How does a Credit Default Swap (CDS) Work?

Credit Default Swaps (CDS) are fast becoming the dominant vehicle for trading credit risk. In this piece, I'll go over the basic features of a standard CDS contract and why they are easier for many traders to utilize over cash bonds.

At the onset of a CDS contract, the buyer of the contract agrees to pay a fixed spread to the seller of the contract. For corporate CDS, the spread is paid quarterly, so if the spread agreed to is 40bps, the seller would pay 10bps per quarter. In exchange, the seller agrees to buy a specified bond (or other instrument) from the buyer at par in the event of a default. Most CDS contracts have a five-year term, but other terms are possible. If CDS are quoted without a term specified, assume its five years.

A CDS is a lot like an insurance policy. This is why CDS are also called protection. The spread paid by the buyer is like the insurance premium. If there is a default, the buyer is essentially made whole because s/he gets par for the bonds. Just like if you have homeowners insurance and you have a fire, the insurance policy pays you whatever your stuff is worth.

The CDS contract references a specific bond (or bank loan). For example, the 5-year Alltel CDS references the AT 7% '12. In the event of a default, the seller will be buying some bond which is pari-passu with the reference bond. The buyer of protection doesn't have to actually own this bond. In fact, the buyer of Alltel protection might be a bank with which Alltel has a credit line. The bank knows that if Alltel gets into trouble that the credit line will be drawn down. But they also know that the CDS contract spread will widen substantially, and they will have a profit in the contract. If Alltel actually defaults, they can buy the bond in the secondary market at a steep discount, then sell it to the seller of protection at par and make a huge profit.

CDS are also a vehicle for speculating on a credit. The buyer of protection is essentially short the credit, while the seller is long. Buying protection may be easier than actually finding the bonds to short. Similarly, selling protection may allow one to get exposure to a credit with greater leverage than would otherwise be the case. CDS also have no interest rate exposure, so someone who wants get get long or short a credit can do so without needing to work about hedging credit risk on either the short or long side.

In fact, selling CDS protection in consort with owning a LIBOR floating asset is exactly like being long a 5-year FRN (if you ignore things like financing costs). Think about it, with the 5-year FRN, you'd get paid LIBOR plus some spread so long as the credit doesn't default. If it does default, you suffer the difference between par and the recovery rate. The CDS/LIBOR combination has exactly the same payout structure. For that matter, selling protection is also very much like buying a 5-year fixed corporate and heding with a 5-year LIBOR swap. You wind up just collecting the spread. For this reason, the CDS should have a similar spread as cash bonds when compared to LIBOR swaps.

In practice, CDS can divert from cash bonds materially for a couple reasons. First, the CDS may be deeper than the cash bonds, and in a fast moving market, the CDS may appear to lead cash bonds. In reality, this may be that the cash bonds aren't trading. More recently, we've seen CDS widen in LBO situations while the cash bonds tighten. This is normally because of covenants in the cash bonds which will result in a make-whole call. This post on the Equity Office Properties transaction describes this possibility in more detail. The same thing happened recently with First Data Corp's LBO. Even with companies rumored, like Alltel, to be possible LBO candidates and have attractive covenants, the CDS tend to under perform cash bonds.

Another reason is a cute little arbitrage involving discount priced bonds. Take Ryland 5.375% '15. This bond was issued in January of 2005 when the 10-year was 50bps lower, and troubles in the housing market has pushed the spread about 75bps wider. As a result, this bond has a dollar price of $92. So let's say I buy the bond and buy protection on it. If it defaults, I get paid par for my bond. I make 8 points. If it doesn't default, and the carry isn't negative, nothing happens. I get paid the yield on the bond, but probably gave up all the spread buying the CDS. So basically its a trade that has a high probability of doing nothing, but a small possibility of producing a nice return. Upside with no downside = arbitrage. Anyway, so CDS that reference discount priced bonds tend to be wider than those referencing premium priced bonds.

Feel free to comment if you have questions, or e-mail me at accruedint AT gmail.com.