Hi, I am trying to understand copulas. Can someone please clarify this issue i am having. many times in analyzing basket of credit derivaties gaussian coupla is used as follows: from the individual names we can obtain the marginal distributions we now put some correlation structure that we want (or may be observed from the market) and then we use the gaussian copula to compute the multivariate probability distribution with our specified correlation. now my doubt is: even though i will get the marginal distribution and the correlation correctly (becuase these are my inputs to the formalism) from the coupla how am i ensured that the multivariate distribution that i observed in the market is gaussian ? i have used gaussian coupla so it means means most probably i observe multivariate gaussian density in the market data ? is the above statement true ? have i made some mistake in my reasoning ? thanks a lot in advance

Hi babu12,If your marginals are normal and you use gaussian copula, then the output is going to be very similar to estimating a Multivariate Normal Distribution. However, if you have non-normal marginals, then the output will be completly different. There are tests out there to test whether or not your sample follows a Multivariate Normal Distribution. Let me see if I am able to find the documents...