Université de Sherbrooke, 2500 Boul. de l'Université, Quebec, QC, J1K 2R1, CA
Generalized simulated method-of-moments for copula parameters of arbitrary dimension.
In this talk, a new methodology for estimating a vector of parameters of an arbitrary dimension in multivariate copula models will be presented. The proposed estimator is based on the first p moments of the multivariate probability integral transformation that one can associate to a given parametric copula model. An unbiased estimator of these p moments is first described, from which a method-of-moments estimator of the unknown vector of parameters is defined. In order that the method be applicable even when explicit expressions for the theoretical moments are not available, a simulated version of the estimator is developed as well. Interestingly, the latter can be performed as long as one is able to simulate from a given copula model. The consistency and asymptotic normality of these estimators are formally established under standard and mild conditions. The performance of these estimators in terms of bias and mean-squared errors is investigated through an extensive simulation study.