The Greek CDS situation is sort of puzzling, but it's possible, and popular, to overstate its puzzlingness. We have probably been guilty of doing so in the past. In brief: if you hold Greek bonds, you sort of have to hand them over and get back other, shinier Greek bonds with half the face value. How sort of? The text of the statement is "we invite Greece, private investors and all parties concerned to develop a voluntary bond exchange with a nominal discount of 50% on notional Greek debt held by private investors," which is an attractive invitation although it does not exactly indicate that the party is occurring right now. But that sentence is code; it was negotiated by the banks' trade group and is a sort of quid pro quo for bank recaps and general regulatory approval so you'd expect most - not necessarily all - of the banks to be onside. The fact that the statement was released suggests that everyone thinks there are soft commitments to exchange from the banks holding the large majority of Greece's debt, though they've thought that before.

If everyone who holds Greek bonds does the exchange, then Greece never defaults. They just did a voluntary exchange. This presents a problem for Greek CDS: if there's no default, there's no credit event, and CDS never pays off even though bondholders lost 50% of principal. This is ISDA's official conclusion and it's just sort of self-evidently right, although some people disagree.

Felix Salmon sums up the general outrage:

[O]n one level, the ISDA statement that this still isn’t a Greek default, for CDS purposes, makes some sense. I’d probably make the same decision myself. But on the other hand, this does make a farce of the idea that credit default swaps constitute default protection, at least in the sovereign arena. If they don’t protect you against this, what earthly use are they?

Well, with most derivatives, it's important to remember that the marginal investor isn't buying them for payoff-at-maturity but for the market moves along the way. People equate CDS to insurance but it's not. If you buy life insurance, it pays out if you die and it doesn't if you don't; if you just decide to take up drunk cliff-diving you don't get any interim payment. Most CDS never pays out because defaults are rare, but it's still a healthy market. Most investors don't primarily care if CDS pays off when they crystallize a loss by handing in a bond in a pseudo-voluntary pseudo-default, because they're unlikely to do that. They care if their CDS mark goes up, in a realizable way, while their mark on the bond goes down. And it sort of does:

So you can exit your position today with a gain on the CDS that looks directionally like the loss on your debt.

Of course that argument shouldn't be taken too far, because it's a backwards induction argument: someone will buy it because they think (someone will buy it because they think [recurse]) it has a payout schedule corresponding in a predictable way to a possible reality. If the thing doesn't pay out when it's supposed to, the backward induction doesn't work. For someone to be willing to close out your CDS today, they have to think it's worth something - that it would pay out in the circumstances where it's supposed to pay out. And why would they think that?

Well, because it probably would. If you wanted to be a jerk, you could buy $1000 of Greek bonds and $1000 of CDS and wait. The voluntary exchange would happen, you'd ignore it, and your debt would come due. Then you'd hand it in and expect your 100 cents on the dollar. If Greece honored your debt, you'd get 100 cents on the dollar on the debt and zero on the CDS. If it didn't, then that definitely would be a credit event, and you'd get back X on the debt and 100 - X on the CDS. It is not entirely easy to predict which of these things would happen: if this purported exchange (or a future iteration of it) goes very well and leaves only a little stub of old debt, Greece will probably honor it; if not so much, then you'll be looking to your CDS. Because you have no idea which will happen, you buy both.

All that is obvious. But you have to be a jerk to do it, because think of all the Greeks who will face ouzo shortages due to your dastardly market manipulations. And in particular, you probably have to not be a big bank for whom European regulators and/or the IIF can make life unpleasant. Peter Tchir:

I would be unwinding basis packages for all sovereign debt. If you are at a bank or a bank hedging desk, I would be selling bonds/loans and selling protection. Everything you thought about CDS and how the hedges would work is potentially irrelevant.

But if you're selling that basis package, someone must be buying it, and the guy buying it is a jerk: a hedge fund who is willing to stare down the IIF, the EU, and Greek protesters and say "give me my money back or I will go complain to ISDA."

This is all fine and obvious and straightforward and actually probably not a reason to think that the Greek exchange will spell the end of sovereign CDS markets, because the Greek CDS market is still trading and basis is not horrific despite this basic structure of "voluntary" exchange having been on the table for months.

So why is it so much fun to freak out about this? Part of it is that it is sort of optically complicated: the ISDA determination process seems opaque, and is divorced from both economic reality and things like ratings agency determinations, which take a dimmer view of "voluntary" principal writedowns. Part of it is that the EU's determination not to trigger CDS does seem of a piece with their annoying determination to assault all markets, everywhere, out of an irrational hatred of speculators and anyone who would profit on Europe's incompetence, and in the aggregate that determination to shut down short selling, derivatives, etc. etc. probably will make European capital markets less useful.

But I think the freakout also reflects how counterintuitive it is that you need so few jerks to make CDS work. It seems likely that a large majority of Greek debt will be exchanged at fifty cents on the dollar (or less - this isn't done yet), by investors who will "voluntarily" give up a chance to call a default, collect on any derivatives they may have, and be made whole. That doesn't matter. They can offload those derivatives. And you only need a very few people to refuse: there's something under $4 billion of Greek CDS net notional outstanding, versus $245bn of Greek government debt, meaning that you'd need to get less than 2% of the debt market to make the simple scheme above work.* Everyone else can trade in CDS to their heart's content, before, during and after this exchange, because the expectation is that 2% of Greece's debt is likely to eventually find its way into the hands of investors who will stand on their contractual rights rather than chip in to help keep the European experiment alive.

This is counterintuitive but it's why people can go around talking about zero-arbitrage models and efficient markets with an almost straight face. You don't need everyone buying CDS to expect it to pay out, you just need a buyer of last resort who'll make it pay out. You don't need tons of short sellers to root out fraud, but you do need to allow short selling so that one or two clever and capitalized short sellers can bet against the frauds. You don't need all the buyers to think the price is right, just the marginal buyer.

Greek CDS "works" only in the limit case, only for a non-bank investor who's willing to be a jerk and run a certain amount of politico-PR risk. But that doesn't mean it mostly doesn't work. It means it entirely works.

* This sentence is loose: default on some de minimis amount of Greek debt would trigger all the Greek CDS in the universe; it's not tied to notional amounts. However, given delivery obligation auction mechanics, a de minimis default would screw with recovery values. In any case it should be clear that keeping $1 of Greek debt out of the exchange would be insufficient to protect CDS buyers and keeping $10bn out is unnecessary.