A1684
Title: High-dimensional semiparametric skew-elliptical copula graphical models
Authors: Gabriele Di Luzio - Sapienza, University of Rome (Italy) [presenting]
Giacomo Morelli - Sapienza University of Rome (Italy)
Abstract: A semiparametric approach called elliptical skew-(S)KEPTIC is proposed for efficiently and robustly estimating non-Gaussian graphical models. Relaxing the assumption of meta-elliptical distribution into the family of meta skew-elliptical distributions that accommodates a skewness component, we derive a new estimator that is an extension of the SKEPTIC estimator. This extension is based on semiparametric Gaussian copula graphical models and applies to skew-elliptical copula graphical models. Theoretically, we demonstrate that the elliptical skew-(S)KEPTIC estimator achieves robust parametric convergence rates in both graph recovery and parameter estimation. We conduct numerical simulations to verify the reliable graph recovery performance of the elliptical skew-(S)KEPTIC estimator. Finally, the new method is applied to the daily log-returns of the stocks of the S\&P 500 index and shows better interpretability compared to Gaussian copula graphical models.
