Why might the equity tranche of CDO benefit from a more correlated reference pool, while the senior tranche prefers lower correlation amongst assets?

(cont.) .....if the above is true, it's to me counter-intuitive as the equity holds the most defaults -- higher correlation in the subordinate tranche would equate to a higher default probability, and thus less for holders of equity. Under this logic?, the inverse would be true of the senior tranche.Any insight on this topic is helpful.

default prob is the same no matter which tranche you're talking abouthigh correlation in the equity tranche = greater likelyhood of 0 defaultshigh correlation in the senior tranche = greater likelyhood of the tranche being touched

Yes, but I'm still unclear as to why "high correlation in the equity tranche = greater likelihood of 0 defaults" and why "high correlation in the senior tranche = greater likelihood of the tranche being impacted"Thanks in advance!

QuoteYes, but I'm still unclear as to why "high correlation in the equity tranche = greater likelihood of 0 defaults" and why "high correlation in the senior tranche = greater likelihood of the tranche being impacted"High correlation means that whatever event takes place a lot of the obligors are likely to do the same thing so either many can default or few can default. So these extremes become more likely with high correlation. So we get the greater liklihood of 0 defaults but also a greater liklihood of many defaults however, this affects the equity tranche more than it affects the senior tranche because even a few defaults can make the equity tranche worthless so it matters less to the equity tranche whether there are few defaults or many. With the senior tranche it matters whether there are many defaults because few defaults dont affect the senior tranche as much.Sorry if this sounds a bit convoluted but should clear it upMinal

Am I wrong to relate the diversity score to correlation, where the greater the diversity score, the lower is the correlation of the entire reference pool? If not, higher correlation (lower diversity score), whether in the senior or equity tranche, should result in a higher probability of defaults, not lower.Thanks for your help on this point.WCM

QuoteAm I wrong to relate the diversity score to correlation, where the greater the diversity score, the lower is the correlation of the entire reference pool? If not, higher correlation (lower diversity score), whether in the senior or equity tranche, should result in a higher probability of defaults, not lower.Yes in a way because higher corr=>higher probability of many defaults and also of no defaults.Diversity score is a measure of the number of 'independent' obligors, so in a crude way -- the higher the diversity score, the lower the corr.

Another way I like to think abt this is of equity as a call option on the underlying.Increasing correlation among the default pool is equivalent to increasing the variance of the default distribution (hence, extremes become more likely than before), hence, increase the 'vol' of the option is associated with a higher price.Rgds,R.

I found the JP Morgan 'Cat metaphore' (or whatever it is named) very useful in order to understand the tranche behavior as a function of the correlation. Imagine that you are a cat willing to cross a room in which a lot of traps are randomly scattered. If you are a standard cat, thus with only one life the "correlated" situation is the best for you because traps are grouped in clusters and there is the most chance that there are safe pathways for crossing the room. If you are a fairy tales cat, i.e. a nine-lives-cat, then the best situation is the one in which the traps are uniformly distribuited in the room(low correlation). You have a high probability of falling into a trap but a very low one of encountering nine traps in a row. Similarly when you are the owner of a super-senior tranche the event you must fear the most is the avalanche of defaults, which is the only event that may cause a loss in a ss tranche. The probability of such an event becomes higher in the case of high correlation. Cheers

great analogy!Has anyone seen any research out there comparing modern mathematical models incorporating correlation between individual loan defaults in CMBS and traditional banking portfolio analysis? thanks much