Download as pdf or txt
Download as pdf or txt
You are on page 1of 54

Modeling Dependence in Econometrics

Selected Papers of the Seventh


International Conference of the
Thailand Econometric Society Faculty
of Economics Chiang Mai University
Thailand January 8 10 2014 1st Edition
Van-Nam Huynh
Visit to download the full and correct content document:
https://textbookfull.com/product/modeling-dependence-in-econometrics-selected-pap
ers-of-the-seventh-international-conference-of-the-thailand-econometric-society-facult
y-of-economics-chiang-mai-university-thailand-january-8-10-2014-1s/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Semantic Technology 4th Joint International Conference


JIST 2014 Chiang Mai Thailand November 9 11 2014
Revised Selected Papers 1st Edition Thepchai Supnithi

https://textbookfull.com/product/semantic-technology-4th-joint-
international-conference-jist-2014-chiang-mai-thailand-
november-9-11-2014-revised-selected-papers-1st-edition-thepchai-
supnithi/

Advances in Web Based Learning ICWL 2018 17th


International Conference Chiang Mai Thailand August 22
24 2018 Proceedings Gerhard Hancke

https://textbookfull.com/product/advances-in-web-based-learning-
icwl-2018-17th-international-conference-chiang-mai-thailand-
august-22-24-2018-proceedings-gerhard-hancke/

Data Mining and Big Data 4th International Conference


DMBD 2019 Chiang Mai Thailand July 26 30 2019
Proceedings Ying Tan

https://textbookfull.com/product/data-mining-and-big-data-4th-
international-conference-dmbd-2019-chiang-mai-thailand-
july-26-30-2019-proceedings-ying-tan/

Multi disciplinary Trends in Artificial Intelligence


10th International Workshop MIWAI 2016 Chiang Mai
Thailand December 7 9 2016 Proceedings 1st Edition
Chattrakul Sombattheera
https://textbookfull.com/product/multi-disciplinary-trends-in-
artificial-intelligence-10th-international-workshop-
miwai-2016-chiang-mai-thailand-december-7-9-2016-proceedings-1st-
Operations Research Proceedings 2014 Selected Papers of
the Annual International Conference of the German
Operations Research Society GOR RWTH Aachen University
Germany September 2 5 2014 1st Edition Marco Lübbecke
https://textbookfull.com/product/operations-research-
proceedings-2014-selected-papers-of-the-annual-international-
conference-of-the-german-operations-research-society-gor-rwth-
aachen-university-germany-september-2-5-2014-1st-edition/

Formal Modeling and Analysis of Timed Systems 12th


International Conference FORMATS 2014 Florence Italy
September 8 10 2014 Proceedings 1st Edition Axel Legay

https://textbookfull.com/product/formal-modeling-and-analysis-of-
timed-systems-12th-international-conference-
formats-2014-florence-italy-september-8-10-2014-proceedings-1st-
edition-axel-legay/

MultiMedia Modeling 20th Anniversary International


Conference MMM 2014 Dublin Ireland January 6 10 2014
Proceedings Part II 1st Edition Sema Alaçam

https://textbookfull.com/product/multimedia-modeling-20th-
anniversary-international-conference-mmm-2014-dublin-ireland-
january-6-10-2014-proceedings-part-ii-1st-edition-sema-alacam/

Birds of Thailand Michael Webster

https://textbookfull.com/product/birds-of-thailand-michael-
webster/

Information and Communication Technologies in Tourism


2014 Proceedings of the International Conference in
Dublin Ireland January 21 24 2014 1st Edition Ander
Garcia
https://textbookfull.com/product/information-and-communication-
technologies-in-tourism-2014-proceedings-of-the-international-
conference-in-dublin-ireland-january-21-24-2014-1st-edition-
Advances in Intelligent Systems and Computing 251

Van-Nam Huynh
Vladik Kreinovich
Songsak Sriboonchitta Editors

Modeling Dependence
in Econometrics
Advances in Intelligent Systems and Computing

Volume 251

Series Editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: kacprzyk@ibspan.waw.pl

For further volumes:


http://www.springer.com/series/11156
About this Series

The series “Advances in Intelligent Systems and Computing” contains publications on theory,
applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually
all disciplines such as engineering, natural sciences, computer and information science, ICT,
economics, business, e-commerce, environment, healthcare, life science are covered. The list
of topics spans all the areas of modern intelligent systems and computing.
The publications within “Advances in Intelligent Systems and Computing” are primarily
textbooks and proceedings of important conferences, symposia and congresses. They cover
significant recent developments in the field, both of a foundational and applicable character.
An important characteristic feature of the series is the short publication time and world-wide
distribution. This permits a rapid and broad dissemination of research results.

Advisory Board

Chairman

Nikhil R. Pal, Indian Statistical Institute, Kolkata, India


e-mail: nikhil@isical.ac.in

Members

Emilio S. Corchado, University of Salamanca, Salamanca, Spain


e-mail: escorchado@usal.es

Hani Hagras, University of Essex, Colchester, UK


e-mail: hani@essex.ac.uk

László T. Kóczy, Széchenyi István University, Győr, Hungary


e-mail: koczy@sze.hu

Vladik Kreinovich, University of Texas at El Paso, El Paso, USA


e-mail: vladik@utep.edu

Chin-Teng Lin, National Chiao Tung University, Hsinchu, Taiwan


e-mail: ctlin@mail.nctu.edu.tw

Jie Lu, University of Technology, Sydney, Australia


e-mail: Jie.Lu@uts.edu.au

Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexico


e-mail: epmelin@hafsamx.org

Nadia Nedjah, State University of Rio de Janeiro, Rio de Janeiro, Brazil


e-mail: nadia@eng.uerj.br

Ngoc Thanh Nguyen, Wroclaw University of Technology, Wroclaw, Poland


e-mail: Ngoc-Thanh.Nguyen@pwr.edu.pl

Jun Wang, The Chinese University of Hong Kong, Shatin, Hong Kong
e-mail: jwang@mae.cuhk.edu.hk
Van-Nam Huynh · Vladik Kreinovich
Songsak Sriboonchitta
Editors

Modeling Dependence
in Econometrics

ABC
Editors
Van-Nam Huynh Songsak Sriboonchitta
Japan Advanced Institute of Science Faculty of Economics
and Technology Chiangmai University
Ishikawa Chiangmai
Japan Thailand

Vladik Kreinovich
Department of Computer Science
University of Texas at El Paso
El Paso, Texas
USA

ISSN 2194-5357 ISSN 2194-5365 (electronic)


ISBN 978-3-319-03394-5 ISBN 978-3-319-03395-2 (eBook)
DOI 10.1007/978-3-319-03395-2
Springer Cham Heidelberg New York Dordrecht London

Library of Congress Control Number: 2013954015

c Springer International Publishing Switzerland 2014


This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection
with reviews or scholarly analysis or material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of
this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publisher’s location, in its current version, and permission for use must always be obtained from Springer.
Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations
are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, 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.
While the advice and information in this book are believed to be true and accurate at the date of pub-
lication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any
errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect
to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)


Preface

The Seventh International Conference of the Thailand Econometric Society (TES


2014) is held in Chiang Mai, Thailand, during January 8th –10th , 2014, and hosted
by the School of Economics, Chiang Mai University, Thailand.
The aim of this conference is to provide an international forum for researchers
and practitioners in the areas of econometrics and quantitative analysis in economics
to stimulate new ideas, present and discuss fundamental and applied research prob-
lems, as well as to foster research collaborations. The main theme of the TES 2014
conference is: “Modeling Dependence in Econometrics”.
This volume contains papers presented at the TES 2014 conference, which were
carefully evaluated and recommended for publication by the Scientific Committee.
We appreciate the efforts of all the authors who submitted papers and regret that not
all of them can be included. The volume begins with a keynote paper by Christian
Francq and Jean-Michel Zakoı̈an, and is followed by invited and contributed papers
arranged into two parts: Fundamental Theory and Applications.
The TES 2014 conference is financially supported by the Chiang Mai School
of Economics (CMSE), Thailand. Thanks to Dean Pisit Leeahtam and CMSE for
providing crucial support throughout the organization of TES 2014. Special thanks
to Prof. Hung T. Nguyen for his valuable advice and constant support.
We sincerely wish to express our appreciation to all the members of Advi-
sory Board, Administrative Committee, Scientific Committee and Local Organiz-
ing Committee for their great help and support. We would also like to thank
Prof. Janusz Kacprzyk (Series Editor) and Dr. Thomas Ditzinger (Senior Editor,
Engineering/Applied Sciences) for their support and cooperation in this publication.
Last, but not the least, we wish to thank all the authors and participants for their
contributions and fruitful discussions that made this conference a success.

Chiang Mai, Thailand Van-Nam Huynh


January 2014 Vladik Kreinovich
Songsak Sriboonchitta
Contents

Part I: Keynote Paper

Multi-level Conditional VaR Estimation in Dynamic Models . . . . . . . . . . 3


Christian Francq, Jean-Michel Zakoı̈an
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Two-Step VaR Estimation in Volatility Models . . . . . . . . . . . . . . . . 5
2.1 Conditional VaR in a General Model . . . . . . . . . . . . . . . . 5
2.2 Asymptotic Properties of the Multi-level Two-Step
VaR Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Constructing Confidence Intervals for the VaR’s . . . . . . . 9
2.4 A Portfolio of VaR’s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.5 Choosing the Weights to Approximate DRMs . . . . . . . . . 10
3 NonGaussian QML Estimation of VaR’s . . . . . . . . . . . . . . . . . . . . . 12
3.1 Reparameterization and VaR Parameter . . . . . . . . . . . . . . 12
3.2 Asymptotic Joint Distribution of the VaR Parameter
Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Empirical Illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Part II: Fundamental Theory

The Effects of Management and Provision Accounts on Hedge Fund


Returns – Part I: The High Water Mark Scheme . . . . . . . . . . . . . . . . . . . . 23
Serge Darolles, Christian Gourieroux
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2 High-Water Mark Allocation Scheme . . . . . . . . . . . . . . . . . . . . . . . 25
2.1 Allocation between A and B Accounts . . . . . . . . . . . . . . . 25
2.2 Discussion of the High-Water Mark Scheme . . . . . . . . . . 28
2.3 The Returns and Effective Performance Fees . . . . . . . . . . 30
3 The Effects of the Scheme on i.i.d. Gaussian Portfolio
Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
VIII Contents

4 Endogeneous Portfolio Management . . . . . . . . . . . . . . . . . . . . . . . . 36


5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
The Effects of Management and Provision Accounts on Hedge Fund
Returns – Part II: The Loss Carry Forward Scheme . . . . . . . . . . . . . . . . . 47
Serge Darolles, Christian Gourieroux
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2 The Loss Carry Forward Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.1 The Basic Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.2 An Allocation Scheme with Provision Account . . . . . . . . 52
3 The Effects of the Scheme on i.i.d. Gaussian Portfolio
Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.1 The Loss Carry Forward Allocation Scheme
(Without Provision Account) . . . . . . . . . . . . . . . . . . . . . . . 54
3.2 The Allocation Scheme with Provision Account . . . . . . . 55
4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

