B0221
Title: Modelling high dimensional stock dependence using factor copulas
Authors: Hoang Nguyen - Orebro University (Sweden) [presenting]
Concepcion Ausin - Universidad Carlos III de Madrid (Spain)
Pedro Galeano - Universidad Carlos III de Madrid (Spain)
Abstract: Multivariate GARCH models are widely used to derive the dynamic dependence structure of financial time series. However, the number of parameters becomes explosive in high dimensions which results in most of the models in the literature being static. Alternatively, factor copulas as the truncated C-vines rooted at the latent variables are proposed for tackling the problem. As in any factor model setting, it is assumed that each financial time series is affected by some common latent factors. Factor loadings are modelled as generalized autoregressive score (GAS) processes imposing a dynamic dependence structure in their densities. We employ the Bayesian approach to estimate the different specifications of the factor copula models. Condition on the latent factors, time series become independent which allows the algorithm to run in a parallel setting. Besides, the independence assumption of the latent models also reduces the computational burden for the conditional posterior distribution. A simulation study shows the performance for each models. We suggest DIC criteria for model selection. Finally, we provide an illustration on the stock price of companies listed in S\&P100 and compare the prediction power of models using Value-at-Risk.
Journal