Which of the following statements about correlation and copula are correct?

I. Copula enables the structures of correlation between variables to be calculated separately from their marginal distributions.

II. Transformation of variables does not change their correlation structure.

III. Correlation can be a useful measure of the relationship between variables drawn from a distribution without a defined variance.

IV. Correlation is a good measure of dependence when the measured variables are distributed as multivariate elliptical.