How to Detect Linear Dependence on the Copula Level? . . . . . . . . . . . . . . 63


Vladik Kreinovich, Hung T. Nguyen, Songsak Sriboonchitta
1 Introduction: Traditional Approach to Detecting Linear
Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2 Detecting Linear Dependence Based on a Copula: Formulation
of the First Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3 Detecting Linear Dependence Based on a Copula: Main Idea
and the Resulting Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4 How to Actually Compute L− and L+ . . . . . . . . . . . . . . . . . . . . . . . 70
5 Case of Heavy-Tailed Distribution: Second Related Problem . . . . 75
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
An Innovative Financial Time Series Model: The Geometric Process
Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Jennifer S.K. Chan, Connie P.Y. Lam, S.T. Boris Choy
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
2 The GP Model and Its Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
2.1 Extension to Covariate Effect . . . . . . . . . . . . . . . . . . . . . . . 84
2.2 Extension to Threshold Time . . . . . . . . . . . . . . . . . . . . . . . 84
2.3 Extension to Threshold Outcome . . . . . . . . . . . . . . . . . . . 85
3 Methodology of Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4 Empirical Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.1 The Intra-day Range Data . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.2 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.3 Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Contents IX

4.4 Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Residual Based Cusum Test for Parameter Change in AR-GARCH
Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Sangyeol Lee, Jiyeon Lee
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
2 Residual Based Cusum Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3 Empirical Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.1 Simulation Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.2 Real Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Dependence and Association Concepts through Copulas . . . . . . . . . . . . . . 113
Zheng Wei, Tonghui Wang, Wararit Panichkitkosolkul
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
2 Basic Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3 Invariance of Affiliation of Subcopula through Bilinear
Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4 Average and Local Measures of Affiliations . . . . . . . . . . . . . . . . . . 121
5 Conditions on Affiliation in the Bivariate Skew
Normal Family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Pairs Trading via Three-Regime Threshold Autoregressive GARCH
Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Cathy W.S. Chen, Max Chen, Shu-Yu Chen
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
2.1 Threshold AR Model with GARCH Effect . . . . . . . . . . . 129
2.2 Priors and Likelihood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
2.3 Pairs Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Testing Dependencies in Term Structure of Interest Rates . . . . . . . . . . . . . 141
Kian-Guan Lim
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
2 Deriving the Spot and Forward Rates . . . . . . . . . . . . . . . . . . . . . . . . 142
3 Term Structure of Interest Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4 Tests of Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
X Contents

Joint Distributions of Random Sets and Their Relation to Copulas . . . . . 155


Bernhard Schmelzer
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
2 Random Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
3 Joint Distributions of Random Sets . . . . . . . . . . . . . . . . . . . . . . . . . . 160
4 Random Sets and Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5 Conclusion and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Vine Copulas As a Way to Describe and Analyze Multi-Variate
Dependence in Econometrics: Computational Motivation and
Comparison with Bayesian Networks and Fuzzy Approaches . . . . . . . . . . 169
Songsak Sriboonchitta, Jianxu Liu, Vladik Kreinovich, Hung T. Nguyen
1 Copulas – A Useful Tool in Econometrics: Motivations and
Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
2 From General Copulas to Vine Copulas: Motivations and
Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
3 Comparing Vine Copulas with Other Techniques for
Describing Multi-Variate Dependence . . . . . . . . . . . . . . . . . . . . . . . 178
4 How Vine Copulas Are Used in Econometrics . . . . . . . . . . . . . . . . 181
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

Part III: Applications

Extreme Value Copula Analysis of Dependences between Exchange


Rates and Exports of Thailand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Chakorn Praprom, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
2 Review Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
3 Data and Model Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
3.1 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
3.2 Bivariate Extreme Value and Generalized Pareto
Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
3.3 Extreme Value Copulas and Joint Tail Estimation . . . . . . 192
3.4 Copula Function and Extreme Value Copulas . . . . . . . . . 192
3.5 Joint Tail Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
3.6 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
4.1 Parameter Estimation of Bivariate Generalized
Pareto Distribution (BGPD) Model . . . . . . . . . . . . . . . . . . 195
4.2 Results of Parameter Estimation of Copulas and
Related Dependence Function . . . . . . . . . . . . . . . . . . . . . . 195
4.3 Results of Joint Tail Estimation . . . . . . . . . . . . . . . . . . . . . 196
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Contents XI

Analysis of Volatility of and Dependence between Exchange Rate


and Inflation Rate in Lao People’s Democratic Republic Using
Copula-Based GARCH Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Tongvang Xiongtoua, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
3 Econometric Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
3.1 Models for Marginal Distribution . . . . . . . . . . . . . . . . . . . 203
3.2 Skewed Student-t Distribution . . . . . . . . . . . . . . . . . . . . . . 203
3.3 Copula Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
3.4 Goodness-of-Fit Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
4 Descriptive Data and Empirical Results . . . . . . . . . . . . . . . . . . . . . . 207
4.1 Data Descriptions and Statistics . . . . . . . . . . . . . . . . . . . . . 207
4.2 Estimates of Marginal Distribution of Growth Rates
of Exchange Rate and Inflation Rate . . . . . . . . . . . . . . . . . 207
4.3 KS and Box-Ljung Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
4.4 Static Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
4.5 Goodness-of-Fit Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
4.6 Time-Varying Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
5 Policy Implication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Modeling Dependence of Accident-Related Outcomes Using Pair
Copula Constructions for Discrete Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Jirakom Sirisrisakulchai, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2 Statistical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.1 Negative Binomial Regression . . . . . . . . . . . . . . . . . . . . . . 218
2.2 Copula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.3 Vine Pair Copula Constructions . . . . . . . . . . . . . . . . . . . . . 219
3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
4.1 Negative Binomial Regressions . . . . . . . . . . . . . . . . . . . . . 222
4.2 Discrete Vine PCC Results . . . . . . . . . . . . . . . . . . . . . . . . 223
5 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Dependence Analysis of Exchange Rate and International Trade of
Thailand: Application of Vine Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Chakorn Praprom, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
2 Review Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
3 Data and Model Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
3.1 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
XII Contents

4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237


4.1 Specifications of C-vine and D-vine Copula Models . . . . 237
4.2 Estimation of C-vine and D-vine Copula Models . . . . . . 238
4.3 Estimation Using Time-Varying Gaussian Copula of
All Pair Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
A Vine Copula Approach for Analyzing Financial Risk and
Co-movement of the Indonesian, Philippine and Thailand Stock
Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Songsak Sriboonchitta, Jianxu Liu, Vladik Kreinovich, Hung T. Nguyen
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
2.1 A GJR Model for Marginal Distributions . . . . . . . . . . . . . 248
2.2 Vine Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
2.3 Parameter Estimation Method . . . . . . . . . . . . . . . . . . . . . . 250
3 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
4 Economic Application of Risk Measures . . . . . . . . . . . . . . . . . . . . . 254
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
Studying Volatility and Dependency of Chinese Outbound Tourism
Demand in Singapore, Malaysia, and Thailand: A Vine Copula
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Jianxu Liu, Songsak Sriboonchitta, Hung T. Nguyen, Vladik Kreinovich
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
3 Copula Based ARMA-GARCH Model . . . . . . . . . . . . . . . . . . . . . . 263
3.1 ARMA-GARCH Model for Margins . . . . . . . . . . . . . . . . 263
3.2 Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
3.3 Vines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
4.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
4.2 Estimation Results of ARMA-GARCH Model . . . . . . . . 267
4.3 Estimation Results of Vine Copulas . . . . . . . . . . . . . . . . . 268
4.4 Application of Dynamic Dependence Structure . . . . . . . . 270
5 Policy Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
Vine Copula-Cross Entropy Evaluation of Dependence Structure and
Financial Risk in Agricultural Commodity Index Returns . . . . . . . . . . . . 275
Songsak Sriboonchitta, Jianxu Liu, Aree Wiboonpongse
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Contents XIII

2.1 C-vine and D-vine Copulas . . . . . . . . . . . . . . . . . . . . . . . . 277


2.2 Minimum Cross Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . 279
2.3 ES and Optimal Portfolio . . . . . . . . . . . . . . . . . . . . . . . . . . 280
3 Data and Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
3.1 Data, KS, and LM Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
3.2 The Ordering of Vine Copulas Based Cross Entropy . . . 281
3.3 Estimation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
A Study on Whether Economic Development and Urbanization of
Areas Are Associated with Prevalence of Obesity in Chinese Adults:
Findings from 2009 China Health and Nutrition Surveys . . . . . . . . . . . . . 289
Jing Dai, Songsak Sriboonchitta, Cheng Zi, Yunjuan Yang
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
3 Data Sets Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
4.1 Dependent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
4.2 Independent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Statistical Analysis of Political Cycles in Australian Stock Market
Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
S.T. Boris Choy, Celestine M. Bond
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
2.1 Data and Variable Specification . . . . . . . . . . . . . . . . . . . . . 309
2.2 Descriptive Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
3 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
3.1 Generalised Autoregressive Heteroskedastic
(GARCH) Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
3.2 Stochastic Volatility (SV) Models . . . . . . . . . . . . . . . . . . . 313
3.3 Error Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
4 Model Implementation and Estimation . . . . . . . . . . . . . . . . . . . . . . 315
5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
5.1 GARCH Model Estimation and Comparison . . . . . . . . . . 317
5.2 SV Model Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
5.3 Parameter Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
5.4 Before and after WWII Analysis . . . . . . . . . . . . . . . . . . . . 325
6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
XIV Contents

Dependence Structure between Crude Oil, Soybeans, and Palm Oil in


ASEAN Region: Energy and Food Security Context . . . . . . . . . . . . . . . . . 329
Teera Kiatmanaroch, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
3 Data and Empirical Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
3.1 Results of C-vine Copula Analysis . . . . . . . . . . . . . . . . . . 335
4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Copula Based GARCH Dependence Model of Chinese and Korean
Tourist Arrivals to Thailand: Implications for Risk Management . . . . . . 343
Ornanong Puarattanaarunkorn, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
3.1 Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
3.2 Characteristics of Copula Families . . . . . . . . . . . . . . . . . . 349
3.3 Tail Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
3.4 Maximum Likelihood Estimation . . . . . . . . . . . . . . . . . . . 353
3.5 Selection of Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
5.1 Results of ARMA-GARCH Model for Marginal
Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
5.2 Results of Copula Estimations . . . . . . . . . . . . . . . . . . . . . . 357
6 Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
7 Conclusion and Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
Analyzing Relationship between Tourist Arrivals from China and
India to Thailand Using Copula Based GARCH
and Seasonal Pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Ornanong Puarattanaarunkorn, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
3.1 Average Seasonal Index . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
3.2 ARMA-GARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
3.3 Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
3.4 Characteristics of Copula Families . . . . . . . . . . . . . . . . . . 372
3.5 Maximum Likelihood Estimation . . . . . . . . . . . . . . . . . . . 374
3.6 Selection of Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
4 Data and Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
4.1 Seasonal Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
4.2 Copula Based GARCH . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
Contents XV

