Parameterization for Spectral Copula Model

Thomas M. Menino Convention & Exhibition Center 

We explore different parameterization of spectral matrices of multiple time series. with given. marginal spectral densities. Such spectral copula models are useful in multiple time series modeling when marginal. univariate specification is available for some or all of the coordinate processes. The models. involve. parameterizing the coherence.s and novel. parameterization of. the matrix function honoring the natural constraints are investigated. Estimation under the new parameterization is. also investigated and the benefits illustrated with numerical examples.

Multiple time series

coherence

Univariate spectral density