Athens University Economics and Business (AUEB)
High dimensional time-varying covariance matrices with applications in finance
Abstract
dc:descriptionScope of this Thesis is to provide contribution in the area of Multivariate Volatility modeling, which involves developing models that can adequately describe the Covariance matrix process of Multivariate financial time series. Development of efficient algorithms for Bayesian model estimation using Markov Chain Monte Carlo (MCMC) and Nested Laplace approximations is the main objective in order to provide parsimonious and flexible volatility models. Detailed review of Univariate Volatility models for financial time series is first introduced. The historical background of each model proposed is illustrated and its properties and advantages are discussed. The several estimation methods that have emerged are also commented. A comparative analysis via a small simulation example for the dominant models in the literature is provided. Continuing the review from the univariate models, the multivariate case follows and competing models for Covariance matrices are extensively presented. Main argument is that currently no model captures the dynamics of higher dimensional Covariance matrices fully, but their relative performance and applicability depends on the dataset and problem of interest. Problems are mainly due to the positive definiteness constraints required by most models as well as lack of interpretability of the model parameters in terms of the characteristics of the financial series. In addition, model development so far focus mostly in parameter estimation and in sample fit; the out-of-sample fit perspective of these models is now also examined. The review section concludes by proposing some small improvements for existing models towards more efficient parameter estimates, faster estimation methods and accurate forecasts. A new class of multivariate volatility models based on the Spectral decomposition of the time changing covariance matrix and the incorporation of autocorrelation modeling for the time changing elements. All elements of the covariance matrix are now a priori allowed to be time changing as independent autoregressive processes and then for any given dataset the prior information is updated and the number of time changing elements is decided. Theoretical properties of the new model are presented along with a series of estimation methods, Bayesian and classical. To estimate these models it is suggested to use an MCMC method for small dimension portfolios in terms of the size of the covariance matrix. For higher dimensions a Nested Laplace approximation approach should be used that provides results much faster with small loss in accuracy. The performance of the proposed model against competing ones in simulated and real datasets is discussed. Results indicate that the new model provides significantly better estimates and projections than current models in the majority of example datasets. In addition, the new model allows for parameter interpretability and parsimony which is of huge importance due to the dimensionality curse. The Thesis concludes by providing some proposals for future research.
Degree
thesis:*- Grantor dc:publisher
- Athens University Economics and Business (AUEB)
- Year dc:date
- 2011
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Plataniotis, Anastasios
- Πλατανιώτης, Αναστάσιος
Subjects
dc:subject × 20- Πολυμεταβλητή στοχαστική μεταβλητότητα
- Πίνακας διακύμανσης
- Αυτοπαλίνδρομα μοντέλα δεσμευμένης ετεροσκεδαστικότητας
- Αλυσίδες Μαρκόφ με χρήση μεθόδων Μόντε Κάρλο
- Μέθοδος Laplace
- Κανονικά κατανεμημένο τυχαίο πεδίο Μαρκόφ
- Φίλτρο Kalman
- Μπεϋζιανή μέθοδος εκτίμησης
- Multivariate stochastic volatibility
- Covariance Matrix
- Garch
- MCMC
- Laplace
- Gaussian Markov random field
- Kalman filters
- Bayesian
- Social Sciences
- Economics and Business
- Κοινωνικές Επιστήμες
- Οικονομικά και Επιχειρήσεις
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/26164