4.3 Results of ARMA-GARCH Model for Marginal


Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
4.4 Results of Copula Estimations . . . . . . . . . . . . . . . . . . . . . . 377
5 Conclusions and Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . 380
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Modeling Dependency in Tourist Arrivals to Thailand from China,
Korea, and Japan Using Vine Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
Ornanong Puarattanaarunkorn, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
2.1 ARMA-GARCH Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
2.2 Multivariate Copula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
2.3 Vine Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
2.4 Vine Copula Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
2.5 Copula Families . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
3.1 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
3.2 Marginal Distributions by ARMA-GARCH Model . . . . . 390
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
4.1 Results of C-vine and D-vine Copula Analysis . . . . . . . . 392
4.2 Time-Varying Copula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
5 Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
Relationship between Exchange Rates, Palm Oil Prices, and Crude
Oil Prices: A Vine Copula Based GARCH Approach . . . . . . . . . . . . . . . . . 399
Teera Kiatmanaroch, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
2.1 Marginal Distribution Model . . . . . . . . . . . . . . . . . . . . . . . 401
2.2 Copula Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
2.3 Vine Copula Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
2.4 Vine Copula Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
3 Data and Empirical Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
3.1 Results of C-vine Copula . . . . . . . . . . . . . . . . . . . . . . . . . . 407
4 Conclusions and Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . 409
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
An Analysis of Interdependencies among Energy, Biofuel, and
Agricultural Markets Using Vine Copula Model . . . . . . . . . . . . . . . . . . . . . 415
Phattanan Boonyanuphong, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
2 Econometrics Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
2.1 Copula Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
XVI Contents

2.2 Vine Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419


2.3 Dynamic C-vine Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
2.4 Marginal Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
3 The Data and Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
3.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
3.2 Results of Marginal Models . . . . . . . . . . . . . . . . . . . . . . . . 422
3.3 Results of Copula Models . . . . . . . . . . . . . . . . . . . . . . . . . 422
4 Applications for Portfolio Management . . . . . . . . . . . . . . . . . . . . . . 425
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
An Analysis of Volatility and Dependence between Rubber Spot and
Futures Prices Using Copula-Extreme Value Theory . . . . . . . . . . . . . . . . . 431
Phattanan Boonyanuphong, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
2 Econometrics Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
2.1 Copula Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
2.2 Marginal Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
2.3 Estimation and Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
4.1 Results for Marginal Models . . . . . . . . . . . . . . . . . . . . . . . 438
4.2 Results for Copula Models . . . . . . . . . . . . . . . . . . . . . . . . . 439
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
Effect of Markets Temperature on Stock-Price: Monte Carlo
Simulation on Spin Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
Arjaree Thongon, Songsak Sriboonchitta, Yongyut Laosiritaworn
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
2 Ising Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 446
3 Monte Carlo Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
4 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
5 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
An Analysis of Relationship between Gold Price and U.S. Dollar Index
by Using Bivariate Extreme Value Copulas . . . . . . . . . . . . . . . . . . . . . . . . . 455
Mutita Kaewkheaw, Pisit Leeahtam, Chukiat Chaiboosri
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
2 Bivariate Extreme Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
2.1 Bivariate Block Maxima . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
3 Copulas and Extreme Value Copulas . . . . . . . . . . . . . . . . . . . . . . . . 459
4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
5 Empirical Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
Contents XVII

6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
An Integration of Eco-Health One-Health Transdisciplinary
Approach and Bayesian Belief Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
Chalisa Kallayanamitra, Pisit Leeahtam, Manoj Potapohn,
Bruce A. Wilcox, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
3.1 Population and Sampling Design . . . . . . . . . . . . . . . . . . . . 465
3.2 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
3.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.1 Benefit of Integration of Eco-HealthOne-Health
Transdisciplinary Approach and Bayesian Belief
Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.2 Constraint of Bayes Risk Minimization Using
Symmetric Loss Function . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.3 Lack of Optimal Sample Size Determination . . . . . . . . . . 472
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
Factors Affecting Hospital Stay Involving Drunk Driving and
Non-Drunk Driving in Phuket, Thailand . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
Jirakom Sirisrisakulchai, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
2 Switching Regression Model for Hospital Stay . . . . . . . . . . . . . . . . 481
3 Copula Approach for Modeling Switching Regression . . . . . . . . . 482
4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
5.1 Binary Choice Equation for Alcohol Consumption . . . . . 485
5.2 Binary Outcome for Fatality or Injury . . . . . . . . . . . . . . . 486
5.3 Zero-Inflated Negative Binomial Models for the
Length of Stay in the Hospital . . . . . . . . . . . . . . . . . . . . . . 486
6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
How Macroeconomic Factors and International Prices Affect
Agriculture Prices Volatility?-Evidence from GARCH-X Model . . . . . . . 491
Gong Xue, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
2.1 China Agricultural Commodity Prices,
Macroeconomic Variables and International
Price Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
XVIII Contents

2.2 International Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494


3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
3.1 Method of GARCH and GARCH-X Models . . . . . . . . . . 495
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
4.1 Data Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
4.2 Unit Root Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
4.3 Cointegration Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
4.4 Error Correction Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
4.5 GARCH-X Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
5 Conclusions and Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . 502
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
Co-movement of Prices of Energy and Agricultural Commodities in
Biofuel Era: A Period-GARCH Copula Approach . . . . . . . . . . . . . . . . . . . 505
Gong Xue, Songsak Sriboonchitta
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505
2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
3.1 Period-GARCH Modeling for Marginal . . . . . . . . . . . . . . 508
3.2 Copula Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
4.1 Data and Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . 512
4.2 Estimation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
5 Conclusion and Policy Implication . . . . . . . . . . . . . . . . . . . . . . . . . . 517
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517
Wage Determination and Compensating Wage Differentials in the
Informal Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
Pisit Leeahtam, Supanika Leurcharusmee, Peerapat Jatukannyaprateep
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
2 Literature Reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
4 Model and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
5 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
5.1 The Sample Selection Regression . . . . . . . . . . . . . . . . . . . 529
5.2 The Wage Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Optimal Combination of Energy Sources for Electricity Generation in
Thailand with Lessons from Japan Using Maximum Entropy . . . . . . . . . 539
Tatcha Sudtasan, Komsan Suriya
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
2 Literature Reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546
Contents XIX

6 Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
Valuation of Interest Rate Derivatives under CSA Discounting . . . . . . . . 551
Amy R. Daniels, Coenraad C.A. Labuschagne,
Theresa M. Offwood-le Roux
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
2 Classical vs OIS Swap Pricing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554
2.1 Classic Approach to Price and Value Interest
Rate Swaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
2.2 CSA Approach to Pricing and Valuing Interest
Rate Swaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556
3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
Systemic Knowledge Synthesis for Product Recommendation . . . . . . . . . 561
Yoshiteru Nakamori
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
2 Theory of Knowledge Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
3 Product Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
4 Data Collection and Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
4.1 Correspondence Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 566
4.2 Fuzzy-Set Theoretical Data Processing . . . . . . . . . . . . . . . 567
5 Information Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
6 Knowledge Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 574

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575


Part I
Keynote Paper
Multi-level Conditional VaR Estimation
in Dynamic Models

Christian Francq and Jean-Michel Zakoı̈an

Abstract. We consider joint estimation of conditional Value-at-Risk (VaR) at sev-


eral levels, in the framework of general conditional heteroskedastic models. The
volatility is estimated by Quasi-Maximum Likelihood (QML) in a first step, and the
residuals are used to estimate the innovations quantiles in a second step. The joint
limiting distribution of the volatility parameter and a vector of residual quantiles
is derived. We deduce confidence intervals for general Distortion Risk Measures
(DRM) which can be approximated by a finite number of VaR’s. We also propose
an alternative approach based on non Gaussian QML which, although numerically
more cumbersome, has interest when the innovations distribution is fat tailed. An
empirical study based on stock indices illustrates the theoretical findings.

1 Introduction
Under the regulations introduced in Finance since Basel 2, bank capital is risk-
sensitive. Financial institutions are required to measure the riskiness of their assets
and, for instance, to hold more capital to compensate more risk. While the Value-
at-Risk (VaR), defined as a quantile of some loss distribution, continues to play
Christian Francq
CREST and University Lille 3 (EQUIPPE),
BP 60149, 59653 Villeneuve d’Ascq cedex, France
e-mail: christian.francq@univ-lille3.fr
Jean-Michel Zakoı̈an
EQUIPPE (University Lille 3) and CREST,
15 boulevard Gabriel Péri, 92245 Malakoff Cedex, France
e-mail: zakoian@ensae.fr
 The authors gratefully acknowledge financial support of the ANR via the Project
ECONOM&RISK (ANR 2010 blanc 1804 03). The second author gratefully thanks the
IDR ”Risques systemiques” for financial support.
 Corresponding author.

V.-N. Huynh et al. (eds.), Modeling Dependence in Econometrics, 3


Advances in Intelligent Systems and Computing 251,
DOI: 10.1007/978-3-319-03395-2_1,  c Springer International Publishing Switzerland 2014
4 C. Francq and J.-M. Zakoı̈an

a prominent role in the mainstream financial risk management, a variety of alter-


native risk measures have been introduced and studied in recent years. The Ex-
pected Shortfall (ES), and more generally the Distortion Risk Measures (DRM), are
quantile-based measures which, by comparison with the VaR at a given level, give
1
further insight on the shape of the loss distribution .
Whatever the choice of a risk measure, it depends on unknown characteristics of
the loss distribution which, for practical use, have to be estimated. In the so-called
standard approach, the quantity of interest is a parameter, defined as a characteristic
of the marginal loss distribution. In the so-called advanced approaches, the focus is
on conditional characteristics of the loss distributions, that is, characteristics which,
at the current date, take into account the available past information. The conditional
VaR, and more generally conditional risk measures, are stochastic processes which
are not directly observable, just like volatility. This complicates the statistical in-
ference of risk measures. The problem is not only to get consistent estimators of
2
conditional risks but also to evaluate the accuracy of such estimators .
Confidence intervals for conditional VaR’s were derived, in the recent economet-
ric literature, using different approaches. Chan, Deng, Peng, Xia (2007) constructed
confidence intervals under the assumption that the errors have heavy tails, using the
Extreme-Value Theory, while Spierdijk (2013) proposed a residual subsample boot-
strap approach. Francq and Zakoı̈an (2012) used a QML approach. They showed
that the problem of estimating a conditional risk measure, for instance a VaR at a
given level, in GARCH-type models reduced to the estimation of a parameter, called
risk parameter.
In the present article we extend those results to the joint estimation of several con-
ditional risks. In practice, it is often important to handle several risk levels, in order
to have a better view on the tail properties of the conditional distribution. We will
provide statistical tools for jointly estimating conditional VaR’s corresponding to
different levels, in a general GARCH-type framework which does not impose a spe-
cific form for the volatility, and for estimating the accuracy of such VaR estimates.
Our approach is aimed at, not only providing VaR estimates, but also confidence
intervals based on asymptotic results. A tractable risk measure based on a vector of
risk levels can be defined by weighting the corresponding VaR’s, that is, by defining
a portfolio of VaR’s. This approach can be connected with DRMs through an appro-
priate choice of the weights. For a given DRM, our asymptotic results allow us to
construct upper and lower bounds based on a finite number of VaR’s.
1
These measures are also advocated because, contrary to the VaR, they satisfy a set of
”coherence requirements” for a large family of distributions.
2
In July 2009, the Basel Committee issued a directive requiring that financial institutions
quantify ”model risk”. The Committee states that ”Banks must explicitly assess the need
for valuation adjustments to reflect two forms of model risk: the model risk associated
with using a possibly incorrect valuation methodology; and the risk associated with using
unobservable (and possibly incorrect) calibration parameters in the valuation model.” For
instance, an important issue in determining the reserves of a financial institution is whether
VaR estimates remain reliable in very hectic periods.
Multi-level Conditional VaR Estimation in Dynamic Models 5

This paper is organized as follows. In Section 2, we start by introducing a general


class of GARCH-type models. Then we derive the asymptotic joint distribution of
the Quasi-Maximum Likelihood Estimator (QMLE) and a vector of empirical quan-
tiles of the residuals. We deduce asymptotic confidence intervals for the VaR’s and
for VaR portfolios. Section 3 proposes another approach for conditional VaR esti-
mation based on non Gaussian QMLEs. An empirical illustration based on major
stock indices is proposed in Section 4. Section 5 concludes.

