Last week ISDA, who are in charge of credit default swaps, circulated some proposed changes to CDS to account for all the Greek, Argentine, SNS, everything unpleasantness. This prompted me to try out my one journalistic technique - calling1 ISDA and asking them to send me a copy - but they declined, so we'll just rely on this research note from JPMorgan's Saul Doctor and Danny White. Here's the gist:2
ISDA will publish a list of “Package Observable Bonds” (POBs) based on size, liquidity, maturity and governing law. The proposals suggest that there could be one domestic and one international law bond in each of the following silos – a) 1-3 years, b) 3-12 years, c) 12-30 years – based on a set of rules that determine the largest and most frequently traded bond in each silo. An initial POB will remain as such unless, prior to the Credit Event, it no longer meets the deliverability criteria, is called/matures, or is reduced below a threshold. New bonds would be added when a particular bucket is empty.
If a Credit Event occurs (Restructuring or other Credit Event) and a POB has been restructured into a package, then that package, in its entirety, will be deliverable into the auction. For example if a POB with a notional of $100m is written down by 50% and the remaining portion converted into 50 shares, then the 50 shares could be delivered against $100m of CDS. If there is more than one package on offer, then the one that has the highest subscribers will be chosen. All obligations meeting the deliverability criteria remain deliverable as long as they were issued prior to the Credit Event.
So lots of people have been calling for this for a long time - me least of all, but also real people like the Managed Funds Association and Darrell Duffie. But you get a sense from that summary of how it's more complicated than dopes like me think.
The basic weirdness of CDS is that it lets the protection buyer deliver any bond of the issuer to the protection seller, in exchange for the par amount.3 In some abstract sense this lets you buy $100mm of generic CDS to "insure" $100mm of any bond or bonds of the issuer that you like, without having your "insurance" match your bonds, which promotes greater liquidity and price discovery and usefulness in CDS. Concretely, though, it gives investors a cheapest-to-deliver option that can be quite valuable: in a straight-up bankruptcy a $100 30-year 2% bond might be worth as much, as a bankruptcy claim, as a $100 1-year 5% bond, but in most flavors of restructuring the 30-year will be worth much less. So an investor who is long the 1-year bond, hedged with CDS, may end up with a windfall by settling the CDS against the 30-year.
But that's life. My simplistic expectation would be that:
- Issuer has 26 bonds, Bond A through Bond Z.
- CDS on a payment default would settle by reference to the cheapest-to-deliver bond, say Bond Q.
- If the issuer restructures its bonds, then Bond A is restructured into Package A, Bond B into Package B, etc.
- CDS on a post-restructuring credit event would settle into the cheapest-to-deliver package, maybe Package Q or maybe some other Package - Package R say - because the packages change the relative values.
But that's not what ISDA does: they use the "Package Observable Bonds" definition to limit the potential packages to just Packages, like, A through F. And they do that before the restructuring even happens: you pick Bonds A through F as the POBs, because they're the most liquid, and then on any restructuring they're the package you look to.
The point of this, I suppose, is to limit the opportunities for investors and issuers to restructure into windfalls at the expense of CDS writers? You could imagine a variety of games otherwise, all of them more or less predicated on really really screwing one set of bonds so as to improve the CDS payoffs of holders of all the bonds. Hypothetically, a hedged holder of Bonds A through Z might prefer a restructuring that pays 60 cents on the dollar in restructured value to Bonds A through Y, and one cent to Bond Z, over a package that pays 70 cents to all of the bonds. (Because the hedged package in the first deal is worth 60 + 99 = 159 cents on the dollar for Package A, 159 cents for Package B, ... 1 + 99 = 100 cents on the dollar for Package Z; while the second deal is worth 70 + 30 = 100 cents on the dollar for each bond.) To some extent that's inherent in the cheapest-to-deliver concept, but you can understand why dealers and ISDA would be worried about adding opportunities for gamesmanship. Especially gamesmanship that would essentially allow sovereigns to pay bondholders with CDS writers' money.
It's not clear what the next steps are: as JPMorgan write, "Exactly how the new proposals are implemented – through a market wide protocol, or as a change to future trading standards – is still to be decided." Strangely, too, these reforms apply only to sovereign CDS.4 The MFA and Duffie proposals were also sovereign-only, and I suppose there's good reason to leave it at that. Corporates don't have the same ability as sovereigns to poof bonds into different bonds or things wholly other than bonds. (Corporate bonds, for instance, tend not to have collective action clauses that allow principal write-down or subordination/equityification of principal: if a corporate tried to do to you what Greece did to its bondholders, you'd just say no, and hang on to bonds that will either pay off at par or get an appropriate CDS payoff.) Corporates seem to cause fewer really knotty CDS headaches than sovereigns.
But not none! We've talked about SNS Reaal, the Dutch baank that was naationalized and whose subordinaated bonds were poofed into "nothing" or "claims" or "bonds, but you can't have them." Lisa Pollaack has a post today about SNS's CDS, which is a mess: basically, the sub bonds' CDS won't get paid off, because you can't deliver the sub bonds into the CDS auction, because the Dutch government has locked them up or evaporated them or whatever your preferred metaphor is. And ISDA's proposals, even if they were enacted tomorrow, would seem to do nothing for those sub bondholders. The modern, sovereign-flavored forms of credit defaults extend beyond just sovereign defaults, and continue to be a headache for CDS.
2.This exchange-property issue is the most interesting one to me. There are three other things being addressed: broadening the scope of Qualifying Guarantees that allow an obligation to be delivered into CDS to align with the cash market, amending successor provisions to capture the case where "Company A moves all debt to Company B and ceases to exist" and no one notices for 90 days (which sounds funny but happened to UnityMedia), and standardizing reference obligations so that all CDS on one Reference Entity will have a single Standard Reference Obligation.
3.Really CDS is cash settled by auctions, but I like to pretend that the old-timey physical settlement mechanics get you to the same place. That might be true. In any case they're easier to visualize.
4.As Doctor and White note:
Given the changing nature of Corporate and Financial debt instruments, we expect to see further proposed changes to the current Definitions, particularly with respect to Financial CDS and upcoming EU bail-in regime.