From Matt Levine's "Money Stuff" email. What I have done here will not qualify for the Fair Use Doctrine and there is no attempt at plagiarism here, I repeat "This is Matt Levine's writing"; if requested by Bloomberg, I will delete the post. Did I mention that Matt Levine wrote this?
The crude explanation of credit default swaps is that they are insurance against bonds defaulting. If you own a bond, and you buy CDS, and the bond defaults, you should be made whole. If it’s a $100 bond, you should get $100 back, no matter what. A bond plus CDS should be a (credit-)risk-free combination.
It is a bit complicated to get this to work in practice. Defaulted bonds are rarely completely worthless. The issuer restructures, in or out of bankruptcy, and the defaulted bonds get paid off in part or exchanged for some new securities. If you get back $20 or $40 or $75 of stuff for your defaulted bonds, CDS would ideally pay off $80 or $60 or $25, so that you end up with $100 for the bond-plus-CDS package.
One way to make this work would be: If a bond defaults, the CDS buyer delivers the bond to the CDS writer, and the CDS writer hands the buyer $100. The buyer is made whole—it has exactly $100—and the CDS writer has to deal with figuring out how much the bond is worth. This is called “physical settlement,” and it’s pretty straightforward—as long as CDS is bought mainly by bond owners as insurance. But of course CDS is used for lots of other purposes too. People will buy CDS on a company not only because they own that company’s bonds, but also as a way to speculate on its credit, or as a hedge for other credit exposures to the company (or its stock, etc.), or as a hedge for other credit exposures to different but correlated companies. People will buy index CDS—a pot of CDS on a bunch of bonds, which pays off a bit if any of those bonds default—as a way to hedge or speculate on general credit risk. CDS is in a narrow sense “insurance against bonds defaulting,” but in a broader sense it has been, historically, the main liquid way to make all sorts of bets on credit.
What this means is that a lot of people own CDS without owning the underlying bonds that they could deliver for physical settlement. Still you could ignore that. You could say “anyone who buys CDS has to deliver the bond to get back $100.” Then people who own CDS but not the relevant bonds would have to buy the bonds, after the default, in order to get paid off on their CDS. In some cases, this would drive the price of the bonds way up after default. Before default, everyone says “this company may not pay these bonds back,” so they trade at $60. After default, all the CDS owners say “hey we need to get some bonds to get paid off on our CDS,” so they go out looking for bonds and bid them up to $70 or $90 or, why not, $99.99. (Probably not $105, because the CDS only pays off $100.)
This is not exactly the system that is used in reality, but it’s not exactly not that system either.
Another way for it to work would be: If a bond defaults, the CDS writer pays a lump of cash to the CDS buyer, and no bonds change hands. This is called “cash settlement.” Ideally the cash settlement would be $100 minus exactly what the bond is worth after the default, but how do you measure that?
You could wait until the restructuring is completed, see what bondholders get, and then have the CDS pay off $100 minus that. This is administratively annoying: Restructurings take a long time, lots of bondholders can’t or don’t want to hold their defaulted bonds all the way through bankruptcy, and if bondholders get back other securities you still have the problem of valuing them. In practice people want CDS to pay out more promptly than “after the bankruptcy is over,” so this approach isn’t really used.
You could have an arbitrary payout. You could have a rule that CDS pays $100 on a default, no matter what the bonds’ recovery is. (Or you could have it pay $60 on a default, on the old rule of thumb that recoveries are typically around 40 cents on the dollar.) That would be … fine, really; it might be a good product for people looking to speculate on credit generally. People suggest it a lot, and it does exist (“binary” or “fixed-recovery” CDS), though it is not that common. The problem is that it doesn’t work that well for people who do want insurance against default on actual bonds that they own; they would be overpaying for their insurance, buying something that pays out $100 even if their loss is only $60.
So the usual intuition is that the cash-settlement payout should be based on the market price of the bonds shortly after the default. If the company defaults, and the bonds trade down to $45, then the CDS should pay out $55. That way if you have a bond and CDS, you get back $100, $45 for your bond and $55 for your CDS. But it’s not always easy to measure the market price on a defaulted bond; it may not trade very much, people may have idiosyncratic reasons for holding on to or buying or selling it, and trading prices might not reflect “real” value.
So the way normal CDS works is that there is an auction. After a default, ISDA—the International Swaps and Derivatives Association, which administers CDS—declares a default and sets an auction date, and then there is a bond auction that does two things. One, it allows for physical settlement: If you own a bond and CDS, and want to get rid of your bond and get back $100 on your CDS, you can do that (by selling your bonds in the auction); if you wrote CDS and want to pay $100 and get back a bond, you can do that (by buying bonds in the auction). Two, it fixes a price for cash settlement: The auction coordinates liquidity, so that anyone who wants to buy or sell the bond (whether or not they own or wrote CDS) can do so, so you get a good market-clearing price for the bonds, which you can use to set the payoff for CDS cash settlement. If the market-clearing price for the bonds is $45, CDS will cash settle at $55.
This is all oversimplified and the actual auction mechanics are kind of complicated, but this is good enough for our purposes.
Once you know the basic structure you can think about how to manipulate it. There are two basic ways: You can try to make CDS pay off too much (because you own CDS), or too little (because you wrote CDS). If you want CDS to pay off too much, what you need is for the auction to clear at a very low price. If you are just a CDS owner, this is hard to do on your own; people who own defaulted bonds are going to want to sell them in the auction. But if you can enlist the company to help you, you’ve got a shot. Basically the method is to have the company issue some new bonds that are deliverable into the CDS auction and that are worth very little money. This will drive down the clearing price of the auction—which usually reflects the cheapest-to-deliver bonds—and thus drive up the CDS payout.
