future, which leads to the expression.

d

p

d

d

= d + 1 + r + (1 + r)2 + ... + (1 + r)H

where H is the time-horizon of a typical investor.

(2.1)

191

Stock market bubbles

The geometrical sum can be expressed as:

1

1

G = 1+ 1+r

1

+ (1 + r)2 + ... + (1 + r)H = G=

1 - (1/(1 + r))H+1 1 - (1/(1 + r))

(1 + ~) [1 - C! r) H+ll

Since 1/(1 + r) < 1 if H is large enough the term (1/(1 negligible and one gets the simple result:

+ r))H+1 becomes (2.2)

On this expression one can read the adverse effect an increase in interest rate can have on stock prices: as r increases, p decreases as l/r. But, as already noted, such a price decrease is not observed during a bull market at least in the first increments of interest rate. Thus, there must be some other mechanism at work. Let us introduce the assumption that the dividend d is expected to grow at a rate 8. The expression (2.1) is then changed into:

d p=

+

d(l + 8) 1+r

+

d(l + 8)2 (1 + r)2

+ ... +

d(l + 8)H (1 + r)H

By using the same summation formula than before one obtains:

1+1+8+(1+8)2+ ... +(1+8)H = l+r l+r l+r that is:

1

l+r p-- d - - [(1+8)H+l --1, 8-r

l+r

_ ( 1+8)H+l 1+r 1 1+8 _ l+r

1

(2.3)

Simple as it is this model encapsulates several of the characteristics of stock market bubbles that we observed in the previous section namely: (1) During bull markets there is a widespread feeling that the boom will last indefinitely which means that H is large. (2) Bull markets are driven by a number of high-growth sectors, which means that 8 is large. (3) During the rising phase PERs increase substantially, which means that prices go up while dividends remain stable (or even fall).

192

ChapterS

(4) Bull markets are often brought to an end by repeated interest rate increases. These observations are described by equation (2.3) through the following mechanisms. During a bull market 0 is usually much larger than r especially for high-growth sectors. For instance in the 1990s, r was of the order of 3 percent in the United States (4.S percent nominal interest rate minus an inflation rate of 1.S percent) while high-growth companies grew at a rate of at least 20 percent. Under such circumstances, for the price in (2.3) to be large, H has to be large which means that the expectation prevailing at time t = 0 is supposed to hold for many years. The belief in the permanence of the bull market (point 1) is thus a major element, not only from a (fairly vague) psychological perspective, but also from the perspective of value theory. The fact that p grows exponentially with H explains that even if the dividend d is small the price and the PDR (price dividend ratio) can reach high levels, which is point 3 (remember that the PER and PDR are roughly proportional). On formula (2.2) it was easy to see that an increase in interest rate would lower p; on formula (2.3) that conclusion is not longer obvious for an increase in r will increase 1/ (0 - r) but decrease (1 + 0) / (1 + r). However, provided the power H is large enough one expects the effect of the last factor to dominate; this is indeed confirmed by numerical computation. Thus p decreases as r increases. More specifically one can note that in relative terms a given interest rate increase will have a greater effect on the share prices of highgrowth companies than on those of slow-growth companies. For instance an increase from 3 percent to 8 percent will have the following effects: High-growth companies: Slow-growth companies:

0=50% 0=2.9%

==} ==}

P decreases by P decreases by

31% 20%

Thus, one can understand that a sufficiently large interest rate increase can break the driving force of high-growth sectors. That point is considered in more detail in the next paragraph.

2.3 Role of discount rate increases Fig.8.Sa,b shows two instances of stock price slides apparently brought about by successive discount rate increases. These cases are less spectacular than the 1929 example (Fig.8.3) for the following reasons: (i) The slide in (Sa) is fairly moderate: only 30 percent in one and a half years. (ii) In (Sb) the slide was more important (about 60 percent) but it began at the very beginning of the series of discount rate increases. These increases were probably aimed at

193

Stock market bubbles

f200

Discount rate 6

.l:l

.~1l00

j

5

hooo

~ ~

4

~

900

3

2 700 I ,

. I

1967.5

1968

1968.5

1969

1969.5

1970

1970.5

1

Fig.8.S a Discount rate versus Dow Jones index: 1967-1970. The bull market ofthe 1950s and 1960s came to an end in late 1968, but it was not a sudden drop as in 1929, rather a period of high volatility which would last during the whole decade amid high inflation and strong increases in prime rates destined to kill inflation. Sources: Wyckoff ( 1972), Farrel (1972).

tI

60

~ 'iIl

....

55

;;

50

]

6 --

5

45 -

4

:i!:

40

3

35

2

30 25

1

20

0 I

15 1989.5

1990

1990.5

1991

1991.5

1992

1992.5

Fig.8.S b Discount rate versus Nikkei index: 1989-1992. In the decline that started in 1989 the market lost about 60 percent of its capitalization. At the same time there was a huge property crash. By 1995 the discount rate had been lowered almost to level zero, but even such a huge decrease could not revive the stock market. Source: Japan Statistical Yearbook (various years).

194

ChapterS

the property bubble rather than at the stock market bubble. One can note that another case of discount rate increase was observed during the period of double-digit inflation of the early 1970s. Between the beginning of 1972 and mid-1974 the American prime rate increased from 4.75 to 12 percent; not surprisingly, there was a 50 percent drop in the Dow Jones Industrial index between mid-1974 and early 1975 (The Economist, 30 August 1975).

2.4

Role of the margin debt

In April 2000 the margin debt on the NYSE amounted to 265 billion dollars. However, this huge amount represented only about 2 percent of the capitalization which was at that time of the order of 12,000 billion dollars. Thus the risk that margin calls could trigger a downward spiral was far more moderate than in 1929. Nevertheless the effect of margin calls was clearly visible on the daily price course during the first semester 2000. During slight (less than 1.5 percent) daily decreases on the NASDAQ one observed corresponding increases on the NYSE, which could be interpreted as a "flight to quality" (remember in that respect that the PER of the NASDAQ was much larger than the PER of the NYSE). In contrast, whenever the fall on the NASDAQ reached 3 percent or more the NYSE experienced a parallel fall, an effect probably due to the fact that margin calls on the NASDAQ had triggered additional sales on the NYSE.

2.5

What is the bottom line after a stock market crash?

Once a bubble has begun to burst where does the fall stop? As explained in chapters 3 and 5 a bubble is characterized not only by a widespread fancy for investment in high-growth companies, but also by a general agreement about basic economic rules and values. In short, during that phase the word of the day is "full steam ahead". After the downturn, unanimity falters and the correlation length of the market begins to plummet. In a former paper (Roehner 2000c) it was shown that the growing uncertainty that engulfs the market can be gauged by the increase in the range of interest rates for a sample of bonds. That variable which is often referred to as the interest rate spread can be considered as a statistical measure for the disparity in lenders' opinions about the future. Furthermore, it is of interest to note that the market begins to recover precisely when the

Stock market bubbles

195

interest rate spread begins to diminish. That effect was confirmed for 8 major stock crashes in the United States between 1857 and 1987. The correlation between P3/Pl and P2/Pl that we pointed out in section one provides another mean for assessing bottom price levels after a downward slide; it has been more fully documented in a separate paper (Roehner 2000d).

PART IV

THE TRIUMPH OF NEOLIBERALISM

Chapter

9

Triumph of neoliberalism in economics

1 Neoliberalism: an overview Broadly speaking, neoliberal policies tend to encourage free trade and to reduce the role of government which is why they are often referred to as freemarket policies. Among the consequences of such policies one can mention the following effects. • Privatization may lead to control by foreign companies; this point is discussed later on in connection with Latin America and Eastern Europe. • Because free-trade is also interpreted as free circulation of manpower, huge amounts of cheap labor become available in industrialized economies which in turn depresses wages and reduces the incentives for technical innovation in production processes. A side-effect of cheap labor imports is to erode the role of unions. For instance in the US private sector the unionization rate decreased from 30% in the 1950s to about 8% in 2008. • Because taxes are seen as an hindrance to free-market mechanisms the neoliberal agenda requires sharp cuts in tax rates for both households and companies and light tax rates for financial revenue. Needless to say, such policies mainly benefit to the wealthiest people. This reduction in income redistribution led to an erosion of social solidarity. A subsequent chapter will discuss this point in greater detail. • In industrialized countries down-sizing of the role of the government leads to under-investment in basic infrastructures such as roads, bridges, dams, dikes, public transportation facilities. From a methodological perspective it is often helpful to consider cases in which the effects that one wants to study appear most clearly. The emirate of Dubai epitomizes many of the positive as well as negative aspects of neoliberal policies. One can mention: massive imports of cheap labor, negligible tax rates for Emiratis and companies, non-unionized workforce, a bold B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_9, © Springer-Verlag Berlin Heidelberg 2009

197

198

Chapter 9

project of creating in the middle of a desert a metropolis based on tourism and financial services. It will be interesting to see how Dubai weathers the economic crisis1 .

1.1 Roots of the neoliberal program As shown by the tables below, the word “neoliberalism” has not been much used in the past and is still sparsely used at the time of writing. The first occurrence of the word “neoliberalism” in the New York Times was in 1983 (6 October, section B, p. 8). Before 2000 it was used once every 5 months, and between 2000 and February 2009, about once every three months2 . Instead alternative expressions have been used. For instance, on 27 February 2009 there was an article entitled: “Obama’s budget plan sweeps away Reagan ideas”. The article refers to the fact that after two decades during which taxes on high income brackets had been lowered, the budget for 2010 was planning tax increases on the top 1%. Yet, “Reagan’s ideas” were not limited to tax policy; they brought about many other changes such as deregulation, reduction of the role played by government agencies, reduction in social welfare programs and so on. Many of these ideas had been advocated by economist Milton Friedman who became a close adviser to president Reagan. Before being implemented by president Reagan, neoliberal ideas had been introduced and applied in Latin American countries. One of the first experiments carried out by the so-called “Chicago Boys” was conducted in Chile after General Pinochet’s coup in September 1973, that is to say about 8 years before President Reagan began to implement a similar program in the United States. Although the word “neoliberalism” may be fairly new the notions that it covers are not. The main ideas of the neoliberal creed have been a permanent and central theme in the program of the American Chamber of Commerce and of the National Association of Manufacturers for decades. For instance after the election of president Roosevelt fierce campaigns were conducted by 1

The sociological consequences of this project are described and analyzed in an article published in the British newspaper “The Independent” (Hari 2009). 2 Incidentally, a similar keyword search for the Venezuelan newspaper GloboVision gave 3 occurrences per month in 2008. Given the frequency with which the expression is used by President Hugo Chavez one would have expected a higher rate of occurrence. However, as we are not familiar with the political orientation of the papers in this country it is difficult at this point to draw a reliable conclusion.

Triumph of neoliberalism in economics

199

Table 9.1a Occurrences of the term “neoliberalism” in newspapers: number of articles per month in 2000 and 2008 Newspaper

2000

2008

Britain Guardian The Independent Average

1 0.08 0.54

4 0.6 2.3

not available 1.7 1.7

0.8 2.5 1.6

United States Christian Science Monitor New York Times Washington Post Average

0.08 0.0 0.2 0.09

0.08 0.4 0.5 0.3

Global average (3 countries)

0.8

1.4

France Figaro Le Monde Average

Notes: According to the definition given on Wikipedia, the term “neoliberalism” is most often used by critics of the doctrine. Proponents prefer to use the terms “market economy” or “market liberalism”. This observation is indeed confirmed by the data given in the table in the sense that “neoliberalism” is used more often by left-wing oriented newspapers than by right-wing papers. However, beyond this left-right difference there is also a clear difference between countries: the frequency is markedly higher in Britain than in the United States. The word “Reaganomics” was mainly used during the two terms of president Reagan (1981-1989); after 1990 the word was used less than once a month in the New York Times. Source: The data have been obtained by using the key-word search engines available on the websites of the newspapers.

these organizations against his “socialist program” (an expression frequently used at the time) and in favor of a “free-enterprise” agenda. The participation of business in the war program was seen as an opportunity to regain some of the lost ground. As an illustration, one can mention the following declaration made on 17 September 1942 by Lammot DuPont, then Chairman of the Board of the

Chapter 9

200

Table 9.1b Occurrence of the word “neoliberalism” on the Internet Language Word

English French German Spanish

neoliberalism neolib´eralisme neoliberalismus neoliberalismo

Number Normalization Normalized of factor number of websites websites [million] [100 million] 2.3 4 0.4 1.8

1.8 0.8 0.3 0.5

1.3 5.2 1.3 3.6

Notes: Because the number of existing websites depends on the language, the numbers of relevant websites must be normalized in some way to make them comparable. As normalization factors we used the number of websites obtained for the keyword “economics” expressed in the respective languages that is to say: economics, ¨ e´ conomie, Wirthschaftswissenschaft or Okonomie, Economia. In terms of normalized numbers there seems to be a marked difference between French and Spanish websites on the one hand and English and German websites on the other hand. Source: The data have been obtained by using the Google search engine.

DuPont Company 3 : “We will win the war by reducing taxes on corporations, high income brackets, and increasing taxes on lower incomes, by removing unions from any power to tell industry how to produce, how to deal with their employees or anything else, by destroying any and all government agencies that stand in the way of free enterprise.” It took about 40 years for the objectives outlined by Lammot DuPont to be fully realized. Naturally, these goals were not specific to American business but were probably shared by corporate leaders in other industrialized countries. What was really new with the version of neoliberalism which emerged in the late twentieth century is the fact that, mainly thanks to the Chicago School, it became an economic theory as well as a political program. So 3

This excerpt of DuPont’s speech before the resolution committee of the National Association of Manufacturers is cited in Selden (1943, p. 99). The primary source is not given specifically for this excerpt but it is indicated that the whole section is based on the following sources: Lobby Investigation Report (senator Blake); Committee on Education and Labor (La Follette Committee), Senate Report No 6, part 6; Investigation of Concentration of economic power, Temporary National Economic Committee (TNEC) monograph 26. It can be noted that Lammot DuPont (1880-1952) was a staunch supporter of the anti-New Deal “American Liberty League”.

Triumph of neoliberalism in economics

201

many Nobel prize awards were bestowed on neoliberal economists that the Chicago School became the center of mainstream economic theory and policy. In spite of the fact that the present crisis has clearly shown that markets do not regulate themselves, the neoliberal conception will probably remain the mainstream paradigm for many years until being replaced (if ever) by an alternative vision which will take years to emerge and spread.

1.2 Neoliberalism and the financial crisis In his testimony of 23 October 2008 before a House Oversight and Government Reform Committee, Alan Greenspan declared: “It was the failure to properly price such risky assets that precipitated the crisis. In recent decades, a vast risk management and pricing system has evolved, combining the best insights of mathematicians and finance experts supported by major advances in computer and communications technology. This modern risk management paradigm held sway for decades. The whole intellectual edifice, however, collapsed in the summer of last year [i.e. 2007] because the data inputted into the risk management models generally covered only the past two decades, a period of euphoria. Had instead the models been fitted more appropriately to historic periods of stress, capital requirements would have been much higher and the financial world would be in far better shape today. When in August 2007 markets eventually trashed the credit agencies’ rosy ratings, a blanket of uncertainty descended on the investment community. It would probably be a mistake to think that the present crisis will bring down the neoliberal ideology. Basically, this ideology is the expression of corporate interests. The mathematical models mentioned by Alan Greenspan were largely ad hoc models set up with the objective of obtaining “rosy ratings” from rating agencies. In this respect one should recall that what makes a model scientific is not the fact that it is expressed through mathematical formulas. Mathematics is only a language which by itself does not carry any truth. What makes a model scientific is the way it has been tested in various conditions. Let us give two illustrations. • Galileo’s law of free fall was stated by him in plain language but it was based on extensive experiments and it provided the starting point of classical mechanics.

202

Chapter 9

• The fact that Einstein’s theory of gravitation uses sophisticated mathematical tools does not have any implication as to its truth. Nowadays it is accepted by most physicists because during 90 years it has been tested in various experiments and astronomical observations and was found to stand these tests with success. There can be little doubt that it will be possible to revise these models in a way which makes them more consistent with the present business situation but without calling into question the “free market” assumptions on which neoliberalism relies. As a matter of fact this is the direction indicated by Greenspan. He does not suggest any oversight of derivatives markets by the Securities and Exchange Commission. Instead he suggests that simply by adequately changing risk parameters in financial models they will again be working as well as they have been working during the past three decades. It is likely that most investors wish to keep a system which has been so profitable. According to the New York Times (4 March 2007), during the three decades up to 1999 Mr. Soros’s fund returned more than 30% a year on average. This represents an overall multiplication by a factor 1.330 = 2620. In the same article, one also learns that in managing the money of King’s College of Cambridge University from 1928 to 1945 (less than two decades) John Maynard Keynes earned the College an average annualized return of 13%, i.e. an overall multiplication by 1.1218 = 9.0. This is certainly a good performance especially during a period of time marked by the Great Depression and the war, but nevertheless between 30% and 13% there is a big difference. When Alan Greenspan speaks of the collapse of the intellectual edifice of financial mathematics he has mainly in mind the pricing system. It is quite possible that exotic financial products had been priced too low, a problem which can probably be remedied by changing a few parameters in the pricing models. In contrast the failure of the risk management edifice calls into question the soundness of sophisticated hedging methods. To explain this point let us start from a concrete example. The University of Harvard relies on its endowment (i.e. its capital assets) for paying the salaries of faculty and staff members and for its extension programs. This endowment which amounted to $37 billion on 1 July 2008 is the largest of any US university. Yet, in the 3 months between July and October 2008 it had shrunk by 22%. The endowments of other universities experienced a similar fate with declines of 13% and 11% for Yale and Princeton respectively (New York Times 22 Feb 2009). One

Triumph of neoliberalism in economics

203

of the main justifications for introducing a broad set of new financial products was to provide more flexibility and efficiency in hedging operations. These new hedging tools, it was claimed, would allow fund managers to avoid major losses even in the case of unexpected changes. Such hedging strategies were particularly suited for university endowments for which the main goal was to protect the underlying capital in a long-term perspective. The large losses experienced by Harvard, Yale and Princeton show that hedging and risk control did not work. Needless to say, the same observation can be repeated for the management of many other funds from money funds to pension funds. In all these cases Greenspan’s “infectious greed” argument does not apply because these managers did not (and by law were not allowed to) pursuit risky strategies; their duty was to use hedging and diversification strategies to protect their capital against any financial upheaval. Yet, it seems they were let down precisely by the sophisticated tools designed to provide protection.

1.3 Neoliberal ideology and social fragmentation In a general way the implementation of neoliberal policies results in processes of social fragmentation. In subsequent chapters we will illustrate this fragmentation by describing examples from the economic, financial and social spheres. One may wonder how this connection between neoliberalism and social segmentation can be explained. As a first possible explanation one may mention the underlying economic theory. The classical and neoclassical view of economics is based on the notion of individuals making rational decisions in order to maximize their utility function. Collective factors such as for instance interactions between employees or “herd effects” in times of financial panics are not taken into consideration. Thus, one can hardly be surprised that neoliberal policies result in processes of social fragmentation. It can be noted that the axioms of neoclassical economics are implemented in a fairly selective way. The existence of monopolies or oligopolies is certainly at odds with neoclassical principles. Yet, nowadays this issues are almost never mentioned by the proponents of free-market. Monopolies such as Microsoft or oligopolies such as Boeing and Airbus are well accepted. This omission is all the more surprising when one realizes that the decade from 1995 to 2005 has been marked by a huge number of mergers and acquisitions which markedly increased economic concentration 4 . 4

The wave of bailouts in the fall of 2008 also resulted in increased concentration because

204

Chapter 9

The lack of attention given to this question stands in contrast to the scrutiny that it received during the period of the New Deal5 . Similarly, questions about lobbying groups which received so much attention in the 1930s are almost never mentioned nowadays. Do such lobbying groups not interfere with a “rational allocation of factors”?

1.4 The promotion of neoliberalism In May 2008 the University of Chicago set up the project of establishing the Milton Friedman Institute. This is a $200 million project and the majority of the funds will be raised from alumni and business leaders. An initial set of donors at the $1 and $2 million level will be invited to become members of the Milton Friedman Society. The initiators of this project claim that the Institute will be a center for path-breaking research. Can such a claim really be trusted? Can one seriously think that the sponsors would be happy if this research leads to the conclusion that free markets do not regulate themselves and need some kind of supervision to ensure long-term sustainability? In this connection one should keep in mind the well-known sentence by Upton Sinclair: “It is difficult to get a man to understand something when his salary depends upon his not understanding it6 .” As we will see in this chapter and in the next, neoliberalism has far reaching implications not only in finance but also in the economic and social spheres. In the last chapter we will examine if there is a way out of the neoliberal ideology. If it ever happens it will probably be a rocky and arduous road. As an illustration one can consider the case of railroad privatization. Such experiments in Argentina, Britain and the United States ended in failure. In Britain privatization resulted in security failures which lead to many accidents and eventually to partial re-nationalization. In 2006 Network Rail, a state-backed organization took over the private company Railtrack. As a result, taxpayers faced unlimited fine for the errors that led to the Ladbroke Grove disaster in which 31 persons died and 400 were injured (“The Independent” 1 November 2006). Yet, three years later and in spite of such clear evidence, privatization of national railroads is still on the agenda of the reforms promoted by the European Commission. many failed firms were bought up by competitors. 5 See for instance the “Investigation of Concentration of Economic Power” by the Temporary National Economic Committee. 6 This statement was quoted by former Vice-President Al Gore during his decade-long campaign against carbon dioxide emissions.

Triumph of neoliberalism in economics

205

Fig. 9.1 The future home of the Friedman Institute at the University of Chicago. The project of the Milton Friedman Institute was launched in May 2008. The university provided half a million dollars in seed money and is seeking $200 million in private donations of $1 million or more. The building aimed for the Institute currently houses the Theological Seminary. One of the main purposes of the Institute will be to host outstanding young scholars, post-doctoral fellows, visiting researchers in order (in the words of George Pratt Shultz) “to prepare them to lead enterprises in all sectors and thereby strengthen market-oriented economies”. In August 2008 more than 100 tenured faculty members have signed a petition opposing the institute. Critics say that they are concerned the institute will be a partisan organization.

2 State power and cracks at grass-root level At the end of the twentieth century the United States was not only the world’s only superpower but also a highly successful nation in terms of scientific innovation, cultural creativity and ideological dominance. These achievements largely contributed to the quasi-universal acceptance of the neoliberal creed. Let us give a closer look at such achievements. • In the decade 1901-1910 the United States won only 2 Nobel prizes (including one for Peace to President Theodor Roosevelt) which represented 3.2% of the prizes awarded. In the decade 1919-1928 the American share doubled to 7.8% but remained modest with respect to other industrialized

206

Chapter 9

countries (e.g. Germany: 22%, France: 12%, UK: 12%). If we now move up to the decade 1981-1990 we see a situation where the United States completely outranks all other countries; its share of Nobel awards has jumped to 49% while the share of Germany which comes in second position is only 8.2%. The situation is almost the same in the decade 1999-2008 with a share of 52% for the United States and only 8.0% for the UK which comes in second position. Moreover, in all academic ranking of world universities (whatever the criteria used) there is an overwhelming proportion of American universities in the top 50. • With corporations such as Microsoft, Google, eBay or Wikipedia7 the United States is in a monopolistic situation for the computer and Internet industry. There is a similar situation in agribusiness and biotechnology where firms such as Monsanto enjoy a near monopoly. • The American film industry and TV channels are present worldwide8 . • The American industry has more world class brands than any other country, e.g. Coca-Cola, Disney movies and recreation parks, MacDonald fast food restaurants, Starbuck coffeehouses, Nike and so on. General Motors and Ford are present through their subsidiaries in many countries worldwide9 . • In terms of military capability the United States enjoys an overwhelming domination. In 2007 the military spending of the United States which reached $500 billions represented 45% of the world total and was about 10 times the spending of China which came in second position. With the advances made in Eastern and Central Europe during the past 20 years the network of American military bases (over 700 worldwide) is more extensive than ever. • During the whole period of the Cold War the United States was able to suppress Communist movements worldwide. It is true that the Vietnam War marked a setback but this failure remained isolated in the sense that 7

The founders of Wikipedia make the claim that their foundation is based on international cooperation which is probably true; nevertheless, close to 100% of its donors are American, at least for those who are not anonymous as is the case of over 50%. 8 For instance in the cafeterias or meeting places of many Japanese universities there is a wide TV screen which exclusively broadcasts CNN programs. 9 The non-military aircraft division of Boeing has tried a similar operation by sharing the manufacturing of the 787 Dreamliner among half a dozen countries. This policy in which the core company provides only the design has made the success of Coca-Cola for decades; it is obviously more difficult to implement for complex products such as cars, ships or aircraft.

Triumph of neoliberalism in economics

207

there was no domino effect. Toward the end of the 1980s a clever policy was able to win over several East European countries and to outmaneuver the Soviet Union. As a testimony of the wide support the United States was able to gather even without the approval of the UN Security Council, it can be mentioned that 38 countries took part in the occupation of Iraq in 2003. Of these, there were 18 East European countries but only 3 from Latin America; whereas most of the 18 stayed until 2008, the Latin American countries were among the first to withdraw their troops in early 2004. In short, by 2000 the United States was highly successful in what can be called high end achievements. Yet, at grass-root level the picture was fairly different. Top level research is thriving but high school education is struggling. Although American corporations were highly profitable for decades the real wages of their employees had been declining since 1975. The worldwide extension of US armed forces is unparalleled but at the same time it has become a separate, closed-off and self-sufficient entity within the American society. Although foreign US policy leads the world the relations between the Federal government and several of the 50 states are becoming increasingly strained. Such a picture would not be surprising in the case of a dwindling colonial empire when falling income from colonial possessions brings about a fiscal and social crisis. Such was more or less the case of the UK after World War II. In the present case, we are facing a completely different and, to my best knowledge, fairly unique situation.

