The Copula GARCH model for time varying
betas in the banking sector.
Dimitris Nikolaidis
2010
Copula functions have become an increasingly popular tool in nance when the
distribution of asset returns is of extreme importance. The main features of copulas
are that they seperate a multivariate distribution into the dependence structure
and the margins, thus allowing two step estimation procedures for the distributional
parameters that minimize the computational burden and also add exibility to the
distribution since the dependence governed by the copula and the margins do not have
to belong to the same parametric family, unlike standard multivariate distributions.
The aim of this study is twofold. In the rst part, the statistical attributes of
copulas are discussed in full detail while in the second part an empirical investigation
of the evolution of stock betas during the modern global nancial crisis period is
conducted. In the empirical part, it is evident that copula models clearly outperform
other, traditional models, in terms of both statistical validity and accuracy in risk
calculations
Collections
Show Collections
