dc.contributor.advisor |
Billio, Monica |
|
dc.contributor.author |
Cavicchioli, Maddalena <1985> |
it_IT |
dc.date.accessioned |
2014-04-05T10:33:37Z |
|
dc.date.available |
2015-04-07T13:58:32Z |
|
dc.date.issued |
2014-03-21 |
|
dc.identifier.uri |
http://hdl.handle.net/10579/4602 |
|
dc.description.abstract |
In questa tesi studiamo alcuni problemi legati a modelli Markov Switching (MS) e alle loro applicazioni in Economia e Finanza. Lo scopo del nostro studio è proporre soluzioni per la selezione del modello e per la stima di serie storiche multivariate soggette a cambiamenti di regime. Nel primo Capitolo presentiamo la letteratura che tratta di sistemi dinamici per modellare serie storiche con cambiamenti di regime. Nel secondo Capitolo studiamo il problema della determinazione del numero di regimi nell’ambito di modelli MS-VARMA e proponiamo alcuni metodi per la scelta del modello basati sulla funzione di autocovarianza e sulla rappresentazione stabile del sistema. Questi metodi sono poi applicati all’analisi del ciclo economico. Nel Capitolo 3 introduciamo modelli a cambiamento di regime per la volatilità di dati finanziari e proponiamo un metodo unificato per la stima di modelli MS-GARCH e modelli a volatilità stocastica con MS (teorema di dualità). Nel quarto Capitolo esploriamo altre questioni che riguardano i modelli MS come la stima e la loro rappresentazione spettrale. Riguardo al problema della stima, otteniamo semplici formule matriciali per la stima di massima verosimiglianza dei parametri per modelli MS-VAR e MS-VAR con effetti ARCH. Questo permette di determinare in maniera esplicita la matrice di varianza-covarianza degli stimatori, e quindi offre una possibilità concreta per l’uso dei test statistici classici. Riguardo al secondo aspetto, studiamo varie proprietà della funzione di densità spettrale di modelli MS-VAR e otteniamo espressioni in forma chiusa per la densità spettrale. La tesi è completata da diversi esercizi di simulazione e applicazioni a dati macroeconomici e finanziari. |
it_IT |
dc.description.abstract |
In this thesis we discuss problems emerging in the application of Markov Switching (MS) models both in Economics and Finance. The aim of the study is to propose solutions for model selection and estimation of multiple time series subject to regime shifts. In Chapter 1 we review the literature about dynamic systems for modeling time series with changes in regimes. In the second Chapter we investigate the problem of determining the number of regimes in MS-VARMA models and describe methods for model selection based on the autocovariance function and on stable representation of the system. Application to business cycle analysis is conducted. In Chapter 3 we introduce MS models for volatility of financial data and propose a unified framework for estimating MS-GARCH and MS-Stochastic Volatility models (duality result). In the fourth Chapter we explore other questions concerning with MS models as estimation and spectral representation. With regards to the first, we obtain simple matrix formulae for maximum likelihood estimates of parameters in the class of MS-VAR and conditional heteroskedastic models. This allows us to determine explicitly the asymptotic variance-covariance matrix of the estimators, thus giving a concrete possibility for the use of classical testing procedure. Concerning the second, we study the properties of spectral density function for MS-VAR models and derive close-form formulae for the spectral density. Several simulation exercises and applications to macroeconomic and financial data complete the work. |
