random normal (0,1) correlated by copulas

Hi,

I created a simulation of random variables correlated by copulas in Matlab, using ksdensity function.

Now I am trying to simulate several N(0,1) distributions, but I want them to be intercorrelated by the same copulas.

How can I insert the copula parameter in normrnd function?

Thank you very much,

Answer 1

[Tom Lane]

The usual idea is that you apply your inverse probability distributions to the marginals (each column) of the copularnd results. In your case, if all variables are to have a standard normal distribution, you can just apply norminv to the entire array.

[Alexandra]

Thank you Tom, I know what you mean. I am trying to build a Copula-GARCH model. I thought that was the way I had to do it. In GARCH, the stochastic part is N(0,1) in the variance equation. So I wanted the N(0,1) to be correlated with copulas to simulate random correlated variables. But I think I was wrong. I need to fit the copula to the residuals of the GARCH model and then multiply the random copulas with the variance equation of GARCH to obtain the correlated returns. I will attempt this.