Flexible copula models with dynamic dependence and application to financial data

A new class of copula models with dynamic dependence is introduced; it can be used when one can assume that there exist a common latent factor that affects all of the observed variables. Conditional on this factor, the distribution of these variables is given by the Gaussian copula with a time-varying correlation matrix, and some observed driving variables can be used to model dynamic correlations. This structure allows one to build flexible and parsimonious models for multivariate data with non-Gaussian dependence that changes over time. The model is computationally tractable in high dimensions and the numerical maximum likelihood estimation is feasible. The proposed class of models is applied to analyze three financial data sets of bond yields, CDS spreads and stock returns. The estimated model is used to construct projected distributions and, for the bond yield and CDS spread datasets, compute the expected maximum number of investments in distress under different scenarios.