Bollerslev (1990) proposes an n-dimensional GARCH model that comprises n univariate GARCH processes tWi related to one another with a constant conditional correlation matrix ρ. We call this the constant conditional correlation GARCH or CCC-GARCH model. It has form
where ρ is a correlation matrix, and
with conditional variances modeled as with a univariate GARCH(p,q) process:
These may be estimated using maximum likelihood.