Sampling from hierarchical Kendall copulas

Brechmann, Eike Christian

Journal de la société française de statistique, Numéro spécial sur les copules, Tome 154 (2013) no. 1, pp. 192-209.

Les copules étant fréquemment utilisées pour modéliser la dépendance, il est crucial de disposer d’algorithmes efficaces permettant de générer des échantillons pseudo-aléatoires à partir de ces distributions. Dans le cas des copules de Kendall hiérarchiques, une stratégie d’échantillonnage descendante repose sur la simulation d’un vecteur aléatoire sachant qu’il prend ses valeurs dans un ensemble de niveau de la copule. Après avoir rappelé les solutions explicites dans le cas où la construction repose sur les copules Archimédiennes, cet article présente de nouveaux résultats pour la copule de Plackett et pour les copules Archimax contenant la classe des copules de valeurs extrêmes. En complément, de nouveaux algorithmes approchés de génération d’échantillons pseudo-aléatoires pour les copules de Kendall hiérarchiques sont proposés et évalués par le biais de simulations.

As copulas are frequently used to model dependence in statistical models, it is of central importance to be able to accurately and efficiently sample from them. In the case of hierarchical Kendall copulas, a top-down sampling strategy involves simulation of a random vector given that it lies in a particular level set. While explicit solutions are available when hierarchical Kendall copulas are built from Archimedean copulas, this paper presents new results for the Plackett copula and for Archimax copulas, which also include the class of extreme-value copulas. Additionally, new approximate sampling procedures for hierarchical Kendall copulas are proposed and evaluated in a simulation study.

[1] Aas, Kjersti; Czado, Claudia; Frigessi, Arnoldo; Bakken, Henrik Pair-copula constructions of multiple dependence, Insurance: Mathematics and Economics, Volume 44 (2009) no. 2, pp. 182-198 | MR | Zbl

[2] Arbenz, Philipp; Hummel, Christoph; Mainik, Georg Copula based hierarchical risk aggregation through sample reordering, Insurance: Mathematics and Economics, Volume 51 (2012) no. 1, pp. 122-133 | MR | Zbl

[3] Barbe, Philippe; Genest, Christian; Ghoudi, Kilani; Rémillard, Bruno On Kendall’s Process, Journal of Multivariate Analysis, Volume 58 (1996) no. 2, pp. 197-229 | MR | Zbl

[4] Brechmann, Eike C. Hierarchical Kendall copulas: Properties and inference (2012) (Preprint, http://arxiv.org/abs/1202.1998) | MR

[5] Brechmann, Eike C.; Schepsmeier, Ulf Modeling dependence with C- and D-vine copulas: The R-package CDVine, Journal of Statistical Software, Volume 52 (2013) no. 3, pp. 1-27

[6] Capéraà, Philippe; Fougères, Anne-Laure; Genest, Christian Bivariate distributions with given extreme value attractor, Journal of Multivariate Analysis, Volume 72 (2000) no. 1, pp. 30-49 | MR | Zbl

[7] Cherubini, Umberto; Luciano, Elisa; Vecchiato, Walter Copula Methods in Finance, John Wiley & Sons, Chichester, 2004 | MR | Zbl

[8] Dißmann, Jeffrey; Brechmann, Eike C.; Czado, Claudia; Kurowicka, Dorota Selecting and estimating regular vine copulae and application to financial returns, Computational Statistics & Data Analysis, Volume 59 (2013) no. 1, pp. 52-69 | MR | Zbl

[9] Ghoudi, Kilani; Khoudraji, Abdelhaq; Rivest, Louis-Paul Propriétés statistiques des copules de valeurs extrêmes bidimensionnelles, Canadian Journal of Statistics, Volume 26 (1998) no. 1, pp. 187-197 | MR | Zbl

[10] Genest, Christian; Rivest, Louis-Paul A characterization of Gumbel’s family of extreme value distributions, Statistics & Probability Letters, Volume 8 (1989) no. 3, pp. 207-211 | MR | Zbl

[11] Genest, Christian; Rivest, Louis-Paul Statistical inference procedures for bivariate Archimedean copulas, Journal of the American Statistical Association, Volume 88 (1993) no. 423, pp. 1034-1043 | MR | Zbl