2 Two-Step VaR Estimation in Volatility Models


2.1 Conditional VaR in a General Model
Consider a GARCH-type model of the form

εt = σt ηt
(1)
σt = σ (εt−1 , εt−2 , . . . ; θ 0 )

where (ηt ) is a sequence of iid random variables, ηt is independent of {εu , u < t},
θ 0 ∈ Rd is a parameter belonging to a parameter space Θ , and σ : R∞ × Θ → (0, ∞).
The most widely used specifications of volatility belong to this class, in particular
the GARCH(p, q) model of Engle (1982) and Bollerslev (1986),

εt = σt ηt ,
(2)
σt2 = ω0 + ∑qi=1 a0i εt−i
2 + p b σ2 ,
∑ j=1 0 j t− j

where θ 0 = (ω0 , a01 , . . . , b0p ) satisfies ω0 > 0, a0i ≥ 0, b0 j ≥ 0. For this model, if
the lag polynomial β (L) = 1 − ∑ pj=1 b0 j L j has its roots outside the unit disk, we
have a representation of the form (1) given by

σt2 = β (1)−1 ω0 + ∑ γi εt−i
2
,
i=1

where β (L)−1 ∑qi=1 ai Li = ∑∞


i=1 γi L . Other classical examples of models belonging
i

to the class (1) are the EGARCH, GJR-GARCH, TGARCH, QGARCH, APARCH,
Log-GARCH, models introduced, respectively, by Nelson (1991), Glosten, Jagan-
nathan and Runkle (1993), Zakoı̈an (1994), Sentana (1995), Ding, Granger and En-
gle (1993), and for the log-GARCH, under slightly different forms, by Geweke
(1986), Pantula (1986) and Milhøj (1987). See Francq and Zakoı̈an (2010) for an
overview on GARCH models.
The conditional VaR of a process (εt ) at risk level α ∈ (0, 1), denoted by
VaRt (α ), is defined by
Pt−1 [εt < −VaRt (α )] = α ,
where Pt−1 denotes the historical distribution conditional on {εu , u < t}. When (εt )
satisfies (1), the theoretical VaR is then given by
6 C. Francq and J.-M. Zakoı̈an

VaRt (α ) = −σ (εt−1 , εt−2 , . . . ; θ 0 )ξα (3)

where ξα is the α -quantile of ηt .

Remark 1. It can be noted that in the standard GARCH(p, q) model, the conditional
VaR at level α automatically satisfies the stochastic recurrence equation
q p
VaRt2 (α ) = ω0 ξα2 + ∑ a0i ξα2 εt−i
2
+ ∑ b0 j VaRt−
2
j (α ).
i=1 j=1

Direct modelling of the conditional VaR has been proposed in several papers, for
instance Engle and Manganelli (2004), Koenker and Xiao (2006), Gouriéroux and
Jasiak (2008). A difficulty in this approach is to constrain the model so as to guar-
antee the monotonicity of the conditional VaR as a function of the risk level. Mono-
tonicity is automatically satisfied in our approach.

2.2 Asymptotic Properties of the Multi-level Two-Step VaR


Estimator
A two-step standard method for evaluating the VaR at different levels αi ∈ (0, 1), for
i = 1, . . . , m consists in estimating the volatility parameter θ 0 by Gaussian QMLE,
and then estimating the ξαi by the corresponding empirical quantiles of the residuals;
see, for instance, Chapter 2 in McNeil, Frey and Embrechts (2005). For a compar-
ison of alternative strategies based on residuals following a preliminary volatility
estimation, see Kuester, Mittnik and Paolella (2006).
Given observations ε1 , . . . , εn , and arbitrary initial values ε̃i for i ≤ 0, we define,
under assumptions given below,

σ̃t (θ ) = σ (εt−1 , εt−2 , . . . , ε1 , ε̃0 , ε̃−1 , . . . ; θ ),

which is used to approximate σt (θ ) = σ (εt−1 , εt−2 , . . . , ε1 , ε0 , ε−1 , . . . ; θ ). A QMLE


of θ 0 in Model (1) is defined as any measurable solution θ n of

θ n = arg min Q
n (θ ), (4)
θ ∈Θ

with
n
n (θ ) = n−1 ∑ t (θ ), εt2
Q t (θ ) = t2 (θ ).
+ log σ
t=1 σt2 (θ )
The following assumptions are required to derive the asymptotic properties of the
QMLE θ n .
A1: (εt ) is a strictly stationary and ergodic solution of Model (1). Moreover,
E|ε0 |s < ∞ for some s > 0.
A2: For any real sequence (xi ), the function θ → σ (x1 , x2 , . . . ; θ ) is continu-
ous. Almost surely, σt (θ ) ∈ (ω , ∞] for any θ ∈ Θ and for some ω > 0.
Multi-level Conditional VaR Estimation in Dynamic Models 7

A3: The function θ → σ (x1 , x2 , . . . ; θ ) has continuous second-order deriva-


tives, and
    
 ∂ σt (θ ) ∂ σ̃t (θ )   ∂ 2 σt (θ ) ∂ 2 σ̃t (θ ) 
sup |σt (θ ) − σ̃t (θ )| + 
 ∂θ − + −  ≤ C1 ρ t ,
θ ∈Θ ∂θ   ∂θ∂θ ∂θ∂θ 

where C1 is a random variable which is measurable with respect to {εu , u < 0}


and ρ ∈ (0, 1) is a constant.
A4 (θ ∗0 ): θ ∗0 belongs to the interior of Θ and σt (θ ∗0 )/σt (θ ) = 1 a.s. iff θ = θ ∗0 .
∗ ∂ σ (θ ∗ )
A5 (θ 0 ): There exist no non-zero x ∈ Rd such that x ∂t θ 0 = 0, a.s.
A6(θ ∗0 ): There exists a neighborhood V (θ ∗0 ) of θ ∗0 such that the following
variables have finite expectation:
     
 1 ∂ σt (θ ) 4  1 ∂ 2 σt (θ ) 2  σt (θ ∗0 ) 2δ
sup   , sup   , sup    .
∗ σt (θ ) ∂ θ 
θ ∈V (θ 0 ) ∗ σt (θ ) ∂ θ ∂ θ  
θ ∈V (θ 0 ) θ ∈V (θ ∗ ) σt (θ )
0


Note that Assumptions A2, A3, A5 and A6 can be simplified for specific forms of
σt : for instance if the model is the GARCH (p, q) Model (2), A2 reduces to standard
assumptions on the lag polynomials of the volatility and A3, A5, A6 can be directly
verified. Note also that the only moment assumption on the observed process is the
existence of a small moment in A1, which is automatically satisfied for standard
models such as the classical GARCH(p, q).
Now let the residuals of the QML estimation
εt
η̂t = , t = 1, . . . , n,
σ̃t (θ n )

and let ξn,αi denote the empirical αi -quantile of η̂1 , . . . , η̂n . Let α = (α1 , . . . , αm ) ,
ξ n,α = (ξn,α1 , . . . , ξn,αm ) and let ξ α = (ξα1 , . . . , ξαm ) denote the vector of popula-
tion quantiles.
Remark 2. The derivation of the joint asymptotic properties of sample quantiles
goes back to Cramér (1946) in the iid case. Different articles have extended these re-
sults for the marginal quantiles of stationary processes, under different dependence
assumptions. See Dominicy, Hörmann, Ogata and Veredas (2013) and the references
therein. We cannot apply their results because (η̂t ) is not a stationary process.

The next result gives the joint asymptotic distributions of (θ n , ξ n,α ). Let Dt (θ ) =


−1
σt (θ )∂ σt (θ )/∂ θ .
Theorem 1. Assume ξαi < 0, for i = 1, . . . , m, E ηt2 = 1 and κ4 := E ηt4 < ∞. Suppose
that η1 admits a density f which is continuous and strictly positive in a neighbor-
hood of ξαi , for i = 1, . . . , m. Let A1-A3 and A4(θ 0 )-A6(θ 0 ) hold. Then
√
n θ n − θ 0 L
κ4 −1 −1
4 J λ α ⊗ J −1 Ω
√ → N (0, Σ α ), Σα =  −1 ,
n(ξ n,α − ξ α ) λα ⊗Ω J ζα
8 C. Francq and J.-M. Zakoı̈an

where Ω = E(D Dt Dt ) with Dt = Dt (θ 0 ), λ α = (λα1 , . . . , λαm ) , ζ α =


Dt ), J = E(D
(ζi j )1≤i, j≤m and

κ4 − 1 p αi
λ α i = ξα i + ,
4 2 f (ξαi )
κ4 − 1 ξα i p α j ξα j p α i (αi ∧ α j )(1 − αi ∨ α j )
ζi j = ξαi ξα j + + + ,
4 2 f (ξα j ) 2 f (ξαi ) f (ξαi ) f (ξα j )
 
with pα = E η12 1{η1 <ξα } − α .
Proof. In view of Francq and Zakoı̈an (Proof of Theorem 4, 2012), we have, for
i = 1, . . . , m,

√ √ 1 1 n
n(ξαi − ξn,αi ) = ξαi Ω  n(θ n − θ 0 ) + √ ∑ (1{ηt <ξα } − αi ) + oP(1),
f (ξαi ) n t=1 i

and
√  −JJ −1 n
n(θ n − θ 0 ) = √ ∑ (1 − ηt2)D
Dt + oP(1).
2 n t=1

Hence

√  1 n 1
Covas n(θ n − θ 0 ), √ ∑ (1{ηt <ξα } − αi ) = pα J −1 Ω .
n t=1 i 2 i

It follows that, for i ≤ j,


√ √
Covas { n(ξαi − ξn,αi ), n(ξα j − ξn,α j )}
 
κ4 − 1 ξα i p α j ξα j p α i αi (1 − α j )
= ξα i ξα j + + Ω  J −1 Ω + ,
4 2 f (ξα j ) 2 f (ξαi ) f (ξαi ) f (ξα j )
√  √
Covas n(θ n − θ 0 ), n(ξαi − ξn,αi )
= λαi J −1 Ω .

