How all CDS are at risk of not paying out
My wonky post last week on how Greece’s default could kill the sovereign CDS market turns out to have been surprisingly popular, especially among policymakers who are worried about whether there’s a serious flaw in the CDS architecture. So today I had a fascinating chat with David Geen, ISDA’s general counsel, to double-check whether there was something I was missing. And it turns out that I was wrong: I wasn’t pessimistic enough. The problem I identified with Greece’s default isn’t just a problem for sovereign CDS: it’s a problem for all CDS.
At heart, the problem is what happens when an issuer swaps out all of its bonds for some new bonds. There’s no reason at all why the new bonds should trade at a massive discount to par — indeed, issuers often like it when their new bonds trade at or near 100 cents on the dollar. But if the CDS auction happens after the bond exchange, and if all of the old bonds are exchanged, then holders of the new bonds are forced to tender new bonds into the exchange, even if they’re trading at 100 cents on the dollar. Which means that holders of old bonds could suffer a huge haircut in the value of their bonds, but still get no payout from their CDS.
This has been an issue in the past. When Anglo Irish Bank restructured its bonds, it amended the old bonds to include a call option which allowed the bank to buy back every €1,000 of bonds for €0.01. That was an effective way of wiping out the value of the old bonds — but it also risked serious damage to the CDS market, since in a CDS auction, the value of a bond is calculated as the price of the bond considered as a percentage of the outstanding principal — and the outstanding principal is considered to be not the face value of the bond but rather the amount of the call option. If Anglo Irish had done the exchange quickly, before a CDS auction was possible, then bondholders would have had to tender bonds with a call option at €0.01 — which would mean that they couldn’t claim any payout on their CDS at all.
In the end, Anglo Irish took pity on the CDS holders, and staggered its restructuring so that there was enough time for ISDA to conduct an auction before the bonds got changed out of all recognition. But hoping that the issuer will act in a friendly manner is not exactly an optimum strategy — especially since, by definition, the issuer will be in the process of going bust.
This is a known issue. In a primer on sovereign state restructurings and credit default swaps dated October 2011, ISDA’s own counsel, Allen & Overy, said as much on page 20.
If an issuer has inserted a right to call at 20%, the outstanding principal would be considered to be 20% of the face value: if it were likely that the call would be exercised, the current value of a bond with a face value of USD100 would be approximately USD20, ie close to 100% of the outstanding principal. If an auction final price was based on this bond, a buyer of protection would receive only a marginal payment in settlement of its credit default swap, rather than a payment reflecting the full diminution in value caused by the restructuring.
The buyer of protection would also lose out in the more straightforward case where all deliverable obligations are redeemed prior to settlement.
This seems unsatisfactory at first blush, particularly as it is effectively the very thing for which protection is bought (the restructuring event) which thwarts the buyer.
This doesn’t just seem unsatisfactory at first blush; it is unsatisfactory. And there is no second blush. Essentially, CDS holders are reduced to hoping that the issuer will be nice, and structure the exchange in such a way as to let them get paid out. But there’s no particular reason why the issuer should do that, especially seeing as how the CDS holders were the people who were effectively shorting the issuer as it tumbled into bankruptcy.
Nearly all issuers in Europe have collective action clauses in their bonds: any of them could ask their bondholders to agree to change the payment terms on all bonds so that call options were introduced or principal amounts reduced. If a supermajority of bondholders did that, then the issuer could change the payment terms immediately, before a CDS auction could be held, and buyers of protection could find that CDS protection to be worthless.
In the past, ISDA has found a slightly kludgy way to deal with this. If you look at the documentation for the CIT auction in 2009, you’ll find this piece of crystalline prose:
If, after the date that is two Business Days after the Notice of Physical Settlement Date, but prior to Delivery, a Deliverable Obligation specified in the Notice of Physical Settlement (or NOPS Amendment Notice, as the case may be) is redeemed and/or cancelled in whole or in part (either in accordance with the terms of the relevant Deliverable Obligation, or otherwise) and, in connection with such redemption and/or cancellation, holders of the Deliverable Obligation receive cash, securities, rights and/or other assets (whether tangible or otherwise) (in each case, whether of the relevant Reference Entity or of a third party) (together, the “Assets”) (such event, an “Asset Exchange”) then, notwithstanding Section 3.4, Buyer’s right to Deliver each relevant Deliverable Obligation shall, to the extent it is the subject of an Asset Exchange, be replaced by a right to Deliver an amount of Relevant Assets equal to, or greater than, but not less than, the Asset Entitlement Amount (together with any part of the Deliverable Obligation that has not been the subject of such Asset Exchange).
I’ll translate that into English for you: the auction might happen after your bonds have been exchanged into something else, which we’ll call Relevant Assets. If that happens, then you’re allowed to consider whatever Relevant Assets you got to be deliverables as far as the auction is concerned. Even if the Relevant Assets are equity rather than debt.
But there’s no indication that this kind of language is going to appear with respect to the Greek restructuring. And there’s no reason for CDS holders to believe that it will appear when bond exchanges happen. And there’s certainly nothing in the existing CDS boilerplate indicating that any such language should ever appear.
If you own protection on a credit, then, you’re very much in a world of caveat emptor. You can trade in and out of CDS and make a good living; these things are, first and foremost, trading vehicles. That’s why they’re more liquid than bonds. But if you have a strategy which involves actually getting paid out on your CDS in the event of default, then you should definitely worry that the payout might not happen, even if the event of default is clear and declared. What’s more, there’s really no good way to hedge that risk.
So far, the CDS market has managed to muddle through, when it comes to restructurings and bond exchanges. But it’s sure to blow up sooner or later. And I’m far from convinced that ISDA and Allen & Overy are capable of reworking the standard documentation to make it more robust to these risks.