WhyStockMarketsCrash This page intentionally left blank WhyStockMarketsCrash Critical Events in Complex Financial Systems D i d i e r S o r n e t t e PRINCETON UNIVERSITY PRESS Princeton and Oxford Copyright©2003byPrincetonUniversityPress Published by Princeton University Press, 41 William Street, Princeton,NewJersey08540 IntheUnitedKingdom:PrincetonUniversityPress,3Market Place,Woodstock,OxfordshireOX201SY AllRightsReserved LibraryofCongressCataloging-in-PublicationData Sornette,D. Whystockmarketscrash:criticaleventsincomplex financialsystems/DidierSornette. p. cm. Includesbibliographicalreferencesandindex. ISBN0-691-09630-9(alk.paper) 1.Financialcrises—History.2.Stocks—Prices—History. 3.Financialcrises—UnitedStates—History. 4.Stockexchanges—UnitedStates—History. 5. Critical phenomena (Physics). 6. Complexity (Philosophy). I.Title. HB3722.S662002 332.63(cid:2)222–dc212002024336 BritishLibraryCataloging-in-PublicationDataisavailable ThisbookhasbeencomposedinTimes Printedonacid-freepaper.(cid:3) www.pupress.princeton.edu PrintedintheUnitedStatesofAmerica 10987654321 Contents xiii Preface Chapter 1 3 What Are Crashes, and Why financial crashes: Do We Care? what, how, why, 5 The Crash of October 1987 and when? 3 7 Historical Crashes 7 The Tulip Mania 9 The South Sea Bubble 12 The Great Crash of October 1929 15 Extreme Events in Complex Systems 20 Is Prediction Possible? A Working Hypothesis Chapter 2 27 The Basics fundamentals of 27 Price Trajectories financial markets 30 Return Trajectories 26 33 Return Distributions and Return Correlation 38 The Efficient Market Hypothesis and the Random Walk 38 The Random Walk vi contents 42 A Parable: How Information Is Incorporated in Prices, Thus Destroying Potential “Free Lunches” 45 Prices Are Unpredictable, or Are They? 47 Risk–Return Trade-Off Chapter 3 49 What Are “Abnormal” Returns? financial crashes 51 Drawdowns (Runs) are “outliers” 51 Definition of Drawdowns 49 54 Drawdowns and the Detection of “Outliers” 56 Expected Distribution of “Normal” Drawdowns 60 Drawdown Distributions of Stock Market Indices 60 The Dow Jones Industrial Average 62 The Nasdaq Composite Index 65 Further Tests 69 The Presence of Outliers Is a General Phenomenon 70 Main Stock Market Indices, Currencies, and Gold 73 Largest U.S. Companies 75 Synthesis 76 Symmetry-Breaking on Crash and Rally Days 77 Implications for Safety Regulations of Stock Markets Chapter 4 82 Feedbacks and Self-Organization positive feedbacks in Economics 81 89 Hedging Derivatives, Insurance Portfolios, and Rational Panics 91 “Herd” Behavior and “Crowd” Effect 91 Behavioral Economics contents vii 94 Herding 96 Empirical Evidence of Financial Analysts’ Herding 99 Forces of Imitation 99 It Is Optimal to Imitate When Lacking Information 104 Mimetic Contagion and the Urn Models 106 Imitation from Evolutionary Psychology 108 Rumors 111 The Survival of the Fittest Idea 112 Gambling Spirits 114 “Anti-Imitation” and Self-Organization 114 Why It May Pay to Be in the Minority 115 El-Farol’s Bar Problem 117 Minority Games 118 Imitation versus Contrarian Behavior 121 Cooperative Behaviors Resulting from Imitation 122 The Ising Model of Cooperative Behavior 130 Complex Evolutionary Adaptive Systems of Boundedly Rational Agents Chapter 5 134 What Is a Model? modeling financial 135 Strategy for Model Construction bubbles and in Finance market crashes 135 Basic Principles 134 136 The Principle of Absence of Arbitrage Opportunity 137 Existence of Rational Agents 139 “Rational Bubbles” and Goldstone Modes of the Price “Parity Symmetry” Breaking 140 Price Parity Symmetry viii contents 144 Speculation as Spontaneous Symmetry Breaking 148 Basic Ingredients of the Two Models 150 The Risk-Driven Model 150 Summary of the Main Properties of the Model 152 The Crash Hazard Rate Drives the Market Price 155 Imitation and Herding Drive the Crash Hazard Rate 162 The Price-Driven Model 162 Imitation and Herding Drive the Market Price 164 The Price Return Drives the Crash Hazard Rate 168 Risk-Driven versus Price-Driven Models Chapter 6 173 Critical Phenomena by Imitation on hierarchies, Hierarchical Networks complex fractal 173 The Underlying Hierarchical dimensions, and Structure of Social Networks log-periodicity 177 Critical Behavior in Hierarchical 171 Networks 181 A Hierarchical Model of Financial Bubbles 186 Origin of Log-Periodicity in Hierarchical Systems 186 Discrete Scale Invariance 188 Fractal Dimensions 192 Organization Scale by Scale: The Renormalization Group 192 Principle and Illustration of the Renormalization Group 195 The Fractal Weierstrass Function: A Singular Time-Dependent Solution of the Renormalization Group contents ix 198 Complex Fractal Dimensions and Log-Periodicity 208 Importance and Usefulness of Discrete Scale Invariance 208 Existence of Relevant Length Scales 209 Prediction 210 Scenarios Leading to Discrete Scale Invariance and Log-Periodicity 211 Newcomb–Benford Law of First Digits and the Arithmetic System 213 The Log-Periodic Law of the Evolution of Life? 217 Nonlinear Trend-Following versus Nonlinear Fundamental Analysis Dynamics 218 Trend Following: Positive Nonlinear Feedback and Finite-Time Singularity 220 Reversal to the Fundamental Value: Negative Nonlinear Feedback 223 Some Characteristics of the Price Dynamics of the Nonlinear Dynamical Model Chapter 7 228 The Crash of October 1987 autopsy of major 231 Precursory Pattern crashes: universal 236 Aftershock Patterns exponents and log- 239 The Crash of October 1929 periodicity 228 242 The Three Hong Kong Crashes of 1987, 1994, and 1997 242 The Hong Kong Crashes 246 The Crash of October 1997 and Its Resonance on the U.S. Market 254 Currency Crashes 259 The Crash of August 1998 263 Nonparametric Test of Log-Periodicity