We have Ω  J −1 Ω = 1 (see Remark 3.1 in Francq and Zakoı̈an, 2013) and thus we
obtain
√ √
Covas { n(ξαi − ξn,αi ), n(ξα j − ξn,α j )} = ζi j .

By the CLT for martingale differences, we get the announced result. 2


Let VaRt (α ) = (VaRt (α1 ), . . . , VaRt (αm )) , the vector of VaR’s at levels αi . We
have
VaRt (α ) = −σ (εt−1 , εt−2 , . . . ; θ 0 )ξ α . (5)
Multi-level Conditional VaR Estimation in Dynamic Models 9

A natural estimator of VaRt (α ) is thus

 t (α ) = −σ̃t (θ n )ξ α .
VaR n,

Remark 3. A classical problem, called quantile crossing, in quantile regression is


that two or more estimated conditional quantile functions can cross or overlap. This
drawback occurs because each conditional quantile function is independently es-
timated (see Koenker (2005)). It is thus worth noting that our estimation proce-
dure does not face this problem. By construction, the estimated conditional VaR are
monotonous functions of the α ’s.

Remark 4. For the standard GARCH(p, q) model, we have J −1 Ω = 2θ 0 , where


[1:q+1]
θ0 = θ0 ,
[1:q+1]
θ0 = (ω0 , a01 , . . . , a0q ) ,
0p

(see Francq and Zakoı̈an (2013)), and the asymptotic variance in Theorem 1 takes
the more explicit form

κ4 −1 −1
4 J 2λ α ⊗ θ 0
Σα =  .
2λ α ⊗ θ 0 ζα

2.3 Constructing Confidence Intervals for the VaR’s


Let Σ α denote a consistent estimator of the asymptotic variance Σ α . Such an esti-
mator can be constructed by i) replacing J by ĴJ = n−1 ∑t=1
n
Dt (θ n )D
Dt (θ n ) ; ii) using
the residuals η̂t to construct an estimator fˆ of the density function f of the innova-
tion, and to replace the theoretical moments of the process (ηt ) by their empirical
counterpart.
The delta method thus suggests a (1 − α0 )% confidence interval (CI) for the
VaRt (αi ) whose bounds are
−1
Φ1−  
1/2
α /2
−σ̃t (θ n,αi )ξn,αi ± √0 Δ α Σ α Δ α , (6)
n ii

where 
∂ σ̃t (θ n,α )
Δ α = ξ n,α , σ̃t (θ n )II m ,
∂θ 

Φα−1
0
denotes the α0 -quantile of the standard Gaussian distribution, and I m denotes
the m × m identity matrix. Note that the choice of α0 (the risk estimation level) is
independent from that of the αi ’s (the financial risk levels). Drawing such CI allows
to underline the importance of the estimation risk for VaR evaluation.
10 C. Francq and J.-M. Zakoı̈an

2.4 A Portfolio of VaR’s


Focusing only on VaR at a given level for measuring risk can be misleading since it
gives a limited view of the distribution, which may result in lack of robustness for
risk management and risk control. To circumvent this problem, several risk measures
have to be jointly considered in practice. To this aim, Distortion Risk Measures
(DRM) have been introduced in the insurance literature, in a series of papers by
Wang and coauthors [see Wang (2000) and the references therein]. A particular case
is the conditional expected shortfall (ES) which, at level α ∈ (0, 1), can be written
as 
1 α
ESt (α ) = −Et−1 [εt | εt < −VaRt (α )] = VaRt (u)du.
α 0
More general DRM take the form
 1
DRMt = VaRt (u)dG(u), (7)
0

where the distortion function, G, is a given cumulative distribution function (cdf) on


3
[0, 1] . It follows from (3) that, for Model (1),
 1
DRMt = −σ (εt−1 , εt−2 , . . . ; θ 0 ) ξu dG(u). (8)
0

In the spirit of DRM, a risque measure which can be interpreted as a portfolio of


VaR’s at different levels is defined by
m
p  VaRt (α ) = ∑ pi VaRt (αi )
i=1

where p = (p1 , . . . , pm ) with pi ≥ 0 for i = 1, . . . , m and ∑m


i=1 pi = 1. This risk mea-
sure can be interpreted as a special DRM with associated distortion function corre-
sponding to Dirac masses at the points αi . In view of (6), an asymptotic CI at level
α0 for this risk measure is
−1
Φ1−  1/2
−σ̃t (θ n,αi )pp ξ n,α ± √0
α /2
pΔ  p
 α Σ α Δ . (9)
α
n

2.5 Choosing the Weights to Approximate DRMs


An estimator of the DRM in (8) can be constructed as follows:
3
Examples of DRM are the Proportional Hazard DRM, defined with G(u) = ur , and the
Exponential DRM defined with G(u) = (1 − eru )/(1 − er ), both of them defined for r >
0. These distortion functions are concave for 0 < r < 1 and r > 0, respectively, which
corresponds to coherent risk measure in the sense of Artzner, Delbaen, Eber and Heath
(1999) [see e.g. Wirch and Hardy (1999)].
Multi-level Conditional VaR Estimation in Dynamic Models 11

n  
 t = −σ (εt−1 , εt−2 , . . . ; θ n ) ∑ G i i−1
DRM −G η̂n,i , (10)
i=1 n n

where (η̂n,n−i ) denotes the order statistics, obtained by ranking the η̂t in ascending
order: η̂n,1 < · · · < η̂n,n .
However, deriving the asymptotic distribution of this estimator, might be a
formidable task. To our knowledge such results do not exist in the literature. In
this section, we use VaR portfolios to obtain lower and upper bounds for a class of
DRM, leading to (approximate) asymptotic CI’s for such DRM.
It is not restrictive to assume α1 < α2 < . . . < αm . Suppose that the support of
the distortion cdf G is [α1 , αm ], that is
 αm
DRMt = VaRt (u)dG(u). (11)
α1

In other words, we focus on ”moderate risks”: we do not consider extreme risks,


corresponding to values of α approaching 0. An example of class of such DRM,
parameterized by the coefficient r > 0 and adapted from the so-called ”proportional
hazard” DRM, is defined by

u − α1 r
G(u) = 1u∈(α1 ,αm ) + 1u∈(αm,1) , (12)
αm − α1

where 1A denotes the indicator function of any set A.


Lower and upper bounds for the DRM in (11), can be constructed as follows.
Because u → VaRt (u) is decreasing we have, noting that G(α1 ) = 0 and G(αm ) = 1,

p L VaRt (α ) ≤ DRMt (α ) ≤ pU VaRt (α )

where

p L = (0, G(α2 ), G(α3 ) − G(α2 ), . . . , 1 − G(αm−1)) ,


pU = (G(α2 ), G(α3 ) − G(α2 ), . . . , 1 − G(αm−1 ), 0) .

It follows that a CI at significance level α0∗ ≤ α0 for this risk measure is


 −1
Φ1−  1/2
−σ̃t (θ n,αi )ppL ξ n,α − √0
α /2  p
 α Σ α Δ
p L Δ ,
α L
n

−1
Φ1−  1/2
α0 /2    
−σ̃t (θ n,αi )ppU ξ n,α + √ pU Δ α Σ α Δ α p U . (13)
n
Another random document with
no related content on Scribd:
silmä oli menestystä ehkäisemässä, missä vihamies väijyi
voimakkaat sanat kielellään tai milloin näkymätön pahan haltijan isku
oli kohtaava. Pelon alaista oli elämä: joka askeleella, joka paikassa
sai olla varuillaan, vähänkin liikkuessaan tuli varustaa itseään niin
voimakkailla taljoilla ja loitsuilla kuin kukin taisi ja pieninkin
onnettomuus todisti pahojen voimien olevan liikkeessä ja pyrki mieltä
masentamaan.
MIELIKUVITUS JA TODELLISUUS
KALEVALASSA

Esittänyt F.A. Hästesko

Tuntunee omituiselta puhua mielikuvituksesta Kalevalassa, joka


kokonaan on mielikuvituksen varaan rakennettu. On siis jotenkin
rajoitettava tätä kysymystä. Kun asetan mielikuvituksen
todellisuuden vastakohdaksi ymmärrän silloin sellaisia
mielikuvituksen luomia, jotka ovat todellisuuden ulkopuolella. Näin
jää tarkastelusta pois esim. miten kansa mielikuvituksessaan on
muodostanut sellaisia tapahtumasarjoja, jotka nyt koottuna ja
järjestettynä muodostavat yhtämittaisen Kalevala-runoelman.
Samoin se vilkas mielikuvitus, mikä on ollut tarpeen sellaisen rikkaan
kuvakielen syntymiseen, mikä tässä runoelmassa esiintyy j.n.e.

Aivan sivumennen tahdon vain mainita kansan uskonnolliset


käsitykset, jotka tosin mielikuvituksen avulla, mutta kuitenkin todella
vallinneen uskon pohjalla ovat syntyneet. Vain mielikuvituksessaan
he tietysti näkivät ne monet haltijat, jotka heitä auttoivat tai
vahingoittivat jokapäiväisessä elämässä. Miten läheiseksi
jumaluusolentojen ja ihmisten väli oli kuviteltu, näkyy useista kohdin
Kalevalaa. Kun Lemminkäisen Pohjolan matkallaan tulee mennä
tulisen kuopan yli, joka oli täynnä kuumia kiviä, niin hän rukoilee
Ukkoa satamaan lunta ja Ukko sataakin sitä sauvan varren
korkeudelta. Ainon äiti taasen lohduttaakseen itkevää tytärtänsä
tarjoaa hänelle ihanaa pukua, jonka hän itse ennen nuorena tyttönä
oli saanut Kuuttarelta ja joka sitäpaitsi oli Kuuttaren oma kutoma.
Jopa niin läheiseksi on tuo suhde ajateltu, että Päivä, Kuu ja Tähti
käyvät saaren kuulua Kyllikkiä kosimassa pojillensa, vaikka Kyllikki
ei heille antanutkaan suostumustaan. Saman Kyllikin sitten
Lemminkäinen sai puolisokseen. — Mielikuvituksessaan kansa loi
itselleen kuvan Tuonelasta, asuinpaikasta kuoleman jälkeen.
Tuonelan musta kuohukoskinen joki ympäröi kuolleitten valtakuntaa.
Tuonen tumma tytti saattaa tulijan joen yli. Elämä muuten siellä on
maanpäällisen elämän kaltaista, siellä juodaan olutta, pestään
vaatteita Tuonelan joessa, kuten ennen elävien mailla. Huonompaa,
kamalampaa se kuitenkin on. Syyllisten vuoteet siellä ovat palavista
kivistä, peitteet kyillä ja käärmeillä pujotettuja.

Voimakkaan mielikuvituksen avulla on kansamme nähnyt luonnon


niin elävänä edessään, kuin se sille on ollut. Eläimet, kasvit ja
kuolleet esineetkin saivat sen ajatuksissa hengen ja näyttäytyivät
sille toimivina, puhuvina ja tuntevina olentoina. Edellä olleessa
vihkosessa on siitä runsaasti esimerkkejä, joten tämäkin kohta tässä
jääköön syrjään.