3 Income inequality: the watershed of 1975 First we consider four graphs which document a major change in the income pattern. • The first graph shows that there was a drastic change around 1979. In the 3 decades before 1979 the growth of earnings was shared fairly uniformly by all income groups. As a matter of fact, the lowest income group benefited slightly more than the highest income group. It can be recalled that prior to 1980 the marginal tax rate in the highest tax bracket was 70%; subsequently it was lowered to 35%. • The second graph presents basically the same evidence. We see that between 1965 and 1978 the income of workers and chief executives progressed at the same pace but after that the two lines separated. The earnings of chief executives doubled while the income of employees stagnated. Note

Chapter 9

208

Household income growth by quintile, 1947−1979

140%

120%

120%

101%

107%

114%

20−40%

40−60%

60−80%

94%

100% 80% 60% 40% 20% 00% 0−20% Lowest income quintile

80−100% Highest income quintile

Household income growth, 1977−1994

90%

72%

70% 50% 25% 30% 9%

−10% −30%

14%

4%

10% −8%

−1%

−16%

0−20% Lowest income quintile

20−40%

40−60%

60−80%

80−90%

90−95%

95−99%

99−100% Top 1% centile

Fig. 9.2 a Change in the income pattern in the United States. The data refer to real (i.e. adjusted for inflation) incomes. After 1979 substantial income increases were confined to the highest income groups and particularly to the top 1%. Source:“Wealth and Democracy” by Kevin Phillips (2002, p. 138), primary sources: Economic Policy Institute, Congressional Budget Office.

that in the following 20 years the growth of executive pay was even faster. As an illustration, in 2007 Barclays paid its president more than $40 million (The Independent 9 November 2008). In relative terms such levels of earnings are similar to the income of dukes and princes in former times. Moreover executives are prepared to go at great length to keep such privileges even to the point of putting their company at risk. The article of the Independent cited above mentions that Barclay’s shunned aid by the British

Triumph of neoliberalism in economics

209

Real incomes in the United States (1968=100)

government to avoid being forced to curb bonuses. Instead the bank preferred to solicit the (hypothetical) aid of Middle Eastern investors.

200

Real income of corporate chief executives Real income of hourly-paid production workers

180 160 140 120 100 80 60 1965 1967.5 1970 1972.5 1975 1977.5 1980 1982.5 1985 1987.5 1990

Fig. 9.2 b Increasing income inequality in US corporations. The graph refers to real (i.e. adjusted for inflation) annual incomes. A growing gap developed after 1979 which comes in confirmation of the previous graph to suggest that there was a deep change in US income patterns in the mid-1970s. In 1999 the average (pre-tax) compensation of the chief executives of the 200 largest US firms was $8.3 million which represented 419 times the wage of the average factory worker; in 1980 the ratio was 42. Source: Adapted from “Wealth and Democracy” by Kevin Phillips (2002, chart 3.22); primary source: “The Economist” June 1989

The two previous graphs were taken from a book by Kevin Phillips entitled “Wealth and Democracy”. It is interesting to note that Phillips was a close adviser of President Nixon. • The third graph compares wage level and strike frequency. We see that after 1975 real wages decreased while at the same time the number of strikes fell dramatically. But this evolution started already after World War II. Several Laws were passed in Congress which strengthened the position of employers and at the same time curtailed the rights of unions. Because this policy was to some extent shelved during the second term of President Eisenhower and because collective labor contracts with employers were usu-

Chapter 9

700

Taft-Hartley Act (1947)

10 2 Wagner Act (1935)

Number of strikes, USA (per million population)

210

600

500

Strikes

10

400

300

Weekly earnings of production workers (dollars of 2000) 1

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

Fig. 9.3 Consequence in terms of earnings of the loss of influence of American labor unions. The Wagner Act was not the first pro-union law of the New Deal era nor was the Taft-Hartley Act the last anti-union law in the decade following 1945. The Wagner Act in fact replaced (and extended) the labor legislation contained in the “National Industrial Recovery Act” (NIRA) of 1933 after the latter was declared unconstitutional by the Supreme Court in 1935. The Taft-Hartley Act was passed in June 1947 shortly after the Loyalty Executive Order (No 9835 of March, 1947) which submitted all federal employees to a loyalty oath. It was followed by numerous indictments of union leaders based on the so-called “Thought Control Smith Act” (1940), by the McCarran Internal Security Act (September 1950), the McCarran-Walter Immigration and Nationality Act (1952) and the Communist Control Act (August 1954). Under the cover of fighting Communism, these laws in fact considerably weakened labor unions. Sources: Statistical Abstract of the United States; Website of the US Department of Labor.

Triumph of neoliberalism in economics

211

ally signed for a duration of 5 years, the effects of this legislation were felt only progressively. Since 1970 there has been a dramatic decline in the number of strikes. Real wages reached their maximum in 1973. After this date the largest part of GDP growth was absorbed by the growth of non-salary earnings (e.g. income from financial assets, real estate profits, etc.) the share of which increased from 35% in 1950 to 55% in 2004. • The fourth graph provides a comparison between the United States and three other industrialized countries. It shows that with the possible exception of the UK which presents a similar (but less marked) trend the other countries do not show the same evolution.

4 Ownership by “absentee landlords” Between 1947 and 2000 the assets of American mutual funds as a proportion of GDP has been multiplied by a factor of the order of one hundred10 . Mutual fund giants such as Fidelity Management and Research (FMR), Vanguard Group, Capital Research and Management or State Street, are the major shareholders (with percentages over 5% and up to 15%) of many large American corporations. The term “absentee landlord” became widely used in the 19th century in relation with English owners of large Irish estates. “Absentee” refers to the fact that these owners spent most of their time in London or in vacation resorts of the French and Italian Riviera. More fundamentally, it refers to the fact that they were only interested in short-term financial returns. Because they had no knowledge of and no interest in the problems of their estates they were reluctant to spend money on investments that would have brought a return only in the long-term. The stewards who managed the estates had of course a better knowledge but they were powerless as far as investment decisions were concerned. The situation is similar for holding companies, mutual funds or private buyout companies. Their interest in the companies that they own is restricted to immediate financial returns11 . Moreover, because of their lack of empathy 10

More details can be found in Roehner (2006, p. 270). This situation is plainly apparent in economic and financial newspapers. Very little attention is given to the technical issues faced by companies. Ninety percent of the content is devoted to questions about mergers, acquisitions, changes of top executives, stock prices, bond issuance and other financial issues. Moreover all these topics are considered in a highly compartmentalized way. Thus, until recently it was not realized that problems in the real estate sector can have an impact on financial markets. Even nowadays it is hardly ever realized 11

Chapter 9

Income share of top 0.1 percent households (%)

212

7 6

U.S. U.K. France Japan

5 4 3 2 1 0

1950

1960

1970

1980

1990

2000

Fig. 9.4 Income inequality in several countries. Vertical scale: share of national income earned by the 0.1% of the households with highest income. The data are based on incomes reported to the fiscal administration; they represent incomes before the payment of income taxes and exclude capital gains. The thick curves are moving averages of the data points. Under an egalitarian distribution of income the top 0.1% would earn 0.1% of national income. According to the present graph their share is 20 larger in 1970 and 60 times larger in 1998. Sources: US, UK, France: Piketty and Saez (2003); Japan: Moriguchi and Saez (2004).

they are in the same position as Stanley Milgram’s instructors (see Roehner that the main engine of an economy is the income earned by employees. If this income is depressed through large scale imports of cheap labor there can be no sustainable domestic demand.

Triumph of neoliberalism in economics

213

2007, chapter 8) who inflicted painful shocks on their subjects without much hesitation. This can be illustrated by the attitude of employers with respect to fatal accidents in the workplace. According to an investigation performed by journalists of the New York Times (Barstow 2003) it appears that over the period 1982-2002 there were 2,197 cases in which fatal accidents were due to deliberate violations of safety laws by employers12 . Even more revealing of the Zeitgeist which prevailed during this period is the attitude of the federal government. • Only 1,242 of these 2,197 deaths were ever investigated by the federal “Occupational Safety and Health Administration”. Moreover, in 93% of the cases that were investigated, OSHA declined to seek prosecution. • Former OSHA officials say that those who were pushing for prosecution were hardly ever rewarded and in some cases they were penalized. • When Congress established OSHA in 1970 it decided to make it a misdemeanor13 to cause the death of a worker by willfully violating safety laws. The maximum sentence, six months in jail, is half the maximum for harassing a wild burro (small donkey) on federal lands. As a result, the average prison term per worker’s death was only 5 days. The fracture between management and workforce is not only harmful for employees, it also results in poor global efficiency. This means that in the long term such a system can hardly survive if competitors are allowed to emerge. Even though the previous observations point to a deepening of the divide between employees and employers in recent decades this should not be interpreted as a steady and linear trend. There was already a broad divide in the 19th century which to some extent narrowed in the first half of the 20th century. As an illustration of the turn of mind which prevailed in the early 20th century one can cite the following excerpt from a Federal report on industrial relations14 : The lives of millions of wage earners are subject to the dictation of a 12

The total annual number of fatal workplace accidents in the same time interval was about 5,000. 13 As opposed to a felony, a more serious criminal act. Petty theft, prostitution, vandalism, public intoxication are misdemeanors. 14 The excerpt is from the “Final Report of the Commission on Industrial Relations” and is cited in Boyer and Morris (1955). As a case in point the report includes the copy of a cable addressed by J.P. Morgan from his castle at Aix Les Bains (France) as to the necessity of a low wage scale.

214

Chapter 9 relatively small number of men who are totally ignorant of every aspect of the industries which they control, except the finances, and are totally unconcerned with regard to the working and living conditions of the employees in those industries.

5 Weakened national solidarity One of the main roles of government whether at federal or state level is to improve collective living conditions for instance by providing schools, roads, bridges, police. Strengthening national solidarity is another important function of governments. In the past decades two factors contributed to reducing this role. The first factor was the influence of the neoliberal ideology. As one knows it proclaimed that the role of government should be limited to its most basic functions (army, diplomacy, police) and that the poor should care for themselves or be helped by non government organizations. Such a narrow conception of social solidarity was already prevalent in 19th century England. Revealing of this orientation was the institution of the workhouses. In order to get shelter and meals, poor people had to carry out absurdly useless tasks such as breaking stones. Similarly in India during periods of famine men,women and children had to break stones in order to get relief. For more details see Nash (1900) and Longmate (1974). The second factor was the erosion in tax-income at both federal and state level. This erosion can be attributed to the change in tax rules (also brought about by the same ideology as already discussed). However one must also mention the development of tax havens. In a report issued in 2005 the “Tax Justice Network” estimated that global tax revenue lost to tax havens exceeded $255 billion per year15 . Actually, when compared with the budget of the United States which is of the order of 2,000 billions, 225 billions is not a big amount. The main point is rather that there has been a rapid development of tax havens during the past 20 years and that this evolution will continue unless the government of the United States changes its stance on this topic. The position of the Obama administration on this important topic will be discussed in a subsequent chapter. 15

The source is: “The price of offshore”, Briefing paper (March 2005). It is true that this figure was contested by the “Center for Freedom and Prosperity”. However, the fact that this organization was created to lobby legislators in favor of market liberalization and offshore financial centers suggests that it cannot entirely be trusted.

Triumph of neoliberalism in economics

215

6 Appendix: Possible scenario for the economic crisis In this appendix we describe a scenario which shows why and how the triumph of neoliberalism may be connected with the financial crisis. It is quite by purpose that we make this point in an appendix. One should keep in mind that there can be no scientific explanation of single events. In other words, this scenario can neither be proved nor disproved. Depending on their turn of mind, some readers will find this scenario plausible or on the contrary unlikely. The only way to get closer to a scientific perspective is to find and compare other realizations of the mechanisms which were at work in this crisis (more on this in a subsequent chapter). Before describing the scenario let us list the main characteristics of neoliberal policies16 . • Privatization of profitable public services • Lower tax rates for companies and affluent people • Deregulation of existing markets and abstaining from regulating new markets (Internet, new financial products) • Liquidation (or disabling) of federal agencies concerned with environment issues, with health hazards in workplaces or with the control of financial transactions • Facilitating rather than opposing economic concentration by fusion, acquisition and buyout • Outsourcing • Substitution of short-term employment contracts to contracts providing long-term employment guarantees • Globalization of financial markets, including (to say the least) a great tolerance for off-shore banking and tax heavens17 • Globalization of labor markets; for industrialized countries this meant importing cheap labor. 16

As all lists of this kind this one is fairly subjective; other people with different interests and backgrounds would list other items. 17 The position of market liberals on tax havens can be assessed fairly clearly by the opposition generated by candidate, president-elect and then president Obama when he claimed that he would attack tax evasion and tax havens; see for instance on the Internet the comments made by Bob Bauman calling the legislation planned by the new administration a “pernicious legislation” and even using the terms “more fascist powers”. It can be recalled that president Roosevelt was frequently accused of organizing a fascist state. In a cartoon that appeared in the New York World-Telegram on 12 November 1935, a businessman writes on a wall: “The New Deal is a combination of George III [who fought Americans in the War of Independence], Hitler, Mussolini, Stalin”.

216

Chapter 9

• Discarding unions; it may be recalled that unionization is banned at Wal-Mart, the largest US company in terms of number of employees. • Drawing higher education and research into the field of market economy. This in particular implies markedly increased tuition fees. A possible scenario would consist in the following steps. 1 During World War II the financial aid provided by the United States took a form which was very different from what it had been during World War I. Through the Lend-Lease Act a total of $50 billion (equivalent to nearly $700 billion at 2007 prices) worth of supplies were shipped to the Allies: $ 31 billion to Britain, $ 11 billion to the Soviet Union, $3 billion to France and $1.6 billion to China. No repayment of these supplies was required. From an economic point of view Lend-Lease amounted to huge subsidies to the American industry. This was followed by the Marshall Plan. The funds provided to European countries had to be used to pay imports of American products. On the contrary during the First World War the United States made loans to the Allies whose repayment create endless problems after the war. After 1950 the policy of lavish credit supply continued through a succession of small wars (Korea, Cold War military expenses, Vietnam, Iraq) which amounted to subsidized orders for US companies. This set the scene of a world in which money supply and credit were plentiful.The way the Federal government responded to the financial crisis was again by increasing the money supply. 2 In the late 1960s American companies experienced an erosion in profitability to which they responded in two ways: by investing abroad in countries where salaries were lower and by limiting wages in the United States through imports of cheap labor. 3 Confronted to an unlimited supply of imported labor US workers and unions became increasingly unable to win strike contests. As a result the number of strikes plummeted and wages began to decrease (as documented in a graph given above). 4 In the decades after 1975 the share of wages and salaries in the US GDP decreased steadily from 76% to 45% with a parallel increase of capital gains (e.g. appreciation of stocks or real estate) and corporate profits. 5 As the US economy could hardly be driven by exports (there was a substantial deficit of the commercial balance in part due to the relocation of factories abroad) the only possible engine was domestic consumption and because of dwindling salaries consumption had to be propped up by cheap and easy credit. Thus, the scene was set for the financial crisis.

Triumph of neoliberalism in economics

217

6 Deregulation of traditional financial sectors and non-regulation of new financial products (derivatives, collateralized debt obligations, and so on) had created the tools for generating abundant credit. 7 In 1994 (Orange County bankruptcy), 1998 (Long Term Capital Management bankruptcy), 2001 (Enron Corporation bankruptcy), 2002 (Global Crossing and WorldCom bankruptcies) there had already been serious crises which showed the risks of derivatives trading. In most cases the problem was surmounted by increasing the money supply. 8 The real estate boom of 1996-2007 was yet another episode of economic growth based on an abundant supply of money. It was made possible by cheap loans and in return housing equity provided collateral value for consumption-oriented loans. 9 Then, in mid-2007 “something” happened (that we do not yet understand) which changed the attitude of major mutual funds and investment funds with respect to the sophisticated financial products that they had welcomed for almost two decades. Of course, credit crunches had already occurred in the past in the wake of the crises mentioned above, but not to the point of affecting almost all financial sectors (bonds, short-term loans, longterm loans, mortgages). In other words, this was something new, perhaps not in nature but certainly in scale. A mechanism which may explain the severity of the credit crunch that began in 2007 would be a growing divide between the strategic interests of investment funds (such as Fidelity, State Street Corporation, Vanguard Group, Washington Mutual Investment Funds, Wellington Management Company, Capital Growth Management) and investment banks. It may be that the rush toward financial products of ever increasing complexity and opacity eventually run against the interests of such funds by depriving them of the market control that the sheer amount of their capital assets would otherwise give them. The fact that up to 2009 the crisis caused the failure of several big banks but was much less disruptive for investment funds would lend some support to this interpretation. Further confirmation may emerge as the crisis unfolds.

Chapter

10

Triumph of neoliberalism in society

The concept of social capital was developed by Harvard political scientist Robert D. Putnam. In his book entitled “Bowling alone” he suggests that the more people have opportunities to meet one another the “healthiest” a community is. The main problem is to define terms such as “healthiest” and “social capital” in a fairly objective way1 . Needless to say, more work is required to better understand the role of social interaction. The triumph of neoliberalism in many countries provides a unique opportunity to analyze the effects of a drastic change in social interaction. Why? In the previous chapter we have already emphasized that neoliberalism is based on an individualistic conception. Not surprisingly, therefore, loss of social solidarity is one the most obvious consequences whenever neoliberal policies are implemented. How does such a loss of social solidarity affect our societies? The answer is not obvious because many effects are indirect rather than direct. It may seem that free-market policies do not prevent people from taking part in bowling associations, in bridge clubs or in astronomy workshops. Yet, if people must work longer, if the ultimate rational and pervasive purpose of our societies is to maximize the consumption of goods, then of course such activities will hardly be encouraged. In the present chapter we discuss some effects of social fragmentation that can be observed in present-days societies. In some instances, such as the changing trends in education, the connection with neoliberal reforms will be fairly clear. In others fields such as infant mortality or homicide rates this connection will be less obvious. Many of the views put forward in this chapter are presented tentatively. 1

That is what we tried to do in a recent study entitled: “Maximization of interaction as an evolution principle for social systems” (2009). B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_10, © Springer-Verlag Berlin Heidelberg 2009

219

Chapter 10

220

1 Leveling off in education trends

PhD degrees conferred annually (per million pop.)

The next graph highlights abrupt changes in educational trends. The lowest line refers to the percentage of high school graduates, the other to the percentages of PhD degrees conferred annually. How can we interpret this graph? The stagnation in PhDs may have several causes but a fairly obvious factor is the inability of middle-class families to pay such long studies. The decrease in high-school graduates may be related to another effect namely the high inflow of uneducated immigrants.

100 PhD

10 2

1973

90

High School graduates (%)

80 1968

70 60

10

50 40 30 20

1

10 1880

1900

1920

1940

1960

1980

2000

0

Fig. 10.1 Number of high school graduates and PhDs in the United States. A marked change in the trend of both curves occurred around 1968-1975. These changes suggest that two transformations took place simultaneously: (i) A contraction (at least relative to total population) of the upper middle class who previously had furnished a large part of PhD graduates. (ii) A widening of the poorest segment whose children leave school without graduating. It can be observed that in contrast with high school and doctorates, the proportion of Bachelors has continued to increase albeit at a rate that is smaller than during the decades before 1970. Sources: PhD: Statistical Abstract of the United States; high school graduates: Barton (P.E.) 2005: One-third of a nation. Policy Information Center Educational Testing Service.

Triumph of neoliberalism in society

221

2 Abrupt increase in the trend of incarceration rate The next graph documents a rapid increase in the incarceration rate after 1975. The homicide rate is showed mainly for the purpose of comparison. It is true that the homicide rate was high in 1972 when incarceration began to pick up but it was also high in 1932. It is probably a general rule that crime tends to increase when economic conditions become difficult. In the years after 1932 the homicide rate decreased by itself (without a jump in incarcerations) as economic conditions became better. In recent years incarceration rates continued to increase even after homicides had fallen back to their level of the 1960s. Another possible factor may be the privatization of many state prisons. Constructing and managing prisons has become a business whose “customers” are the inmates.

3 Change in the trend of infant mortality Fig. 10.3 shows the evolution of infant mortality that is to say mortality between birth and one year of age. Infant mortality is a kind of synthetic indicator of various social conditions such as the health of the mother, her working conditions, the healthiness of the home and so on. This graph shows the US rate relative to a sample of three other industrialized countries chosen fairly randomly. Surprisingly, in the present case the change in the trend occurred as early as 1950. It is in the 1960s that the rate became higher than in the reference countries. In 1975 the ratio was equal to 1.45 but there is no change of trend as in the previous graphs.

4 Increased segmentation of the American society 4.1 Why should one focus on social segmentation? It was already mentioned that the federal tax rate was greatly reduced for rich people in the late 1970s that is to say approximately at the time when many trend changes took place. The increase in income inequality is a sign of greater social segmentation. The changing attitude of the federal government was certainly an important factor. During the New Deal it can be said that the Federal Government sided with the workers and the middle class. It is in his famous address of

US incarceration rate (percent of population)

1

Incarceration rate Homicide rate

0.9 0.8

14

0.7 12

0.6 0.5

10 0.4 8 0.3 6 0.2 4

Homicide rate (per 100,000 population)

Chapter 10

222

2

0.1 1920

1930

1940

1950

1960

1970

1980

1990

2000

2010

0

Fig. 10.2 Rate of incarceration and homicide rate. The marked increase in the rate of incarceration after 1975 resulted in a reduction in the homicide rate which began around 1990. Between 1930 and 1955 the homicide rate fell fairly steadily; this fall seems to have little or no connection with the (small) increase in the incarceration rate that occurred between 1925 and 1935. It should probably be attributed to the improvement of living conditions during the New Deal era. In 2008 the average cost of housing a single prisoner was $ 46,000 a year (i.e. $ 3,800 a month, a figure which is higher than the median personal income of full time workers in the United States (The Independent, 15 February 2009). Sources: Statistical Abstract of the United States, various years.

7 April 1932 (that is to say at the beginning of his first electoral campaign) that Franklin Roosevelt introduced the notion of the “Forgotten Man”: “These unhappy times call for the building of plans that rest upon the forgotten man at the bottom of the economic pyramid.”

Infant mortality: (US or UK)/(France-Japan-Sweden)

Triumph of neoliberalism in society

223

2

1.8

US

1.6

1.4

UK 1.2

1

0.8

0.6 1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

2010

Fig. 10.3 Ratio of infant mortality rate in the US (or UK) to the rate in three other countries. Infant mortality refers to the mortality of babies between birth and one year of age. Because this variable reflects health care and living conditions in all social strata, it is considered a significant indicator of social welfare. The graph shows that in the 1950s there was a change in the trend. The new trend displays a growing gulf between the US (as well as the UK albeit to smaller degree) and other developed countries. Sources: Statistical Yearbooks of the respective countries.

This was not only rhetoric but was matched by sweeping reforms. Many of them were blocked by the Supreme Court especially during Roosevelt’s first presidency. For years the Wagner Act which gave so many new rights to workers and unions was not accepted as law by the employers because they

224

Chapter 10

expected it to be nullified by the Supreme Court. However, such an explanation would suggest that the whole matter depended upon the personality and political choices of American presidents. In short Roosevelt versus Reagan. It would be a very anthropomorphic explanation. It might be tempting to attribute some of the changes that we listed above (i.e. lower educational achievements, higher infant mortality) to the increase in income inequality. Strictly speaking that would be a mistake however. Indeed, it is rather the fact that real wages stagnated or decreased which made living conditions more difficult for low income people. If all incomes had been increasing (even though at different rates) that would have made greater inequality almost bearable. A closer inspection shows that the segmentation due to income inequality is only one (albeit a major one) of many symptoms of segmentation in the American society. From the perspective of physics it makes sense to examine the degree of segmentation for we know that greater segmentation basically means less overall interaction. Listing segmentation factors is a fairly indirect method for assessing interaction strength but it has the great advantage of being easy to use.

4.2 Persistence of the Black-White divide One might think that in the wake of the Civil Rights movements the divide between African-Americans and the rest of the country progressively narrowed. Complete integration can be considered to be achieved when the statistical indicators of the minority population become the same as those of the total population, that is to say same average income, same infant mortality, same percentage of college graduates and so on. Here we examine one of these indicators, namely infant mortality. Fig. 10.4 shows that instead of decreasing the gap rather increased. This is a surprising result because for other aspects (e.g. access of African-Americans to higher education) the gap narrowed. Perhaps is this outcome in some way connected with the factor that we examine in the next section

4.3 “White flight” For over 5 decades a phenomenon commonly referred to as “White flight” has lead middle-class people to move from city centers to distant suburban

Ratio of infant mortality rates: minority/total pop

Triumph of neoliberalism in society

5 4

225

Australia New Zealand USA

3

2

1

1920

1940

1960

1980

2000

Fig. 10.4 Ratio of infant mortality rate of the minority to the total population for three countries. The purpose of the graph is to assess the progress in the degree of integration of the minority. Somewhat surprisingly, in spite of the success of the Civil Rights movement, there was little progress in the United States (at least in terms of infant mortality). Sources: Statistical Yearbooks of respective countries.

areas. This phenomenon has transformed many city centers (e.g. Atlanta, Cincinnati, Detroit) into distressed areas. In fact, rather than a White flight the phenomenon is more a move away from poverty. Indeed, it concerns all middle-class groups whether White, Hispanic or Black. This point is explained in an article by Jego and Roehner (2006).

4.4 Gated communities Simultaneously, the development of so-called “gated communities” has provided a new form of self-chosen segregation. Gated communities are private areas in which new owners are accepted only if they fulfill certain conditions. In large-scale gated areas like retirement communities, there may be local stores, restaurants and businesses all within the secured area of the complex.

226

Chapter 10

As this notion is fairly difficult to define precisely the statistics provided by the American Housing Survey are uncertain. However, an Internet key-word search (performed in mid-November 2008) shows that the expression “gated Community” is used quite commonly as a commercial argument by property developers. Here are the results obtained for different key-words: • “gated community”: 3,870,000 websites found by Google • “exclusive gated community”: 137,000 websites • “upscale gated community”: 25,300 websites • “charming gated community”: 4,600

4.5 The self-sufficient world of US Armed forces Since the end of World War II military and veterans live in a world where many services are provided by the army: homes, stores, schools, hospitals, sport installations, vacation resorts, funeral services. In short, the personnel of the US armed forces lives in a kind of huge gated community. The root of this privileged status can be traced to the Servicemen’s Readjustment Act of 1944 (also called GI Bill). Its main objective was to provide college education for the 14 million World War II veterans. Tuition was paid directly by the government to the chosen institutions of higher education. The veterans also received one year of unemployment compensation. In addition they were eligible for low interest, zero down payment loans to buy homes and start businesses. Although the GI Bill served as a blueprint for subsequent legislation regarding the Korean or Vietnam wars those later bills were more restrictive. Nevertheless, they contributed to create a section of the American society which is a world in its own and they gave great influence to veteran associations such as the American Legion.

