2013/01/27 by Jan Beran, Beran, Jan, Georg Mainik +1
Economics, Econometrics and Finance · Mathematics · #Financial Risk and Volatility Modeling #Statistical Methods and Inference #Monetary Policy and Economic Impact
paper · pdf · doi:10.48550/arxiv.1301.6376
Estimation of extreme value copulas is often required in situations where\navailable data are sparse. Parametric methods may then be the preferred\napproach. A possible way of defining parametric families that are simple and,\nat the same time, cover a large variety of multivariate extremal dependence\nstructures is to build models based on spectral measures. This approach is\nconsidered here. Parametric families of spectral measures are defined as convex\nhulls of suitable basis elements, and parameters are estimated by projecting an\ninitial nonparametric estimator on these finite-dimensional spaces. Asymptotic\ndistributions are derived for the estimated parameters and the resulting\nestimates of the spectral measure and the extreme value copula. Finite sample\nproperties are illustrated by a simulation study.\n