Läheisesti luonnon elolliseksi ajattelemiseen liittyy se mielikuvitus,


mikä on luonut sellaiset loitsut, etenkin syntyrunot, ja loitsuille
omistetun voiman, mistä edellä on jo ollut puhetta. Tosin tämäkin
kaikki on syntynyt varman uskon kannattamana, mutta niissä on
kuitenkin mielikuvituksellakin niin suuri osa, että otan tähän vielä
muutaman esimerkin.
Syntyrunoista on suurimpia esim. raudan synnyn kuvaus: Ukko
ilmoinen Jumala synnytti kämmeniään vastakkain hieroen kolme
Luonnotarta. Ne kävivät aikojen alussa pilven äärtä myöten ja
pudottivat maahan maitoansa. Siitä ol raudan alku. Rauta tahtoi
tavata veljeänsä tulta ja meni sen luo, mutta tuli oli polttaa raudan
kuoliaaksi, niin että se töin tuskin pääsi pakenemaan suohon. Siellä
se piili vuoden, kaksi, kolme. Susi juoksi suota myöten, painoi
maahan syvän jälen ja siihen kohosi rautaruostetta. Seppo Ilmarinen
syntyy tällä välin, sattuu suden jälille, näkee raudan kurjan tilan ja
arvelee viedä sen tulen yhteyteen. Kuullessaan tulta mainittavan
rauta pelästyy, mutta Ilmarinen lohduttaa sitä sanoen sen siellä vain
kauniiksi kasvavan. Seppä lietsoo rautaa kuumassa ahjossa. Tukala
on raudan asema ja se pyrkii pois sieltä. Mutta Ilmarinen aavistaa jo
sen pahat tavat ja kysyy, kasvaisiko se ehkä kauheaksi, kovin
raivoksi rupeaisi, jos pääsisi tulesta pois. Siinä silloin vannoo rauta
raukka vaikean valansa ahjolla ja alaisimella ei koskaan syövänsä
omaa sukuaan, ihmistä:

onpa puuta purrakseni, kiven syäntä syöäkseni, etten veistä


veikkoani, lastua emoni lasta.

Siitä seppo Ilmarinen takoo raudan keihäiksi, kirveiksi ja


kaikenlaisiksi kaluiksi. Mutta terästämistä se vielä kaipaa. Mehiläistä
hän pyytää tuomaan karkaisuaineita. Mutta mehiläisen niitä etsiessä
herhiläinen hiiden lintu äkkiä tuokin käärmehen kähyjä, madon
mustia mujuja. Ilmarinen pettyy, luulee niitä Mehiläisen mesiksi ja
kastaa niihin raudan. Siitäpä rauta raivoksi rupesi:

petti vaivainen valansa, veisti raukka veljeänsä, söi kuin


koira kunniansa, sukuansa suin piteli.
Vetisillä mailla tavattava raudan ruoste, raudan alku ja sen
taipumus tehdä ihmiselle vahinkoa on siinä mielikuvituksen avulla
koetettu selittää ja sovittaa yhteen.

Yhtä suuria ellei suurempiakin hyppäyksiä tekee kansan


mielikuvitus esittäessään, mitä kaikkia asioita loitsun avulla oli
mahdollista toimittaa. Olkoon siitäkin vielä esimerkkinä Väinämöisen
ja Joukahaisen kilpalaulanta. Kun Väinämöinen suuttuneena
Joukahaiseen alkoi loitsia, niin koko luontokin hänen edessään
vapisi:

järvet läikkyi, maa järisi, vuoret vaskiset vapisi, paaet


vahvat paukahteli, kalliot kaheksi lenti, kivet rannoilla rakoili.

Kurja oli Joukahaisenkin asema: pajupehkon Väinämöinen lauloi


hänen länkiinsä kasvamaan, reen kultalaidan lauloi lampiin haoiksi,
helmiletkun ruoskan ruo'oksi meren rannalle, laukkipään hevosen
kosken rannalle kiviksi. Miekan kultakahvan hän lauloi salamoiksi
taivaalle, jousen kaariksi vesien päälle, nuolet haukkoina ilmassa
kiitämään, koiran maahan maakiviksi. Lakki miehen päästä kohosi
pilven kokaksi, viitta taivaalle hattaroiksi, vyö linnunrata-tähdistöksi,
kintaat painuivat järveen lumpeiksi. Itse Joukahaisen hän lauloi
suohon yhä alemmas ja alemmas, niin että virta jo jalkoja veti,
hiekka silmiä hioi..

Esimerkkinä taikojen kuvitellusta voimasta olkoon seuraava. Kun


Kalevalan miehet Pohjolassa ilmoittivat Sammon jaolle saapuneensa
eivätkä pohjolaiset siihen suostuneet, täytyi Väinämöisen tarttua
taikatoimeen. Nukutettuaan ensin soitollansa Pohjolan kansan
uneen, hän unineulalla siteli ja univoiteella siveli nukkuvien silmät,
niin että he painuivat pitkään sikeään uneen. Sillä aikaa oli sammon
ryöstäjillä hyvä tilaisuus mielin määrin mellastaa kivimäessä ja
korjata Sampo veneeseensä.

Mutta käykäämme vihdoin tarkastamaan sellaisia kohtia, missä


kansan mielikuvitus ilman minkäänlaista luulottelunkaan tukea
vapaana ja laajana lainehtii. Mielikuvituksen esteetön liitely tuntuu
laulajasta miellyttävältä ja kevyeltä, ei ihme että silloin joskus
säkeitten läpi pilkistää partasuun laulajan hymyilevä, leikillinen mieli.
Laulaja tuntee itsekin laulavansa satua, jolle ei mikään taikausko
anna uskottavuuden leimaa.

Ensinnä silloin johtuu mieleen Väinämöisen Vipusessa käynti.


Väinämöinenhän oli taivaankaarella istuvalle Pohjan immelle
luvannut veistää veneen värttinän palasista. Laulamalla hän sen laati
ja vesille sen tuli mennä käsillä siihen koskematta. Mutta kesken
veneen tekoa häneltä puuttui kolme sanaa; ilman niitä vene ei
valmistunut. Turhaan Väinämöinen etsi sanoja pääskysen päälaelta,
kesäpeuran kielen alta, suusta valkean oravan. Turhaan hän kävi
Tuonelassa niitä hakemassa, sieltä ei niitä annettu. Vihdoin hänen
täytyi turvautua vanhaan kuuluisaan loitsijaan Antero Vipuseen, joka
kuitenkin kauvan oli kuolleena maassa maannut ja vahva metsä jo
peitti hänen hautansa:

haapa kasvoi hartioilla, koivu kulmilla yleni, leppä leukaluun


nenässä, pajupehko parran päällä, otsalla oravikuusi,
havuhonka hampahilla.

Mutta Väinämöinen ei sittenkään epäillyt mennä hänen luokseen.


Hän kävi ensin seppo Ilmarisella teettämässä rautaisen korennon,
johon sisälle pantiin terästä ja päälle vedettiin pehmeämpi rauta.
Miekallaan hän kaatoi puut Vipusen haudalta, syöksi rautaisen
korennon hänen suuhunsa. Siitä virsikäs Vipunen heti heräsi unesta
ja puraisi korentoa hampaillaan, mutta ei tiennyt terästä purra sen
sydämestä. Silloin Vipunen

avoi suunsa suuremmaksi, nieli miehen miekkoinensa,


kulahutti kulkkuhunsa tuon on vanhan Väinämöisen.

Vaikeaksi alkoi kuitenkin Vipusen olo käydä: hän muisteli


eläissään syöneensä uuhta ja vuohta, lehmää ja sikaa, vaan ei vielä
tällaista palaa. Pelästyi myös Väinämöinen ja arveli jo tuhonsa
tulleen. Mutta kekseliäisyyttään hän ei menettänyt:

veitsi on vyöllä Väinämöisen, pää visainen veitsessänsä,


tuosta hän teki venosen, — soutelevi, luitelevi suolen päästä
suolen päähän, souteli joka sopukan, joka supun suikerteli.

Ei vielä käynyt Vipunen laulamaan. Silloin Väinämöinen

löihen itsensä sepoksi —


painoi paitansa pajaksi,
hiat paian palkehiksi,
turkkinsa tuhottimeksi,
housut hormiksi rakensi,
sukat hormin suulliseksi,
polvensa alaisimeksi,
vasaraksi kyynäspäänsä.
Siinä hän takoi yön lepäämättä, päivän pysähtymättä. Vipunen jo
hätääntyi. »Mi sinä lienet miehiäsi?» hän kysyi. »Olen jo syönyt sata
urosta, tuhonnut tuhannen miestä, mutta en ole vielä moista syönyt:

syet suuhuni tulevat, kekälehet kielelleni, rauan kuonat


kulkkuhuni.»
Ja nyt alkoi hän loitsia pois tuota pahaa. Luki hän siinä tautien
synnyn tiedustelut, luovutussanat, apusanat, pelotussanat. Mutta
Väinämöinen oli olevinaan vain hyvin rauhallinen, väitti itsellään
olevan hyvän olon ja ruokaakin kylliksi, vieläpä uhkasi lujemmin
lyödä moukarillaan, ellei saisi kuulla sanoja, oikeita syntysanoja, sillä
»ei sanat salahan joua, eikä luottehet lovehen».

Silloin virsikäs Vipunen, — jonk oli suussa suuri tieto, mahti


ponneton povessa, aukaisi sanaisen arkun, virsilippahan
levitti, lauloaksensa hyviä, — noita syntyjä syviä, ajan
alkuluottehia, joit ei laula kaikki lapset ymmärrä yhet urohot.

Ja hän »lauloi synnyt syitä myöten, kuinka luojansa luvalla


itsestänsä ilma syntyi, ilmasta vesi erosi, veestä manner maatelihe,
manterelle kasvut kaikki». Vielä hän »lauloi kuun kuvoannasta,
auringon asetannasta, ilman pielten pistännästä, taivosen
tähytännästä.» Voimakasta oli hänen laulunsa, ei oltu sellaista ennen
kuultu. Pääksytysten hän päivät lauloi, yhytysten yöt saneli ja »suu
se syyteli sanoja, kieli laski lausehia kuin on sälkö sääriänsä, ratsu
jalkoja jaloja». — Kun Väinämöinen oli kylliksensä saanut sanoja
kuulla, palasi hän veneellensä ja voitonvarmana ilmaisi seppo
Ilmariselle hyvin suoritetun matkansa tulokset: »Sain sanat salasta
ilmi, julki luottehet lovesta.»

Hupia vailla ei ole sekään tapaus, mikä samporetkeläisille sattuu.


Laskea karittelevat Kalevalan miehet purrellaan Pohjolaa kohti,
kulkevat onnellisesti jo monien koskien läpi, monien karisalmien
kautta, niin keskellä merta, väljillä vesillä, pursi seisahtuu
juoksemasta Mikä siellä esteenä?

Se on Seppo Ilmarinen, toinen lieto Lemminkäinen, pistivät


melan merehen, lastun kuusen lainehesen, päästeä
nytystelevät tuota purtta puutoksesta.

