Confusion in cross-currency fallback! What about basis swap confusion?
Risk.Net published a recent article about the confusion related to the fallbacks in the cross-currency swap market. The article is "Cross-currency confusion stalks FCA announcements" (subscription required).
Confusion and winner and loser have been the common component of the fallback process up to now. I certainly don't disagree with the article but as I have said in a similar context recently, I'm surprised by the surprise.
One of the elements put forward in the article is "opposing legs of a cross-currency swap could be fixed up to a year apart". One of the main cross-currency market is USD-LIBOR/EUR-EURIBOR, and in that case, the opposite legs could fix 5 or more years apart. In the cross-currency case, I'm not impressed by one year.
In the single currency case, opposite legs of basis swaps may fix at the same time, but will fix on data from different periods. The periods will be obtained with the most recent 5 year period for which the ON in arrears is fully known. This means that the LIBOR fixing dates will start for LIBOR-1M 5Y and 1M in the past and for LIBOR-12M 6Y in the past . The computation is based on a 5Y look-back; this means that the fixing periods for the above tenors on the LIBOR side have a 4Y1M overlap and a 11M non-overlap. The non-overlap is around 18% of the period, not something negligible and not qualitatively different from the issue mentioned on the cross-currency case.
If as hinted by the FCA, the announcement is done at the end of 2020, the 1M computation will include the recent Covid-19 credit crisis. The 12M computation will not include that period for the LIBOR fixing side but will include the rate cut on the ON fixing side. That means strong distortion on the rates and significant discrepancies between tenors. I will try to add some graph of the data comparison between the different tenors in the coming days (time permitting).
Note that the offset periods on the LIBOR fixing is an arbitrary choice, one could have fixed on the same LIBOR period for all tenors with the last date the day before the announcement. That would have meant a known spread adjustment known at the end of the first period where it is used. As in-arrear fixing supporters pretend that an in-arrear fixing is not a problem for the RFR, I'm sure they would agree that this is not a problem either for the spread adjustment. Especially that this issue will appear only once for the spread adjustment and for the next 50 years for the RFR in-arrears fixing. The choice of lock-back period is to my opinion an unfortunate choice for the LIBOR basis market. I know that the LIBOR basis market will die with LIBOR, but it would have been courteous to let this long serving member die peacefully.
Confusion and winner and loser have been the common component of the fallback process up to now. I certainly don't disagree with the article but as I have said in a similar context recently, I'm surprised by the surprise.
One of the elements put forward in the article is "opposing legs of a cross-currency swap could be fixed up to a year apart". One of the main cross-currency market is USD-LIBOR/EUR-EURIBOR, and in that case, the opposite legs could fix 5 or more years apart. In the cross-currency case, I'm not impressed by one year.
In the single currency case, opposite legs of basis swaps may fix at the same time, but will fix on data from different periods. The periods will be obtained with the most recent 5 year period for which the ON in arrears is fully known. This means that the LIBOR fixing dates will start for LIBOR-1M 5Y and 1M in the past and for LIBOR-12M 6Y in the past . The computation is based on a 5Y look-back; this means that the fixing periods for the above tenors on the LIBOR side have a 4Y1M overlap and a 11M non-overlap. The non-overlap is around 18% of the period, not something negligible and not qualitatively different from the issue mentioned on the cross-currency case.
If as hinted by the FCA, the announcement is done at the end of 2020, the 1M computation will include the recent Covid-19 credit crisis. The 12M computation will not include that period for the LIBOR fixing side but will include the rate cut on the ON fixing side. That means strong distortion on the rates and significant discrepancies between tenors. I will try to add some graph of the data comparison between the different tenors in the coming days (time permitting).
Note that the offset periods on the LIBOR fixing is an arbitrary choice, one could have fixed on the same LIBOR period for all tenors with the last date the day before the announcement. That would have meant a known spread adjustment known at the end of the first period where it is used. As in-arrear fixing supporters pretend that an in-arrear fixing is not a problem for the RFR, I'm sure they would agree that this is not a problem either for the spread adjustment. Especially that this issue will appear only once for the spread adjustment and for the next 50 years for the RFR in-arrears fixing. The choice of lock-back period is to my opinion an unfortunate choice for the LIBOR basis market. I know that the LIBOR basis market will die with LIBOR, but it would have been courteous to let this long serving member die peacefully.
Comments
Post a Comment