Modeling dependence in high dimensions using Factor Copulas
T his paper builds a model for the dependence between the top 10 stocks of S&P 100. We follow the method introduced by Oh and Patton (2017) in building a Latent Factor model for the Copulas of a set of random variables. To estimate the Copula models parameters we use a Simulation Method of Moments (SMM). A simulation study shows that the SMM provides accurate estimates of the Factor Copula models. We consider restrictive models that force equidependence and do not allow for tail dependence between variables and than relax those restrictions to compare the results. In the works of Oh and Patton (2017) they used the Spearman correlation and quantile tail dependence. We introduce in this paper another moment condition calculated on the base of the Spectral measure. This new moment condition seems to perform adequately.