OK, here's another question:For a single-currency swap, does the choice of discounting curve depend on the currency of the collateral posted on that swap?Say for example I have a straightforward, single currency, vanilla EUR IRS with counterparty A. Let's also assume that my CSA agreement with counterparty A specifies that all collateral should be posted in USD. My question is this: does this affect my choice of curve for discounting the cash-flows on the swap, i.e. should I use the domestic EONIA curve (as normal), or a cross-currency curve (in order to bring it back to Fed Funds)?I used to think that the answer was yes, but now I'm starting to think that it's no. A discussion with anyone who has experience on this subject would be very useful.Ta

Hi thedoc,Were you able to learn anything more about your last question?I always assumed that for sure for a swap where the collateral is in USD for example we have to consider cross-currency basis and dollar ois curve or swap curve (depending on whether we are using ois or libor discounting).In yen where the basis well before 2008 was always significant for long-dated stuff especially, and dollar collateral was standard, I think some banks used to have a model consistently pricing cross-currency, single-currency and FX forwards, with libor discounting.

This paper and other written by those authors should solve your questions.

QuoteOriginally posted by: gammaslideHi thedoc,Were you able to learn anything more about your last question?I always assumed that for sure for a swap where the collateral is in USD for example we have to consider cross-currency basis and dollar ois curve or swap curve (depending on whether we are using ois or libor discounting).In yen where the basis well before 2008 was always significant for long-dated stuff especially, and dollar collateral was standard, I think some banks used to have a model consistently pricing cross-currency, single-currency and FX forwards, with libor discounting.I thought about it some more, and using a xccy basis is definitely the correct method.The doubts I had were over what happens in the case where the counterparty defaults. In that case, (naively) you could simply convert the USD collateral into EUR via an FX trade, and this would appear to validate the use of a non-xccy curve for discounting (i.e. EONIA in theis case). However of course, this argument is flawed. When the collateral and swap are in different currencies, then you effectively have xccy risk (which ideally you would hedge, but in practice you wouldn't), and when you factor this in, the FX argument doesn't hold water.Thanks

Does it really make sense to combine a cross-currency basis swap spread (which is derived from Libor rates) with an OIS curve?