In the question mentioned above, when using the probabilities of default in the copula functions, the answers substitute 0.05 and 0.15. In doing this, I presume we then looking at the joint distribution in the lower tail? Is there a reason why we would use this instead of 0.95 and 0.85 (and then manipulating in terms of a survival copula to answer the question ) and look at the upper tail? Does the choice of upper tail or lower tail just depend on whether we are thinking of positive or negative losses? In the second part of the question the answers say a Gumbel copula may be most appropriate because of the upper tail dependence in the case of recession (ie as they are both more likely to default simultaneously). This seems contradictory to the first part, where it looks to me that if we are treating the defaults to be in the lower tail, a Clayton would be more appropriate. Am I missing something here? Thanks!