4.6 Deepening gap between employers and employees In the private sector the unionization rate dropped from about 30% in the 1950s to 8% in 2007. As noted earlier, the role of labor unions has declined in spite of a fall in real wages. In other words the sharp decrease in the number of strikes can hardly be interpreted as an expression of satisfaction. From the perspective of social segmentation it can be observed that the unions provided a bridge and means of communication between employees and employers. Their quasi-disappearance has cut off this link. In a sense the present trend can be seen as a return to the situation which prevailed during the 19th century. We do not yet see the high level of confrontation between

Triumph of neoliberalism in society

227

workers and employers which marked this century but it is probably in the making.

4.7 Aging and underfunded transportation infrastructures The Penn Central’s bankruptcy in 1970 was the final blow to long-haul privatesector passenger train service in the United States. The troubled line abandoned most of its remaining passenger rail service, causing a chain reaction among other railroads. The federal government stepped in and, in 1971, created Amtrak, a virtual government agency, which began to operate a skeleton service on the tracks of Penn Central and other US railroads. After privatesector reorganization efforts failed, Congress nationalized the Penn Central under the terms of the “Railroad Revitalization and Regulatory Reform Act” of 1976. A subsequent attempt to create Comrail in 1987 lead to failure in 1999. In short, there remain only few (fairly obsolete) lines for passenger transportation by train in the United States. As a result, transportation by car (including buses) and by air became the two most common forms of transportation. The nation-wide bus system operated by the Greyhound company is efficient but it is slow and for some reason it is little used by middle-class people. A good part of the road infrastructure is aging and deficient: 13% of bridges in the United States (i.e. about 80,000) share the same “structurally deficient” rating as the bridge that collapsed in Minneapolis in August 2007. In addition, an equal number are “functionally obsolete”. Of course, to make sense of such data one would need to compare them with similar data in other countries.

4.8 Decline in the membership of social organizations According to the website of a Christian organization that tries to encourage social gatherings2 , in the late 1960s and early 1970s there has been a drop of about 40% to 50% in membership in all sorts of organizations: the PTA3 , the 2

http://www.sccquest.org/gatherings/fall05.htm • PTA: “Parent-Teacher Association”, a voluntary organization bringing together parents and teachers. • Elks Club: “The Benevolent and Protective Order of Elks” (also often known as the “Elks Lodge” or simply “The Elks”) is an American fraternal order and social club founded in 1868. It is one of the leading fraternal orders in the US, claiming over one million members. • “Kiwanis International” is a global organization of volunteers headquartered in Indianapo3

228

Chapter 10

Elks Club, the Kiwanis Club, the League of Women Voters and the NAACP. In “Bowling alone” Robert Putnam documents the loss in membership of many civic organizations. Putnam uses bowling as an example. Although the number of people who bowl has increased in the last 20 years, the number of people who bowl in leagues has decreased. Putnam found that in time of economic depression as for instance in the 1930s, membership in social organizations experienced a marked decline.

lis, Indiana. It comprises approximately 8,000 clubs in 96 countries with over 260,000 adult members (2008). Kiwanis emphasizes service to children and youth. • NAACP (pronounced N-double-A-C-P), the “National Association for the Advancement of Colored People” is one of the oldest and most influential civil rights organizations in the United States.

Chapter

11

Triumph of neoliberalism in finance

“Hoarding has immobilized about $1.5 billion [representing 18 billion in dollars of 2000] of the national gold supply and has caused drastic deflation and credit contraction. A dollar hoarded destroys 5 to 12 dollars in potential credit. A nation-wide pledge was given to President Hoover by business organizations to convince their membership that investment in government bonds are as safe as the government itself and that with the creation of the Reconstruction Finance Corporation [on 22 January 1932] there is less likelihood of bank failures.” (New York Times 7 February 1932 p. 1) “Credit crunch pinches struggling companies. Corporations are in the grip of the worst credit crunch in a decade. The credit crunch is a collision between a slowing economy and the go-go days of just a few months ago, when it looked like almost any loan to a business was safe. The ability to raise cash has dried up for all but the most financially sound companies. It is likely to get worse before it gets better.” (USA Today, 5 January 2001) The excerpts of these articles, one written during the Great Depression the other published during the recession of 2001 show that: • Credit contractions are a major characteristic of most depressions and recessions. • Credit crunches start unexpectedly and abruptly; this is particularly clear in the second excerpt. It was also true in February-March 1933 when a new wave of hoarding swept the United States in the weeks preceding the inauguration of President Roosevelt. (see below for more detail) During the past two decades there have been (at least) 4 credit crunch episodes, namely: 1994, 1998, 2001-2002, and finally the one which started in August 2007 (more details below). Yet, the previous crises were much less B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_11, © Springer-Verlag Berlin Heidelberg 2009

229

230

Chapter 11

serious as the last one. Of prime importance was the fact that the tricks implemented by the Treasury (e.g. injecting liquidity in banks, buying shares, lowering the Federal Reserve interest rate) which seemed to work so well in all previous crises, did no prevent big banks (such as Sears or Lehman Brothers) and financial institutions (such as Fannie Mae and Freddie mac) from facing bankruptcy. In other words, one should not ask why there was a crisis but why it turned out to be so severe. In this chapter we examine the changes brought about in the financial sector by the neoliberal agenda. We will make this part fairly short for two reasons. (i) Much has already been written on the financial aspects of the crisis (ii) Because financial transactions involve many technicalities for which the required data are not always available, it is difficult to reach reliable conclusions. Financial instruments are tools which can serve different purposes. Ultimately, of course, the main objective is to secure a high return rate, but this involves many different steps. • Finance can be a powerful control tool. This is clear for holding companies. in this form of ownership a variety of firms are bought up, aggregated to larger corporations or partitioned into smaller companies and sold. Most often, this kind of strategy is dictated by the perspective of short-term financial profit. • At macroeconomic level one can mention the requirements imposed on indebted countries by international organizations such as the World Bank or the International Monetary Fund. Usually, to get new loans, indebted countries were compelled to reduce budget deficit either by compressing expenses or by selling and privatizing profitable state-owned companies. • Macroeconomic constraints often go hand in hand with control over the state’s monetary policy. Because such mechanisms are of ever increasing importance, we devote several sections to them. This leads us to a broader discussion of how countries can influence one another.

1 Aggravating factors It can be argued that neoliberal reforms made the present crisis more serious than previous ones. Among the aggravating factors, one can list the following.

Triumph of neoliberalism in finance

231

• The deregulation of existing financial markets that began in the 1980s compounded by the refusal to regulate in any way the new derivatives markets which came into existence after 1985 created an opaque world (largely based on deceit and graft as we will see); this was an important factor in the collapse of confidence in the summer of 2007. • Over the past 25 years the number of tax havens has more than doubled to around 70 worldwide. Over the same period the the volume of bank deposits held offshore in tax shelters has risen to $ 11 trillion (The Observer 17 November 2002, Business section, p. 3). • According to the recommendations of the Basle Committee, the debtto-equity ratio1 of banks should not exceed 11 which corresponds approximately to a capital adequacy ratio2 of 8%. Nevertheless, major British or American banks had a ratio which was substantially higher: the median debt-to-equity ratio of UK banks was 33 with a range from 18 to 65 (Financial Times 12 Dec 2008). In 2006, that is to say at least one year before the fall in their stock prices began, the average debt-to-equity ratio of the five major Wall Street Investment banks was 25, the highest was Morgan Stanley at 31, the lowest Merrill Lynch at 213 . Even these inflated figures do not truly represent the level of indebtedness of banks because they were able to upload much of their debt to Special Purpose Entities that is to say basically straw institutions which they controlled but whose debt did no longer appear on their balance sheet, a technique that was also used by Enron Corporation with the consequences that one knows. The present crisis is often referred to as the “subprime crisis” but this expression is inadequate and misleading. Subprime loans were also made during previous housing booms yet without the devastating consequences that we have seen this time. This statement can be illustrated and supported be sev1 An important point is how equity should be estimated? For banks the standard definition is: equity=assets-debt. As an example, consider a bank who holds the following assets (in million of euros): (i) Cash: 10 (ii) Government bonds: 15, (iii) Mortgage loans: 20 (iv) Other loans: 55; and whose debt to depositors is 95. then the equity would be: 100 − 95 = 5. In this example the debt-to-equity ratio would be 95/5 = 19 and the equity-to-assets would be 5/100 = 5%. For publicly traded banks shareholder equity (that is to say stockmarket capitalization) can also be a fairly straightforward measure of equity. However, in times of panic when share prices of banks are in free fall such a measure becomes less reliable. 2 The capital adequacy ratio is the ratio of equity to an estimate of the assets at risk. In our previous example the risk for cash and government bonds would be considered zero, the mortgage loans at risk may be considered 50% of the total, while for “Other loans” 100% may be considered at risk. This gives a total amount at risk of 65 and a capital adequacy ratio of 5/65 = 7.7%. 3 Source: Company Annual Reports, SEC Form 10k

232

Chapter 11

eral observations. • The run on “Northern Rock”, a major English bank involved in mortgage loans, occurred in September 2007 that is to say at a time when UK housing prices were still in the up-going phase. Even US housing prices had at that time declined by only 5.8%. This suggests that the crisis was more a financial crisis than a housing crisis. • What we wrote about credit card laxness at the beginning of chapter 3 (of the edition published in January 2001) became even more pronounced in subsequent years. Between 2000 and 2006, even though Americans’ real income was essentially stagnant credit-card borrowing rose by about 30% ; between 2003 and 2008 the profits of credit-card companies jumped 45%. Since credit-card companies, unlike most lenders, are allowed to change the terms of their loans at any time, people who borrowed money at, say, 9% may be required to pay 17% some time later. (The New Yorker 16 March 2009, article entitled: “House of cards”). • During the real estate crisis of 1990-1995 housing prices in the west and north-east of the United States fell by about 40%, yet without major consequences for Fannie Mae and Freddie Mac the two big government sponsored entities. There was only a 20% fall in their stock prices. On the contrary, during the present crisis their stock prices began to fall sharply in October 2007 at a time when real estate prices had fallen less than 7% and they were put in conservatorship (a status similar to receivership for banks) one year later in September 2008. At that time US real estate had experienced a fall of 9.1% that is to say much less than during the crisis of 1990-1995. • Even banks which had no exposure to US mortgage securities (e.g. the Icelandic banks) experienced great problems which lead to their nationalization. • By the end of 2008 for funds and other financial institutions the most “toxic” assets were OTC derivatives (OTC=Over-the-Counter) assets that is to say derivatives which were not standardized products issued by an exchange but products designed, created and sold by banks on a private basis. • The globalization of financial risk sharing explains that the failure of a major American or British bank had worldwide consequences. Although the previous housing boom (and bust) was also shared by several industrial countries (Britain, France, Japan, Sweden, the United States) the bank failure remained confined. For instance, the failure of the Saving and Loans thrifts did hardly affect financial institutions outside of the United States. These factors gave rise to numerous sale abuses. They all shared one common feature, namely that during a period which lasted several years and in

Triumph of neoliberalism in finance

233

some cases extended over several decades, such sales brought huge profits4 . In the following paragraphs we describe such a case in more detail. Abusive sales of loan insurances This case is described in an article published in the English newspaper “The Independent” (17 March 2009); it illustrates how customers were deliberately mislead by financial institutions. This problem was aggravated by the fact that regulatory agencies failed to take corrective action. During the real estate boom in Britain well established financial institutions5 sold protection insurance along with real estate loans. These insurances were supposed to cover house buyers in case of repayment difficulties due for instance to unemployment or illness. There were two main abuses in these insurance contracts: (i) the premiums (to be paid along with loan repayment) were very high compared with average market prices; it was estimated that their cost was five to ten times higher than what it should have been. For a mortgage a typical annual premium was about 400 pounds. (ii) the contracts contained conditions which in practice would prevent any claim. For instance, people were warned that any pre-existing health problem would invalidate the guarantee even if the claim was made for a reason not related to health problems. Thousands of complaints have been filled (at a rate of 800 per week) in 2008-2009. The Financial Service Authority (the British financial watchdog) did not take any action until early 2009. This is all the more revealing because the problem was identified by the FSA as early as November 2005 (see the FSA press release of 5 November 2005 particularly the section entitled “FSA calls on industry to improve sales practices urgently”). Some institutions were fined (e.g. a subsidiary of HSBC in January 2008 and Alliance & Leicester in October 2008) by the FSA6 . Yet no decisive action was taken until February 2009 when the FSA prohibited the selling of loan insurances in which the premium is paid in total at the start, the so-called single premium contracts7 . Many of the problems listed above have been identified by lucid economists 4

With rates over 50% and in some cases, such as the one described below, over 500%. They are not identified in this article; however another source names: Alliance & Leicester, Barclays, the Lloyds Banking Group which includes Lord’s, Halifax and Royal Bank of Scotland, Regency Mortgage Corporation. 6 These fines were in the million pound range whereas the earnings from single premium insurances were in the billion pound range. 7 More specifically, in such policies the entire premium is added to the loan as a lump sum and interest is charged on it. 5

234

Chapter 11

or even by the media. For instance the hefty bonuses and “golden parachutes” (bonuses for executives who leave a company after it experienced serious problems) doled out to top employees have raised protests in newspapers since 2001 but without any real effect. Similarly, the Sarbanes-Oxley Act of 30 July 2002 tried to correct some of the abuses revealed by the Enron bankruptcy. On account of what happened in 2007 and 2008 (which had many similarities with the Enron case in terms of opacity and deceit) it seems that this law was largely ineffective. This episode suggests that the problem will not be solved just by passing new laws. Unless the balance of power between reforming forces and lobbying groups changes, little will be achieved. In summary, it can be said that the crisis which began in July 2007 was not qualitatively different from previous crises. Credit crunches, subprime mortgage crises, real estate downturns or collapses in some derivatives markets had occurred in the past. What was new was the fact that these problems occurred simultaneously.

2 Greater segmentation The changes that we described in the previous section may seem fairly diverse. Yet they have one common characteristic. They brought greater segmentation in financial markets in the sense that the “distance” between the agents who subscribed a contract and the institution supposed to carry out the service described in the contract increased sometimes to the point of leaving the realization of the contract suspended in thin air. Several examples of such segmentations are given in a very interesting book by Frank Partnoy (2003). • Remember that credit default swaps are a kind of insurance against the default of a borrower. Because there were no disclosure requirements it was often impossible to know who ultimately held the contract (on the secondary market the contract could be sold just like a bond) and would fulfill the service specified in the contract8 . • In a sense structured finance and securitization are modern versions of a fairly common practice namely transferring the rights to payments owed by borrowers. 8

In the case of a bond this can hardly happen because the persons who ultimately hold the bonds when they come to maturity know which institutions have issued the bond and they will demand payment. For CDS one has the opposite situation in the sense that the last holder has to pay for the defaulted loan.

Triumph of neoliberalism in finance

235

As an illustration let us consider the case of General Motors which grants low interest loans to its customers. When the rights to such car-lease payments (i.e. what the car owner will repay each month) are transferred to another company or investor (let us call it C) this becomes a securitization process. For General Motors the big advantage is that these loans can be taken off its books. There is a downside however. General Motors has some information regarding its customers which may allow it to estimate their capacity to repay the loan. Ideally, one may think that this information can also be transferred to C. Yet, observation shows that a good deal of information is lost in the process. For instance, C may be a financial company which has not working understanding of the specifics of car-lease loans. The result will be that the ability of car buyers to repay their loans will become an abstract financial notion. The link between the quality of the debt and the financial situation of the debtors has been severed. Of course, this problem is amplified when the rights to payment are resold by C to another company D, by D to E and so on. • Following the bankruptcies of Enron (end of 2001) or WorldCom (July 2002), the bankruptcy fillings showed that these companies had several hundred creditors. Twenty years earlier companies of similar size would have had at most a dozen creditors (Partnoy 2003, p. 377). This was the direct result of the fact that new financial tools allowed risk to be spread on a larger scale. But again this resulted in loser links between borrowers and creditors. These three examples illustrate a general trend toward greater segmentation. Similar trends have been described in previous chapters in the social and economic spheres. In the last two sections we listed a number of factors which explain at least qualitatively why the present crisis is more serious. However, one has to recognize that qualitative explanations are not really satisfactory. The real goal would be to be able to predict the magnitude of the crisis. To do this, one has to rely on a comparative analysis of other major crises with the handicap that there are of course only few crises of such a magnitude.

236

Chapter 11

3 How can one follow the buildup of a speculative episode? Speculative price peaks for stock or housing prices can be identified fairly easily.

3.1 Stocks For stocks it is well known that in a long-term historical perspective the timeaverage of the ratio of share prices to annual dividends9 which are distributed to shareholders is on average around 20. Once the average price-to-dividend ratio of a stock market becomes higher than 30 or 40 and continues to increase one can be almost certain that such a situation will sooner or later become unstable and lead to a downturn. Although it is difficult to predict at what point the downturn will occur 10 one can be almost sure that it will happen.

3.2 Real estate In real estate markets the analog of the price to dividend ratio is the ratio of the price of a house or an apartment to the annual rent. It turns out that the long-term historical average of the price to rent ratio (PRR) is also in the range 15-20. During speculative episodes prices usually increase much more quickly than rent levels which means that the PRR will go up. During a speculative episode, housing prices can double11 which implies (if we assume that the rent did not increase faster than the consumer price index) that the PRR may have jumped to about 35. Another illustration was provided by Hong Kong between 1993 and 2001. From 1993 apartment prices were multiplied by 2.3 In this case data for rent changes are available and they show that the PRR increased from 15 to 25. After the downturn of 1998, both the price and the PRR fell back to their initial level of 199212 . 9

The so-called price-to-earnings ratio (or PER) is also commonly used; however, if one wishes to consider this question not from the perspective of the company but rather from the investors’s point of view, then it is more natural to use the price to dividend ratio (PDR). 10 One can remember that just before the downturn in March 2000 the PER of the NASDAQ market was close to 200. 11 In London between 1996 and 2007 housing prices adjusted for inflation were multiplied by 3. This is a fairly exceptional case however. 12 A graph of these variables can be found in Maslov et al. (2003, p. 1443).

Triumph of neoliberalism in finance

237

Can this level of 20 for the PDR or PRR be given a more intuitive interpretation? A PRR of 20 means that the annual rent represents 1/20=5% of the price. In other words, the yield of the investment will be 5%. What gives a special importance to this figure is the fact that the long-term historical yield of investment grade bonds is also about 5% 13 . On the basis of this 5% yield of bonds, the 5% yield of shares or apartments can be sen as a result of an arbitrage process. When the yield of real estate falls below 5% investors may prefer to hold low risk bonds (for instance Treasury bonds). In other words, the 5% yield will represent a kind of (long-term) equilibrium state. For a while the speculative episode may feed itself on capital gains, but capital gains are inherently unstable in the sense that any price fluctuation may be amplified and bring about the collapse of the bull market. Pushing the question somewhat further one may wonder what reason may explain the fact that the long term average of bond yields is of the order of 5%? If one admits that the inflation rate is low enough to be neglected in our calculation, then an interest rate of 5% will provide an annual return of 5 cents on an euro. To get a return of 100 cents on an euro will take 20 years. If the inflation is not low enough to be neglected but not over 2% then it may take about 30 years to get a return of 100 cents on a euro. In short for a 20- or 30-year bond the total of the coupon payment will be almost equal to the capital. Twenty or thirty years represents the time length between two generations. At this point it is difficult to give a more accurate justification of this connection. One can only speculate that in a virtual population in which the time length between two generations is 2 years the “natural” interest rate would be 50% rather than 5%. Coming back to the financial crisis, one may wonder what insight can be gained from the previous considerations. There were two major components in this crisis, real estate speculation and derivatives trading. • For real estate the situation if fairly clear in the sense that there were recurrent price peaks as shown in Fig. 11.1. The mechanism outlined previously provides a fairly natural explanation for the succession of these episodes. Everytime that for some reason (low interest rate, substantial rent inflation14 ) there was an acceleration in real estate 13 As the price of investment grade bonds, that is to say low risk bonds, is usually not very different from 100, this means that the coupon rate is of the order of 5%. For more details see Macaulay (1938). 14 For instance, the rapid development of the buy-to-let market was an important factor in the London housing bubble. It pushed up prices but at the same time it depressed rents which resulted in a low return and a high PRR.

Chapter 11

Median price of new houses in the West (1980 dollar)

238

200 (1.90, 0.73 ?)

(1.50, 0.81) (1.64, 0.78)

100 90 80

(?, 0.82)

70

ear

y per 8% 7 . 1

60 50 1970

1980

1990

2000

2010

Fig. 11.1: Median price of new houses in the West of the United States. The two numbers above the peaks give the amplitude of the peak (the ratio of peak price to initial price) and the amplitude of the fall in the downgoing phase (ratio of trough price to peak price). The whole curve has been smoothed using a three-year centered moving average. The value 0.73 for the last peak is based on a prediction made in Roehner (2006, p. 179). Source: U.S. Bureau of Census

trading this opened the way for a new speculative episode. It can be noted that this graph is restricted to the West of the United States; this is because prior to 2000 the real estate markets in different regions of the United States were not synchronized; thus the national price average does not show a clear pattern as does Fig. 11.1. By the way, the synchronization which came about after 2000 was certainly a factor which made the crisis more serious.

3.3 Bonds For bonds the analog of the price to rent ratio is simply what is called the yield which is defined as the ratio of the nominal rate of the bond (also called coupon rate because this percentage is printed on the bond’s coupons) to the price of the bond on the secondary market (it is called “secondary” to emphasize the difference with the “primary market” when the bond is offered to investors for the first time).

Triumph of neoliberalism in finance

239

For investment grade bonds the price usually remains confined within fairly narrow limits (e.g. 95 to 105). For this reason neither the price nor the yield are good indicators for stress in bond markets. A better indicator is the difference between the yield of corporate Baa bonds and the interest rate of Treasury bonds (of same duration). This indicator is referred to as the corporate bond - Treasury spread. Moody’s rating scale which range from AAA for the most secure (e.g. Treasury bonds) to C corresponding to a bond almost in default is divided into two classes: the upper class from AAA to Baa corresponds to investment grade bonds while the lower class from Ba to C corresponds to speculative grade bonds. Thus the Baa – Treasury spread is an estimate of how more uncertainty Baa bonds carry as compared with Treasury bonds. The graph below shows that every time there was stress in financial markets (for insistence as a result of a series of major company failures) the spread became larger. The peak which occurred in the fall of 2008 was particularly high. Over the period 1962 to 2008 the highest peak had been in early 1983 when the spread reached 4%. The fact that the peak of fall 2008 was 2 percentage points higher underlines the gravity of the present crisis. In early 2009 the spread was narrowing but this does not necessarily mean that it will not climb again in coming months. For instance during the period 1980-1984 there were several ups and downs: in mid-1980 the spread reached 3%, then it decreased to 1.5% ((mid-1981) before climbing to 4% and then falling to 1.2% in mid-1984.

3.4 Derivatives A common feeling is that the present crisis has been aggravated by distrust in complex derivatives. Thus, a natural question is whether one can find an indicator which would permit to estimate the degree of speculation that prevailed in the derivative market. This not an easy question however. There are many sorts of derivatives and they are traded in many different places especially if one includes over-the-market derivatives. In addition, because many of these products are fairly new there are few good sources. This is why it is difficult to get a clear picture of the return of such transactions. Although it is possible to give some quantitative elements, it will be seen that they raise additional questions. As is well-known there was a multiplication by 10 in total world derivatives outstanding contracts during the 10 years from 1998 to 2007. How should

4 3

LTCM

5

Enron + WorldCom

California banks + Savings and Loans

Yield spread: 10-year Treasury - Baa corporate bond

6

Lehman Brothers + others

Chapter 11

240

2 1 0

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

Fig. 11.2 Spread of investment grade bonds over Treasury bonds. The spread is the difference between the yield of Baa corporate bonds and the yield of 10year Treasury bonds. As Treasury bonds can be considered a securities with no uncertainty, the spread represents the equivalent in terms of yield of the uncertainty attached to Baa corporate bonds. The graph for speculative grade bond yields minus 10-year Treasury yields has the same shape. It reached 20% in the fall of 2008 and then declined to 16% in early March 2009. Sources: Graphs and data for spread (Internet)

one interpret such a rapid development? One factor was certainly that derivatives were were profitable especially in the 1990s. But why were these products so profitable? After all they were only traded by trained persons (e.g. fund managers, hedge fund traders); why were they willing to buy at a price which gave a return of 30% (or even higher) to sellers. This is hardly the mark of an efficient market. One reason may have been that trading in complex derivatives allowed pension funds or insurance companies to make (in a oblique way) transactions in risky products they were not allowed to make by law. Thus, one can understand that to some extent they were willing to overpay the derivatives. According to another “theory”, the huge increase in the volume of transactions was a parade against the decline in the profit rate of individual transac-

Triumph of neoliberalism in finance

241

tions. But before such a thesis can be accepted one would need at least some data documenting the decline of profit rates in the course of time. It is true that the ratio of market values to notional amounts fell from 3.1% in 2001 to 2.2% in 2007 (Triennial Central Bank Survey of Foreign Exchange and Derivatives Market Activity in 2007, 19 December 2007), but this cannot be considered as a convincing proof of a fall in the rate of profit.

4 Spreading neoliberalism: overlords and vassals According to the Collins online dictionary a tribute is a payment made by a state to another, usually as an acknowledgment of submission. In feudal societies it was the amount of money paid by a vassal to his lord.

