One of the central premises of CDS is that the “basis” package should work. An investor should be able to buy a bond, and buy CDS to the same maturity and expect to get paid close to par – either by the bond being repaid at par and the CDS expiring worthless, or through a Credit Event, where the price of the bonds the investor owns plus the CDS settlement amount add up to close to par.
Thinking of the “basis” is useful in many ways as it focuses you on the similarities of cash positions and CDS. It also helps explain why some things are Credit Events and others are not. The Credit Event and what it took to trigger one (use of Collective Action Clauses) worked well.
The settlement of the CDS contracts worked well, but that was pure dumb luck. It should be fixed going forward, but does make playing the basis in sovereign credits a risky proposition.
Old Greek bonds were trading around 22% of par prior to being exchanged for a package of EFSF bonds and new Greek bonds. The CDS settlement wound up using 21.5%, so all seems good, but it really was just luck.
Typically the CDS gets settled against bonds that existed at the time of the Credit Event. In this case all the old Greek law bonds had already been exchanged. The only “old” bonds outstanding were those held by central banks (allegedly worth par) and some English law bonds that delayed their PSI until the end of this week. This situation is very unusual but it leaves “basis” package holders with a lot of uncertainty, as they own a package of assets from the exchange, but aren’t settling against that package of assets.
In this case, bondholders received 15 in EFSF bonds and 31.5 in new Greek bonds. The new Greek bonds were trading at about 22% of par, so the value of the package was 15 + 31.5% * 22 = 21.93
What if the package had been different? What if the terms had been better, and bondholders received 25 in EFSF bonds plus the 31.5% in new Greek bonds. Let’s assume the new Greek bonds still trade at 22% of par (they would actually likely trade lower since Greece has to pay the EFSF back). In this case the value of the exchange package would have been 25 + 31.5% * 22 = 31.93. But CDS would still have settled at 78.5% value since it was done purely on the value of the new Greek bonds, not on the value of the package. Basis holders would have made over 110 on the package in this case instead of the 100 they hope for.
On the other hand, what if the deal was more about bailing out Greece and less about helping the banks. Assume no EFSF money comes as part of the PSI, instead it is given to Greece to reduce debt, and let’s assume that the face amount of new Greek bonds was only 25%. That would be a much greater amount of debt reduction for Greece. It is highly likely that these new Greek bonds would trade higher than 22% (since there is so much less debt). Using 30% of par as the price in this case, the exchange package is worth 25% * 30 = 7.5 (a massive 92.5% NPV loss). The CDS would settle at 30 (so a payment of 70 points) and the basis package would have only paid out 77.5, a massive loss compared to the 100 expected.
If Greece had made the coupon higher, shortened the maturity, added GDP kickers so that the new bonds would trade close to 100, then a total disaster for CDS holders. The PSI package would still be worth only 25 points in this example 25% * 100. The CDS would have had zero value.
The outcome was completely dependent on the structure of the PSI and the trading price of the new bonds in a way that should scare basis holders, and possibly all CDS Buyers and Sellers. The outcome is bizarre because the value of the CDS settlement had almost NOTHING to do with the value of the PSI exchange package total value. Even in Greece, an amortizing bond rather than a series of bonds would have impacted settlement by at least 1% and had the GDP kicker been part of the bond rather than detachable that too would have impacted settlement. Had the EFSF contribution been built into the Greek bonds, it would still have qualified as Sovereign Restructured Obligation, but would have impacted CDS settlement dramatically.
It is hard to do much for old contract, but ISDA should work to correct this randomness. There are two fixes, one better than the other. ISDA could have the auction accelerated so it could be done before the old bonds are exchanged into new bonds. This is a bit problematic because the bond exchange could be done same day as CAC, and it creates more systematic risk as counterparties have less time to prepare their payment or close out trades that left them net flat.
The better solution is to have the auction on the value of the “exchanged” assets. If 100 of bonds got 15 of EFSF bonds and 31.5 of new bonds, then the auction should be on the value of 15 of EFSF bonds and 31.5 of new bonds, rather than on 100 of new bonds (which in this case, somewhat randomly, gave about the same answer). That would ensure that the “basis” package worked, but would also give buyers and sellers of CDS more comfort that the payout on CDS reflects the value of the old bonds.
I don’t think I would play in Portuguese CDS at 36 points up front. I am quite convinced Portugal will go down the path of a Greece and experience a Credit Event. So although I’m fairly certain Portugal will have a Credit Event, I would not be willing to bet that they don’t find a way to make the PSI come out in such a way that new bonds traded much closer to par, potentially negating the value of the CDS. So until ISDA changes, I think the risk/reward here is too large.
In Spain and Italy, where CDS trades at 410 and 355 respectively, the risk/reward is still in your favor for several reasons. The downside, even if there is a Credit Event with 100% recovery you lose far less. Italy and Spain are both so big, it will be harder to restructure in a way that makes the recovery so high – Portugal is small enough, a lot more could be done. Also, CDS trades in $’s, so although a Portuguese default might not have much impact on the FX rates, it is hard to imagine that fears of a Spanish or Italian Credit Event wouldn’t have a big negative impact on the currency. So I am less concerned about those, but would be trying to find borrows on longer dated Spanish bonds and play it simply as the CDS uncertainty is real, and potentially dangerous. I would be just as uncomfortable having sold CDS because the outcome could be equally dangerous where more of the value of the package you get would have created a much smaller loss than writing CDS.
These concerns are very specific to sovereign CDS and don’t look like they apply to corporate CDS.