abstract
We investigate the common
practice of estimating the dependence structure between credit default swap prices on multi-name credit instruments
from the dependence structure of the equity returns of the underlying firms. We find convincing evidence that the
practice is inappropriate for high-yield instruments and that
it may even be flawed for instruments containing only firms
within a sector. To do this, we model individual credit ratings
by univariate continuous time Markov chains, and their joint
dynamics by copulas. The use of copulas allows us to incorporate
our knowledge of the modeling of univariate processes, into a
multivariate framework. However, our test and results are robust to
the choice of copula.
|