4.1 Dollarization A modern version of the tribute is the seigniorage paid to an overlord country A by a country B who uses the currency of A. B is usually either a small country who is politically under the influence of A or a country which experienced economic troubles (e.g. high inflation rates or inability to repay its debt) and had to ask for the help of A. The notions of seigniorage, full dollarization, unofficial dollarization, full currency board that will be explained in this section provide an insight into the international implications of free capital markets at the level of central banks and national currencies. But first of all, in order to make this more concrete let us give an illustrative example. Dollarization in Ecuador • March 1999: Ecuador is in trouble. The country’s 39 commercial banks, the targets last week of huge withdrawals by depositors who feared the banks were on the verge of collapse, had to close until the situation improves. Earlier austerity moves had frozen salaries of public workers and ended subsidies on gasoline. Since January 1999 the value of the sucre, the national currency, has fallen in dollar terms by more than 40 percent, with most of the decline occurring last week. President Mahuad, a Harvardeducated lawyer, plans to increase the value-added tax from 10% to 15%, a measure that has drawn strong opposition from the Chamber of Commerce whose president declared: “You do not build a country with more tax increases”. There had been speculation that the president might follow the path taken by Argentina and decree either parity between the sucre and the dollar or outright “dollarization,” in effect abandoning the sucre altogether. He did not take such a step, but said it is possible he may do so later. (New York Times: 13 March 1999)

242

Chapter 11 • Ecuador, January 2000 In 1999 the economy contracted more than 7% and the country defaulted on about half of its $13 billion foreign debt. From Jan. 5, 1996, through Jan. 14, 2000 the Ecuadorean currency declined 88% against the dollar President’s Mahuad’s decision to abandon the currency, the sucre, in favor of the dollar won the support of banks and many business groups. The essence of the government’s program is to replace the sucre, named for Ecuador’s national hero, Marshal Antonio Jose de Sucre, with the US dollar. Yet, in the rest of the country there is little support for this decision. Even the military, which has supported Mr. Mahuad in past crises, has offered only lukewarm support. The president of the Central Bank, his top two deputies as well as the bank’s assistant manager resigned to protest the president’s decision. The new currency plan is a major setback for the International Monetary Fund. An agreement was in sight for a $250 million loan. After the president’s decision the managing director of the IMF, Michel Camdessus, issued a terse declaration. Apart from Panama whose currency is the dollar since 1903, many Latin American nations already rely on dollars, to one degree or another. In Peru customers in many supermarkets can pay in Peruvian soles or in dollars. According to a World Bank official, in Argentina, a lot of the costs of dollarization have already been incurred, so the step to full dollarization would not be great. According to an economist full dollarization eliminates devaluation risk, and, consequently, will likely result in interest rates which are both lower and less sensitive to crises elsewhere. Dollarization means that the interest rate will be set by the Federal Reserve in the same way as the interest rate in the eurozone will be set by the European Central Bank. Alan Greenspan, the chairman of the Federal Reserve, made clear that the monetary policy of the Fed would not take into account cyclical differences between dollarized countries and the United States. For instance an interest rate hike, intended to slow expansion of the American economy, might send Ecuador into recession. In times of conflict the United States may use the dollar as an economic weapon, withholding currency from an intractable government to bend it to Washington’s will. Indeed, the Bush administration took just that step against Panama as part of its effort to overthrow General Manuel Noriega. Seigniorage is the difference between the production cost of coins and bills (which is low especially for bills) and the revenue obtained by the central bank by selling the currency to banks. In other words it is the revenue generated by issuing new currency. In the case of Ecuador, seigniorage represents $30 million a year. With dollarization in effect the Ecuadorean government will have to buy the dollars from the Federal Reserve. (New York Times: 11 January 2000, 12 January 2000, 18 January 2000)

Triumph of neoliberalism in finance

243

• The collapse of the Mahuad government began on 21 February 2000 when Indian protesters marched on the national Congress in Quito. A military unit assigned to guard the building instead stepped aside, allowing Mr. Vargas and his supporters to enter and proclaim the overthrow of Mr. Mahuad and the dissolution of Congress. Deposed president Mahuad was replaced by the the vice-president, Mr. Noboa who also favors dollarization. (New York Times: 23 January 2000)

The episode of the shift to the dollar in Ecuador clearly suggests that such a move may not be welcomed by the population. Yet, for different reasons the period 1999-2001 seemed a good time for the dollarization of Latin American countries. 1) The euro was introduced in banking and foreign exchange transactions in January 1999 and for the general public in January 2002. This set an example which, quite understandably, the United States were tempted to follow by creating an integrated American market in which the dollar would be universally accepted. NAFTA (North American Free Trade Agreement) came in effect in 1994. In this respect, one should recall that between 1946 and 1956 the United States greatly encouraged the formation of a political and economic union in the wake of the formation of NATO which was itself a response to the Warsaw Pact15 . 2) The crisis of 1997-1998 in South-West Asia and in Russia has shown the devastating effects from speculation in currencies especially for small economies (one may recall that China was little affected). Dollarization has many drawbacks but one cannot deny that it provides a protection against speculation. 3) Back in 1999 just before the decline of the NASDAQ market American prosperity was at a peak. This was the time of the “New Economy”. As a result, the free-market ideology advocated by US economists was still in favor with most Latin American economists and political leaders. The collapse of the Argentinian economy was still in the making. Ecuador had shifted to dollarization in 2000, El Salvador in January 2001, and it was expected that Argentina, a much larger economy, would soon follow suit. As observed in an article published in the New York Times on 25 November 2000, these events reflected “a growing trend in Latin America to embrace the dollar as official tender”16 . Indonesia explored setting up a currency board based on 15 The formation of an European Defense Community was energetically encouraged by Secretary of State John F. Dulles. To no avail eventually. 16 The article notes that “El Salvador’s decision to adopt the US dollar as its currency won

Chapter 11

244

the dollar in February 1998 but was convinced not to do so by the International Monetary Fund (New York Times 20 February 1998). In January 1999, Rudiger Dornbusch, an economist at the Massachusetts Institute of Technology made the following observation: ”In this environment it is increasingly ridiculous to argue that every country must have its own central bank. Currency sovereignty is the right to have stagnant growth”. The same article also says that Argentina was better positioned than its neighbors to rebound quickly, in part because outside investors have faith that if they invest they will not face the risk of devaluation before they get their money back. At that time some Brazilian economists were considering whether to follow Argentina’s lead by linking the real to the dollar. (New York Times 2 January 1999). One year later Professor Steve H. Hanke, an advocate of dollarization who served as a senior economist on president Reagan’s Council of Economic Advisors, observed: “The only way to extinguish the frequency and reduce the magnitude of recurring currency crises is to put these little half-baked central banks out of business.” (New York Times 23 January 2000). Table 11.1 Articles about dollarization in the New York Times (1981-2008) 1981 − 90 1991 − 2000 2001 − 05 06 − 08 article/year article/year article/year article/year Dollarization Currency board

1.0 0.2

2.7 9.1

1.8 3.0

0.31 0.30

Notes: The table shows that both dollarization (i.e. adopting the dollar as national currency) and the establishment of currency boards (a milder version of dollarization) attracted US interest in the period 1991-2005. After 2006 these issues were dropped. Source: The data have been obtained by using the key-word search engines available on the website of the New York Times.

4.2 Currency board Hong Kong did it in 1984. Argentina in 1991. Estonia in 1992 and Lithuania last year. Soon, El Salvador and Jamaica might do immediate support from the United States Treasury Department”, an attitude in sharp contrast with the neutral wait-and-see attitude about dollarization in Ecuador one year earlier.

Triumph of neoliberalism in finance

245

it. Brazil is studying it. And even Mexico sees it as a possible path to financial stability. “It” is the adoption of the currency board. Last year, Lithuania pegged its lita to the United States dollar and has seen its foreign exchange reserves increase by 10 times. A currency board makes sense for Mexico because of its economic integration with the United States through the North American Free Trade Agreement. (New York Times: article of 5 February 1995 entitled “A Strong Leash for Currencies on a Rampage”) Under the currency board system a country establishes a peg with an anchorcurrency (e.g. dollar, euro or yuan). Countries that adopt such a system abandon monetary sovereignty to the central bank of the anchor-country which means that interest rates and supply of money are decided abroad. The sole responsibility of the country is to defend the value of the local currency at the fixed exchange rate, which can only be changed by Congressional action (in the case of a currency pegged to the US dollar). In short, the only difference with dollarization is that the country nominally retains its currency and the central bank (which becomes a completely passive institution) still receives seigniorage revenue by selling coins and bills to banks. For obvious reasons the US administration has been very reluctant to openly advocate currency board systems. Yet, because their adoption strengthened the power of the Federal Reserve, it was regarded with favor by neoliberal economists as attested by the following excerpt from an editorial published in the “National Review” (22 December 1997): Instead of abdicating responsibility, Treasury Secretary Robert Rubin should be promoting American interests. • First, he should press the self-strangulated Asian economies to adopt market-liberalization measures (as, indeed, Taiwan is doing). • Next, he should endeavor to restore long-term monetary stability in Asia. Ideally, that would mean persuading smaller Asian economies to adopt a full-scale currency board, in effect unifying their currencies with the dollar. That would be better than IMF loans pegged to tax hikes. The tone of the article of 1994 cited at the beginning of this section suggests that the multiplication of currency board pegged on the dollar is seen as a

246

Chapter 11

favorable development by the author17 .

4.3 The aftermath of the Argentinian collapse The decade 1990-1999 marked the high tide of the adoption of the neoliberal agenda in Latin America under leaders such as Carlos Menem (Argentina) who introduced a de facto dollarization in 1992, Fernando Color (Brazil), Patricio Alwyn (Chile), Carlos Salinas de Gortari (1988-1994), Ernesto Zedillo (1994-2000), Vicente Fox (2000-2006), three successive presidents of Mexico18 , Luis Alberta Lacalle (Urugay), Carlos Andres Perez (Venezuela). In June 1990 US President George H. W. Bush announced the “Enterprise for the Americas Initiative” with the goal of achieving hemispheric free trade by 2000. NAFTA (North American Free Trade Agreement) which came into force in 1994 marked the first major step in that direction. The collapse of the economy of Argentina was a watershed. The implementation of the neoliberal agenda had begun after the military took power in March 1976. On the agenda were wage freezes, deregulation, free trade, privatization19 , endless borrowing (and escalating foreign debt). For a while de facto dollarization (1991) lowered inflation, attracted foreign investors and boosted growth. Argentina became a show-piece of neoliberalism and globalization. As the IMF’s star pupil, the country received huge credits (in the early 1990s it was the world’s fourth largest recipient of foreign funds). But the debt-driven economy began to crash in 2000. In a sense the mechanism of the crash was not very different from what happened in 2008 in Britain, California, Dubai, Iceland, Ireland and East 17

The article goes on to say that “after the Hong Kong dollar was pegged to the United States currency calm was restored and capital flight arrested”. Well, this was written in 1994, but in the fall of 1997 a huge wave of speculation tried to break the peg between the Hong Kong dollar and the US dollar. This episode suggests that currency boards are not a sufficient safeguard against speculators. In an article published on the website of the Cato Institute (4 June 2003) Professor Hanke goes to the point of recommending an “orthodox currency-board rule” for China (PRC). Under such a proposition supply of money and exchange rates would be fixed mechanically without leaving any room for monetary policy. 18 Salinas was a graduate of Harvard, Zedillo of Yale, Fox of Harvard; Fox had also been an executive and President of Coca Cola Mexico from 1964 to 1979. These personal links with the United States translated into friendly diplomatic relations; it is also an illustration of the importance of universities like Harvard, Columbia and Yale as “nurseries” of heads of foreign governments. 19 On 6 September 1992 the New York Times run an article entitles: “The big push toward privatization in Argentina”

Triumph of neoliberalism in finance

247

European countries: credit scarcity slowed down the debt-driven economy and lead to the failure of over-leveraged banks; this, in turn, brought about withdrawal of funds invested on a short-time basis and pushed unemployment to high levels. During the Argentinian crisis unemployment reached 22%.

4.4 Historical roots The first attempt to establish a currency board seems to have occurred in 1849 in the British colony of Mauritius in the Indian Ocean. In 1912 the system was extended to Western Africa and in particular to Ghana and Nigeria with the creation of the West African Currency Board. Such systems in which the monetary policy of a country is under the control of another was not confined to colonies, it was also used when countries defaulted on their debt. In such cases one (or several) representatives of the creditor nation got a seat (and a de facto veto right) in the central bank committee of the debtor nation. Such a procedure can be seen as an adaptation to the case of a whole country of the receivership procedure which is used in the United States for bankrupted companies. In the receivership procedure an administrator who represents the creditors takes the direction of the board until the company is able to emerge from bankruptcy. Such a system was used recurrently by Britain and the United States in Latin America. Several cases of this kind are described in an interesting and well documented book by Nearing and Freeman (1926).

4.5 Other forms of oversight and control The establishment of a currency board gives a country A control over the monetary policy of a country B. Of course, other forms of oversight and control are possible particularly in the political, cultural and social spheres. The question is of interest for our purpose in so far that it helps to explain how the neoliberal agenda has been adopted on such a large scale and in a relatively short time span. The thesis that we present below could be discarded as being nothing more than a new version of conspiracy theory. It is true that in a general way it is difficult to separate exogenous influence from endogenous factors. What makes it possible here is the fact that we can study not one but several different cases. Thus, it becomes possible to identify regularities and pat-

248

Chapter 11

terns. These regularities in turn will lead to predictions. If the predictions are confirmed by subsequent observations it will give scientific credence to the present thesis. To develop such a test here would draw us too far away from our main topic but it is important to realize that it can be done. The phenomenon has its roots in the years after World War II. In the wake of the American occupations that took place in many defeated countries as well as in a few Allied or neutral countries20 the State Department was able to favor the emergence of governments, political parties, newspapers, radios (and later TV channels) who were both anti-Communist and friendly to American interests. In defeated countries such a Germany or Japan this was done through the tight control imposed on the governments which emerged during the occupation and behind the curtain of censorship. The purges performed in Germany and Japan offered an excellent opportunity to select people and ensure their lasting loyalty. In countries such as South Korea (or the Philippines) which were liberated countries this was done under the cover of martial law and military administration established to counter the Communist threat21 . The occupations lasted about 6 years22 . It should be noted that this was a much longer time span than in previous conflicts. After the defeat of France in 1815 and 1871 the foreign occupation lasted only 1 or two years and during this time there were very few (if any) attempts to control the French administration. How was it possible to extend the post-war control to the period after the occupation. An obvious answer is that even after the occupied countries regained their independence, they remained dependent on the presence of American troops for their defense. This was of course the case in Germany, Japan and South Korea but also in Britain, France (until 1967), Italy and Turkey. It is well known that in Japan the Liberal Democratic Party (LDP) has been in power almost continuously since its formation in 1955. It is also known that from the 1950s through the 1970s, the American Central Intelligence Agency spent millions of dollars in attempting to influence elections in Japan to favor the LDP against more leftist, less pro-American parties, such as the Socialists and the Communists, although this was not revealed 20

e.g.: Germany, Iceland, Italy, Japan, South Korea, Turkey. This is true for South Korea but even in the Philippines there were guerrilla groups which fought the domination of big landlords. In China a similar struggle took place which however had an opposite outcome. 22 More details can be found in Roehner (2008), a book about the occupation of Japan. 21

Triumph of neoliberalism in finance

249

until the mid-1990s when the New York Times exposed it23 . In fact, to invoke the influence of the CIA is to take a narrow view. The bottom line is that a country which relies on another for its defense must follow the lead of the latter at critical junctures. This was the gist of the partnership between vassals and their overlord. An episode which occurred in France in 1946 is quite revealing in this respect. During a crucial political meeting for the formation of a new government a motorcyclist delivered a message from General Billotte, deputy Chief of Staff, which said that “a Socialo-Communist government would be seen as a threat by our Allies; as a result they may consider reducing their commitment to guarantee our security” (Demory 1995). Similar pressure was applied during the crisis which preceded the return to power of General de Gaulle in 1958 at least until this return was eventually endorsed first by President Eisenhower and finally by the State Department. There are similar examples in Germany, Japan and the United Kingdom. For instance since the end of World War II the UK has been relying on US-made Polaris and Trident missiles for its nuclear submarines. Moreover, the information sharing agreement between British and US intelligence agencies translates into an unequal partnership because of superior means and funding on the American side24 . In countries such as Austria, Germany, Italy, Japan or South Korea which were occupied by Allied forces25 two of the most effective means to maintain a lasting influence beyond the occupation period were the following. 1 One of the first objectives of the occupation forces was to identify, investigate, neutralize and control the intelligence and counter-intelligence agencies of the defeated countries. Subsequently, when the country emerged from occupation status and tried to built up a new intelligence structure the close links established with American (or British) intelligence ensured that this structure would have close links with US intelligence. But this was an asymmetric relation which bordered on outright dependence. This point is illustrated in an article by Siegfried Beer (2003, section “Intelligence cooperation or dependence”). Beer’s article concerns Austria but there is little doubt that the same situation prevailed in other occupied countries26 . 23

Source: http://www.fact-archive.com/encyclopedia/Liberal Democratic Party (Japan) Other ways and means for keeping a handle on public opinions are described in Roehner (2007, p. 145-149). 25 Naturally, a similar argument can be made for countries occupied by the USSR but in order to get a clear insight into these cases one must be able to read Russian. 26 Beer writes: “There can be absolutely no doubt about the fact that the beginnings of Aus24

250

Chapter 11

2 It often comes as a surprise that German or Austrian people who had played a substantial role during the Third Reich nevertheless came to occupy important positions in the post-war era in organizations controlled by the US. As is well known German people had to fill in questionnaires (called “Fragebogen” the German word for questionnaire) In the American Occupation zone, the requirement to complete Fragebogen extended to all those over the age of eighteen 27 . This gave the Allies an extensive data base about positions occupied by Germans in Nazi or military organizations. Moreover, records of war crime trials was another source of information. Needless to say, that information could be used by the Allies to exercise pressure on individuals. Incriminating information could be “forgotten” or “withheld” for pliant people or it could be “reactivated” for the purpose of embarrassing persons who had a tendency to go in the wring direction. In other words, the existence of self-incriminating files provide a means of control As illustrations one can mention (i) Kurt Waldheim who was Secretary General of the United Nations from 1972 to 1981 and President of Austria from 1986 to 1992 but whose military activities in the Wehrmacht unexpectedly came to the attention of the medias in 1986. (ii) Otto Schulmeister who as editor of “Die Presse” and “Wort und Wahrheit” was an important figure in post-war Austrian journalism. In 2009 it was revealed that over two decades he had been working for the CIA; it is claimed that in exchange his activity during the Third Reich were kept under the rug. During the Cold War it was relatively easy to keep right-wing governments in power under the requirement of fighting Communism. After 1990, a new strategy had to be defined. As far as one can know, it relied on supporting pro-western youth groups. Such youth groups were established in Belarus (Zubr), Georgia (Kmara), Kyrgystan (Kelkel), Russia (Nat-Bols of Edward Limonov), Serbia (Otpor), Ukraine (Pora). In the case of the Otpor youth group this can be illustrated by the following excerpts from the New York Times. American assistance to Otpor and the 18 parties that ultimately ousted Milosevic is still a highly sensitive subject. But Paul B. McCarthy, an official with the Washington-based “National Endowment for Democracy” is ready to divulge some details. Of the almost $3 million spent by his group in Serbia since September 1998, he says that Otpor was certainly the largest trian intelligence after World War II were marked by heavy, almost intolerable dependence on foreign influence. 27 In the British zone only candidates for public or semi-public jobs had to complete it; by December 1946 about 1.5 million Fragebogen had been completed.

Triumph of neoliberalism in finance

251

recipient. Money went into Otpor accounts outside Serbia and McCarthy held a series of meetings with the movement’s leaders in Podgirica, the capital of Montenegro, in Budapest [in the north of Hungary] and in Szeged [in the south of Hungary near the Serbian border]. At the “International Republican Institute, another Washington group financed by AID [USAID: United States Agency for International Development, closely related to the Department of State.] an official says that some of the $ 1.8 million the Institute spent in Serbia in the last year was provided direct to Otpor. (NYT, article by Roger Cohen (2000)).

The fact that the opponents to Milosevic received $ 25 million in American aid between July 1999 and August 2000 (NYT 23 September 2000) does of course not prove that this money played a “crucial” role in the uprising28 . A more convincing test is provided by the fact that the article of 23 September described in advance what would indeed happen two weeks later on 5 October. The article explicitly says that whatever the results of the election, they will not be accepted by the population, that “angry voters will take to the streets”, and that the same “people power” which spelled the end of the Marcos regime in Manila in 1986 will similarly bring down Milosevic. The article recognizes that the crucial factor would be the attitude of the police and army. According to a subsequent article (NYT 9 October 2000) it was Velimir Ilic, mayor of the town of Cacak, who was in charge of infiltrating the police with paratroopers of the Yugoslav Army and young policemen from Cacak. After such groups won political contests, pro-American governments came to power. In the case of Serbia, William D. Montgomery US ambassador to Croatia made no mystery of this objective when he declared “We hope the new generation of leaders will come from Otpor’s ranks. That would be a fair return on America’s investment in the movement.” (NYT 26 Nov 2000)

4.6 Privatization and economic return Once these governments were in power different reforms were introduced which strengthened the ties with the United States: privatization of state owned companies, training of the police and army by American instructors, establishment of bases especially for the US Air Force. One of the clear28

As a matter of fact for single events it is impossible to prove anything regarding the respective roles of exogenous versus endogenous factors; it is only for a collection of events that a scientific assessment can be made. The larger the sample of events under consideration, the sharper the proof can be made.

252

Chapter 11

est illustrations is provided by Georgia, but a similar trend was observed in Poland, the Czech Republic, Romania or Kosovo. It can be argued that privatization was the most important part of that program. A case in point was provided by the fate of the Prime minister in Slovakia. In 1989, Vladimir Meciar was chosen as Prime minister of the Slovakian part of Czechoslovakia by the new political party “Public Against Violence” (Verejnost Proti Nasiliu, VPN). In 1992 with his Czech colleague Vaclav Klaus he was at the origin of the division of Czechoslovakia into two separate countries. But in contrast to Klaus who was an “ardent free-market reformer”, Mr. Meciar was a hard-line nationalist and opposed to the privatization of big state-owned companies (New York Times articles of 2 January 1993, 13 February 1994). The article of 1994 was entitled “Slovakia is balking at privatization”. Less than 3 months later Mr. Meciar was ousted from his position as prime minister with the support of the movement OK’98.

4.7 Role of the armed forces This kind of influence appears in its clearest form in countries such as Afghanistan, Egypt, Pakistan or Turkey and more recently Iraq. In these countries the army and police were trained by American advisers and funded by the US government and they are the strongest organized forces in the country29 . In the past three decades whenever instability threatened, the government of such countries was routinely taken over by the military. As we know, this happened recurrently in Pakistan and Turkey. In a sense, in such countries there have been successful implementations of the scenario which was planned (and carried out unsuccessfully) in China30 and Vietnam.

4.8 The future Will the present economic crisis result in a weakening of this kind of influence? It is unlikely. The cost of running such operations is fairly low especially when compared to the cost of US armed forces; therefore the funding of such programs will not be much affected by the crisis. 29 Shortly after its inauguration the Obama administration was planning to spend about $ 2 billion a year over the next 5 years to build an Afghan armed force (army and police) of about 400,000. This funding would represent twice the budget of the Afghan government. 30 During World War II there was an extensive training program of Nationalist troops by American advisers. This operation pioneered a strategy that would be re-enacted many times in the post-war era..

Triumph of neoliberalism in finance

253

If the economic situation becomes more difficult it can become tempting to extract more economic advantages from countries which are ruled by proAmerican governments. In this perspective, toppling local dictators (e.g. Milosevic or Sadam Hussein) can be seen as a means for replacing them by governments who are friendly to US interests and willing to concede economic privileges to American corporations. There have been many historical examples of this kind in Latin American countries. Naturally, the United States is not the only country that pursues similar objectives. A clear illustration was given by the economic exploitation of China through the concession system (which prevailed from the mid-19th century until 1942 and was followed in November 1946 by an unequal economic trade agreement with the United States) in which almost all western countries took part. What makes the United States special is: (i) its unparalleled expertise in implementing such policies (ii) its worldwide dominance. Such policies complement the neoliberal agenda in the field of foreign relations in the sense that free trade and open door policies are necessarily to the advantage of the dominant partner. Although in recent years this policy was very successful in Eastern Europe, one may argue that it was less successful in Latin America. Indeed, in the wake of the economic crisis in Argentina governments were elected (for instance in Bolivia, Brazil, Venezuela) which openly questioned the neoliberal agenda and pledged to keep American interests at bay. Will this be a lasting trend. It is still too early to know for sure.

4.9 Why does privatization offer inroad opportunities to foreign companies? In many Latin American countries the privatizations carried out by liberal governments in the 1990s left major national companies in the hands of foreign corporations, especially in promising sectors such as telecommunication, water distribution, electricity production and distribution, oil production and distribution and so on. A similar scenario took place in Eastern European countries. Such an outcome was of course hardly surprising because Western countries could rely on a sophisticated banking sector which enabled them make competitive bids. However, this gave rise to what can be called an absentee landlord system (see Roehner 2007, chapter 8) which resulted in under-investment and mismanagement. Eventually such policies generated much public discontentment.

254

Chapter 11

An illustration of the absentee landownership to which privatization leads is provided by Mongolia. Although the country experienced a democratic revolution in 1989 the former communist party remained in power until 1996. The program of the party which came to power included the privatization of 60% of all state property. What were the results31 ? • The country’s largest government owned bank was bought (a 76% ownership by a consortium from Connecticut and Switzerland for $ 12 million. • Another major Mongolian bank was bought for $ 7 million by a Japanese group but it is managed by American executives. It was renamed XAAN. • In 2000 the government handed management of a Mongolian bank called Ag Bank to the US Agency for International Development which outsourced it to a Washington consulting firm32 . • Over the following two years (i.e. 2004-2005) other privatizations were planed, for instance the national airline, the largest petroleum importer, the nation’s largest insurance company. At the end of the privatization process most of these companies will be in foreign hands and owned by Western companies which have very little knowledge about Asia and more specifically about Mongolia. Such links are even weaker than between English landlords and their Irish estates back in the 19th century. The lessons learned from historical episodes suggest that in the long run absentee landlordship is an inefficient form of management. Yet, such a system can hold on for a long time if it is backed by state power. In the case of Ireland the system lasted for over three centuries.