Mutta ei pääse vene irti. Väinämöinen kehoittaa Lemminkäistä


tarkastamaan, mihin vene on tarttunut. Lemminkäinen katsoo
veneen alle:

ei ole veno kivellä, ei kivellä, ei haolla, vene on hauin


hartioilla, veen koiran konkkaluilla.

Lemminkäinen tempaa miekkansa ja iskee veneen alle, mutta


silloin »itse vierähti vetehen, kourin aaltoihin kohahti». Ilmarinen
vetää Lemminkäisen tukasta takaisin veneeseen ja pilkkaa hänen
onnettomuuttaan, mutta itse iskiessään kalaa hänen miekkansa
murenee kappaleiksi ja vene yhä on paikoillaan. Väinämöinen
vihdoin melallaan lyö haukea; sen pää saadaan veneeseen, mutta
pursto putoaa pohjaan. Hauin leukaluusta Väinämöinen saa itselleen
kanteleen.

Lisättäköön vielä Ilmarisen Tuonelan hauin pyyntö. Pohjolan


tyttären saadakseen täytyy Ilmarisen Tuonelan joesta pyytää hauki.
Voidakseen työnsä suorittaa hän morsiamensa neuvosta takoo
suuren kokon, »vaakalinnun», jolle »kourat rauasta kuvasi,
teräksestä temmottimet, siiviksi venehen vieret». Itse hän nousee
kokkonsa selkään ja neuvoo sitä menemään Tuonelan joelle. Kun
Ilmarinen haroo Tuonelan jokea, hyökkää vetehinen hänen
kimppuunsa, mutta kokko pelastaa hänet painamalla vetehisen
takaisin jokeen. Ja siinäpä »jo tulevi tuonen hauki, veen koira
vengottavi», eikä olekaan se mitätön elävä: hampahat ovat haravan
varren pituiset, kita kuin kolme koskea, selkä kuin seitsemän
venettä. Se tahtoo hyökätä Ilmarisen kimppuun, mutta samalla
huomaa kokkokin sen ja iskee kyntensä siihen vasten suomuja.
Syntyy ankara kaksintaistelu. Hauki painautuu veden alle vetäen
kokon mukanaan, mutta silloin kokko ponnistaa voimiaan, kohoaa
ylös, niin että mustat mudat veden päälle nousee, iskee uudelleen
toisen kyntensä haukeen toisen rannan kallioon. Kilpistyy kynsi
kalliosta, uudelleen painuu hauki pohjaan ammottavine arpineen.
Vielä kolmannen kerran kokko ponnistaa, saa hauin kynsiinsä ja
nostaa kuin nostaakin sen »päälle selvien vesien». Niin lähti kalasta
suomuksia ja kokosta höyheniä, ettei vettä vedeksi eikä ilmaa
ilmaksi olisi tuntenut. »Omenatammen» oksalle kokko kaloineen
istahtaa ja vaivansa palkaksi maistelee saalistaan. Mutta kun
Ilmarinen kokkoa siitä moittii, niin se sydäntyy, kohoaa ilmaan ja
lentää pilvenreunalle, niin että

pilvet liikkui, taivot naukui, ilman kannet kallistihe, katkesi


Ukolta kaari, kuulta sarviset sakarat.

Pyydän enää viitata leikillisen ja vapaan mielikuvituksen sellaisiin


luomiin kuin esim. Ilmarisen yritykseen takoa itselleen kullasta
morsianta (37 runo) ja Väinämöisen soittoon, jolla hän koko luonnon
lumoaa (41 runo).

Jo tässä esityksessä ja siellä täällä muissa tämän sarjan


vihkosissa on tullut mainituksi esimerkkejä siitä mielikuvituksesta,
joka muuttelee esineitä ja asioita todellisuutta suuremmiksi tai
pienemmiksi. Sellainen oli äsken mainittu Tuonelan hauki ja sitä
tavoitteleva kotka, sellainen se iso tammi, joka kasvoi taivaalle, esti
pilvet juoksemasta, auringon paistamasta ja niin synnytti ainaisen
yön maailmaan. Sellainen oli se suuri härkä, joka Pohjolan häihin
tapettiin, sen kun

Hämehessä häntä häilyi, pää keikkui Kemijoella, sata syltä


sarvet pitkät, puolta toista turpa paksu, viikon kärppä
kääntelihe yhen kytkyen sijalla, päivän lenti pääskyläinen
härän sarvien väliä, — kuun juoksi kesäorava häpähältä
hännän päähän. —

Niinpä runokin siitä sanoo: »ei ollut suuri, eikä pieni, olihan oikea
vasikka». — Hirveä oli se aitakin, joka ympäröi Pohjolan pihaa, kun
Lemminkäinen sinne matkasi häitten viettoon:

aita oli rautainen rakettu, teräksinen tarha tehty, sata syltä


maa emähän, tuhat syltä taivosehen, keihäillä seivästetty,
maan mauilla ajastettu, alahatse kyy matavi, alemmaksi aita
pantu, ylähätse lintu lensi, ylemmäksi aita pantu.

Toiset esineet taasen ovat mitättömän pieniä kuten se pieni, mutta


sitten suureksi jättiläiseksi kasvava mies, joka kaatoi ison tammen.
Samanlainen oli se mies, joka Pohjolan häihin tuon hirveän sonnin
tappoi, hänkin näet »alle maljan maata mahtui, alle seulan
seisomahan.»

Muutakin liioittelua tavataan Kalevalan yksityiskohdissa hyvin


runsaasti. Sellaista on täynnänsä esim. kuvaus Lemminkäisen hirven
hiihdännästä. Jo sukset, joilla hän hirveä ajoi takaa, olivat
erinomaiset, niitä kun Lyylikki lylyjen seppä vuoden ajan valmisteli:

sykysyn lylyä laati, talven kalhua kaverti, päivän vuoli


sauvan vartta, toisen sompoa sovitti.

Mutta silloin niistä tulikin niin arvokkaat, että

saukon maksoi sauvan varsi, sompa ruskean reposen.

Kun Lemminkäinen sitten näillä suksillaan ponnisteli hirvensä


jälessä, niin
tuli suihki suksiloista, savu sauvojen nenistä.

Ja jo lähtiessään kun hän »ensi kerran potkaisihe», niin pääsi jo


»silmän siintämättömähän»; kun »kerran toisen kuopaisihe», oli jo
korvan kuulemattomissa. — Toisen kerran tuli Lemminkäisen pyytää
hiiden ruuna, jonka »tukka tulta tuiski, harja suihkivi savua». Mutta
senkin hän saavuttaa, kun Ukko sataa sen päälle rakeita, jotka olivat

pienemmät hevosen päätä, päätä ihmisen isommat.

Ja Pohjolan häät sitten, joihin tuo suunnaton härkä teurastettiin?


Jos oli syötävätä yllin kyllin, niin olipa juotavatakin, koska olutta
tehtäessä

kuut kiviä kuumettihin, kesät vettä keitettihin, salot puita


poltettihin, kaivot vettä kannettihin.

Olipa sulhasenkin kotona syötävätä niin vahvasti, että


Väinämöisen täytyi mainita, miten heiltä syödessä

terät vieri veitsistänsä, päät putoili puukoistansa.

Ja se tupakin, missä tällaisia häitä vietettiin, oli sellainen, että

kukko kun laessa lauloi ei sen ääni maahan kuulu, penin


haukunta perässä ei kuulu ovehen asti.

Niin laaja oli morsiamen sukukin, että jos

kappa ois kylveä papuja jyvä kullenki tulisi, kappa panna


pellavia kuitu kullenki tulisi.

Ja komea vihdoin oli morsiamen auttajakin (saajanainen), niin että


Väinämöinen
luuli päivän paistavaksi,
kun sen paistoi paian kaulus,
luuli laivan läikkyväksi,
kun sen läikkyi lakki päässä.

Tällainen liioittelu on mielestäni hiukan erilaista kuin edellä esitetyt


tapaukset. Nämä ovat näet lähellä sellaista vilkkaan ja voimakkaan
mielikuvituksen luomaa puheen höystettä, jossa puhuja ei itsekään
huomaa mitään liikaa sanoneensa ja jota toinen samanlainen
kuuntelija ei myöskään aivan sananmukaisesti ymmärrä. Ken
itäosissa maatamme liikkunut ja siellä kansan kesken puhellut ei olisi
tavannut vilkasta kertojaa, joka häntä tällaisilla mahdottomuuksilla
olisi huvittanut.

*****

Mielihyvällä käännymme tarkastamaan Kalevalan todellisuutta


esittäviin kohtiin. Tässäkin on rajoitettava alaa: ei voi ottaa puheeksi
todellisuutta esim. eri henkilöitten luonteen kehityksessä tai
tapausten juoksussa, vaan ainoastaan katsella, minkä verran
Kalevalan laulajat ovat kyenneet tekemään oikeita havaintoja
ympäriltään ja niitä todenmukaisesti esittämään. Tapausten ja
luonteitten kehityksen todenmukaisuutta täytyy edellyttää itsestään
olevan olemassa, sillä vain sen mukaan ne sellaisina voittavat
kiintymystämme puoleensa.

On itsestään ymmärrettävää, että läheinen seurustelu luonnon


kanssa on kansan johtanut tekemään hyvin tarkkoja havaintoja
luonnonelämästä, vaikka perin usein mielikuvituskin on kuvauksissa
todellisuutta paljon muutellut. Jo toisessa runossa kohtaa lukijaa
tarkka ja yleensä todenmukainen esitys eri puitten kasvupaikasta,
johon ne Sampsa Pellervoinen alkujaan muka kylvikin:
mäet kylvi männiköiksi, kummut kylvi kuusikoiksi, kankahat
kanervikoiksi, notkot nuoriksi vesoiksi, noromaille koivut kylvi,
lepät maille leyhkeille, tuomet kylvi tuorehille, raiat maille
raikkahille, — pajut maille paisuville, katajat karuille maille. —

Väinämöisen kylvämä ohra taasen kasvoi kuten hyvä ohra ainakin:

tähkät kuuella taholla, korret kolmisolmuisena.

Jokainen hevosmies tietää niinikään, että hevonen usein tuntee


edessä olevan vaaran silloin kun ajajalla ei siitä vielä ole
aavistustakaan. Niinpä Lemminkäisen Pohjolaan ajaessa tulisen
kuopan lähestyessä

jo oronen ouostuvi, hevonen hörähtelevi.

Erittäin tarkka ja sattuvapiirteinen on mielestäni Pohjolaisten


vaarinotto koiran haukunnasta, jolla se ilmoittaa vieraitten
saapumisen. Jo ennenkuin ketään näkyykään tai ihmisen korva
mitään kuuleekaan, vainuaa vanha tarkka koira jotain outoa olevan.
Silloin se vielä rauhallisena harvakseen haukkuu, istuu maassa
nojaten etujalkoihinsa, häntä rentona alas laskeutuneena ja kuono
suunnattuna sinnepäin, mistä se pahaa aavistelee:

jopa haukkui hallikoira — pimeässä Pohjolassa; — hiisti


ensin hiljemmältä, harviammalta murahti perän lyöen
pientarehan, hännän maahan torkutellen.

