I've already commented on this,

http://www.noelwatson.com/blog/PermaLink,guid,ca8f0369-3933-4816-a136-b850c306dab1.aspx

but an article in the Independent got me thinking how much the market will be moved in the event of one or more CPDOs unwinding

My initial thoughts were that the unwinding shouldn't move the market a great deal as if this were the case the CPDO would not be a viable product. My reasoning being that if buying large amount of an index moved the market that much, then the twice yearly index rolls, where the CPDOs sell protection on the new series and buy protection on the prior series (to net out their position from the sale six months previous), the costs would be too great. However, looking at the ABN CPDO article

SurfCPDO.pdf (755.98 KB)

they assume a roll cost of only 1bp (see page 29), with no mention of market movements due to large volumes. After some investigation this makes sense. When the CPDO managers trades with a counterparty (typically a flow desk within the same bank as the CPDO structurer) during the roll, the counterparty's net exposure does not change a great deal. They are selling 4.75 year index protection and buying 5.25 year exposure. The index constituents between the two series do not typically change a great deal meaning that the counterparty will not be dramatically changing his risk exposure.

Contrast this with a CPDO unwind where lots of credit protection is sought with no corresponding  buying, and you can see where there may be some trouble ahead. Having said that, it would be good to know how large the CPDO market is in comparison to the amount of Series 8 ITRAXX Europe/CDX issuance to see how large the potential scale of the problem may be - this article thinks $25 billion for CDX and ITRAXX

http://www.reuters.com/article/bondsNews/idUSN2827508620080228