it_IT |
dc.language.iso |
eng |
it_IT |
dc.publisher |
Università Ca' Foscari Venezia |
it |
dc.rights |
© Maddalena Cavicchioli, 2014 |
it_IT |
dc.subject |
Markov Switching |
it_IT |
dc.subject |
MS-VARMA models |
it_IT |
dc.subject |
MS-GARCH models |
it_IT |
dc.subject |
MS-SV models |
it_IT |
dc.subject |
State space representations |
it_IT |
dc.subject |
Spectral representations |
it_IT |
dc.subject |
Maximum likelihood estimates |
it_IT |
dc.title |
Essays on Markov Switching models with applications in economics and finance |
it_IT |
dc.type |
Doctoral Thesis |
en |
dc.degree.name |
Economia |
it_IT |
dc.degree.level |
Dottorato di ricerca |
it |
dc.degree.grantor |
Scuola superiore di Economia |
it_IT |
dc.description.academicyear |
2014 |
it_IT |
dc.description.cycle |
26 |
it_IT |
dc.degree.coordinator |
Bernasconi, Michele |
|
dc.location.shelfmark |
D001322 |
it |
dc.location |
Venezia, Archivio Università Ca' Foscari, Tesi Dottorato |
it |
dc.rights.accessrights |
openAccess |
it_IT |
dc.thesis.matricno |
955787 |
it_IT |
dc.format.pagenumber |
XVI, 182 p. |
it_IT |
dc.subject.miur |
SECS-P/05 ECONOMETRIA |
it_IT |
dc.description.note |
Doctor Europaeus |
|
dc.description.tableofcontent |
Contents
List of Figures vii
List of Tables ix
Introduction xiii
1 Some of Representations of Dynamic Systems.
Modeling Time Series with Changes in Regimes 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 ARMA and ARIMA Representations . . . . . . . . . . . . . 2
1.3 Characterizing ARMA Representations . . . . . . . . . . . . 7
1.3.1 Markov Coe cients . . . . . . . . . . . . . . . . . . . . 7
1.3.2 Hankel Matrix Rank . . . . . . . . . . . . . . . . . . . 8
1.4 State-Space Representation . . . . . . . . . . . . . . . . . . . 9
1.5 Markov Chains . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5.1 The De nition . . . . . . . . . . . . . . . . . . . . . . . 14
1.5.2 Representing a Markov chain by an Autoregression 14
1.5.3 Forecasts . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.4 Reducible Markov chain . . . . . . . . . . . . . . . . . 16
1.5.5 Ergodic Markov chains . . . . . . . . . . . . . . . . . . 17
1.5.6 Periodic Markov chains . . . . . . . . . . . . . . . . . 18
1.6 Time Series Models of Changes in Regime . . . . . . . . . . 19
1.6.1 The model . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6.2 Optimal Inference for the Regime . . . . . . . . . . . 211.6.3 Forecasts and Smoothed Inferences for the Regime 24
1.6.4 Forecasts for the Observed Variables . . . . . . . . . 28
1.6.5 Maximum Likelihood Estimation of Parameters . . 29
1.7 EM algorithm and Likelihood function . . . . . . . . . . . . 30
1.7.1 EM algorithm: general principles . . . . . . . . . . . 31
1.7.2 First characterization of EM algorithm . . . . . . . 33
1.7.3 Second characterization of EM algorithm . . . . . . 34
1.7.4 Explicit form of the EM algorithm . . . . . . . . . . 35
1.8 State-Space Models with Markov Switching . . . . . . . . . 39
1.8.1 Speci cation of the Model . . . . . . . . . . . . . . . 40
1.8.2 Basic Filtering and Estimation . . . . . . . . . . . . . 40
1.8.3 Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.9 Determination of the Number of Regimes . . . . . . . . . . 51
2 Markov-Switching VARMA Models 57
2.1 Determining the Number of Regimes in Markov-Switching
VAR and VMA Models . . . . . . . . . . . . . . . . . . . . . . 57
2.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 58
2.1.2 VARMA Representations . . . . . . . . . . . . . . . . 60
2.1.3 Markov Switching Moving Average Models . . . . . 67
2.1.4 Markov Switching Autoregressive Models . . . . . . 73
2.1.5 Data Simulation . . . . . . . . . . . . . . . . . . . . . . 78
2.1.6 Application on foreign exchange rates . . . . . . . . 81
2.1.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2.1.8 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . 85
2.2 Business Cycle and Markov Switching Models with Dis-
tributed Lags: a Comparison between US and Euro Area 89
2.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.2.2 Markov Switching Models and Business Cycle . . . 92
2.2.3 The MSI(M; r) - VAR(0) Model . . . . . . . . . . . . 94
2.2.4 The MSI(M; r)-VARMA(p; q) Model . . . . . . . . . . 97
2.2.5 Business Cycle Models . . . . . . . . . . . . . . . . . . 99
2.2.6 Empirical Application . . . . . . . . . . . . . . . . . . . 1052.2.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 110
2.2.8 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . 113
3 Markov Switching Models for Volatility: Filtering, Approxima-
tion and Duality 117
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.2 Markov Switching GARCH . . . . . . . . . . . . . . . . . . . 121
3.3 Auxiliary Models for MS-GARCH . . . . . . . . . . . . . . . 122
3.4 State Space Representation and Filtering . . . . . . . . . . 124
3.5 Duality Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
3.6 Markov Switching Stochastic Volatility . . . . . . . . . . . . 129
3.7 Numerical and Empirical Applications . . . . . . . . . . . . 131
3.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.9 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4 Estimation and Spectral Representation 145
4.1 Analysis of the Likelihood Function for Markov Switching
VAR(CH) Models . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.1.2 Time Series Models of Changes in Regime . . . . . . 147
4.1.3 The Basic Markov Switching Model . . . . . . . . . . 152
4.1.4 State-dependent Autoregressive Dynamics . . . . . . 158
4.1.5 State-dependent Multivariate ARCH Models . . . . 163
4.1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 168
4.2 Spectral Density of Regime Switching VAR Models . . . . 170
4.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 170
4.2.2 Markov-switching VAR(0) process . . . . . . . . . . . 171
4.2.3 Markov-switching VAR(p) process . . . . . . . . . . . 176
4.2.4 A numerical example . . . . . . . . . . . . . . . . . . . 177
4.2.5 Long memory or Regime Switching? . . . . . . . . . 178
4.2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 181 |
it_IT |
dc.identifier.bibliographiccitation |
Cavicchioli, Maddalena. "Essays on Markov Switching models with applications in economics and finance", Università Ca' Foscari Venezia, PhD Thesis, 26 cycle, 2014. |
it_IT |