The phenomenon you describe is a well known behaviour in the CDO market. The implied correlation is not unique for mezz tranches, i.e for a contract with sensitivity towards the mid parts of the loss distribution there are two correlations which will produce the same price. An NtD with N > 1 can behave in this way.The reason for this will become obvious to you if you draw a graph of the loss distribution (number of defaults on the x-axis and number of occurrences, or the probability, on the y-axis). Make sure you understand what part of this loss distribution your contract is sensitive to. Draw the probability as the function of number of defaults for a range of correlations. You will see that your contract's area of interest in the graph will have the same size for two different correlations.Hope that helped.

Anupkurup, I don't think so. I think it is a problem of the framework, and constitutes one of the primary criticisms of the single factor Gaussian copula model.I have seen some practitioner explanations for the implied correlation smile that are based around why the mezzanine tranche implied correlation is so low. Included are investor segmentation and hedging pressure of banks. I don't think they hold any true weight, as once you are using base correlations the mezzanine tranche doesn't stick out like it does using implied correlations.

Hi, anupkurupI am currently doing my project about valuation of the default correlation products, particularly n-th to default basket.But i am struggling in pricing the basket with normal or student t copula. I am wondering if you have the matlab code for this part.Thanks in advance!Looking forward to your reply.

I have heard the follow reasons for the 2nd to default and junior mezz CDO "smile" ... albeit I believe it is simply due to market technicals #3 below.1) segmentation amongst investors in tranches all holding different views on correlations. Real$ vs. Fast$ accounts. 2) "model risk" - uncertainty how to model the credit risk. **3) Local demand conditions - the implied corr. on the mezz. tranche may reflect strong interest in banks selling prot. on this segment of the loss distribution4) Other model besides Gaussian ... Student T, relaxing the restriction of constant pairwise corr, changing recovery rates, including macro econ. factors....just my thoughts...

hi anupkurupjust saw this, and was wondering if you could email me a version you have in VBA?my email address is farhoud.moaddel@standardbank.com