5 Fate of abstentee ownership in Eastern Europe during the next decade The pro-democratic revolutions in Eastern Europe can be seen as a repetition of what happened in Latin America in the two decades 1982-2002. In the previous decade (1972-1982) several Latin American countries (e.g. Chile, Argentina, Uruguay, Bolivia) had been ruled by military dictators. Several of them were already pro-American and willing to implement neoliberal re31

The following information is based on an article from the New York Times (21 October 2003) entitled: “In Mongolia, a tilt toward a free market” 32 Since, as is well known, the USAID has close links with the Central Intelligence Agency this decision had certainly a political significance.

Triumph of neoliberalism in finance

255

forms; such was for instance the case of Augusto Pinochet in Chile who had direct contacts with Milton Friedman and other advisers from the University of Chicago. Once brought down, these dictators were replaced by civilian governments which were also pro-American and neoliberal. As a result many public services (in banking, telephone, water distribution) were privatized which often brought them under the control of foreign companies. In many cases these experiences were disappointing in the sense that they resulted in price increases while at the same time the foreign companies did not fulfill their promises regarding long-term investments in basic infrastructure. As already mentioned, this outcome lead to a rejection of neoliberalism and to a more guarded attitude with respect to foreign companies. In a sense, the crisis of 1997-1998 had similar results in Asia. It is not unreasonable to expect a similar evolution in Eastern Europe. At present (early 2009) many of the governments which came to power in recent years are much more pro-American and neoliberal than their electors33 . Once the failure of neoliberal policies will become apparent, it can be expected that people will reject them as they have done in Latin America. Of course, such a statement assumes that electors will be given a real choice. If the main parties in fact propose more or less the same program (as was the case with “New Labour” in Britain) then a growing gulf will develop between voters and the political parties supposed to represent them.

6 Previous speculative crises In many respects the crisis which started in July 2007 was just a repetition of similar episodes that occurred in the past. As illustrated by Fig. 11.3 speculative bubbles have occurred repeatedly since (at least) the nineteenth century.

6.1 Previous “subprime crises” More specifically, previous crises have already featured many of the issues that we have seen in the present crisis, e.g. sharp downturn in real es33

In Poland and in the Czech Republic this was revealed by several surveys on crucial issues such as the war in Iraq or the installation of anti-missile facilities; government policies were disapproved by a wide margin (60% or more).

256

Chapter 11

Fig. 11.3 The way to grow poor; the way to grow rich. This lithography was created in 1875 by the American artists Nathaniel Currier (1813-1888) and James Merritt Ives (1824-1895). It makes probably reference to the burst of a housing and railroad bubble in the fall in 1871 which lead to a financial panic in the fall of 1873. Between 1857 and 1880 the firm Currier & Ives created about 7,500 images. Source: Lithography by Currier & Ives, 1875. Library of Congress Prints and Photographs Division, reproduction number: LC-USZ62-662 (public domain).

tate prices (see Fig. 11.1), no-documentation loans34 , adjustable rate mortgages35 , boom in mortgage-backed-securities, tightened lending standards which led to a credit crunch (New York Times 18 February 1984 p. 1, 14 Jan 1990). The mortgage delinquency rate rose to 4.15 percent of total loans outstanding in the fourth quarter of 199436 , collapses of mortgage insurers37 . 34 Which means that loans were attributed to persons who did not show any official document to back their income statements. 35 Which includes loans whose interest rates were artificially low during the first initial years and were set to increase sharply in the following years. 36 The delinquency rate includes loans that are at least one payment past due but does not include loans in foreclosure. At the end of the third quarter of 2008 the delinquency rate stood at 7.0% of all loans outstanding; the percentage of loans in the foreclosure process was 3.0 percent (Mortgage Bankers Association). 37 Such as the failure of TMIC, a large mortgage insurer in February 1988 (New York Times 3 March 1988).

Triumph of neoliberalism in finance

257

On 11 October 1967 a new expression made its first appearance in the columns of the New York Times, namely the expression “mortgage backed securities”. The beginning of the article summarizes fairly well what was the purpose of these securities: “The Federal Government’s chief mortgage man hopes for mortgage market”. Indeed, back in 1967 there was no integrated national mortgage market in the sense that because of regional surpluses or shortages of credit, mortgage rates used to differ by as much as 2% between cities (NYT 22 Jan. 1984). In other words there were local credit crunches which lead to high credit rates. This lack of integration was due to insufficient communication links between local markets. It can be recalled that there was a similar situation in commodity markets before the advent of railroads. Back in the early nineteenth century in France wheat price differentials in distant cities were of the order of 50% (more details can be found in Drame (1991))38 . Even booms in mortgage-backed securities had been experienced many times in the 1970s and 1980s. In an article of 1977 one can read: “One of the fastest-growing innovations in the credit markets this year has been mortgagebacked securities” (New York Times 20 November 1977, p. F5). Detailed articles published in the New York Times (Arenson (1979), Berg (1984)) explain how mortgage-backed securities work and delineate some of the problems that may arise. According to a figure given in Wikipedia (article “Savings and Loans crisis”) the Savings and Loans crisis of the early 1990s cost taxpayers about $150 billion which represents 2.5% of the GDP of 1992. Transposed to 2008, 2.5% of GDP would represent about $300 billion. At the time of writing (March 2009) the cumulative amount of successive bailout plans for banks represents (at least) one trillion dollars that is to say three times more than the cost of the previous real estate crisis. This shows once again that the burst of the housing bubble is only one of the components of the present crisis.

6.2 Regulation, self-regulation or no regulation at all The article of 1979 about mortgage-backed securities mentioned above put great emphasis on the necessity of regulation. In 1979 real estate activity reached a peak (at least in the West) and there was much talk about possible federal regulation. Not surprisingly, in order to avoid federal regu38 The situation was basically the same in other European countries and it had dramatic consequences in times of dearth.

258

Chapter 11

lation mortgage dealers claimed that they would be able to introduce selfregulation. The “Government National Mortgage Association” was skeptical. One of its executives made the following comment. “We still have some doubts about whether it is possible to put together an effective self-regulation program. It is necessary for Federal agencies to consider alternatives to self-regulation in case the program is not put into place or is not effective”. Responding to this concern a mortgage dealer at Salomon Brothers declared: “We have been given a chance to self-regulate; I believe it is a last chance. If we don’t get something in place very soon, the game is over”. In fact, in the following years free-market ideology gained greater momentum with the result that no effective regulation, whether federal or industrybased, was ever introduced.

6.3 A highly profitable business Back in 1983-1984, mortgage trading had already became a very profitable business which is quite remarkable because interest rates were fairly high at this time. In his article of January 1984, Eric Berg gives the following illustrations. • The rise of the mortgage market has produced a bonanza for Wall Street. Nearly every major house has expanded its mortgage-trading activity. For example 40% of the net income of Salomon Brothers in 1983 was provided by mortgage-related commissions and trading profits. • In January 1984 the Savings and Loans industry was still recovering from a difficult position due to a depressed housing market from 1979 to 1982. The rapid growth of mortgage-backed securities to a total of $253 billion or 14% of the $1.7 trillion in mortgages outstanding gave them the possibility to jettison old, low-paying mortgages even sometimes at losses of as much as 50%. In other words, the mortgage-backed securities market enabled some savings and loans to stave off bankruptcy. • One of the problems of the mortgage industry has been the possibility given to home buyers to pay off their loans early. Freddie Mac addressed that problem by creating “Collateralized Mortgage Obligations” which reduced the risk of prepayment. All these observations suggest that the main financial innovations for mortgage markets had already been introduced in the early 1980s and were at

Triumph of neoliberalism in finance

259

that time working fairly well. In other words, the tools were probably sound and it is rather the abuses to which they gave rise which brought about the problems which lead to the financial crisis.

Chapter

12

Is there a way out?

It was said that Reaganomics was a Trojan horse for the rich. Under the cover of a new tax policy, tax rates on top brackets were drastically reduced from 70% to 35%. In a book published in 1986, David Stockman, Ronald Reagan’s budget director, writes that the main purpose of the tax reform was to bring down the tax rate on upper income brackets and that to make the reform palatable as a political matter it had to be presented as a new bold economic approach called supply-side economics1 . In the same line of thought as suggested in previous chapters the main themes of neoliberalism (which is the global version of Thatcherism and Reaganomics) included many demands which have been on the corporate agenda for decades. If we accept this assumption it must be possible to find other historical episodes during which neoliberal policies were either dropped or re-introduced after having been abandoned for a while. It is the purpose of this chapter to present and briefly discuss such episodes. We will describe five episodes. 1) The crisis of 1893-1897. This crisis is interesting because it has some similarities with the present crisis. It started as a financial crisis, became an economic crisis and ended in what is called the progressive era which was marked by a kind of New Deal. It was the time of Theodore Roosevelt’s “Square Deal”2 . After World War I the decade of the 1920s was again a period of frenzy and 1

In contrast Keynesian policies rely on stimulating demand and consumption and because lower-income earners spent a much larger proportion of their income than high-income earners they recommend to lower tax rates for cash-strapped households. Supply-side economics relied on the development of entrepreneurship and on the implicit assumption that “a rising tide will lift all boats”. 2 In many respects T. Roosevelt’s Square Deal prefigured F. Roosevelt’s New Deal. In 1906 he convinced Congress to create the “Interstate Commerce Commission” to regulate interstate railroad rates. It was the first true federal regulatory agency. B.M. Roehner, Hidden Collective Factors in Speculative Trading: A Study in Analytical Economics, DOI 10.1007/978-3-642-03048-2_12, © Springer-Verlag Berlin Heidelberg 2009

261

262

Chapter 12

speculation marked by an advance of neoliberalism. 2) The transition from the neoliberal agenda to the New Deal agenda in 1932-1933 in the wake of the Great Depression. 3 The resistance against the implementation of New Deal policies in the period 1933-1938. 4) The post-war period after the death of President Roosevelt in 1945 (at the beginning of its fourth term) and the end of the war economy was marked by a series of defeats for unions and labor. After 1947 with the buildup of the Cold War the progressive forces who had supported the New Deal suffered even greater blows. The economic effects of these setbacks were not immediately apparent because globally it was a period of great prosperity. For instance real wages continued to increase for about 20 years. 5) At the beginning of his term in the spring of 2009, president Obama gave several signals which suggested that neoliberal policies were no longer favored. At the time of writing (April 2009) it is still too early to say if these signals will open the way to new policies. However, the evidence currently available suggests that the president has opted for going along with business lobbies rather than to oppose them.

1 Crisis of 1893-1900 The depression which began in 1893 had some common characteristics with the depression which started in 2008. • In 2008 many US banks had to be bailed out by the government. In 1893 many banks and railroad companies had to be bailed out by the government. For instance on 16 August 1893 the Northern Pacific Railroad was placed in the hands of receivers (after having been declared in bankruptcy) for the third time. By February 1994 one sixth of the railroad mileage had passed into the hands of receivers (NYT 27 Feb 1894 p. 4). • In 1894 as in 2008 the situation of many banks was beleaguered by recurrent writedowns. An article of the New York Times of 13 August 1893 observes: “Commercial bank directors are not hopeful. A good deal of the paper is reputed worthless”. Two months later one reads the following judgment: “The most reckless and unscrupulous methods may be pursued by the directors of a bank until the institution has been plunged into bankruptcy” (NYT 25 Oct 1893 p. 4). • As can be seen from Fig. 12.1 the initial increase rates of unemployment were fairly parallel for the two depressions.

Is there a way out?

263

Unemployment rate (%)

Needless to say, it would be pointless to compare the two depressions in a global way. What would make sense is to compare specific aspects. For instance it could be interesting to compare the evolution of delinquent mortgages in the two episodes. It is not possible to develop such studies within the framework of the present publication, but we intend to do so in a subsequent paper.

20 18

2006

2008

2010

2012

2014

2016

1894

1896

1898

1900

Depression of 1893 Depression of 2008

16 14 12 10 8 6 4 2 0

1890

1892

Fig. 12.1 Unemployment in 1890-1900 on the one hand and in 2006-2009 on the other hand. As explained in the text the depression of 1893 had several common points with the depression of 2008. A 3-point moving average was used to smooth the curve for the depression of 1893. Source: Unemployment data for 18901900 are from Lebergott and Romer series cited in Wikipedia, entry: “Panic of 1893”.

Chapter 12

264

2 1932-1933: Runs on banks and closing of the Stock Exchange President Roosevelt’s inauguration took place on Saturday 4 March 1933. The New York Stock Exchange closed on Friday night3 and remained shut until Tuesday 14 March. During these 10 days there was also a banking moratorium which means that most banks were closed. This, in a nutshell, illustrates the dramatic circumstances under which the new administration began to work. Chronology 12.1

Events before and after President Roosevelt’s inauguration on 4 March 1933

Runs on banks and hoarding of currencies have been common features since (at least) mid-1931. However, in 1931 and 1932 only small amounts of gold were sent abroad. On the contrary the new run on the banks which started in mid-February 1933 resulted in a huge outflow of gold. As always it is difficult to say what triggered this movement. Was it a reaction of defiance against the reforms planned by the new administration or was it a panic due to deteriorating financial conditions? In response to this situation President Roosevelt took drastic emergency measures in the first days after his inauguration. Hoarding of gold was forbidden under threat of heavy penalties. The prohibition against holding gold was only lifted by President Gerald Ford. 1932, Feb 4 President Hoover appealed to the country to cease the hoarding of currency, as a patriotic move toward loosening credit. Hoarded funds were estimated by the President at $ 1,300 million every dollar of which, he said, meant the destruction of from $ 5 to $ 10 of credit. (NYT 4 February 1932, p. 1) 1933, Feb 13 Situation in real estate. Efforts to help the real estate situation which is one of the worst problems now besetting the United States resulted this week in the formation of the “Realty Stabilization Corporation” for New York city which is to borrow $100 million from the “Reconstruction Finance Corporation” in order to finance mortgages. (Times 13 February 1933 p. 19) 1933, Feb 15 With more bank failures there has been a recurrence of hoarding by nervous people. All the banks in the state of Michigan have been closed for one week which means a moratorium for all business transactions. (Times 15 February 1933 p. 19) 3 At this time the NYSE was normally open on Saturday; the only weekly closing day was Sunday.

Is there a way out?

265

1933, Feb 27 Runs on banks in different parts of the United States increased in number. Ten states have already enacted moratorium legislation. Congress has amended the “National Bank Act” so as to allow national banks under condition of approval by the Secretary of Treasury to withhold from depositors (for a period of 6 months) all but a small fraction of their balance. Hoarding did not confine itself to currency (as was the case in last year’s crisis) but also resulted in the hoarding of gold. This is probably due to the feeling that there will be a depreciation of currency. (Times 27 February 1933 p. 19) 1933, Mar 2 Injection of public money into the financial system. The “Reconstruction Finance Corporation was formed in February 1932 to protect banks, insurance companies, railways and other corporations in difficulty by the use of government credit. The amount of the advances that it was called upon to make has exceeded all expectations; it totaled $2 billion [in dollars of 2000 this would represent about $28 billion]. Among the factors which brought about this crisis one can mention: (i) Prices of shares and land were pushed to heights out of all relation to any return which could be expected from them. (ii) American banks lent money freely and themselves embarked upon highly speculative operations. (iii) Everyone was encouraged to speculate on margin that is to say with borrowed money. (Times 2 March 1933, p. 13) [When one reads the factors to which the Times ascribes the financial crisis which lead to the Great Depression it is tempting to draw a parallel with the current crisis.] 1933, Mar 4 The New York Stock Exchange closed on this day which is a Saturday whereas in normal times it used to be open on Saturdays. It will remain closed during 10 days that is to say until the end of the Banking moratorium on March 15. (Farrell 1972) [Similarly, there were high expectations after the inauguration of president Obama in January 2009.] 1933, Mar 4 Americans are a people of invincible hope. But seldom can their eagerness to see a new President inaugurated have equaled that of this year. It is a delayed realization of what they have been carrying in their hearts since last November. (NYT 4 March 1933, p. C12) 1933, Mar 4 Excerpt of the inaugural address of President Roosevelt. “Practices of the unscrupulous money changers stand indicted in the court of public opinion, rejected by the hearts and minds of men. Faced by failure of credit they have proposed only the lending of more money. Stripped of the lure of profit by which to induce our people to follow their false leadership,

266

Chapter 12 they have resorted to exhortations, pleading tearfully for restored confidence. They know only the rules of a generation of self-seekers. They have no vision, and when there is no vision the people perish. The measure of the restoration lies in the extent to which we apply social values more noble than mere monetary profit. Happiness lies not in the mere possession of money; it lies in the joy of achievement, in the thrill of creative effort. The joy and moral stimulation of work no longer must be forgotten in the mad chase of evanescent profits [. . .] There must be a strict supervision of all banking and credits so that there will be an end to speculation with other people’s money. [. . .] In the event that the Congress shall fail to take one of these two courses, I shall ask the Congress for broad Executive power to wage a war against the emergency, as great as the power that would be given to me if we were in fact invaded by a foreign foe.” [It is not uncommon for a president to make similar pledges. Because the “practices of unscrupulous money changers” have never been very popular in any country, it is an easy way to rally a broad support. What differentiated President Roosevelt from his predecessors is that these promises were seconded by acts. During the first “Hundred Days” 15 major laws were passed, among which one can mention the Farm Relief Bill, the Civilian Conservation Corps which would provide work to 250,000 young men or the law establishing the Tennessee Valley Authority to bring hydro-electric power and flood control to 7 southern states. Other laws addressed the problem of home mortgage foreclosures which was causing Americans to lose homes at a rate of 360,000 per year. In the last sentence the president warns the Congress that if needed he is ready to ask for special war time executive power.]

1933, Mar 5, 11 pm The President’s Bank proclamation. A proclamation by the President of the United States WHEREAS, there have been heavy and unwarranted withdrawals of gold and currency from our banking institutions for the purpose of hoarding; WHEREAS continuous and increasingly intensive speculative activity abroad in foreign exchanges has resulted in severe drains on the nation’s stocks of gold; Because of this state of emergency the law of 6 October 1917 will again be put in force. Under this law the penalty for hoarding by natural persons is imprisonment for not more than 10 years. (NYT 6 March 1933 p. 1) 1933, Mar 6, 9 The two steps in the prohibition of gold hoarding. (i) On March 6, taking advantage of a wartime statute that had not been repealed, president Roosevelt forbade the hoarding of gold or silver (in bullion or coins) under penalty of 10 years of imprisonment or/and $ 10,000 fine. (ii) On 9 March Congress incorporated the penalties of the wartime statute into the Emergency Banking Act. (NYT 6 Apr 1933 p. 16)

Is there a way out?

267

1933, Mar 27 For several past months there has been a substantial transfer of assets from New York to London which can be estimated at about $ 160 million. (NYT 27 March 1933 p. 23) 1933, Mar 28 During February 1933 there has been an alarming drop of the gold reserves of the banks of $558 million. In the first days of the Roosevelt administration, orders were sent to the Treasury (under the Emergency Banking Act) to compile a list of all persons who had withdrawn large amounts of gold in the last two years without apparent business reasons for doing so. The Treasury first asked the commercial banks through the Federal Reserve Banks to forward the list by March 13. Subsequently the deadline was extended to March 27. What action the government will take toward the persons who have not returned their gold has not yet been announced. It is expected that the government define what it considers to be hoarding after what a rule will be laid down by which the Department of Justice may prosecute hoarders. (NYT 28 March 1933 p. 27) [On 27 March 1933, the Federal Reserve Banks had a store of gold of only $503 million; this shows the magnitude of the withdrawal.] 1933, Apr 6 The amount of gold still in hoarding can be estimated to about $1,000. How much of this is in foreign countries and how much is in the United States can hardly be known. (NYT 6 April 1933 p. 1) 1933, Apr 6 The executive order issued on April 6 affirms that the national emergency still continues to exist [in spite of the end of gold outflow] and sets out in more detail the government prohibition against hoarding. All persons are required on or before May 1 to deliver to a Federal Reserve Bank gold in coins and bullion or gold certificates now owned by them. They will receive lawful money in return. The penalties remain as set by the President’s first proclamation. (NYT 6 April 1933 p. 16) [In the Donald Duck cartoon stories there is the character of dear Uncle Scrooge4 (“oncle Picsou” in the French version) who swims in a pool full of gold coins. This character was created in 1947 at a time when it was still illegal to hold gold.]

3 1933-1938: Opposition to New Deal policy In most historical accounts of how the New Deal policy was implemented one aspect is often under-reported, namely the bitterness of the resistance 4 A scrooge is an ungenerous person; Ebenezer Scrooge was the protagonist of a novel by Charles Dickens.

268

Chapter 12

put up by business interests and in particular by the “National Association of Manufacturers” (NAM) and the “United States Chamber of Commerce” which were the two most powerful associations lobbying for business interests. As a matter of fact, pro-union regulations were not implemented for years because employees were expecting that they would be voided by the US Supreme Court in the same way as the “National Industrial Recovery Act” was declared unconstitutional in 1935. Resistance to the New Deal policy one the one hand and increasing popular support for the reforms on the other hand lead around 1935 to what is called the “Second New Deal”. It was markedly more pro-labor than the first New Deal. There can be little doubt that by 2009 similar reforms would elicit even more powerful opposition. The greater concentration of newspapers, TV and radio channels in the hands of a few big corporations would make the make the struggle for public opinion more asymmetrical and unbalanced than in the 1930s. As will be explained in the next section, the pendulum began to move in the opposite direction almost immediately after the death of President Roosevelt in April 1945.

4 Anti New Deal reforms in the aftermath of World War II As a general rule, wars put workers in a better bargaining position for the simple reason that they bring about a shortage of labor. This was particularly true during World War II in the United States. Some 12.5 million American joined the armed services in a time when plants were operating at full capacity to fulfill war supply plans. The same effect had been observed during the Civil War and during the First World War. What made it particularly strong in World War II is the fact that President Roosevelt supported the advancement of the rights of workers.

4.1 Broken promises On 11 January 1944 in a speech delivered in Congress President Roosevelt defined an “Economic Bill of Rights for the American People”5 : “In our days these economic truths have become accepted as self-evident: 5

This speech was made one year before President Roosevelt’s death.

Is there a way out?

269

• The right to a useful and remunerative job. • The right of every family to a decent home. • The right to adequate medical care. • The right to adequate protection from economic fears of old age, sickness, accident, and unemployment. • The right to a good education.” Many of these objectives were dropped within a few years. So radical has been the change in the Zeitgeist that by 2009 the very notion of a bill of economic rights seems somewhat weird and utopian. There is still much talk about “human rights” but what is to be understood under this expression seems to have been seriously squeezed. Already during the war, in June 1943 Congress passed the repressive SmithConolly Labor Act, a forerunner of the Taft-Hartley Law, over the veto of President Roosevelt. After the war business organizations tried to turn back the clock. The Cold War which began almost immediately after 1945 was instrumental in this respect. In fact, business began to advocate for an end to the New Deal even before the end of the war. The objections of the NAM were mainly directed against the Wagner Act and the Temporary National Economic Committee (TNEC). One of the main activities of the TNEC. was to enforce anti-trust legislation and in 1942 the NAM published a massive refutation (830 pages) of the arguments of the TNEC (Scoville et al. 1942). In mid-1943 NAM officials called for a government commitment to a return to free enterprise in the postwar period (NYT 22 Jun 1943 p. 27). At the end of 1944, the NAM adopted a 6-point program whose themes were basically those of Friedrich Hayek’s book The road to serfdom (NYT 8 Dec 1944 p. 1). The message against government intervention was repeated relentlessly in the late 1940s. Between 1946 and 1950 the NAM distributed 18 million pamphlets that pushed anti-New Deal, anti-union and anti-communist sentiments. A cartoon service serving more than 3,000 weeklies disseminated cartoons that humorized the NAM theses. For instance, the forgotten man that Roosevelt popularized in his speeches is represented as a tattered taxpayer. The NAM also produced radio programs (e.g. the Family Robinson) and movies and it run vast national billboard campaigns (Ewen 1996) . These campaigns proved indeed effective. On 23 June 1947, The United States Senate followed the House of Representatives in overriding Truman’s

Chapter 12

270

veto and establishing the Taft-Hartley Act as a law. It amended the Wagner Act (which Congress had passed in 1935) in a way which was favorable to business. Moreover, while the prospect of universal, federally insured health coverage seemed close, the project was killed in Congress in November 1949.

4.2 The turning point of the Cold War In some political circles the Soviet Union had remained the main foe even though it was officially a close ally during the war. There had already been a “Red Scarce” in the 1920s. During the 1930s a “House Committee on UnAmerican Activities” (HCUA) hunted down Communists and sympathizers, i.e. members of so-called front organizations. In the weeks before the election of November 1944 at a time when the war was still in full swing there was quite a revealing incident that was mentioned by President Roosevelt in his radio address of 5 October 1944 (see below). Thirteen Republican Congressmen had mailed 3,116,000 copies of a speech in which they alleged that “the Roosevelt Administration is part of a gigantic plot to sell our democracy out to the Communists”6 . It can be said that the Cold War began soon after the death of President Roosevelt on 12 April 1945. Under President Truman, Secretary of the Navy James Forrestal who was strongly anti-Communist assumed an essential role. Starting in 1946 and in collaboration with the FBI, the American Chamber of Commerce published a series of anti-Communist brochures which were based on the assumption that the only objective of American Communists was to overthrow the government7 . The Taft-Hartley Law was a fruit of the Cold War and represented a landmark victory of business in its fight against the power of unions. Document 12.1

President Roosevelt’s radio address of 5 October 1944

My fellow Americans: I am speaking to you tonight from the White House [. . .] 6

The article of the New York Times (4 October 1944, p. 14) which mentions this story says that the speech attacked the “Congress of Industrial Organizations” (C.I.O.) union but it does not allude to the attack against the Roosevelt Administration, probably an effect of self-censorship in war time. 7 The first titles were as follows: Communist infiltration in the United States: its nature and how to combat it; Communists in the government: the facts and a program (1947); Communists within the Labor Movement, facts and countermeasures (1947). Other titles followed subsequently.

Is there a way out?

