Where is the market? [Elektronische Ressource] : three econometric approaches to measure contributions to price discovery / vorgelegt von Franziska Julia Peter

Informations

• Publié par: eberhard_karls_universitat_tubingen
• Publié le: 01 janvier 2011
• Nombre de lectures: 11
• Langue: English
• Poids de l'ouvrage: 4 Mo

Extrait

Where is the Market?
Three Econometric Approaches to Measure Contributions to Price
Discovery
Inaugural-Dissertation
zur Erlangung des Doktorgrades
der Wirtschafts- und Sozialwissenschaftlichen Fakulta¨t
der Eberhard-Karls-Universita¨t Tu¨bingen
vorgelegt von
Franziska Julia Peter
aus Stuttgart
2011Dekan: Professor Dr. rer. soc. Josef Schmid
Erstberichterstatter: Professor Dr. rer. pol. Joachim Grammig
Zweitberichterstatter: Professor Dr. rer. pol. Martin Biewen
Tag der mu¨ndlichen Pru¨fung: 01. April 2011To whom, if not to you.Acknowledgements
Thanks to my parents for their never-ending support and trust,
To my little sister for teaching me how to surf the waves,
To my brothers, for finally remembering my birthday,
And to my grandparents for being wise and open-minded.
Thanks to my supervisor for picking me up from the econometric gutter,
Shoving me into conferences all around the world,
Andmakingmerealize that“Sowhat?” isnot anappropriate answertoastupidquestion.
Thanks to my co-authors for enduring to work with me,
To Jo, for making up new words and teaching me the importance of being persistent,
To Kerstin, my kindred spirit, for the scarves and always calling back.
Thanks to my colleagues,
To Tommy for his heartily hugs,
To Stephan for his glass being always half full,
To Luis for finally attributing some humanity to me,
To Ramona for our joint analysis of the complexity of the past,
To Tobi for uttering pieces of wisdom,
To Oli, for pointing out that sometimes there is nothing more important than the next
soccer game.
To all the student assistants for their delicious cakes and for breathing fresh air into the
daily routines of a working day.
Franziska Julia Peter, Tu¨bingen 03. April 2011Contents
List of Figures i
List of Tables ii
1 Introduction 1
2 Standard Measures for Contributions to Price Discovery 5
3 A Data Driven Approach to Estimate Unique Information Shares 10
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Fat Tails, Tail Dependence, and Unique Information Shares . . . . . . . . . 13
3.2.1 Motivation and Econometric Specification . . . . . . . . . . . . . . . 13
3.2.2 Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.3 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.3 Empirical Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3.1 Credit Default Swaps, Credit Spread, and the Price of Credit Risk . 21
3.3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4 Estimation Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . 24
3.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Appendix A: Propositions and Proofs . . . . . . . . . . . . . . . . . . . . . . . . 31
4 An Intensity Based Information Share 36
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 A New Measuring for Price Discovery in International Stock Markets . . . 39
4.2.1 The Autoregressive Conditional Intensity Model . . . . . . . . . . . 39
4.2.2 The Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.3 Intensity Based Unique Information Shares . . . . . . . . . . . . . . 46
4.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.4 Cross Sectional Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Appendix B1: Adjustment of Intraday Effects . . . . . . . . . . . . . . . . . . . . 58
Appendix B2: Additional Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . 595 Using Transfer Entropy to Measure Information Flows 63
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 The Concept of Transfer Entropy . . . . . . . . . . . . . . . . . . . . . . . . 65
5.3 Empirical Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.3.1 Pricing Credit Risk: Information Flows between the CDS market
and the Corporate Bond Market . . . . . . . . . . . . . . . . . . . . 69
5.3.2 The Information Transfer between Market Risk and Credit Risk . . 79
5.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Appendix C: Additional Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6 Summary and Conclusion 87List of Figures i
List of Figures
2.1 Information share estimates at different frequencies. . . . . . . . . . . . . . . . . . . 8
3.1 Scatter plots of composite price innovations. . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Scatter plots of composite price innovations with DGPs revealed. . . . . . . . . . . . . 17
3.3 Scatter plots of VECM residuals. . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.1 Pooled point process illustration. . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 Intraday pattern of durations.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3 Cumulated impulse response function of a standard deviation innovation shock. . . . . . 47
4.4 Stock specific intensity based and Hasbrouck information shares on TSX. . . . . . . . . 53
5.1 Kernel density plots of Allianz CDS and credit spread first differences. . . . . . . . . . 73
5.2 VIX and iTraxx. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80List of Tables ii
List of Tables
3.1 Data descriptives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 Mixture model estimation results. . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Alternative measures for contributions to price discovery. . . . . . . . . . . . . . . . 27
4.1 Sample stocks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 Descriptive statistics across sample stocks. . . . . . . . . . . . . . . . . . . . . . . 45
4.3 Estimation summary results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.4 Residual diagnostics for the ACI model. . . . . . . . . . . . . . . . . . . . . . . . . 50
4.5 Intensity based information shares – descriptives. . . . . . . . . . . . . . . . . . . . 51
4.6 Cross sectional regression results. . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.7 Stock specific estimation results. . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.8 Stock specific intensity based information shares. . . . . . . . . . . . . . . . . . . . . 62
5.1 Reference entities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2 Effective transfer entropy estimates. . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.3 Information share estimates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.4 Effective transfer entropy for VIX and iTraxx. . . . . . . . . . . . . . . . . . . . . . 81
5.5 Johansen cointegration test statistics. . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.6 Optimal block length selection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 861 Introduction 1
1 Introduction
The question“Where is the market?” has been of interest to researchers for over a decade.
It was first addressed in the early 1990s, when a number of studies examined stocks that
were listed on several regional U.S. exchanges. The simultaneous trading of one stock on
several exchanges gave rise to competition among the different trading venues and mea-
suring the importance of each market for the price discovery process of the common stock
became the subject of research (see Hasbrouck 1995, Harris et al. 1995). In the following
years the competition intensified due to the rapidly growing number of internationally
cross-listings and the question “Where is the market?” went global. In particular for
smaller stock exchanges, the large U.S. markets posed a threat, since they might take over
thepricediscoveryprocessoftheduallylistedstocks andtherebydiminishtheimportance
of the respective home market. In recent years the focus of price discovery studies has
shiftedawayfromstockmarketsandspreadtovariousfieldsinempiricalfinance. Whether
in the case of commodity markets (see Figuerola-Ferrett and Gonzalo 2010), the treasury
market (see Mizrach and Neely 2008) or newly developed derivative markets such as the
oneforcreditdefaultswaps(seeBlanco etal. 2005), thequestionwhichmarketleads price
discovery is always one of the first to be asked.
However, the huge and diversified amount of empirical research into measuring contribu-
tions to price discovery is not mirrored by an adequate number of studies concerning the
methodological aspects. The twoprevalent measures both date back to 1995, Hasbrouck’s
(1995) information shares and the Gonzalo and Granger (1995) approach. In 2002, the
Journal of Financial Markets devoted an issueto measuringcontributionsto price discov-
ery, in which both approaches are compared and critically evaluated (Journal of Financial
Markets, 2002, Vol. 5, Issue 3). Since then only very few advances have been made with
regard to improving the methodologies used or proposing innovative approaches to mea-
sure contributions to price discovery.
This thesis presents three methods that either resolve drawbacks of the standard method-
ologies or offer