Journal of Financial Econometrics Advance Access originally published online on September 15, 2009
Journal of Financial Econometrics 2009 7(4):437-480; doi:10.1093/jjfinec/nbp014
| ||||||||||||||||||||||||||||||||||||||||||||||||||
Modeling International Financial Returns with a Multivariate Regime-switching Copula
Norwegian School of Economics and Business Administration (NHH)
Universidad Carlos III de Madrid
CORE, Université Catholique de Louvain
Address correspondence to Andréas Heinen, Departamento de Estadística, Universidad Carlos III de Madrid, 126 Calle de Madrid, 28903 Getafe (Madrid), Spain, or e-mail: aheinen{at}est-econ.uc3m.es.
JEL Classification: C32, C35, G10
| Abstract |
|---|
In order to capture observed asymmetric dependence in international financial returns, we construct a multivariate regime-switching model of copulas. We model dependence with one Gaussian and one canonical vine copula regime. Canonical vines are constructed from bivariate conditional copulas and provide a very flexible way of characterizing dependence in multivariate settings. We apply the model to returns from the G5 and Latin American regions, and document three main findings. First, we discover that models with canonical vines generally dominate alternative dependence structures. Second, the choice of copula is important for risk management, since it modifies the Value-at-Risk (VaR) of international portfolios and produces a better out-of-sample performance. Third, ignoring asymmetric dependence and regime-switching in portfolio selection leads to significant costs for an investor.
KEYWORDS: asymmetric dependence, canonical vine copula, international returns, portfolio selection, regime-switching, risk management, Value at Risk
We are grateful for comments and suggestions from Jonas Andersson, Luc Bauwens, Claudia Czado, Victor de la Peña, Rob Engle, René Garcia, Andrew Gelman, Bruno Gérard, Christian Hafner, Malika Hamadi, Philipp Hartmann, Chris Heyde, Bob Hodrick, Sébastien Laurent, Jostein Lillestol, Ching-Chih Lu, Thomas Mikosch, Andrew Patton, Jose Scheinkman, Johan Segers, Yongzhao Shao, I-Ling Shen, Assaf Zeevi, and participants at NHH, the Norwegian Central Bank, Pace University, Cornell, the Columbia Risk Seminar, Universidad Carlos III Madrid, ECORE seminar at Université Libre de Bruxelles, the International Conference on Finance in Copenhagen, the Federal Reserve Bank of Boston, and the Federal Reserve Bank of New York, as well as the participants of the Multivariate Volatility Models conference in Faro in October 2007. The comments of two anonymous referees and of the editor helped us improve the paper greatly. Lorán Chollete acknowledges financial support from Finansmarkedsfondet by grant #185339. Lorán Chollete and Andréas Heinen acknowledge financial support from the Institut Europlace de Finance. Andréas Heinen acknowledges support from the Spanish government by grant no. SEJ2006-03919. Alfonso Valdesogo acknowledges financial support from the contract "Projet dActions de Recherche Concertées" by grant no. 07/12-002 of the "Communauté française de Belgique", granted by the "Académie Universitaire Louvain". The usual disclaimers apply.
Received February 20, 2008; revised July 28, 2009; accepted July 29, 2009