La lecture en ligne est gratuite
Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres
Télécharger Lire

Experimental studies of overconfidence in financial markets [Elektronische Ressource] / vorgelegt von Julija Michailova

De
158 pages
Experimental Studies of Overconfidence in Financial Markets Inaugural-Dissertation zur Erlangung des akademischen Grades eines Doktors der Wirtschafts- und Sozialwissenschaften der Wirtschafts- und Sozialwissenschaftlichen Fakultät der Christian-Albrechts-Universität zu Kiel vorgelegt von MA, Julija Michailova aus Vilnius, Litauen Kiel, 2010 Gedruckt mit Genehmigung der Wirtschafts- und Sozialwissenschaftlichen Fakultät der Christian-Albrechts-Universität zu Kiel Dekanin: Prof. Dr. Birgit Friedl Erstberichterstattender: Prof. Dr. Ulrich Schmidt Zweitberichterstattender: Prof. Dr. Till Requate Tag der Abgabe der Arbeit: 13.10.2010 Tag der mündlichen Prüfung: 21.12.2010 To my Family Моей семье Mano šeimai Mojej rodzinie ACKNOWLEDGEMENTS It is a pleasure to thank all those who made this thesis possible! First of all I acknowledge the Stadt-Kiel Scholarship and the German Academic Exchange Office (DAAD) scholarship. I would like to thank the staff of DAAD, especially Ludmila Winter Souhradová, Maria-Luise Nünning and Hans Golombek. I am indebted to my first supervisor Prof. Dr. Ulrich Schmidt for accepting me as a doctoral student, his continuous support and encouragement. I greatly appreciate helpful comments and guidance of my second supervisor Prof. Dr. Till Requate.
Voir plus Voir moins



Experimental Studies of Overconfidence
in Financial Markets






Inaugural-Dissertation
zur Erlangung des akademischen Grades eines Doktors
der Wirtschafts- und Sozialwissenschaften
der Wirtschafts- und Sozialwissenschaftlichen Fakultät
der Christian-Albrechts-Universität zu Kiel




vorgelegt von
MA, Julija Michailova
aus Vilnius, Litauen






Kiel, 2010



Gedruckt mit Genehmigung der
Wirtschafts- und Sozialwissenschaftlichen Fakultät
der Christian-Albrechts-Universität zu Kiel



