CCP Discounting big bang: delaying or not?

In a recent opinion, following to the QuantSummit panel where I was asked a similar question, I indicated that one of the first element I would like to see delayed due to the current crisis and the related emergencies is the CCP discounting big bang.

First, the big bang was not necessary. The original idea, as described in the ARRC paced transition plan, was to have a period with two parallel PAI world and transfer smoothly and over time between them. Later it was decided on the EUR and USD sides to skipped the "paced" part and do a big bang, at least at the CCP level.

The big bang make the transition easier for CCPs. One day they have one approach, the next day (actually 3 days later, as this is done on a week-end) they have another approach. At any moment, they have to deal with only one discounting curve. This is not the case for the users. They have the cleared trade to deal with, so the complexity for them is at least as high than for the CCPs, but also numerous bilateral agreements. For each of those bilateral agreements they need also a transition. At best all the transition will take place on the same date as the CCP big bang. Even in that unrealistic case, they have to deal with many transitions in parallel. But more likely not all the transition will take place at the same time, some may be done with a big bang on another date and some may be more paced. This means that in term of valuation and risk management each user need systems well more sophisticated than the one of the CCPs.

Another issue related to the discounting big bang is the swaption valuation. For swaption with settlement mechanism linked in one way or another to a CCP (cleared physical settlement or cash settlement with collateral discounting referencing a CCP), the change of CCP discounting has a serious impact on the valuation. Why we came to that problematic impact is twofold: 1) the swaptions contract were not very well written (this is about to change with ISDA adding required fields in confirmations) 2) the big bang was decided by CCP without taking the impact on external users into account.

In a recent Risk.Net article on "Discounting delay risks swaptions mess – Eurex" (subscription required), EUREX claims that the big bang should be kept in June because 1) swaption 2) not an issue for the CCP logistically.

Let me use some quotes and comment them:

[...]this would lead to complications for swaptions struck more recently that were priced assuming the current timeline would apply.
The plan for the discounting switch was announced in the fourth quarter of last year, so if people have executed trades they would have factored a value transfer into their pricing. If you now postpone the discounting switch, what does that do to the trades that they’ve already executed between now and the potential delay?

Andreas Franke

Let me reverse the claim. How did the CCP took into account the value of swaption already executed when deciding of the big-bang in the fourth quarter of last year? Were the complications for swaption struck before the big bang announcement that were priced assuming a EONIA discounting would apply taken into account? I found those argument very disingenuous coming from a CCP. In a previous post on Compensation (and lack thereof): swaptions, protocol and swaps, I used the above CCP argument as an irony (see note at the blog end); it seems that my banter was taken at face value.

Although many of our employees are working from home right now, the fact this switch was planned well in advance means it’s not an issue for us logistically speaking, so we’re almost at the point of no return and changing that date would likely increase the operational risks around the switch.

As discussed above, the transition is not an issue from a logistical point of view at CCPs because the issue is way easier for CCPs than for end users. For CCPs this is a one-off over a week-end with two unconnected worlds. For users it is a long term parallel world problem with subtle interaction between them and basis risk management issues. I'm sure it is likely to increase CCP operational risk, in case the date is hard-coded somewhere in the software, but for end users it may not be a "likely operational risk" but a "certain market risk" in term of valuation and permanent risk management.

The CCPs have the power to decide unilaterally to proceed, but the goodwill of the clearing users has already been tested recently with the LIBOR fallback and the value transfer implied by the CCP's decision to use the "at their sole discretion" part of the rules with its material value and risk changes to all legacy trades. Maybe they don't want to stretch that goodwill even further in those testing time.