271

I have just been looking at a statement by a member of the Congress, Representative Anderson, Chairman of the House Committee on Campaign Expenditures, about a document recently sent free, through the mails, by one Senator and twelve Representatives- all of them Republicans. They evidently thought highly of this document, for they had more than three million copies printed free by the Government Printing Office, requiring more than 18 tons of scarce paper, and sent them through the mails all over the country at the taxpayers’ expense. Now let us look at this document to see what made it so important to 13 Republican leaders at this stage of the war when many millions of our men are fighting for freedom. Well, this document says that the “Red specter of Communism is stalking our country from East to West, from North to South” the charge being that the Roosevelt Administration is part of a gigantic plot to sell our democracy out to the Communists. This form of fear propaganda is not new among rabble rousers and fomenters of class hatred who seek to destroy democracy itself. It was used by Mussolini’s black shirts and by Hitler’s brown shirts. It has been used before in this country by the silver shirts and others on the lunatic fringe. But the sound and democratic instincts of the American people rebel against its use, particularly by their own Congressmen and at the taxpayers’ expense. I have never sought, and I do not welcome the support of any person or group committed to Communism, or Fascism, or any other foreign ideology which would undermine the American system of government, or the American system of free competitive enterprise and private property. That does not in the least interfere with the firm and friendly relationship which this Nation has in this war, and will, I hope, continue to have with the people of the Soviet Union. The kind of economy that suits the Russian people, I take it, is their own affair. The American people are glad and proud to be allied with the gallant people of Russia, not only in winning this war but in laying the foundations for the world peace which I hope will follow this war and in keeping that peace. Source: Website of the American Presidency Project.

4.3 The Taft-Hartley Law (1947) Passed by Congress in June 1947, the Taft-Hartley Law nullified many of the rights that workers and unions had gained through the Wagner Act of 1935. In that sense it really marked a turning point in US labor relations. To make this point clear one must explain in some detail its main provisions. • It required union officers to deny under oath any Communist affiliation. As a result, the 232,000 officers who signed the oath were in daily

272

Chapter 12

Fig. 12.2 Advertisement for the “Road to Serfdom”. Hayek’s book presents opinions but contains very little facts and almost no data. At that time the Reader’s Digest had a circulation of one million copies. Published in 1944 jointly in Britain and in the United States, the book benefited from a worldwide advertisement campaign; for more details about this campaign see Roehner (2007, p. 117-119). As a matter of fact, it was part of a broader campaign destined to get rid of the reforms of the New Deal which had raised strong objections from business associations and their lobbying groups.

danger of employer-inspired testimonies which would expose them to perjury indictments and could send them to prison. There were several cases

Is there a way out?

273

of this kind. One of them received broad public attention. In 1954 Clinton Jencks who had played his own role as a union leader in the movie “The Salt of the Earth” was indicted, convicted of perjury and sentenced to 5 years in prison. The conviction rested largely on the testimony of a paid informant for the Federal Bureau of Investigation who later recanted his story. The case was taken up by the Supreme Court in 1957 in a decision which bared a system of prosecution based on paid informants. • The law gave courts the power to fine unions for alleged violations, thus exposing them to million of dollars in legal expenses, fines and damage suits. • The law established a 60-day cooling off period in which no strike could be declared. • By outlawing mass-picketing it gave again employers the opportunity to employ replacement workers.

5 Will Obama’s presidency mark a turning point? Chronology 12.2 shows that on the basis of his first month in office President Obama intends to break away from the assertions of the neoliberal creed. This is particularly clear for the following issues: tax evasion and off-shore tax shelters, unionization, tax rate for the highest income brackets, federal aid to alleviate the cost of higher education. Yet, in politics intentions are not enough. Any substantial change must overcome the resistance power of the social and political forces who took advantage from the previous system. So the real question is rather8 : “Will President Obama and the Democratic Party commit themselves with sufficient energy to the objective of making such a change really happen?” We try to answer this question at the end but before that, one can make three preliminary observations: 1) Proponents of neoliberalism are not willing to question the soundness of their model in the light of the present crisis. They did not do that after the failures of Enron, Global Crossing and WorldCom (which were early warnings of the problems which surfaced in 2007) and they will not do so either in the wake of the current crisis. The common position of banks and 8

In the time of globalization it could seem bizarre to put so much emphasis on the policy of the United States. Yet, at the present time its influence is so overwhelming that even for large countries such as China or India their economic options are modeled on the US particularly as far as labor policy is concerned. A major US shift away from the neoliberal agenda would bring about a global turning point.

274

Chapter 12

neoliberal think tanks is: “We are opposed to any form of federal regulation; whatever mistakes had been made will be corrected by the profession”. In short, the plan is to overcome the present crisis and to resume the activities which had been so profitable during the past decades. Actually this attitude makes sense from their perspective. This can be shown by the following simple calculation. It is said that the fund of George Soros had an average annual return of 30% between 1969 and 1999. Such a return represents a multiplication by a factor K = 1.330 = 2, 620. So even after a fall of 90% the multiplier is still equal to k = 262 which corresponds to an average annual return of 20%. 2) There is already a vocal opposition to President Obama’s agenda but for the time being (March 2009) it comes from hard core of neoliberals such as the Rupert Murdoch medias, the Cato Institute, the Heritage Foundation and other neoliberal think tanks. On the contrary, the Hearst Press is showing moderation and understanding. 3) A major difference with the New Deal is that President Obama had been elected much earlier in the crisis. At the time of President Roosevelt’s inauguration the country had already been in economic recession for 3 years and the unemployment rate was 25% of the civilian labor force (Historical Statistics of the United States p. 135). As a result there was a consensus that “something had to be done” which even extended to business circles. Although this good will did not last very long, in the present case there is nothing of the sort. If one adds to that the fact that the medias are much more powerful (especially due to TV) and more concentrated in a few hands that they may have been in 1933, one would expect that the groups who oppose any change will put up a more determined fight. Chronology 12.2 Will there be a shelving of neoliberalism under the Obama administration? Through decisions taken within a few weeks after his inauguration on a number of important issues, President Obama has shown that the changes that he promised during his campaign imply the calling into question of several of the axioms of the neoliberal creed. As illustrated in this Chronology this concerns in particular the following points: (i) Curtailment of tax havens worldwide (ii) Higher taxes on rich people (iii) Federal aid to students (iv) Halt to the privatization of Medicare. 2009, Jan 27 Against the stimulus plan (1). Cato Institute mounted a full frontal

Is there a way out?

275

attack against [President Obama’s] stimulus plan. (excerpt from the website “Mike Norman Economics”). [The stimulus plan (American Recovery and Reinvestment Act) was intended to be passed by Congress on a bipartisan basis. Incidentally, this is why it was found highly disappointing by progressive economists such as Paul Krugman. Nevertheless: (i) The plan passed with zero Republican support (ii) It drew fierce criticisms from neoliberal think tanks. (iii) On 29 January, the Cato Institute sponsored a nationwide advertisement campaign in newspapers against the plan. For the President this may have provided a clear indication that it was futile to seek bipartisanship.] 2009, Jan 29 Against the stimulus plan (2). The Cato Institute purchased fullpage ads in major national newspapers (e.g. the New York Times or Washington Post). After citing a statement made by Mr. Obama on 9 January (i.e. before his inauguration) which read: “There is no disagreement that we need action by our government, a recovery plan that will help to jumpstart the economy.”, the message said: “With all due respect, Mr. President, that is not true.” Below there was a short declaration which contained in particular the following sentence: “More government spending by Hoover and Roosevelt did not pull the United States economy out of the Great Depression in the 1930s.” After these comments came a list of about 190 names, mostly professors who signed the declaration. The same advertisement was placed in 15 student newspapers on 11 February 2009. [Incidentally, the US production of steel which was 15 million tons in 1932 had climbed to 53 in 1939 (and 83 in 1941, Historical Statistics of the United States p. 693); thus, to say that government spending by President Roosevelt did not pull the country out of the depression does not seem quite correct.] 2009, Feb 6 Executive Orders about the role of unions. President Obama signed an Executive Order, effective immediately, authorizing executive agencies of the federal government to require every contractor on large-scale construction projects to become a party to a Project Labor Agreement (PLA) with unions. This Executive Order, which specifically revokes contrary Executives Orders issued by former President George W. Bush in 2001 and reinstates a Clinton-administration rule, was immediately hailed by organized labor. The reaction of the business community, not surprisingly, was far different. The chief executive officer of the “Associated General Contractors of America” said that this order “has the unfortunate potential to limit contractors’ ability to compete for projects at a time when the government is reporting that over one million construction workers have lost their jobs.” (http://www.jacksonlewis.com/legalupdates/article.cfm?aid=1634) [To the extent that this rule will apply to all contractors it is difficult to understand why it should limit their ability to compete. The real reason of the protest may be that this rule will strengthen unionization in a time when

276

Chapter 12 union-free companies have become the rule. This is the fourth pro-labor Executive Order signed by President Obama since January 30th. Another order bars federal contractors from being reimbursed for expenses incurred in trying to persuade employees not to form a union. As a result of these orders most of the infrastructure projects funded in the $787 billion stimulus plan will have union workers.]

2009, Feb 20 Fight against tax shelter abuses (1). Swiss law prohibits Swiss citizens from revealing any information covered by bank secrecy. This was confirmed when (on Friday night) the Swiss Administrative Tribunal ruled that UBS should not reveal to the American authorities the names of American clients having a secret account. Yet, shortly after this decision was released UBS announced that it had already transmitted the names of about 300 clients (in a total of 52,000 US customers). On Wednesday the bank had received an ultimatum that it had until Thursday to comply or face an indictment which would bar it from banking activity in the United States. One obvious question is: how were the customers selected whose names were transmitted? If they did not give permission they could rightly complain about discrimination. [The importance of this information comes from the fact that the demand for the names of the clients had already been made in June 2008. See in this respect the following articles: (i) Wall Street Journal 14 May 2008: Two charged in tax case. (ii) Spiegel 20 May 2008: Europe, US battle Swiss bank secrecy. (iii) Bloomberg 1 July 2008: UBS may be forced to comply with a request by Federal prosecutors to reveal the names of US clients with secret bank accounts set up to evade income-tax. Under the Bush administration, the matter was left pending. Nothing happened between July 2008 and 20 January 2009. Although UBS said it was cooperating with the IRS it did not release a single name. Then under the new administration the procedure was energetically reactivated. It can be recalled that in February 2007 together with his colleagues Carl Levin and Norm Coleman, Senator Obama had introduced a bill, the “Stop Tax Haven Abuse Act” which contained much stricter regulation for both individuals and companies. The bill was not passed at the time but a similar bill may be introduced once again in 2009 with this time a better chance to become law.] 2009, Feb 26 Higher tax on rich people (1). Obama’s planned tax would hit highest earners hardest. Taking from wealthy people to support universal health care will create enmity between high income people and those less fortunate. It will fracture society. (excerpt from the website of the Cato Institute). [This statement should be compared with the observations presented in an earlier chapter which suggest that the past two decades were marked by a

Is there a way out?

277

broad increase in social segmentation.] 2009, Feb 27 Higher tax on rich people (2). Obama’s budget plan sweeps away Reagan ideas. In a radical departure from recent history, the budget planned by President Obama would sharply raise taxes on the rich mainly by eliminating tax cuts which were enacted under former President George W. Bush. The top income tax rate for couples making more than $ 250,000 would increase from 35 percent to about 40%. The tax increases would be delayed until 2011 (NYT 27 February 2009, San Francisco Chronicle 28 February 2009) [An increase from 35% to 40% represents a small change if one recalls that before 1980 the top rate was 70%.] 2009, Feb 27 Federal aid to students. If adopted the planned budget will markedly increase federal aid for students who want to go to college. Over the last three decades the cost of college tuition climbed while the pay of less-educated workers declined. As a result the United States lost its standing as the country in which the largest share of young adults graduates from college. (NYT 27 February 2009) 2009, Feb 27 Taking control of Citigroup. The US government will swap the $25 billion in preferred stock that it holds in Citigroup from a previous bailout into common stock. This will raise its share in the bank to 36%. To emphasize the significance of this decision it must be recalled that in contrast to common stock preferred stock does not carry any voting right. This conversion will make the Federal government the main shareholder of Citigroup. Although the current chief executive will remain in place for the moment, the boards of directors will be reshaped. While this is not outright nationalization it comes close to it. In the wake of this decision the shares of the bank fell 39% to $ 1.50. (Associated Press, 27 February 2009, article entitled “New Citi plan may serve as model but carries risks”.) [Financial support provided to banks under the Bush administration took mostly the form of purchasing “toxic” assets or preferred stock which did not give the Treasury any control over the banks decisions.] 2009, Feb 27 Budget plan draws broad criticism in the Murdoch medias. As an illustration one can mention the comments in the New York Post: “Obama’s budget schemes to drain staggering amounts of money from people who worked for it and steer it to people who didn’t. This isn’t the free market. It’s the freeloader market”. Even the Federal aid to help select students to pay their tuition is disapproved on the account that it comes on top of Social Security and Medicare which is said to be on “the verge of catastrophic collapse”. The New York Post, one of the oldest American newspapers, belongs to the group of Rupert Murdoch.

278

Chapter 12 This excerpt shows that five weeks after his inauguration, President Obama is already facing a fierce opposition at the forefront of which one finds the Rupert Murdoch medias (which include the Fox News TV channel, the “Weekly Standard” magazine and many others) and neoliberal think tanks such as the Cato Institute. In contrast, the Hearst newspapers (e.g. Houston Chronicle, San Francisco Chronicle, Seattle Post-Intelligencer) are much more moderate and balanced in theirs comments.

Fig. 12.3 Cover of the “Weekly Standard” of 9 March 2009. The “Weekly Standard” magazine belongs to Rupert Murdoch’s News Corporation. The titles of some of the articles published in this issue are quite revealing: “The Return of Big Government”, “Hey, Big Spender”, “A Ph.D. in Every Pot”, “Indefensible”, “Reports of conservatism’s demise have been greatly exaggerated”, “Unions are part of the problem, not part of the solution”. 2009, Feb 28 The “Federal Deposit Insurance Corporation” (a government agency providing a kind of insurance to customers of failed banks) has been authorized to raise an emergency fee of 0.2% of deposits from the 8,305 federally insured financial institutions. In addition to this emergency measure the regular insurance premium will be raised to about 0.14%. In 2008 the average premium paid by banks and thrifts was 0.063%. (Associated Press 28 February 2009, article entitled: “FDIC raising fees on banks.”)

Is there a way out?

279

[During 2008 the FDIC rescued directly several small banks. In the case of large banks the FDIC mainly acted as a middle man. Thus, Washington Mutual (with assets around $300 billion) was acquired by JP Morgan Chase for $1.9 billion in a deal brokered by the FDIC. The capital of the FDIC is only about $20 billion and would have been quickly exhausted if it had attempted large rescue operations.] 2009, Mar 3 Fight against tax shelter abuses (2). Bills against offshore tax havens were introduced by democrats in the Senate and House of Representatives. The Senate bill expands on one co-sponsored last year by then-Senator Barack Obama and Senator Carl Levin, who has sought a broad crackdown on tax dodgers. Texas Democrat Lloyd Doggett who introduced the legislation in the House declared: “These outrageous tax havens add to the soaring budget deficit and shift the tax burden to small businesses and families who play by the rules”. (Reuters 3 March 2009, article entitled: “Foreign tax havens targeted in US bills”) [The article also notes that “a thriving business in tax evasion developed in recent years on Wall Street among consulting firms, hedge funds and other financial players. Some purveyors even sought patent protection for their schemes. This shows that on this issue the Obama administration is intending to change the trend which prevailed in recent years. 2009, Mar 4 Fight against tax shelter abuses (3). UBS wealth management executive Mark Branson, appearing Wednesday 4 March before a Senate Committee, said he objected to an Internal Revenue Service (IRS) lawsuit that seeks to force UBS to reveal the names of up to 52,000 US customers who may have concealed their accounts from tax authorities. He said that UBS cannot disclose information to the IRS that would put its employees at serious risk of criminal prosecution under Swiss law. He also said that as of 30 September 2008 47,000 American clients had failed to file special tax forms known as W-9s with the bank. UBS was supposed to require them to do so. So far, UBS has shut down 14,000 American offshore investment accounts but would not close their deposit accounts. At the same audience IRS Commissioner Doug Shulman said the US government was taking unprecedented measures to combat offshore tax avoidance and promised that “much more is in the works”. Senator Levin, the chairman of the Committee observed that banking secrecy is “part and parcel of a conspiracy to commit a crime under our law”. Cloak-and-dagger tactics said to have been employed by UBS such as coded language in internal e-mails and memos or foreign shell companies were on display at the hearing. [UBS’s refusal to disclose other names in addition to the 250 already transmitted has certainly been weighed carefully. In the worst case scenario it would mean the closing of UBS operations in the United States (25,000 employees out of a total of 78,000) but also perhaps in countries that may follow

280

Chapter 12 the American lead (e.g. Australia, Canada, Mexico). It can be hypothesized that the present decision was taken in agreement with the Swiss government and is based on the fact that the breaking of the secrecy rule would be more serious for Switzerland than a drastic contraction of the activity of UBS. The total assets of American UBS customers is estimated at about $15 billion (Associated Press 4 March 2009, article entitled: “UBS says it had 47,000 accounts for Americans”). This figure is probably an underestimate because it is known that to open an account at UBS the minimum was $1 million which on the basis of 47,000 customers would imply assets of (at least) $47 billion.]

2009, Mar 10 Pro-union legislation. The so-called “Employee Free Choice bill” was introduced in Congress. The bill is meant to make it easier for employees to unionize because instead of a secret ballot, as is currently the practice, they could simply sign a card supporting unionization. The business community is lobbying hard against the legislation with the leadership of Wal-Mart Stores and the US Chamber of Commerce. Wal-Mart is vehemently opposed to the legislation and its spokesman declared that the company feels certain that Congress will not pass this pro-union bill. (Dowjones Business News) [The real problem of unionization seems to be that in 25% of the unionization drives, pro-union workers are just (illegally) fired9 . Currently (2009) the penalty for firing pro-union workers is so weak that it does little to deter law breakers. Whereas many Americans would probably agree that this is not fair, it is likely that the card check debate will not raise much public interest and even less approval. It must be noted that this legislation was already introduced in Congress in 2007 and defeated. In short, it does not appear to be a wise move to have chosen this battleground to start the all-important battle about the respective rights of employers and employees.] 2009, Mar 18 The Federal Reserve follows the UK and Switzerland in their policy of printing money. “The Fed is printing $1 trillion of money, and using those funds to buy bonds. I think quantitative easing [the current way of referring to money printing] is the right way to go”. (New York Times 20 March 2009, article by Paul Krugman) [After that announcement was made many analysts expressed their concern 9

The ballot for unionization may be held if more than 30% of employees sign cards asking for it. But, according to current legislation, employers naturally get to scrutinize and challenge the signatures before the election can be called. This allows companies to use the current process to intimidate, coerce, and even blacklist workers involved in organizing those who signed the cards. In other words, the employer will know how employees are going to vote long before they step into the voting booth. Many employers resort to spying, threats, intimidation, harassment and other illegal activity in their campaigns to oppose unions. (http://www.californiaprogressreport.com, The Washington Independent 14 January 2009)

Is there a way out?

281

that the exchange rate of the dollar may slide and that, as a result, inflation may pick up in the United States with the possibility of a flight of money to more stable currencies. For the time being such fears seem unfounded. As far as the future is concerned, the outcome depends on how much dollars will be printed. One trillion could seem a big amount but compared with the deflationary process that results from the fall of house prices, it is in fact fairly small. In 2007 the total market value of US residential and commercial property was around $30 trillion (22 for residential and 8 for commercial); thus, an annual price fall of about 20% results in deflation amounting to $6 trillion. If annual quantitative easing comes to exceed this amount one may begin to worry about inflation.] 2009, Mar 29 Regulation plans of the financial system. The Obama administration wants to rewrite the rules of the game for the finance industry. This would include forcing derivatives trading on to recognized exchanges, new oversight of hedge funds and additional powers to shut down large firms that threaten to destabilize the global economy. Republican leaders signaled that they would not allow President Obama to bulldoze new legislation through Congress and called for time for reflection. The American Bankers Association immediately threw up objections to the plan. Analysts said that the planned regulation could curtail profitable trading opportunities. (The Independent 27 March 2009) [At the time of writing (March 2009) there was much talk about regulation. However one should not forget that the Sarbanes-Oxley Act of 30 June 2002 already tried to establish enhanced standards for publicly traded companies and public accounting firms. This law was passed in the wake of the Enron and Global Crossing bankruptcies which revealed the broad extent of deceit and corruption. Its objective was to restore public confidence in US capital markets. The very fact that this law was only mildly opposed by financial interests and their lobbying groups suggests that it was not taken very seriously. The crisis which erupted 6 years later indeed confirmed that opacity, deceit and graft were still widespread and pervasive. The Act established a new quasi-public agency, the “Public Company Accounting Oversight Board” (PCAOB) which was charged with overseeing and inspecting accounting firms. In following years very little was heard about this agency.] 2009, Mar 29 Prospects of a landmark pro-union bill. In the early 1950s about 35% of private-sector workers belonged to unions. Today less than 8% do. As a result, wages have lagged behind rising productivity. How President Obama and congressional leaders decide to craft pro-union legislation will determine the post-recession economy. Many expect this question to be a defining showdown of Obama’s presidency. (Washington Post 29 March 2009, p. A05) [At the time of writing there are very few articles on this question. Yet, it

282

Chapter 12 seems clear that there can be no sustainable economic growth in the United States unless wages and salaries again begin to grow instead of declining as was the case during the past 30 years.]

2010 Rosy forecasts. The Obama budget forecasts that, despite the depth of the current recession, the economy will recover and grow by 3.2% in 2010 and then climb to an even more robust 4% in the three following years. (Seattle Post-Intelligencer 26 February 2009)

6 What should be expected in 2010–2011? We have given a fairly detailed account of the first 100 days of the Obama administration. Let us now come back to the key questions addressed in this section: will there be a real break with the neoliberal agenda? Our flat answer is “No”. Why? There is a considerable difference between the 100 days of President Roosevelt and the 100 days of President Obama. Between March and May 1933 15 laws were passed which represented structural changes. Between January and March 2009, the laws which were passed intended mostly to provide short-time relief; very little was done in terms of structural changes. The attempts which were made to introduce major breaks (e.g. in terms of tax havens or labor regulation) were not pursued with enough energy and did not materialize in major laws. In the weeks after his inauguration President Obama enjoyed strong public support which would have enabled him to make major breaks. He did not take advantage of this opportunity. On several important issues President Obama backed down from the promises he made during his campaign, e.g. (i) regarding relief provided to home owners in bankruptcy: the finance industry was adamantly opposed to allowing bankruptcy judges to reduce the mortgage level. “When the time came to stand up to the banking lobbies the White House didn’t” wrote the New York on 4 May 2009. (ii) The introduction pro-union legislation has been given a false start as we explained above (iii) The struggle against UBS and other tax heavens seems to have been shelved. It is doubtful that these projects can be resumed later on especially once the strong public support of the first months has dwindled. So, with the basic rules of the system practically unchanged what can one expect to happen? Financial markets have very short time constants. They collapsed within a

Is there a way out?

283

few months but they can also come back to life within a few months. As soon as credit will flow again the transactions which have been so profitable in the past (e.g. leveraged buyouts, mergers and acquisitions, initial public offerings) will be resumed. For economic markets the time constants are much longer and recovery will take several years. Let us give an illustration. During 2008, industrial production in developed countries fell by about 16%; how long will it take to return to the production level of early 2008? Since in past years the highest (sustained) growth rates were of the order of 4%, it means that it would take 4 years to regain the production level of early 2008. In other words even assuming steady, uninterrupted growth over the next 4 years, the production level of January 2008 will be reached again only in January 2013. If underlying problems such as income distribution are not solved in the meanwhile a recovery in the form of the Japanese “lost decade” of the 1990s seems likely; this means a sputtering growth engine with “good” quarters being followed by quarters posting negative growth. Under such an assumption recovery may take much longer. In Japan the index of industrial production reached a level of 101.6 in 1991 and regained that level only in 2005, that is to say fourteen years later.

7 Some tests for the years to come Will there be a way out of the neoliberal ideology in the years to come? There is no way to answer this question in a scientific way. However, it is possible to define some test-indicators which should allow us to see more clearly in which direction the world is heading.

7.1 Comparison between 1929 and 2008 in terms of synchronicity Financial globalization progressed markedly between 1990 and 2008; therefore it is hardly surprising that the financial crisis assumed a worldwide extension in a matter of months. What is more surprising is how rapidly it affected the production sphere, a feature which is documented in Fig. 12.1c. The severity of the depression which started in 2008 is not due only to its worldwide extension, but also to the number of sectors which experienced drastic declines within a few months. At the time of writing one can mention:

284

Chapter 12

• Banking • Residential real estate • Office and commercial real estate. For instance, the market value of the Canary Wharf property, London’s financial hub fell 26% between December 2007 and December 2008 (The Independent 27 March 2009). • Transport sector. For instance, the port terminal of Portland, Oregon which is a major hub for transpacific trade has seen a sharp reduction in activity. Trans-Pacific cargo traffic fell 32% in 2008 (Portland Business Journal 20 March 2009). • Overproduction problems in the automobile industry spread fairly quickly. Can one use Fig. 12.1 for predictive purposes? Intuitively, one may be tempted to speculate that the more synchronous and widespread a crisis is, the deeper it will be and the longer it will last. be. Let us examine these points more closely. • It is true that the downward momentum shown by graph 12.1c is impressive. The whole question, however, is whether this fall is due to exogenous or endogenous factors. As an example of the first kind one can mention the Spanish flue pandemic of 191810 . Because it killed a substantial number of young people worldwide and disrupted exchanges it certainly had a detrimental impact on economic activity. Yet, the epidemic lasted only a few months, basically from September to November 1918. As soon as it abated exchanges and activity were re-established and growth was resumed Thus, in this case the impact of a sudden and global shock was neither severe nor short-lived. • Fig. 12.1c shows that the impact of the current crisis is not only universal but also fairly strong. Annual falls of industrial production in the range of −10% – − 20% are not common. Is there a connection between the depth of a recession and its duration? In order to find out one can consider the case of the Great Depression in the United States. It is often argued (especially by proponents of neoliberalism) that the New Deal did not work because the industrial production resumed its level of 1929 only in 1939. In such a statement one forgets that the index fell from 124 in 1929 to 67 in 1932, a fall of about 50%. After the New Deal policy was started in 1933 the index grew steadily (with the exception of 1938 which was marked by a decline) but it took of course some time to rescue the American economy from the abyss into which it had fallen. Even 10 It is estimated that anywhere from 20 to 100 million people were killed worldwide, more than double the number killed in World War I.

