Hi,Lets say I am trading a spread range swap against a CMS 5 yr and 3 yr which is currently trading at 78% correlation. My question is if the correlation increases to 79% why is it that my PnL is impacted. Where in the pricing equation is the correlation? Say I am short on this spread. Hence, if Correl increases is it good or bad for me?Mathematical or Intuitive answers would help a lot.Cheers.

Intuitively, if your coupons depend on a CMS spread staying within a given range, then the forward correlation between the 2 CMS rates determines their joint distribution at the fixing dates. Let's look at e.g. 10 vs 2 year spread....if the correlation is high, then the spread will pretty much stay constant...and hence your coupons too (the discounted value of which determine the price). On the other hand, if the correlation low....the cpns will be lower.....etc etc, you get the picture. So correlation actually has a big impact on these structures and it is important to get it right. However, what you need is a term structure of forward correlation...and maybe even an adjustment for sth like a skew.

Hi,I get your point, thanks. I have built a EWMA model to capture correlation of IRS strikes of different tenors. For example KRO 5yr - 3yr correl is around say 95%. I am interested in your point of forward correlation, can you please explain how that makes sense? and what is the skew risk aspect of it.Cheers.

MFEGRAD, if you are trading a bunch of different tenors across the curve, have you tried looking into PCA (pricinpal component analysis) in order to understand relative movements?Regards,Diogo

MFEGRAD Have you looked Rebonato´s book Interest Rate Options Models? There´s a goos discussion about volatilities and correlations beteween the tenors of the term structure.RegardsBob

My two cents on this...When pricing the non-callable spread swap, you indeed need to consider the marginale joint distributions of the two spreads, and thats' where outturn correlations will indeed come into play.But when pricing callables, TARNs, discrete KO, etc... then indeed you need to consider a few more things as well.For instance, when you decide to cancel the swap, the decision will be made from comparing the remaining part of the swap and the value you get from not cancelling the swap. That decision will then depend on the joint dirtbituion of the two CMS rates seen from that point (hence forward correlation and forward skew dependency), and also on the correlation between the swap rate of the funding leg and all the CMS spreads on the structured leg. That last component explains why you may need to consider PCA. In that case, you may want to look at comparing multi-factor BGM with two-factor short rate models for instance, and see what are the various impacts.

QuoteI get your point, thanks. I have built a EWMA model to capture correlation of IRS strikes of different tenors. For example KRO 5yr - 3yr correl is around say 95%. I am interested in your point of forward correlation, can you please explain how that makes sense? and what is the skew risk aspect of it.I am really not an expert on this...i just use the numbers the quants come up with whenver i need them. But intuitively, as a dealer, you should think of forward correlation as being different from historical correlation, just because it moves around and creates risk. So it will naturally have a risk premium to it. The skew is due to the dependence on the level (strike), e.g. spread options are quoted for different correlation values at different strikes. In any case, from what I have seen, those are very very fine points when it comes to the valuation of exotics. At the end of the day, the quoted bid ask ends up being so wide that to me it seems almost silly to be splitting hairs over the exact correlation skew....just my unqualified 2 cts.