Dekanin: Prof. Dr. Birgit Friedl



Erstberichterstattender: Prof. Dr. Ulrich Schmidt

Zweitberichterstattender: Prof. Dr. Till Requate













Tag der Abgabe der Arbeit: 13.10.2010
Tag der mündlichen Prüfung: 21.12.2010





To my Family
Моей семье
Mano šeimai
Mojej rodzinie











ACKNOWLEDGEMENTS
It is a pleasure to thank all those who made this thesis possible!
First of all I acknowledge the Stadt-Kiel Scholarship and the German Academic Exchange
Office (DAAD) scholarship. I would like to thank the staff of DAAD, especially Ludmila
Winter Souhradová, Maria-Luise Nünning and Hans Golombek.
I am indebted to my first supervisor Prof. Dr. Ulrich Schmidt for accepting me as a doctoral
student, his continuous support and encouragement. I greatly appreciate helpful comments
and guidance of my second supervisor Prof. Dr. Till Requate. Their experience and well-
wishing attitude enabled me to conduct this research successfully.
I want to express gratitude to Prof. Arlington W. Williams and Dr. Briony D. Pulford for a
fruitful cooperation, and to thank Prof. Dr. Jürgen Golz for valuable comments. I am grateful
to Prof. Dr. Urs Fischbacher for prompt answers to questions about the software z-Tree. It is a
pleasure to thank my colleague Israel Waichman for sharing his knowledge of experimental
processes. I also acknowledge help of the student research assistants at the chair of
“Innovation, Competition Policy and New Institutional Economics”: Julia Schirrmacher and
Dennis Nissen.
I can’t help but thank my dear friends and colleagues who made my life in Kiel interesting
and full of joy: Emily Sek, Aidan Islyami, Zhanna Kapsalyamova, Inna Melnykovska, Natalia
Maksajda, Maciej Sierżant, Katarzyna Karłowska, Milda Stanionytė, Miriam Schneider, Jaba
Ghonghadze, Patrick Weber, Justyna Mysliwy. A special thank to Christian Radden, who
helped in translating experimental instructions of my first experiment. Not to forget Indrė
Morkunaitė, a close friend who supported me all the way from home. My life is happier when
there are you – my friends!
I owe my deepest gratitude to my parents Natalija Michailova and Dr. Andrejus Michailovas
for their unconditional love, which only parents feel for their children. My parents have
worked hard to give me and my brother the best they could, and I want to thank them for their
sacrifice and for being an example for us to follow. I am indebted to you for supporting my
initiatives and helping me become the person I am. I am grateful to my little brother Kirilas
Michailovas for being my best friend and a trip companion! It is a pleasure to thank my
loving grandmothers Liudmila Michailova and Tamara Macneva who were my biggest fans
through all these years! I want to thank my aunt Natalja Michailova who has always been an
example of the independent career woman to me. Thank you all for the spent hours on the
phone listening to my endless stories with patience and interest!
I would like to show my gratitude to Piotr Jamróg who turned my life in Kiel to one of the
happiest experiences of my life! The rainy days in Kiel were lighted up by his kind and loving
heart! He was the greatest helper, supporter and an inspiration to me!
I want to thank my family for being there for me through good times and bad times, and
dedicate my thesis to them! TABLE OF CONTENTS
Introduction 1
References 4
Development of the Overconfidence measurement instrument for the Economic Expe-
riment 6
1.1 Introduction 7
1.2 Overconfidence in Psychological Research 9
1.2.1 Definition of the Overconfidence Bias 10
1.2.2 Measurement of Overconfidence 11
1.2.3 Factors Influencing the Degree of Overconfidence 13
1.3 Overconfidence in Financial Literature 14
1.3.1 Findings from the Research 14
1.3.2 Measurement of Overconfidence in Empirical and Experimental Studies 15
1.3.3 Problem Statement 19
1.4 Method 21
1.5 Results 23
1.5.1 Statistical Tests 25
1.6 Second Experiment 28
1.7 Conclusions 31
References 32
Appendix 1.A 40
Appendix 1.B 41
Appendix 1.C 42
Appendix 1.D 43
Appendix 1.E 43
Appendix 1.F 44
Appendix 1.G 46
Appendix 1.H 47
Overconfidence and Bubbles in Experimental Asset Markets 50
2.1 Introduction 51
2.2 Overconfidence 54
2.2.1 Overconfidence in Psychological Research 54
2.2.2 Overconfidence in Financial Research and Contributions 55
2.3 Hypotheses 59
2.4 Pre-Experimental Overconfidence Measurement 60
2.5 Experimental Design 62
2.5.1 Participants 62
2.5.2 Experimental Procedure and the Rules of the Game 63
i 2.5.3 Incentives 64
2.6 Experimental Results 65
2.6.1 Numerical Characteristics of the Two Types of the Market 65
2.6.2 Overconfidence Measure from the Forecasting Task 69
2.6.3 Forecasting Precision 71
2.6.4 Bubble Measures 72
2.7 Conclusions 78
References 80
Appendix 2.A 85
Appendix 2.B 86
Appendix 2.C 90
Appendix 2.D 91
Appendix 2.E 95
Overconfidence, Risk Aversion and (Economic) Behavior of Individual Traders in
Experimental Asset Markets 96
3.1 Introduction 97
3.2 Psychological Determinants of Trading Activity and Performance 100
3.2.1 Overconfidence 100
3.2.2 Risk aversion 102
3.2.3 Similar Research and Value Added 104
3.3 Hypotheses 108
3.4 Experimental Procedure and the Rules of the Game 109
3.4.1 Asset Market Experiment 110
3.4.2 Risk aversion Expet 111
3.5 Experimental Results 114
3.5.1 Univariate and Bivariate Analysis 114
3.5.2 Multivariate Analysis 120
3.6 Risk Aversion Analysis: Experimental Results 129
Conclusions 131
References 134
Appendix 3.A 139
Appendix 3.B 140
Appendix 3.C 141
Appendix 3.D 142
Appendix 3.E 144
Appendix 3.F 145
ii LIST OF FIGURES

1.1 Distribution of accuracy (a) and overconfidence per question (b) in test-50 24
1.2 Distribution of the bias score of participants per test: a. test-50 and b. test-18 25
2.1 Average trade prices in both types of markets 65
2.2 Development of the average market price 66
2.3 Volatility of asset prices in both types of markets 67
2.4 Average trading activity (turnover) in both types of markets 68
2.5 Average overconfidence in both types of markets 69
2.6 Comparisons of BS(2-8) and BS(9-15): a. rational market; b. overconfident market 70
2.7 Average forecast and market price: a. rational market; b. overconfident market 72
2.8 Normalized price deviations from FV by trading period and overconfidence level:
a. rational market, b. overconfident market 74
2.9 Normalized absolute price deviations from FV by trading period: a. rational mar-
ket, b. overconfident market 75
3.1 Normalized profits per participant 114
3.2 Distribution of individual profits across trading activity quintiles 115
3.3 The cross-sectional distribution of relative profits by treatment 117
3.4 Distribution of safe choices in the group 128