Is there a way out?

285

Depression of 1929 USA

Mexico

Brazil

20

20

20

0

0

0

-20 1925

-20 1930

1935 1925

Germany

-20 1930

1935 1925

France 20

20

0

0

0

1925

-20 1930

1935 1925

Japan

-20 1930

1935 1925

Sweden 20

20

0

0

0

-20

-20

-20

1930

1935 1925

1930

1930

1935

UK

20

1925

1935

Russia

20

-20

1930

1935 1925

1930

1935

Fig. 12.4a Variation of industrial production in 9 countries before and after 1929. The graph shows annual variations expressed in percent. The zone delimited by the broad vertical line corresponds to the first year of the depression, i.e. the transition from 1929 to 1930. The graph shows that contrary to a common belief the Great Depression was not a worldwide depression. The UK, Japan and Russia were mildly or not at all affected. It can also be seen that during the 1929-1930 time interval, apart from the United States there were only two other countries (Germany and Brazil) where there were sharp negative growth rates. In short, the depression started with a low degree of synchronicity. The fact that over 1929-1930 all the curves are going downward means that the second derivatives (i.e. the accelerations) were all negative. This has a simple interpretation. Once the depression had started in the United States, the reduction of US imports induced a slowdown in the activity of all main economic partners; this brought about reduced growth even in countries where growth rates remained positive. Sources: Mitchell (1978), Mitchell (1982), Mitchell (1983)

with an annual growth rate of 10% it would take 7 years to raise production from 67 to 124. This example clearly suggests that depth and duration of a depression are two related variables because it takes time to recover the lost ground. As an illustration, consider the case of Japan. During the golden decade 1981-1990 the average annual growth rate of Japan’s industrial production

Chapter 12

286

Recession of 2001 10

10

10

USA

Mexico

Brazil

5

5

5

0

0

0

-5

-5 2000

2002

2004

10

-5 2000

2002

2004

10

2000

Germany

France 5

5

0

0

0

-5 2000

2002

2004

10

-5 2000

2002

2004

10

2000

Sweden 5

UK

0

0

0

-5

-5

2002

2004

2004

5

-5 2000

2002

10

Japan 5

2004

Russia

5

-5

2002

10

2000

2002

2004

2000

2002

2004

Fig. 12.4b Variation of industrial production in 9 countries before and after 2001. The graph shows annual variations expressed in percent. Compared with the graph for 2008 there are major differences in terms of amplitude and in terms of synchronicity. (i) The vertical scale shows that it was a recession of low magnitude; The three largest negative growth rates (US, Mexico and Japan) have a magnitude around −5% whereas in the next graph the three largest falls are much larger: −31% (Japan), −19% (Germany), −17% (Sweden),. (ii) During the first year of the recession, industrial production continued to increase in three countries: Germany, France and Russia. Sources: Website of “Trading Economics”: http://www.tradingeconomics.com; Statistical Yearbook of Sweden; Historical Statistics of Japan.

was 4.6%; the highest annual rate occurred in 1988 with 9.7% 11 . Thus, to recover from the 31% fall which occurred in 2008 it would take 31/9.7 = 3.2 years at the fastest possible annual rate and 31/4.6 = 6.7 years at the fastest average rate sustainable over a period of several years.

11 The sharp drop of 1975 (−11%) was followed by a quick recovery but nevertheless the growth rate was only 9.2% in 1976.

Is there a way out?

287

Depression of 2008 USA

10

Mexico

10 0

0

-10

-10

-10

-20

-20

-20

-30

-30 2006

2008

Germany

10

-30 2006

2008

France

10

2006

0

0

0

-10

-10

-20

-20

-20

-30

-30 2008

Japan

10

-30 2006

2008

Sweden

10

2006

0

0

0

-10

-10

-20

-20

-20

-30

-30 2008

2008

UK

10

-10

2006

2008

Russia

10

-10

2006

Brazil

10

0

-30 2006

2008

2006

2008

Fig. 12.4c Variation of industrial production in 9 countries before and during 2008. The graph shows annual variations expressed in percent. The depression started with a high level of synchronicity: all the falls are larger than 10%. It is probably the first time in the history of mankind that a severe recession started worldwide with such synchronicity. Somewhat surprisingly, the falls are smallest in the two countries (UK and US) where the financial crisis originated. As of January 2009, the unemployment rate is almost the same (around 8%) in the European Union, in Russia and in the United States; the increase rate, however, is fastest in the US. As a matter of comparison, the steepest annual decline during Japan’s “lost decade” (1991-2001) was −7% which occurred from 1997 to 1998 and again from 2000 to 2001; in 1974 the oil shock brought about a fall of only 11%. (Historical Statistics of Japan). China (which is not included in this graph because of the lack of data in the 1930s) had still a positive growth of about 8% in 2008, a rate which marked a substantial reduction with respect to earlier growth rates of about 15%. Sources: Website of Trading Economics: http://www.tradingeconomics.com

7.2 Comparison with the Great Depression in terms of social interactions In a previous chapter we documented the decrease in social interaction over the past decades. According to Harvard professor Robert Putnam economic

288

Chapter 12

depressions lead to even greater social isolation. He found that Kiwanis clubs, PTAs12 , and other social groups lost around half their members from 1930 to 1935. Putnam believes that the effect would be more pronounced today. The Depression was a boom time for movies. Today movies are no longer cheap but TV provides a cheap substitute. Yet watching TV is an activity which contributes to social isolation.

8 Supply of labor 8.1 In the 19th century Historically, it had always been a great concern for employers to be able to rely on an abundant supply of cheap labor. Back in the 19th century as slave trade was progressively being abolished, British planters brought indentured workers from India to many parts of the British empire where they were needed to work on tropical plantations; one can mention for instance Fiji, Guyana, Jamaica, Kenya, Mauritius, South Africa, Surinam, Trinidad, Uganda13 . Yet, except during the world wars, no attempt was made at that time to bring workers from the colonies to Britain or other European countries. Apparently, the supply of labor provided by Irish immigrants was deemed sufficient to support the industrial development in England and Wales.

8.2 After 1945 It is only after World War II that increasing numbers of immigrants from Jamaica, India or Pakistan entered Britain. In subsequent decades the globalization of labor supply became one of the main features of the new economic system. In 2009 this remains a crucial question because so long as supply vastly exceeds demand there can be no real improvement in real income and 12

“Kiwanis International” is a global organization of volunteers headquartered in Indianapolis, Indiana. It comprises approximately 8,000 clubs in 96 countries with over 260,000 adult members (2008). Kiwanis emphasizes service to children and youth. PTA: “Parent-Teacher Association”, a voluntary organization bringing together parents and teachers. 13 After these countries became independent in the 1960s, there have in several cases been rejection riots which lead to the departure of a sizable fraction of the Indian population.

Is there a way out?

289

in working conditions. Yet, in all industrialized countries (with the possible exception of Japan) employers are still very committed to their demand for cheap labor and its corollary which is free immigration. For instance in 2009 the vice president of the US Chamber of Commerce declared that the business community remained committed to a significant guest-worker program14 and welcomed an expansion of temporary worker programs. (New York Times 14 April 2009). The European Union has long-term programs for recruiting foreign workers in sectors in which the supply is not sufficient. Needless to say, there would be another obvious way to increase the supply side, namely by raising salaries. If wages for night watchmen, nurses or construction workers were raised by 20% there is little doubt that the scarcity problem would be quickly solved15 . This question arises even for highly qualified jobs such as scientific research. So long as experts in computer science, electrical engineering, mathematics, physics from countries such as China, India, the Philippines, Russia or South Korea are willing to fill positions in American industry and universities the average salary of PhD graduates will stagnate and the number of PhD diplomas earned by American students will continue to decline (see chapter 9 in this respect). In the same line, the development of post-doctoral programs has provided research institutions with a flexible, temporary and fairly cheap supply of scientific researchers from all over the world. But the downside is that highly trained scientists will have to wait until they are between 30 to 40 years old to get a permanent position, a situation which most of them do not happily accept 16 . It is likely that denying permanent positions to young employees adversely affects birth rates. Indeed, does the ability to support a family in a sustainable way not appear as a reasonable precondition for getting married? 14 Such programs are opposed by American unions because the immigrants are tied to one employer and cannot change jobs no matter how abusive working conditions are. In other words, such contracts are fairly similar to the contracts of indentured workers of the 19th century or to the labor contracts of foreign workers in countries such as Dubai or Saudi Arabia. 15 The argument that it would erode the competitiveness of these companies is pointless because these are domestic sectors which are not in competition with other countries. 16 Twenty years ago in France it was possible to obtain a permanent position in a research university at the age of 24; in 2009 one can apply at the age of 28 at the earliest; often young researchers are recruited after 30. The situation is similar in Germany or in the United States.

290

Chapter 12

8.3 Can the trend be reversed? How can this trend be reversed? The first idea which comes to mind is to re-introduce restrictions on crossnational migrations of workers. For historical, cultural and sociological reasons this was the policy followed in Japan. Yet, it would be very difficult to implement a similar policy elsewhere. For instance, to restrict the movements of workers between the countries that compose the European Union would run against the very ideas on which the European construction is based. In addition it would be almost useless because the companies would relocate their plants into Eastern countries where wage-levels are lower. NAFTA, the North American Free Trade Agreement, does not provide for free movements of workers but it makes little difference because US companies can freely establish their plants in the north of Mexico in the socalled maquiladora area17 . The ability to shift production out of the country, and then sell the products back home, gives transnational corporations a crucial edge. In short, unless there is a broad come-back of protectionism, that solution will not work. Is there another solution? Definitely yes. So far, all free-trade agreements whether on a bilateral basis or on a multilateral basis (e.g. under the European Union or World Trade Organization) contained rules which were designed for the benefit of companies but none which would improve the situation of people and workers. There were rules ensuring the free circulation of goods or prescriptions about the respect of patents, brands, intellectual property, but no rules about fiscal policies (regarding tax rates for corporations) or about the rights of unions. As there were no rules about tax rates on the profit of companies, this brought about a race in which each country lowered its tax-rates in order to attract more foreign investments. But at the same time this deprived the governments of part of their receipts and made them unable to provide basic services (e.g. free education) to their population. As low wage countries have an inherent attractiveness, it would make sense to require tax rates on profit 17

The term maquiladora refers to an assembly plant set up in Mexico by a non-Mexican firm near the US-Mexican border. Starting in 1965 many non-Mexican companies were attracted by the low wages, special tax concessions and the proximity to the US market. They usually assemble parts manufactured elsewhere, and by law they must re-export 80% of their production. By 2000 the maquiladoras employed more than a million Mexicans, mostly women. The managers are usually foreigners, whereas the hourly-paid workers, who have little job security and few benefits, are Mexican.

Is there a way out?

291

to be higher (for instance 5% higher) than in industrialized countries. That would have two advantages. (i) It would introduce a kind of balance between low wages and higher tax rates. (ii) It would provide resources to the governments of developing countries for improving infrastructure and education. Otherwise the conditions in these areas will not improve. The maquiladoras came in existence in Mexico around 1965 to take advantage of the low level of wages but more than forty years later there are still no unions in these companies. As a consequence, the condition of the workers has scarcely improved and these areas are still afflicted by third-rate infrastructures (in terms of water, streets, education) because the local government is too poor to make the required investments. So far there were no rules in free-trade agreements about the basic rights of unions. As a result, workers of low-wage countries have been unable to improve their condition. The union-free plants established in the north of Mexico are a case in point. Thus, if the workers in developing countries cannot improve their condition, no equilibrium can ever be reached. There will be a race to the bottom, to ever lower real wages until the economy comes to a standstill because there is no longer enough purchasing power in the system. As we already explained this mechanism may explain the emergence of the debt-driven economy that lead to the recent crisis. What kind of rules could be included in free-trade agreements? To proclaim the basic right to form unions is not enough. Such a right is probably included in the constitutions of many countries. As we already explained, unions are powerless unless their rights are clearly stated. One may wonder why this is so. • There is a fundamental asymmetry between employers and employees which comes from the fact that employers can fire their employees whereas employees cannot fire their employers. The following observations are merely consequences of this situation. • “Company unions” (also called “yellow unions”) are run by the company managers and are not affiliated with an independent trade union. Such unions were popular in the United States during the early 20th century, but were outlawed under the Wagner Act of 193518 ). It seems fairly obvious that yellow unions are not the best way to promote the right of workers. Balanced free-trade agreements would forbid company unions just as the Wagner Act did. • If employees can fire unionized workers they will be able to prevent 18

Company unions are a mainstay of labor organization in Japan.

292

Chapter 12

the formation of unions. Similarly, if during a strike employers can hire replacement workers no strike can possibly succeed. As a result any union will be useless and will be deserted by workers. Thus, free-trade agreements should contain rules prohibiting unfair labor practices such as those that we mentioned. • Whether or not free-trade agreements should contain minimum wage requirements is debatable. If the minimum wage in Romania was set at 2/3 of the minimum wage in France there would still be a strong incentive for business to relocate plants in Romania but it would be a first step in a process of convergence of average wage levels in the two countries. In case such a rule is omitted it may take several decades to Romanian workers for securing such a wage level through the normal bargaining process with employers. The weaker their unions are, the longer it will take. Needless to say, the previous proposals would be completely unacceptable for free-market apologists. As was already the case during the New Deal, any proposal which calls for an extension of the right of unions would be labeled by them as being a concession to socialism, a label which has become almost infamous19 . Unless there is a radical departure from the prevailing Zeitgeist such proposals will remain wishful thinking and vain hopes. Nowadays, the debate about neoliberalism is obscured by purely ideological arguments. In an other publication (Roehner 2009) we have tried to analyze this issue from the point of view of network science. This provides what can be considered as a more scientific insight. It turns out that efficient and creative systems are systems in which there is a high degree of interaction. The three previous chapters suggest that the implementation of the neoliberal agenda resulted in economic and social segmentation, in other words it weakened interactions and at the same time reduced efficiency. To make this argument more compelling one should be able to measure interactions fairly accurately. This will require extensive (comparative) work20 . but we have the feeling that such a research program will shed new light on this problem.

19

The social and economic record of Scandinavian countries where such policies were implemented certainly do not justify such a judgment. 20 In physics it took more than a century to explore the interactions between molecules, atoms or nucleons.

Is there a way out?

293

9 Insight about the future 9.1 Toward a new model? The Chinese ideogram for the word “crisis” is formed of two characters: one (wei) represents “danger” and the other (ji) means a “crucial juncture” (a secondary meaning is “resourcefulness”). It may well be that for China the current crisis indeed represents a crucial juncture and perhaps even an opportunity. Why? Before 2009 Chinese economic growth was largely based on massive exports21 to which foreign companies contributed for about one half. However, in the course of time this may have become a sure recipe for stagnant wages. Indeed, many foreign companies came to China to benefit from low wages and threatened relocation if wages were increased. Stagnant wages compounded with a thriving class of Chinese entrepreneurs would soon have lead to an oligarchic system and a de facto alliance with foreign business interests. This is not pure speculation. • After all, this was already the power structure under the Kuomintang government. • Moreover, the Latin American countries provide several illustrations of such an evolution. For instance, after the Mexican Revolution of 19101920, it took less than 50 years for an oligarchic political class to emerge which monopolized political power with the support and blessing of US business interests. Nowadays, Mexico seems to be torn apart by internal strives much as China was during the concession era. In the fall of 2008 the Chinese government has already signaled its intent to develop the internal market That, of course, means distributing higher salaries as was done in the United States during the New Deal era. If the present crisis lasts long enough (let us say more than 10 years) that move will perhaps lead to the development of a new interesting economic model.

9.2 Is there a way out? When asking this question at the beginning of the chapter we had two issues in mind: (i) Is there a way out of neoliberalism? (ii) Is there a way out of the economic crisis which started in 2008? In the previous chapters we argued 21

In 2008 exports represented about 30% of the GDP of China.

294

Chapter 12

that these questions are closely connected. We observed that real wages in the United States have been stagnant or decreasing over the last three decades. Unless this trend is reversed there will be no sustainable growth in consumption and therefore no long-term increase in GDP. But this trend can hardly be reversed if economic policy continues to follow the neoliberal agenda22 . As we have explained, this agenda leads to increasing social and economic fragmentation which hampers economic efficiency. This is why we devoted this chapter to examining if a change in economic policy is likely to occur under the Obama administration. Our conclusion was that there was a momentum for change but that it was just too weak. Perhaps, within two or three years there will be a “Second Obama” just as after 1935 there has been a “Second New Deal”? This, of course, is impossible to predict. As was explained earlier, credit crunches occur suddenly but can end quite as abruptly. This opens the possibility that financial profit will again start to flow within one or two years. However, on the economic side the best we can hope for is probably a “lost decade” similar to what Japan experienced during the period 1992-2005. If during this lost decade there is a substantial change in the redistribution of national income, one can expect that the economic engine will start to run again.

22 Proponents of neoliberalism may not agree on this point. But is this not what the last decades seem to show not only in industrialized countries but also in Latin America? In the coming years East European countries can be expected to follow the same unsuccessful path. In the previous chapters we avoided discussing this issue from a theoretical perspective because we have the feeling that this would be a waste of time. Unless social interactions are included into the theoretical framework of economics it will remain wanting and incomplete.

References

ABEILLE (L.P.) 1764, 1910: R´eflexions sur la police des grains en France et en Angleterre. Librairie Geuthner. Paris. AMARAL (L.A.N.), BULDYREV (S.V.), HAVLIN (S.), LESHHORN (H.), MAASS (P.), SALINGER (M.A.), STANLEY (H.E.), STANLEY (M.H.R.) 1997: Scaling behavior in economics. I. Empirical results for company growth. Journal de Physique I France 7,621-634. AMERICAN BOOK-PRICES CURRENT. A record of books sold at auction in New York and elsewhere from July 1 [X] to July 1 [Y]. Compiled from the auctioneer’s catalogues. E. P. Dutton. New York. ANDERSSON (I.) 1973: Histoire de la Su`ede des origines a` nos jours. Editions Horvath. Paris. ARBEL (A.) 1983: Crash: ten days in October, will it strike again? Longman Financial Services. Chicago. ARENSON (K.W.) 1979: Worries of mortgage trading. Self-regulation urgency grows. New York Times 7 November 1979 p. D1 and D17. [A recapitulation entitled “How mortgage-backed securities work” signals two possible problems: (i) Mortgage bankers contract with Wall Street investment firms to sell them a pool of mortgages in 4 months. If for some reason activity slows down in the housing market they may not be able to supply the planned mortgages. To make sure that enough mortgages are sold to house buyers it is tempting to soften standard requirements (i.e. no downpayment or no-documentation loans) (ii) There is a similar problem at the other end in the sense that no down payments are required for the investors (insurance companies, credit unions) who buy mortgage-backed securities. If for some reason they slow their purchases Wall Street investment firms will not be able to sell their packages and will incur big losses.]

AVELINE (N.) 1995: La bulle fonci`ere au Japon. Association des Etudes Fonci`eres. Paris.

296

References

BAK (P.) 1996: How nature works: the science of self-organized criticality. Springer-Copernicus. New York. BAK (P.), PACZUSKI (M.), SHUBIK (M.) 1997: Price variations in a stock market with many agents. Physica A 246,430-453. BAULANT (M.), MEUVRET (J.) 1960: Prix des c´er´eales extraits de la mercuriale de Paris (1520-1698). S.E.V.P.E.N. Paris. BECKMAN (R.) 1988: Crashes. Why they happen. What to do. Sidwick Jackson. London. Beitr¨age zur Statistics Mecklenburgs 7, 3-4 (1873) BEER (S.) 2003: “Bound” to cooperate. Austria’s little-known intelligence community since 1945. The Journal of Intelligence History 3,1,19-31. BERG (E.N.) 1984: Rise of a national mortgage market. Trading home loans draws billions in new funds. Uniform [interest] rates emerge. New York Times 22 January 1984, Section 3 (Business). [The article shows that in 1983 the secondary mortgage market (through which mortgages are resold as securities) was already well developed: the percentage of new and old mortgages resold as securities represented 60% of new mortgage loans.]

BENTHAM (J.) 1796: Defence of usury; shewing the impolicy of the present legal restraints on the terms of pecuniary bargains. Mathew Carey. Philadelphia. BIOLLAY (L.) 1885: Le pacte de famine. Guillaumin. Paris. BLUME (M.E.), SIEGEL (J.J.), ROTTENBERG (D.) 1993: Revolution in Wall Street. The rise and decline of the New York Stock Exchange. W.W. Norton and Company. New York. BODART (G.) 1908: Milit¨ar-historisches Kriegs-Lexikon (1618-1905). C.W. Stern. Vienna. BONANNO (G.), VANDEWALLE (N.), MANTEGNA (R.N.) 2000: Taxonomy of stock market indices. Preprint Cond-Mat, 19 January. BORD (G.) 1887: Le pacte de famine. Histoire du bl´e en France. A. Sauton. Paris. BOUCHAUD (J.-P.), POTTERS (M.) 1997: Th´eorie des risques financiers. Al´ea. Saclay. BOUCHAUD (J.-P.), POTTERS (M.) 2000: Theory of financial risks. Cambridge University Press. Cambridge (to appear). BOURGEOIS (R.) 1913: La crise e´ gyptienne. Arthur Rousseau. Paris.

References

297

BOYAJIAN (W.E.) 1988: An economic review of the past decade in diamonds. Gems and Gemology 24,134-153. BRECOURT 1934: Liste alphab´etique et prix des livres adjug´es a` l’hˆotel des ventes de Paris. Brecourt. Paris. BRUCHEY (S.) 1991: Modernization of the American Stock Exchange 19711989. Garland Publishing. New York. BRYCE (R.) 2002: Pipe dreams. Greed, ego and the death of Enron. Public Affairs. New York. BUELENS (N.), GINSBURGH (V.) 1993: Revisiting Baumol’s “art as floating crap game”. European Economic Review 37,1351-1371. CAES (C.J.) 1990: Covered calls. The safest game in the option market. Liberty Hall. Bull Ridge Summit. CAES (C.J.) 1993: Tools of the bear. Probus. Chicago. CAES (C.J.) 1994: Tools of the bull. How to make money in bull markets. Probus. Chicago. CALDARELLI (G.), MARSILI (M.), ZHANG (Y.-C.) 1997: A prototype model of stock exchange. Europhysics Letters 40,479. CALOMIRIS (C.W.), GORTON (G.) 1991: The origins of banking panics: models, facts and bank regulation. in Financial markets and financial crises, R.G. Hubbard ed. University of Chicago Press. Chicago. CAPMAS (M.) 1829: L’int´erˆet de l’argent dans le prˆet ou l’usure: condamn´ee comme contraire au droit naturel, divin et politique par toute l’antiquit´e comme depuis le XIIIe si`ecle. E. Bricon. Paris. CARSWELL (J.) 1960, 1993: The South Sea bubble. Alan Sutton. Dover (N.H.) CELLARD (J.) 1996: John Law et la R´egence: 1715-1729. Plon. Paris. CHALMIN (P.) ed. 1999: Cyclope. Mati`eres premi`eres et commodit´es. Economica. Paris. CLAUSEWITZ (K. von) 1972: La campagne de 1815 en France. Librairie Militaire Chapelot. Paris. CLAYTON (H.H.) 1944: World weather records. Smithsonian Institution. Washington DC. CLOUZOT (M.) 1996: Guide du bibliophile franc¸ais. Librairie Giraud-Badin. Paris. COCKETT (R.) 1994: Thinking the unthinkable. Think tanks and the economic counter-revolution of 1931–1983. Harper Collins. London.

298

References

COHEN (B.) 1997: The edge of chaos. Financial booms, bubbles, crashes and chaos. John Wiley. Chichester (UK). COHEN (R.) 2000: Who really brought down Milosevic? New York Times 26 November 2000. COLEMAN (J.S.) 1964: Introduction to mathematical sociology. Free Press. New York. COMMON STOCK price histories 1910-1986 and logarithmic supplement, 1988: WIT Financial Publishers. Anchorage. CONAN DOYLE (A.) 1914: The valley of fear. A.L. Burt. New York. CONLISK (J.) 1989: Three variants on the Allais example. American Economic Review 79,3,392-407. COWLES (V.) 1960: The great swindle; the story of the South Sea bubble. Harper. New York. CROUCH (R.L.) 1970: The volumes of transactions and price changes on the New York Stock Exchange. Financial Analysts Journal 26,4,104109. CYCLOPE 1990 - : Les grands march´es mondiaux. Economica. Paris DEATON (A.), LAROQUE (G.) 1992: On the behavior of commodity prices. Review of Economic Studies 59,1-23. DELTEIL (L.) ed. 1920-1931: Annuaire des ventes de livres, manuscrits, reliures armori´ees. Guide du bibliophile et du libraire. Delteil. Paris. DEMORY (J.-C.) 1995: Georges Bidault 1899-1983. Julliard. Paris. DESCARTES (R.) 1637,1965: Discourse on method, optics, geometry and meteorology. Boobs-Merill Co. DI VITTORIO (A.), GINSBURGH (V.) 1996: Des ench`eres comme r´ev´elateurs du classement des vins. Les grands crus du Haut-M´edoc. Journal de la Soci´et´e de Statistique de Paris 137,2,19-49. DRAME (S.), GONFALONE (C.), MILLER (J.A.), ROEHNER (B.) 1991: Un si`ecle de commerce du bl´e en France 1825-1913. Les fluctuations du champ des prix. Economica. Paris. EPSTEIN (E.J.) 1982: The diamond invention. An expos´e of the international diamond monopoly. Hutchinson. London. EBELING (D.), IRSIGLER (F.) 1976: Getreideumsatz, Getreide und Brotpreise in K¨oln. Mitteilungen aus dem Stadtarchiv von K¨oln, Heft 65, K¨oln.