Yhä ja yhä se haukkuu ja jo on Pohjolan isäntä vakuutettu, että

ei halli valetta hauku, ikipuol' ei ilman virka ei se honkihin


horise.
Hän menee ulos ottamaan asiasta selvää ja

katsoi koiran suuta myöten, nenävartta valvatteli.

Silloin hän huomaa, että on todellakin vieraita tulossa: kirjokorjin


siellä ajetaan ja toiset laivoin suurin laskevat. Vieraitten tulo harvaan
asuttuun sydänmaahan oli harvinainen tapaus, eikä silloin tiennyt
olivatko ystäviä vaiko vihollisia tulijat. Arvalla usein tiedusteltiin, ketä
olivat ja tällä kertaa sattuivat olemaan suuret sulhot, Väinämöinen ja
Ilmarinen Pohjolan neidon pyynnöllä.

Sellaisilla sydänmailla matkoja tehdessä ei kulkijalla usein ole


tasaiseksi tallattua tietä astuttavanaan, ei edes aina kapeaa,
kiemurtelevaa polkuakaan, joka määräpaikkaan erehdyttämättä
johtaisi, vaan hän saa vain luonnonmerkeistä suuntaa määräten
samota korpien läpi. Tällaisia tien opastuksia on kansamme tottunut
antamaan ja seuraamaan. Miten tutulta monesta tuntuukaan
siniviitan Viidan eukon neuvo Kullervolle tämän etsiessä
vanhempiaan:

astut päivän, tuosta toisen, — kohti luotehesen vaara


vastahan kulevi, sie astut alatse vaaran käy vaaran vasenta
puolta; tuostapa joki tulevi oikealle puolellesi, käy sitä joen
sivua kolmen kosken kuohimitse, tulet niemen tutkamehen,
pääyt päähän pitkän niemen, tupa on niemen tutkamessa.

Metsää kiireellä kahlovan vanhan mummon kuvaus on niinikään


elävä ja todellinen. Kun Lemminkäisen äiti henkensä edestä riensi
vaaraan joutunutta poikaansa auttamaan, niin hän

kourin helmansa kokosi,


käsivarsin vaattehensa, —
mäet mätkyi mennessänsä,
norot nousi, vaarat vaipui.

Elävä kuvaus kiireisestä matkasta on Kalevalan laulajalla ollut


tilaisuus antaa kertoessaan Sammon ryöstäjien pakoyrityksestä
pohjolaisten purren edestä. Ajatelkaamme suurehkoa, mutta
kevytkulkuista venettä, jota voimiensa takaa »souti seppo Ilmarinen,
souti lieto Lemminkäinen, souti kansa kaikenlainen». Todellakin
silloin

lyllyivät melat lylyiset, hangat pinkki pihlajaiset, vene


honkainen vapisi; nenä hyrski hylkehenä, perä koskena
kohisi, vesi kiehui kelloloissa, vaahti palloissa pakeni.

Mutta katsokaamme kotoista elämää. Eikö ole ilmielävänä


edessämme kesäinen kuva talonpoikaiskodista, jos poikkeamme
hakemaan Lemminkäisen kera Tieraa sotaretkelle:

iso päätyi ikkunalla keihäsvartta vuolemassa, emo aitan


kynnyksellä kirnua kolistamassa, veljekset veräjän suussa
laitioita laatimassa, sisarekset sillan päässä vaippoja
vanuttamassa.

Tai jos näemme, miten Aino-neiti talon tyttären tavoin menee


kesäisenä kylpyiltana lehdosta taittamaan kylpyvihtoja kotiväelleen ja

taittoi vastan taatollensa, toisen taittoi maammollensa,


kokoeli kolmannenki verevälle veiollensa.

Ehkäpä tahtoisimme kuulla, miten Kalevala kuvaa kylpemistä ja


sen vaikutusta. Käymme silloin Ilmarisen luo, joka sulhasmiehenä
valmistautuu Pohjolaan lähtemään. Matkalle aikovan varustuksiin
kuului kylpeminen ja sitä enemmän sulhasmiehelle ja sepälle:

siitä seppo Ilmarinen sekä kylpi kylläksensä, valelihe


valkeaksi, pesi silmät sirkeäksi, silmäkulmat kukkeaksi,
kaulansa kananmuniksi, koko varren valkeaksi, tuli saunasta
tupahan, — kasvot vallan kaunihina, poskipäät punertavina.

Puhtaalle ruumiille on nyt mieluista vetää valkea palttinapaita ja


kaatiot kapoiset, sukat, kengät kelvolliset, haljakka ja sarkakauhtana,
vyö ympäri ruumiin, kintaat kirjatut käteen, pystyinen kypärä päähän.
Soma on sitten istua rekeen, missä alla on karhun talja, päällä
turskan talja, edessä virkku hevonen, kulkuset ja kellot, toivossa,
»että kaunot katsahtaisi, impyet ihasteleisi».

Tai mitä arvellaan niistä kokeneista ja oikea-ajatuksisista


neuvoista, joita morsiamelle annetaan hänen kotoa lähtiessään: kun
kostuttelet lattiata sitä laastaksesi, niin nosta lapsi lattialta lautsaselle
ja viihdykkeeksi

anna leipeä kätehen, vuole voita leivän päälle, kun ei leipeä


talossa, anna lastunen kätehen.

Kenties tulee vieras, silloinpa

syötä vierasta sanoilla kunnes keitto kerkiävi.

Ja mikä tarkka mielentilojen ilmaisumuotojen tunteminen näkyy


neuvosta: kun tuot puita tupaan, niin laskiessasi niitä sylistäsi

heitä halko hiljallensa kovasti kolajamatta taikka appi


arveleisi, anoppi ajatteleisi vihoissasi viskelevän, kiukussa
kolistelevan.
Jännittävää odotusta, mikä jo esiintymisessä näkyy, kuvataan
Ilmarisen kotona, kun sinne poikaa odotetaan morsiamineen
saapuvaksi: vanhojen silmät vuotavat vettä pitkästä tuijottamisesta,
nuorten jalat väsyvät ulkona odotellessa ja lasten jalkoja palelee, kun
he vasta sulaneilla seinuksilla seisovat:

silmät vanhoilta valuvi ikkunoissa istuessa, polvet nuorilta


nojuvi veräjillä vuottaessa, lasten jalkoja paleli seinuksilla
seisoessa.

Kansansa suuren heikkouden rakentaa riitaa naapurinsa kanssa


pienistä asioista osoittaa laulaja tuntevansa, kun hän kertoo, mistä
Kalervon ja Untamon veljekset joutuivat sotaan keskenään:

Untamoinen verkot laski Kalervon kalavetehen, Kalervoinen


verkot katsoi, kalat konttihin kokosi; — Kalervoinen kauran
kylvi Untamon tuvan ta'aksi, Untamolan uljas uuhi söi
Kalervon kaurakylvön, Kalervoisen kärtsä koira repi uuhen
Untamolta.

Vaikkakin todellisuutta liioitteleva tai muuten muutteleva


mielikuvitus Kalevalassa onkin hyvin voimakas, ei sieltä kuitenkaan
puutu tarkkaa silmää tekemään havaintoja ulkoisesta elämästä ja
hienoa aistia erittelemään erilaisia sieluntilojakin, ei myöskään kykyä
niitten esittämiseen.
IHMINEN JA LUONTO KALEVALASSA

Esittänyt F.A. Hästesko

Elias Lönnrot sanoo vertaillessaan kansanrunon ja taiderunon


laatua keskenään: »Kummallakin, niin kansan, kun oppineitten
runoilla on oma arvonsa ja etuisuutensa, sitä emme tahdo, emmekä
taida vastustella. Mutta toinen toisensa rinnalla niitä pitäen näemmä
edellisissä luonnon, jälkimäisissä moninaisuuden voittavan.
Kansanrunoja siitä syystä ei juuri saatakaan tehdyiksi sanoa. Niitä ei
tehdä, vaan ne tekeytyvät itsestänsä, syntyvät, kasvavat ja
muodostuvat semmoisiksi ilman erityisettä tekijän huoletta.
Oppineitten runot siinä kohta erotaksen kansanrunoista, etteivät ole
ajatuksesta syntyneitä, vaan ajattelemalla tehtyjä. Tekijää ei pakoteta
sydämensä kyllyydestä runoilemaan, vaan runoilee omalta
päättämältänsä. Siinä on muuttunut työksi, mikä kansanrunoissa oli
ilmauma; elävän käen ääni metsässä kuvakäen kukunnaksi
seinäkelloissa; luonnon puro kaivetuksi vesiojaksi; luontainen metsä
istutetuksi puistoksi. Oppineitten runoelmissa pilvi itsestänsä ei ala
sataa, vaan ensinnä tehdään pilvi, jonka sitte annetaan sataa, se on:
oppinut runoille ruvetessansa ensin kokee mielensä ja sydämmensä
sillä aineella täyttää, josta päätti runoilla. Mikä siis on
kansanrunoissa se omaisuus, jossa enimmästi tekorunoista
erotaksen? Se on luonto ja teeskentelemättömyys; sula mielen
ilmotus ilman mitään salaamatta, mitään ulkoa lisäämättä.» — Ehkei
tämä ero täydelleen pidäkään paikkaansa, niin on se kuitenkin
pohjaltaan oikea. Kansanruno on kaikessa tapauksessa välittömästi
syntynyttä, laulajan mielialan itsestään tapahtuva purkautuminen:

emp' ois outo ollenkana, saattamaton saanutkana, ruvennut


runon tekohon, vaan tuli iso ikävä se minun pani pakolla,
sepä työnti työntämällä.

Luonnon kanssa läheisessä yhteydessä oleminen ja sen


erottamaton liittyminen kansanmiehen ja -naisen elämään taasen
antaa runsaimmat aiheet laululle ja muodostaa sen laajimman
liikkumisalan. Niin tunnustaa runo itsekin, että sen ainekset ovat
luonnosta ja elämästä saatuja: karjassa Muurikin ja Kimmon jälessä
kulkiessaan heiniä ja kanervia polkiessaan ja kuullessaan sateen
pieksännän ja tuulen huminan on ihminen ne löytänyt. Niin ovat siis,
runon omilla sanoilla ilmaistuna, laulajan tunteet ja ajatukset:

tieohesta tempomia, kanervoista katkomia risukoista


riipomia, vesoista vetelemiä, päästä heinän hieromia, raitiolta
ratkomia, paimenessa käyessäni, läsnä karjan laitumilla,
metisillä mättähillä, kultaisilla kunnahilla, mustan Muurikin
jälessä, Kimmon kirjavan kerällä. Vilu mulle virttä virkkoi, säe
saatteli runoja, virttä toista tuulet toivat, meren aaltoset
ajoivat, linnut liitteli sanoja, puien latvat lausehia.

(1 runo 53—70 säkeet).

Toiselta puolen tällaiseen yksinäisyyteen liittynyt raskas mieliala


on antanut alun runoille:

You might also like