iii LIST OF TABLES

1.1 Average confidence, accuracy and bias score for the three levels of question
difficulty of the final overconfidence test (test-18) 23
1.2 Characteristics of the three levels of question difficulty of the test-18
from the experiment on 14.06.09 29
2.1 Bubble measures in each session 76
2.2 Average values of some of the bubble measures from previous studies 77
3.1 The ten paired lottery-choice decisions 111
3.2 Risk aversion classification based on the number of safe lottery choices 112
3.3 Profit and trading activity in the lowest and highest trading activity quartiles:
a. The whole sample; b. Sample without outliers 116
3.4 Trading activity (all errors are heteroskedasticity corrected) 120
3.5 Gains from Trade (all errors are heteroskedasticity corrected) 125

iv INTRODUCTION
Behavioral finance is a relatively new field in economics, which uses findings from
psychology and sociology for explaining investor behavior or market anomalies when rational
models provide no sufficient explanations (Glaser et al., 2003). Traditional decision making
models in economics rest on the premise that the world is populated by rational, calculating,
and unemotional utility maximisers, called homo economicus. However numerous empirical
and experimental studies demonstrated that investors’ behavior does not fit into this
rationality framework.
Overconfidence is one of the psychological biases, which can cause deviations from rational
behavior. The concept of overconfidence is based on the large body of evidence from
cognitive psychological research. Studies conducted by psychologists demonstrate that most
people are overconfident about their abilities, knowledge, precision of their personal
information, and are unreasonably optimistic about their future. Interest of economists in the
consequences of investor and market overconfidence has generated a large body of literature.
The research evidence suggests that overconfidence in financial markets can manifest itself
through high volumes of trade, excessive price volatility, dissemination of speculative
bubbles, etc. On the individual level overconfidence results in higher trade aggressiveness,
portfolio undiversification, risk underestimation and suboptimal performance.
This thesis takes a closer look at the imperfections of human nature and their impact on
behavior in economic situations, by conducting a laboratory experiment in a financial market
setting. The aim of this research was twofold: first, to investigate the role of market
overconfidence in the occurrence of bubbles in asset prices and the emergence of other stylized
facts of financial markets (excessive trade, excessive price volatility); second, to investigate the
influence of behavioral factors, namely the degree of overconfidence and risk aversion, on
financial decision making of economic subjects. This is a challenging research topic,
especially in the face of devastating economic consequences of the recent real estate bubble. It
is important to improve the knowledge of the causes lying behind the formation of bubbles
and their subsequent burst, and get a better understanding of the motivation and reactions of
market participants. Ability to grasp the role of psychological biases on the functioning of
financial markets could result in the development of the mechanisms and implementation of
policies that could, if not hinder the occurrence of the speculative price bubbles, at least
enable control of their magnitudes.
1 In real-world situations no direct way to assess the degree of market overconfidence exists. It
is also a troublesome task to guess who of the investors is overconfident and how strong her
degree of overconfidence is. Not to mention their informational, financial, and preferential
heterogeneity. Necessity to control for too many factors in the field data was the reason to
choose laboratory experiments for the purpose of data collection. Experimental simulations of
financial markets are of high value to research, as they allow isolating the effect of a
particular bias on individual behavior. Moreover according to Friedman and Sunder (1994)
experimental data are relatively easy to interpret. Biais et al. (2005) point out that study of
controlled environments allows more confident inferences about cause and effect relations.
Laboratory market setting also enables overcoming several problems, existing in
determination of the fundamental asset value, namely estimation of expected dividends on the
asset through the time period, determination of the terminal asset value and discount rates for
calculation of the present value. These components can be controlled in the laboratory asset
market, and thus favorable conditions for observing subjects’ behavior are created.
This thesis consists of three papers. In Chapter 1 an instrument is developed that is later used
in economic experiments to measure subjects’ overconfidence. Chapter 2 analyzes the
connection between overconfidence, stock market bubbles and other stylized facts (high trade
and volatility) from an aggregated market perspective. In Chapter 3 economic consequences
of overconfidence and risk aversion are being assessed at the individual level. Each chapter
begins with motivation of the research issue and review of the related literature. After that
description of the experimental design is presented and research objectives are formulated.
Then data analysis follows. Finally conclusions are drawn. Experimental instructions are
provided in the Appendixes to the papers.
Prior to running the economic experiments, described in Chapters 2 and 3, subjects’ degree of
overconfidence had to be measured. However in economic experiments no conventional
method of overconfidence measurement exists. Various proxies, tests and tasks applied for
this purpose do not always offer measures that are satisfactory. These considerations were the
basis upon which the first chapter of this thesis was built. This chapter is aimed at the
development of an instrument (test) that would enable the construction of a comprehensive
measure of individual overconfidence for the use in economic experiments. The developed
instrument should enable assessment of the differences among subjects with respect to their
degree of overconfidence and minimize the measurement error. The final test, consisting of
18 general knowledge questions, was obtained in a two-stage procedure. In the first
experimental phase a pilot test, consisting of fifty general knowledge questions of unknown
2

Un pour Un
Permettre à tous d'accéder à la lecture
Pour chaque accès à la bibliothèque, YouScribe donne un accès à une personne dans le besoin