Události

Po 05.12.2011

Prof. Giovanni Urga: “Dynamic Conditional Correlation Models with Asymmetric Multivariate Laplace Innovations”

Prof. Giovanni Urga

Cass Business School, City University London, United Kingdom

Authors: Giovanni Urga, Juan-Pablo Cajigas, and Alexios Ghalanos

Abstract: In this paper we propose to estimate multivariate GARCH processes and a class of dynamic conditional correlation (DCC) models assuming that the n-dimensional returns series follow the Asymmetric Multivariate Laplace (AML) distribution. The AML distribution is able to capture asymmetry and leptokurtosis which characterise returns from financial assets. Under general conditions, it preserves desirable properties such as finiteness of moments and stability under geometric summation. We prove that the maximum likelihood estimator provides consistent estimates for a variety of DCC models when AML distribution is assumed for standardised residuals. We also prove strict stationary of DCC models. The empirical validity of the proposed framework is tested by fitting 21 FTSE All-World stock indices and 12 bond return indices and evaluate its in-sample performance via alternative risk management measures. We provide clear evidence that in our data set the asymmetric generalised (AGDCC) model with AML distribution overwhelmingly outperforms the variety of DCC models that assume normality of innovations.

Full Text: “Dynamic Conditional Correlation Models with Asymmetric Multivariate Laplace Innovations”