[12] Hofert, Marius Sampling Archimedean copulas, Computational Statistics & Data Analysis, Volume 52 (2008) no. 12, pp. 5163-5174 | MR | Zbl

[13] Hofert, Marius Efficiently sampling nested Archimedean copulas, Computational Statistics & Data Analysis, Volume 55 (2011) no. 1, pp. 57-70 | MR | Zbl

[14] Iman, Ronald L.; Conover, William J. A distribution-free approach to inducing rank correlation among input variables, Communcations in Statistics - Simulation and Computation, Volume 11 (1982) no. 3, pp. 311-334 | Zbl

[15] Joe, Harry Multivariate Models and Dependence Concepts, Chapman & Hall, London, 1997 | MR | Zbl

[16] Kurowicka, Dorota; Cooke, Roger M. Uncertainty Analysis with High Dimensional Dependence Modelling, John Wiley, Chichester, 2006 | MR | Zbl

[17] Longin, François; Solnik, Bruno Extreme correlation of international equity markets, Journal of Finance, Volume 56 (2001) no. 2, pp. 649-676

[18] Mainik, Georg Sampling dependence: empirical copulas, Latin hypercubes, and convergence of sums (2012) (Preprint, http://www.math.ethz.ch/ gmainik/Papers/lhs-dep-sums.pdf)

[19] Mardia, Kantilal V. Families of Bivariate Distributions, Griffin, London, 1970 | MR | Zbl

[20] McNeil, Alexander J. Sampling nested Archimedean copulas, Journal of Statistical Computation and Simulation, Volume 78 (2008) no. 6, pp. 567-581 | MR | Zbl

[21] McNeil, Alexander J.; Nešlehová, Johanna Multivariate Archimedean copulas, $d$-monotone functions and ${\ell }_1$-norm symmetric distributions, Annals of Statistics, Volume 37 (2009) no. 5B, pp. 3059-3097 | MR | Zbl

[22] Mai, Jan-Frederik; Scherer, Matthias Simulating Copulas: Stochastic Models, Sampling Algorithms, and Applications, World Scientific Publishing Co., Singapore, 2012 | MR | Zbl

[23] Nelsen, Roger B. An Introduction to Copulas, Springer, Berlin, 2006 | MR

[24] Pickands, James Multivariate extreme value distributions, Proceedings of the 43rd Session of the International Statistical Institute, Buenos Aires (1981), pp. 859-878 | MR | Zbl

[25] Plackett, Robin L. A class of bivariate distributions, Journal of the American Statistical Association, Volume 60 (1965) no. 310, pp. 516-522 | MR

[26] Rubin, Donald B. A noniterative sampling/importance resampling alternative to the data augmentation algorithm for creating a few imputations when fractions of missing information are modest: the SIR algorithm, Journal of the American Statistical Association, Volume 82 (1987) no. 398, pp. 543-546

[27] Rubin, Donald B. Using the SIR algorithm to simulate posterior distributions, Bayesian Statistics 3 (Bernardo, J. M.; DeGroot, M. H.; Lindley, D. V.; Smith, A. F. M., eds.), Oxford University Press, Oxford, 1988, pp. 395-402 | MR | Zbl

[28] Sklar, Abe Fonctions de répartition à n dimensions et leurs marges, Publications de l’Institut de Statistique de L’Université de Paris, Volume 8 (1959), pp. 229-231 | MR | Zbl

[29] Salvadori, Gianfausto; Michele, Carlo De; Kottegoda, Nathabandu T.; Rosso, Renzo Extremes in Nature: An Approach Using Copulas, Springer, Berlin, 2007

[30] Whelan, Niall Sampling from Archimedean copulas, Quantitative Finance, Volume 4 (2004) no. 3, pp. 339-352 | MR | Zbl

[31] Wu, Florence; Valdez, Emiliano; Sherris, Michael Simulating from Exchangeable Archimedean Copulas, Communications in Statistics - Simulation and Computation, Volume 36 (2007) no. 5, pp. 1019-1034 | MR | Zbl