The classic case here is the Hovnanian trade, where Hovnanian Enterprises Inc. struck a deal with a hedge fund to (1) default on some debt for a bit and (2) also issue some new debt with comical terms that would trade at a very low price, lowering the auction price and so increasing the payoff on CDS. The hedge fund owned a lot of CDS, so it would get a large payoff and use some of it to give Hovnanian attractive financing. This was a clever plan, though it was ultimately abandonedbecause everyone got really mad.
If you want CDS to pay off too little, what you need is for the auction to clear at a very high price, ideally $100. If you are a CDS writer, this is not especially hard to do; all you have to do is buy up all the bonds so that no one can put them up for auction. Just corner the market in the bonds. Of course then you have spent a lot of money buying back the bonds for more than they’re worth, which defeats the purpose of having CDS pay off too little, so you will probably prefer a subtler approach. For instance, sometimes there’s one relatively low-value bond that will set the price of the CDS auction, and other higher-value bonds that will lead to a lower CDS payoff; if you can corner the market in the first bond, perhaps you can save more money on your CDS than you spend on the corner. (This is sort of the Sears Holdings Corp. trade, though there was a lot going on there.) Or sometimes a company will have more CDS outstanding than bonds—perhaps because it is in a lot of credit indexes, but has paid back a lot of debt[3]—so you can corner the bonds relatively cheaply.
I say “manipulate,” but you can get these weird scenarios without any intentional manipulation. Just by accident a company can have one weird bond that is worth very little and that sets the price of the auction too low, or just by accident its bonds can be scarce and the auction price can be too high.
Anyway. Europcar Mobility Group defaulted on its debt in December, failing to make an interest payment, as part of a negotiated restructuring deal. Its CDS was triggered, and on Wednesday there was a settlement auction. The bonds seem to be worth something in the neighborhood of 70 cents on the dollar; that is, that’s roughly the value they expect to get in the restructuring. So you might expect CDS to pay out around 30 cents on the dollar.
But! Back in October, as the restructuring was being negotiated (but before the default), it looked like there might be a weird quirk where the CDS paid too much. From Bloomberg:
An anomaly in credit insurance on Europcar Mobility Group could prove lucrative for some traders as the French rental-car firm seeks to restructure 1.3 billion euros ($1.5 billion) of debt. …
Investors are betting a payout on Europcar would be higher than normal because of an unusual feature -- the contracts reference a loan, as well as the company’s bonds. A 50 million-euro loan from Credit Suisse Group AG last year was designed to be deliverable into credit swaps and traders are betting that will have a lower recovery value.
“For some people, this will have been a superb trade,” said Munich-based Felsenheimer, who saw the package quoted at about 100% of face value a couple of months ago. “It’s a very special situation because of the loan being deliverable and cheaper to deliver than the bonds.”
But it turned out not to be deliverable. And in the event, there turned out to be a weird quirk where the CDS paid too little. Specifically, zero. From the Financial Times yesterday:
A technical quirk has rendered derivatives linked to the defaulted debt of a French rental car company worthless, in the latest embarrassing incident to hit the $9tn credit default swaps market.
Buyers of credit default swaps (CDS) on Europcar were expecting a payout to compensate them for the company failing to honour its debt. But an auction held on Wednesday to determine the level of award left them with nothing.
The farce marks the latest in a string of mishaps that have undermined confidence in CDS — derivatives that are intended to act like insurance against a borrower reneging on its debts.
Part of the problem here was that a lot of the relevant bonds were, not cornered exactly, but locked up:
The problem stemmed from the terms of Europcar’s restructuring, which meant that most holders of its €1bn bonds were restricted from trading, preventing the debt from being included in the crucial auction that determined the swap payout.
If nobody who owns the bonds can sell them—because they’ve agreed not to, as part of their deal to get a recovery in the restructuring—then there won’t be any bonds available at auction, so the auction price will be 100,[4] so the CDS will pay zero. (Which is unfortunate for some bondholders who agreed not to sell the bonds, but who had CDS that they hoped would compensate for their losses.)
But that isn’t quite the whole story; there were actually enough bonds available outside of the lockups. The problem is that … investors … forgot? A JPMorgan Europe Credit Research note yesterday noted that “even accounting for the lock-ups, the deliverable pool was seemingly large enough to cover potential net interest in the Auction,” and pointed out:
The outcome is a missed opportunity for market participants given that anybody can participate in the second of round of the auction (regardless of whether they had a CDS position or not) and submit (through a participating dealer) as many limit orders in the direction of the open interest as they wish. … Given that the company is seemingly offering investors a package valued somewhere in the high 60s, any market participant could have offered the bond into the auction at an elevated price (even significantly above par). ...
If you owned one of these bonds, worth about $70, you could have put in a limit order to sell it in the CDS auction for $99.99, or $120 for that matter. What would it hurt? But not enough people did. The consensus seems to be that there was nothing nefarious; they just assumed that the CDS auction would clear at a normal price—the preliminary indicative price in the initial round of the auction was $73—and didn’t bother submitting their bonds. From the FT again:
While there was still more than enough debt outstanding to settle the swaps properly, not enough holders ultimately sold their bonds into the auction.
“This is an inherent threat in the auction that we always reminded people of,” said Athanassios Diplas, of Diplas Advisors, a former banker regarded as one of the architects of the modern CDS market.
“You cannot rely on someone else to provide liquidity in the auction.”
I don’t know what to tell you. A lot of finance is built around sort of no-arbitrage assumptions: “If X happened, then people could make a risk-free profit, so they’d flock to the opportunity until X did not happen.” In liquid continuous markets those assumptions often more or less work. But sometimes people just forget to take their risk-free profits, and the system breaks down a bit.