Authors: Jin Zhang, Dietmar Maringer
Addresses: Centre for Computational Finance and Economic Agents, School of Computer Science and Electronic Engineering, Wivenhoe Park, Colchester CO4 3SQ, UK. ' Faculty of Economics and Business Administration, University of Basel, Peter Merian-Weg 6 4002 Basel, Switzerland
Abstract: Copulas provide investors with tools to model the dependence structure of financial products. The choice of copulas plays an important role in successful copula applications. This paper discusses the copula selection problem for the so-called |D-vine| decomposition from a perspective of the |safety first| asset allocation. The Joe-Clayton copula and the Student t copula are considered as building blocks for the D-vine structure. As an alternative to conventional approaches, the proposed pair-copula-GARCH model provides simulated asset returns for the optimal asset allocation which is implemented by using a heuristic optimisation approach. When assessing the reliability of portfolio loss prediction, it is found that the EWMA of RiskMetrics performs slightly better than the copula-GARCH models in the study of value-at-risk and expected shortfall minimisation. However, the Joe-Clayton copula model outperforms the EWMA in the case of Omega ratio minimisation.
Keywords: downside risk; AR-TGARCH; pair copulas; asset allocation; differential evolution; modelling; dependence structure; financial products; portfolio loss prediction; portfolio investment.
International Journal of Financial Markets and Derivatives, 2011 Vol.2 No.1/2, pp.121 - 148
Published online: 10 Feb 2011 *Full-text access for editors Access for subscribers Purchase this article Comment on this article