Full text
| |
| Author(s): |
Carlos César Trucíos Maza
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Matemática, Estatística e Computação Científica |
| Defense date: | 2016-08-25 |
| Examining board members: |
Luiz Koodi Hotta;
Mauricio Enrique Zevallos Herencia;
Ronaldo Dias;
Pedro Alberto Morettin;
Marcelo Fernandes
|
| Advisor: | Luiz Koodi Hotta |
| Abstract | |
This Ph.D dissertation consists of three essays that address the problem of obtaining forecast densities in volatility models. The first two essays deal with the problem in a context of univariate volatility models while the latter one addresses the problem in a multivariate context. In all essays the proposed methodologies are implemented and applied to empirical data. The first essay, "Bootstrap prediction in univariate volatility models with leverage effect", proposes an algorithm to construct forecast intervals for returns and volatilities for the EGARCH and GJR-GARCH univariate volatility models. This algorithm is an extension of Pascual Romo and Ruiz (Computational Statistics & Data Analysis, 2006, v 50:2293-2312) (PRR) algorithm developed to obtain forecast intervals for returns and volatilities for the GARCH model. In addition, it is observed that both PRR algorithm and our extensions to the EGARCH and GJR-GARCH models can be drastically affected by the presence of outliers (paper published in Mathematics and Computer in Simulation, 2016, v 120:91-103). The second essay, "Robust bootstrap forecast densities for GARCH returns and volatilities", proposes a robust algorithm to obtain forecast densities for returns and volatilities in GARCH models. This algorithm robustifies the procedure of PRR and shows a good finite sample properties for contaminated and uncontaminated series. Finally, the third essay, "Robust bootstrap densities for dynamic conditional correlations models", presents a robust algorithm to obtain forecasts densities for returns, volatilities and also conditional correlations in the context of cDCC model (corrected dynamic conditional correlation). In addition, the third essay also presents an algorithm to obtain forecast densities for Value-at-Risk of portfolios (AU) | |
| FAPESP's process: | 12/09596-0 - Bootstrap prediction in univariate and multivariate volatility models |
| Grantee: | Carlos Cesar Trucios Maza |
| Support Opportunities: | Scholarships in Brazil - Doctorate |