References

299

FAIGUET DE VILLENEUVE (J.) 1770: L´egitimit´e de l’usure l´egale, o`u l’on prouve son utilit´e, [ . . . ] o`u l’on discute les passages de l’Ancien et du Nouveau Testament sur l’usure. Marc-Michel Rey. Amsterdam. FAMA (E.F.) 1965: The behavior of stock-market prices. The Journal of Business 38,34-105. FARMER (J.D.) 2000: Market force, ecology and evolution. Journal of Economic Behavior and Organization. (to appear). FARREL (M.L.) ed. 1972, 1992: The Dow Jones averages 1885-1970. Dow Jones. Princeton. FAURE (E.) 1977: La banqueroute de Law. Gallimard. Paris. FAY (S.) 1982: The great silver bubble. Hodder and Stoughton. London. FEUILLOLEY (M.) 1996: Sp´ecificit´es des objets d’art et de collection comme actifs patrimoniaux. Thesis. University of Science and Technology of Lille (France). FIERRO (A.) 1991: Histoire de la m´et´eorologie. Deno¨el. Paris. FRIEDMAN (M.) 1953: Essays in positive economics. The University of Chicago Press. Chicago. GALBRAITH (J.K.) 1997: The great crash, 1929. Boston : Houghton Mifflin Co. GALLETLY (G.) 1988: The crash and the coming crisis. Northcote House. Plymouth. GRANGER (C.W.J.) 1991: Reducing self-interest and improving the relevance of economic research. Paper presented at the 9th International Congress of Logic, Methodology and Philosophy of Science. Uppsala (Sweden, August 1991). GRANVILLE (J.E.) 1985: The warning; the coming great crash in the stock market. Freundlich Books. New York. GRENIER (J.-Y.) 1985: S´eries e´ conomiques franc¸aises (XVIe-XVIIIe si`ecles). Editions de l’Ecole des Hautes Etudes en Sciences Sociales. Paris. GRILICHES (Z.) 1957: Hybrid corn: an exploration in the economics of technological change. Econometrica 25,501-522. GROLIER (E. de) 1942-: Guide du bibliophile et du libraire. Rombaldi. Paris. HAMBLIN (R.L.), JACOBSEN (R.B.), MILLER (J.L.L.) 1973: A mathematical theory of social change. John Wiley. New York.

300

References

HARI (J.) 2009: The dark side of Dubai. The Independent 7 April 2009. [Available online on the website of the newspaper.] HARLESS (D.W.), CAMERER (C.F.) 1994: The predictive utility of generalized expected utility theories. Econometrica 62,6,1251-1289. HATTON (T.J.), WILLIAMSON (J.G.) 1998: The age of mass migration: causes and economic impact. Oxford University Press. New York. HERBERT (C.-J.) 1755, 1910: Essai sur la police g´en´erale des grains, sur leurs prix et sur les effets de l’agriculture. P. Geuthner. Paris. HEY (J.D.), ORME (C.) 1994: Investigating generalization of expected utility theory using experimental data. Econometrica 62,61291-1326. IRVING (W.) 1930: The great Mississippi bubble: a time of unexampled prosperity. Random House. New York. JEGO (C.), ROEHNER (B.M.) 2006: White flight or flight from poverty. Journal of Economic Interaction and Coordination 1,75-87. JOHANSEN (A.), SORNETTE (D.) 2000: The NASDAQ crash of April 2000: yet another example of log-periodicity in a speculative bubble ending in a crash. Cond-mat/0004263. KARPOFF (J.M.) 1987: The relation between price changes and trading volume: a survey. Journal of Financial and Quantitative Analysis 22, 1, 109-126. KELLY(M.), O’GRADA (C.) 1999: Market contagion. Evidence from the panics of 1854 and 1857. Workpaper. Centre for Economic Research, Dublin. To appear in the American Economic Review. KINDLEBERGER (C.P.) 1978: Manias, panics and crashes. A history of financial crises. Basic Books. New York. KLAVEREN (J. von) 1961: Rue de Quincampoix und Exchange Alley. Vierteljahrschrift f¨ur Sozial-und Wirtschaftsgeschichte 48,329-359. KRUGMAN (P.) 1995: Dutch tulips and emerging markets. Foreign Affairs 74,4 (Jul/Aug), 28-44. KRUGMAN (P.) 1998: Asia: What went wrong. Fortune 137,4 (Mar 2), 32-34. KURIAN (G.J.) 1994: Datapedia of the United States. Bernam Press. Lanham (MD). LABROUSSE (E.), ROMANO (R.), DREYFUS (F.-G.) 1970: Le prix du froment en France au temps de la monnaie stable (1726-1913). S.E.V.P.E.N. Paris.

References

301

LANGER (W.L.) ed. 1968: An encyclopedia of world history. Houghton Mifflin. Boston. LEVY (M.), LEVY (H.), SOLOMON (S.) 2000: Microscopic simulation of financial markets. Academic Press. New York. LIESNER (T.) ed. 1989: One hundred years of economic statistics. United Kingdom, United States of America, Australia, Canada, France, Germany, Japan, Sweden. Facts on File. New York. LUX (T.) 1995: Herd behavior, bubbles and crashes. The Economic Journal 105,881-896. MACAULAY (F.R.) 1938: The movements of interest rates, bond yields and stock prices in the United States since 1856. National Bureau of Economic Research. New York. LONGMATE (N.) 1974: The workhouse. Temple Smith, London. [According to testimonies cited in this book, workhouses inspired a real terror to poor people. They were developed in their modern form following the law of 1834. A similar institution was introduced in Ireland by the “Irish Poor Law” of 1838. It did not prevent thousands of Irish people from dying of starvation in the mid-nineteenth century.]

LUX (T.), MARCHESE (M.) 1999: Scaling and criticality in a stochastic multi-agent model of a financial market. Nature 397,498-500. MACHINA (M.J.) 1987: Choice under uncertainty. Problems solved and unsolved. Journal of Economic Perspectives 1,1,121-154. MANDELBROT (B.B.) 1997: Fractals and scaling in finance: discontinuity, concentration, risks. Springer-Verlag. New York. MANTEGNA (R.N.) 1991: Levy walks and enhanced diffusion in Milan stock exchange. Physica A 179,232-242. MANTEGNA (R.N.) 1999: Hierarchical structure in financial markets. European Physical Journal B 11,1,193-197. MANTEGNA (R.N.), STANLEY (H.E.) 1995: Scaling behavior in the dynamics of an economic index. Nature 376, 6 July, 46-49. MANTEGNA (R.N.), STANLEY (H.E.) 1999: An introduction to econophysics. Correlation and complexity in finance. Cambridge University Press. Cambridge. MARINI (P.), REMOND (B.) 1976: Sp´eculation et politique fonci`ere. Berger Levrault. Paris. MARION (M.) 1976: Dictionnaire des institutions de la France. Picard. Paris.

302

References

MARSILI (M.), CHALLET (D.) 2000: Trading behavior and excess volatility in toy markets. Preprint Cond-mat/0004376. MARTIN (G.) 1908: Les famines en 1693 et en 1709 et la sp´eculation sur les bl´es. Bulletin du Comit´e des Travaux Historiques et Scientifiques. Section des Sciences Economiques et Sociales 150-172. MASLOV (S.), ROEHNER (B.M.) 2003: Does the price multiplier effect also hold for stocks? International Journal of Modern Physics C 14, 10, 1439-1451. MASSACRIER (A.) 1978: Prix des timbres-poste classiques de 1904 a` 1975. A. Maury. Paris. MASSIE (R.K.) 1980: Peter the Great. His life and world. Little, Brown and Company. London. MATTEY (J.P.) 1990: The timber bubble that burst. Government policy and the bailout of 1984. Oxford University Press. New York. MATUSOW (H.) 1955: False witness. Cameron and Kahn, New York. MEEKER (J.E.) 1922, 1930: The work of the stock exchange. Ronald Press Company. New York. MEUVRET (J.) 1956: Le commerce des grains et des farines a` Paris et les marchands parisiens a` l’´epoque de Louis XIV. Revue d’Histoire Moderne et Contemporaine 3,169-203. MEUVRET (J.) 1971: Etudes d’histoire e´ conomiques. Cahier des Annales No 32. MITCHELL (B.R.) 1978: European historical statistics 1750-1970. Macmillan. London. MITCHELL (B.R.) 1982: International historical statistics: Africa and Asia. New York University Press. New York. MITCHELL (B.R.) 1983: International historical statistics: the Americas and Australasia. Macmillan. London. MODESTE (V.) 1862: De la chert´e des grains et des pr´ejug´es populaires qui d´eterminent les violences dans les temps de disette. Guillaumin. Paris. MONTROLL (E.W.), BADGER (W.W.) 1974: Introduction to quantitative aspects of social phenomena. Gordon and Breach. New York. MURRAY (J.D.) 1989,1993: Mathematical biology. Springer-Verlag. Berlin. NASH (V.) 1900: The Great Famine. Longman, Green and Company. London. [The title refers to the famine in India in 1900. It was one among several famines

References

303

that occurred in India in the late 19th century and the first half of the 20th century (the last one was in 1942-1943). The famine of 1900 took about 1.25 million lives in British districts. For the whole of India the estimated death toll was 3 million. Relief was provided on the workhouse model in use in England and Ireland, which means that Indian people had to break stones in order to get food.]

NEARING (S.), FREEMAN (J.) 1926: Dollar diplomacy. A study in American Imperialism. George Allen and Unwin. London. O’GRADA (C.),WHITE (E.N.) 1999: Who panics during panics? Evidence from a nineteenth century savings bank. NBER Working Paper No 8851. O’ROURKE (K.H.), WILLIAMSON (J.G.) 1999: Globalization and history. the evolution of a nineteenth-century Atlantic economy. MIT Press. Cambridge. OLIVEIRA (S.M. de), OLIVEIRA (P.M.C. de), STAUFFER (D.) 1999: Evolution, money, wars and computers. Teubner. Stuttgart. PACKARD (V.O.) 1957: The hidden persuaders. D. McKay. New York. PALMER (R.R.) 1959: The age of democratic revolution. A political history of Europe and America 1760-1800. Princeton University Press. Princeton. PANDEY (R.B.), STAUFFER (D.) 1999: Search for log-periodic oscillations in stock market simulations. Preprint. PARTNOY (F.) 2003: Infectious greed. How deceit and risk corrupted the financial markets. Profile Books. London. PERSSON (K.G.) 1993: On corn,Turgot and elasticities: the case of deregulation of grain markets in mid-eighteenth century France. Scandinavian Economic History Review 1,37-50. Phillips (K.) 2002: Wealth and democracy. A political history of the American rich. Broadway Books. New York. ¨ (G) 1972: Problems and theorems in analysis. Vol.1. POLYA (G.), SZEGO Springer-Verlag. Berlin. POMMEREHNE (W.W.), FREY (B.S.) 1989: Le placement en peintures. Une e´ tude empirique sur 350 ans (1635-1987). Journal de la Soci´et´e de Statistique de Paris 130,3,171-186. PRESTON (R.L.) 1973: How to prepare for the coming crash. Hawkes Publications. Salt Lake City. PRIBRAM (A.F.) 1938: Materialen zur Geschichte des Preise und L¨ohne in ¨ Osterreich. Carl Ueberreuters Verlag. Vienna.

304

References

PUTNAM (R.D.) 2000: Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster. New York. QUID 1997: [An annual compilation of statistical data] edited by D. Fr´emy and M. Fr´emy. Robert Laffont. Paris. RADETZKI (M.) 1990: A guide to primary commodities in the world economy. Basil Blackwell. Oxford. RAMSES 1981-: Rapport annuel mondial sur le syst`eme e´ conomique et les strat´egies. Economica. Paris. ROEHNER (B.M.) 1989: The decrease of price correlation with distance and the concept of correlation length. Environment and Planning A 21,289298. ROEHNER (B.M.) 1990: Corr´elations entre fluctuations des prix et fluctuations d´emographiques: France, XIXe si`ecle. Population 2,199-326. ROEHNER (B.M.) 1995: Theory of markets. Trade and space-time patterns of price fluctuations. A study in analytical economics. Springer-Verlag. Berlin. ROEHNER (B.M.) 1997a: The comparative way in economics: a reappraisal. Economie Appliqu´ee 50,4,7-32. ROEHNER (B.M.) 1997b: Jesuits and the state: a comparative study of their expulsions (1590-1990). Religion 27,165-182. ROEHNER (B.M.) 1999: Spatial analysis of real estate price bubbles: Paris 1984-1993. Regional Science and Urban Economics 29,73-88. ROEHNER (B.M.) 2000a: The correlation length of commodity markets. 1. Empirical evidence. 2. Theoretical framework. European Physical Journal B 13, 175-187 and 189-200. ROEHNER (B.M.) 2000b: Speculative trading: the price multiplier effect. The European Physical Journal B 14,395-399. ROEHNER (B.M.) 2000c: Identifying the bottom line after a stock market crash. International Journal of Modern Physics C 11,1,91-100. ROEHNER (B.M.) 2000d: Determining bottom price levels after a speculative peak. European Physical Journal (to appear). ROEHNER (B.M.) 2006: Real estate price peaks. A comparative overview. Evolutionary and Institutional Economic Review 2,2,167-182. ROEHNER (B.M.) 2007: Driving forces in physical, biological, and socioeconomic phenomena. A network science investigation of social bonds and interactions. Cambridge University Press, Cambridge (UK).

References

305

ROEHNER (B.M.) 2008 a: Relations between Allied forces and the population of Japan. LPTHE, University of Paris 6. [The book is available in university libraries in China, France, Japan, the UK and the US. It can also be downloaded from the author’s website at the following address: http://www.lpthe.jussieu.fr/∼roehner/occupation.html]

ROEHNER (B.M.) 2008 b: Econophysics: challenges and promises. An observation-based approach. Evolutionary and Institutional Economic Review 4,2,251-266. ROEHNER (B.M.) 2009: Interaction maximization as an evolution principle for social systems. Lectures given in China and Japan in 2008. Working Report, Institute for Theoretical and High Energy Physics (LPTHE), University of Paris 6. [Available at: http://www.lpthe.jussieu.fr/∼roehner/interac.html]

ROEHNER (B.M.), SORNETTE (D.) 1998: The sharp peak – flat trough pattern and critical speculation. The European Physical Journal B 4,387399. ROEHNER (B.M.), SORNETTE (D.) 1999: Analysis of the phenomenon of speculative trading in one of its basic manifestations: postage stamp bubbles. International Journal of Modern Physics C 10,6,1099-1116. ROEHNER (B.M.), SORNETTE (D.) 2000: “Thermometers” of speculative frenzy. The European Physical Journal B 16,729-739. ROGALSKI (R.J.) 1978: The dependence of prices and volume. Review of Economics and Statistics 60,2,268-174. ROSENBERG (H.) 1934: Die Weltwirtschaftskrisis. Kohlhammer. Stuttgart. ROTTA (A.) 1994: Bookselling in a changing world. in: Antiquarian books, P. Bernard et al. eds. Scolar Press. Aldershot. RUBNER (A.) 1966: The economics of gambling. Macmillan. London. RYAN (B.), GROSS (N.C.) 1943: The diffusion of hybrid seed corn in two Iowa communities. Rural Sociology 8,15-24. SAKOLVSKI (A.M.) 1932: The great American land bubble. The amazing story of land-grabing, speculations, and booms from colonial days to the present time. Harper and Brothers. New York. SAMPSON (A.) 1982: Les banquiers dans un monde dangereux. Laffont. Paris. translated from: The money lenders: bankers and a world in turmoil (1981).

306

References

SAVARY DES BRUSLONS (J.) 1723: Dictionnaire universel de commerce. Contenant tout ce qui se fait dans les quatre parties du monde par terre, par mer, de proche en proche, et par les voyages au long cours tant en gros qu’en d´etail. Jacques Etienne. Paris. SCHNELLER (M.I.) 1978: Security price changes and transaction volumes. Comment. American Economic Review 68,4,696-697. SCHWARTZ (A.J.) 1995: An interview with Anna J. Schwartz. The Newsletter of the Cliometric Society 10,2,3-7. SCHWERT (G.W.) 1996: Stock market crash of October 1987. in E.N. White ed.: Stock market crashes and speculative manias. Edwar Elgar. Cheltenham. SCOTT (F.D.) 1977: Sweden. The nation’s history. University of Minnesota Press. Minneapolis. SEUFFERT (G.K.L.) 1857: Statistik des Getreide und Viktualien Handels im K¨onigreiche Bayern. Weisz. Munich. SHATNER (M.), MUSHNICK (L.), LESHNO (M.), SOLOMON (S.) 2000: A continuous time asynchronous model of the stock market. Preprint, Cond-mat/0005430. SHILLER (R.J.) 1990: Speculative prices and popular models. A symposia on bubbles: third contribution. Journal of Economic Perspectives 4,2,13101 SNOOKS (G.D.) 1993: Economics without time: a science blind to the forces of historical change. Macmillan. Basingstoke (UK). SNOOKS (G.D.) 1998: Longrun dynamics: a general economic and political theory. Macmillan. Basingstoke (UK). SORNETTE (D.) 2000: Speculation. Spontaneous symmetry breaking in economic valuation. Preprint (March 17). SORNETTE (D.), JOHANSEN (A.), BOUCHAUD (J.-P.) 1996: Stock market crashes, precursors and replicas. Journal de Physique I France 6,167175. STAUFFER (D.), SORNETTE (D.) 1999: Self-organized percolation model for stock market fluctuations. Physica A 271,496-506. STOCKMAN (D.) 1986: The triumph of politics: why the Reagan revolution failed. G.K. Hall. Boston. [David Stockman argues that the failure came from the fact that spending (and in particular social spending) was not reduced in proportion to tax reduction which created a huge public deficit. He acknowledges that he was somewhat

References

307

naive and failed to understand that the whole scheme was in fact a disguised way for cutting tax rates on high income brackets. This explains the title of a long article “The education of David Stockman” which appeared in the December 1981 issue of the magazine “The Atlantic Monthly”.]

STOMMEL (H.), STOMMEL (E.) 1979: The year without summer. Scientific American, June, 134-140. ¨ (R.), PALMSTIERNA (C.F.) 1944: Histoire de la Su`ede. SVANSTROM Stock. Paris. SWIERENGA (R.P.) 1968: Pioneers and profits: land speculation on the Iowa frontier. Iowa State University Press. Ames. TANIGUCHI (T.) 1993: Japan’s banks and the “bubble economy” of the late 1980s. Monograph Series No.4. Princeton University. TAUCHEN (G.E.), PITTS (M.) 1983: The price variability-volume relationship on speculative markets. Econometrica 51,2,485-505. TVEDE (L.) 1990: The psychology of finance. Norwegian University Press. TYCHONOV (A.N.), SAMARSKI (A.A.) 1964: Partial differential equations of mathematical physics. Vol.1 Holden Day. San Francisco. VANDEWALLE (N.), BOVEROUX (P.), BRISBOIS (F.) 2000: Domino effect for world market fluctuations. Cond-mat/0001293 (January). VIGREUX (P.) 1967: Dynamique du march´e financier. Vol.1. Les Editions d’Organisation. Paris. VINCENT (F.) 1946: Histoire des famines a` Paris. Editions Politiques, Economiques et Sociales. Paris. VON NEUMANN (J.), MORGENSTERN (O.) 1953: Theory of games and economic behavior. Princeton University Press. Princeton. WALVIN (J.) 1992: Black ivory. A history of British slavery. Harper Collins. WHITE (E.N.) ed. 1996: Stock market crashes and speculative manias. Edward Elgar. Cheltenham. WILLIAMS (E.) 1944, 1994: Capitalism and slavery. University of North Carolina Press. Chapel Hill (N.C.). WIRTH (M.) 1883: Geschichte der Handelskrisen. Frankfurt am Main. Sauerl¨ander’s Verlag. WORLD COINS 1976-: Iola (Wisconsin). WOOD (C.) 1992: The bubble economy. The Japanese economic collapse. Sidgwick and Jackson. London.

308

References

WYCKOFF (P.) 1971: Wall Street and the stock markets. A chronology (1644-1971). Chilton Book Company. Philadelphia. YING (C.C.) 1966: Stock market prices and volumes of sales. Econometrica 34,3,676-685. ZHANG (Y.-C.) 1998: Evolving models of financial markets. Europhysics. News 29,51.

Index

Absentee landlords 211-214, 254-255 Accidents (at work) 213 Accounting rules 57-58 American Steel 113 American Telephone and Telegraph 113 Antiquarian book market 31, 134, 137 Argentina (bankruptcy) ix, 246 Art market 79 Banana prices 23 Bear market 151-179 Berlin 154 Black-White divide 224 Bond market 77 Brahe 4 Bull market 151-179 Byron 132-133 California 74-75 Central Intelligence Agency (CIA) 250 Citigroup 277 Cobalt 66, 69 Coefficient of variation 137-141 Coffee 165

Coins 37, 92 Colonial empire 207, 247 Commodity markets 20-25, 163164 Comparative analysis x, 7 Complexity 5 Concave 156 Concession system (in China) 253 Consumer confidence 97 Contagion 83-84, 124 Copper 158 Corner 29, 106 Cournot 5 Credit crunch 229 Crisis of 1893-1857 259-262 Currency board 244 De Beers 67 Debt-to-equity ratio 231 Deregulation 115 Derivatives 39, 202, 239-240 Diamonds 26, 35-36, 66-69, 95 Diffusion model 143-145 Discount rate 179, 186, 192-193 Doctoring (accounts) xi Dollarization 241-242, 244 Dubai 197

310 EASDAQ 41 Economic Bill of Rights 267-268 Econophysics vii Ecuador (dollarization) 241-243 Enron Corporation (bankruptcy) ix, 217, 231 Ensemble dispersion 137-140 Exotic wood 158 Fed (American Federal Reserve) 151, 179 Fielding 80, 132, 133 Fisher 58-59 Flat trough pattern 157, 164, 166 Floating stockpile markets 22, 70 Fragebogen 250 Free-trade agreement 253, 291293 Frequency of transactions 111 Friedman, Milton 204-205 Funds (financial) 217 Galsworthy 132, 133 Gated communities 225-226 General Motors 235 Gold 67-69, 165 Grains 17-21, 139, 152-153, 160161, 175 Greenspan 151, 179 Harvard 85-87 Hayek, Friedrich 271 Health care 52 Harvesting ants 3 High school graduates 220 Hoarding (gold, currency) 229, 261, 265

Index Hong Kong 27, 90 Hunt 68 Incarceration rate 222 Income inequality 207-212 Industrial production 284-288 Infant mortality 221, 223 Inflation 70-71, 86, 130 Infrastructures 227 Investment funds 39, 113 Investors (percent of) 136 Ireland 83 Jesuits 73 Jute 158 Keynes, John Maynard 202 Latin America ix, 198, 253, 294, 295 Lead 158 Lobbying 260 Log-periodic pattern 152, 156 Longitudinal analysis 7 Loyalty oath 210 Levy-Levy-Solomon model 77 Maquiladora 291 Margin debt 180, 187-188 Massacrier 31, 134 McCarran Internal Security Act 210 Mendel 2 Merril Lynch 231 Meteorology 5, 6 Mississippi bubble 73 Morgan Stanley 231

Index Morgenstern 2 Mortgage backed securities 255258 MOTHERS 41 Munster (Ireland) 83 Murdoch medias 277-278 NAFTA 291 NASDAQ 41-42, 115-116, 173, 190 Neoliberalism vii, 197-205, 215, 231 Neuer Markt 41 New Deal 260, 266-268, 294, 295 Newton 2, 4, 9 Nikkei index 78 No-documentation loan 255 Northern Rock 232 Nouveau march´e 41 Obama, Barack (President) 260, 272-282 Orange County (bankruptcy) 217 Otpor 250 Palladium 67, 69 Paintings 37, 80 Pareto 5 Paris 72, 111, 115 Partnoy, Frank ix, 305 PER see Price earnings ratio Phillips, Kevin 208-209, 305 Platinum 67-69 Poe 132-133 Port-Said 26 Postage stamps vii, 28-31, 129131, 165-166, 175

311 Press 89-91 Price earnings ratio (PER) 10, 45, 109, 186 Price multiplier effect 11, 121 125-149 Printing money 281 Privatization ix, xi, 215, 251-254 Property, property values vii, 56, 60 74-75, 78, 119-123, 137, 140-142, 147, 164-165 Quasi-experiments 7 Railroad companies (bailed out) 260 Reagan, Ronald (President) 198, 259 Real estate market vii, 56, 60 7475, 78, 119-123, 137, 140142, 147, 164-165, 232, 236238 Regularities 3, 5, 10 Regulation 257, 281 Resilience pattern 11, 182-185 Rich people xii, 276 Roosevelt Franklin (President) 260, 264-268, 273 Roosevelt, Theodore (President) 259 Run on bank(s) 261, 263 Sarbanes-Oxley Act 234 Savings banks 75 Saw-timber 24 Segmentation 214, 221-228, 234235 Sharp peak - flat trough pattern

312 11, 154-167 Shelley 80, 132 Silver 67-69, 72, 165 Simple systems 2 Singapore 27, 90 Size (of apartments) 129 Slave trade 50-51 Smith Act 210 Social fragmentation 203 (see also segmentation) Sornette viii, 152, 190 South Sea bubble 73 Social organizations 227-228, 288 Spatial propagation 120-122 Special Purpose Entities 231 Speculative bubble 156 Speculative ratio 14 Square Deal 259 Stock markets 37-42 Stock prices 35, 146-148, 167178, 179-195, 236 Subprime crisis viii, 231-232, 255 Sugar 24, 164-165 Supply/demand functions 70 Supply of labor vii, 289-290 Taft-Hartley Act 222, 271-272 Tax havens 231, 275, 279 Tax shelters, see tax havens Tea 165 Thatcherism vii, xi Tokyo 72, 79, 80, 90, 93 Trading volume 97-107, 116 Transactions costs 114-115 Transverse analysis 7 Turn over ratio 43, 45

Index Two-body problem 2 UBS (bank) 275-276 Ulster (Ireland) 83 Unions 210, 271, 275, 282, 283, 293 US Agency for International Development (USAID) 251 US Armed Forces 206, 226 Usury 50 Von Neumann 2 Wagner Act 210, 271 Wealthy people, see Rich people White flight 224 Zeitgeist 213 Zinc 165