How gross and net CDS notionals really work | FT Alphaville

How gross and net CDS notionals really work

According to the DTCC, there is $75bn gross notional outstanding for CDS contracts referencing Greece. However, on a net basis this figure reduces substantially to $3.7bn.

What do these different numbers tell you, and how are they constructed?

Let’s look at a simple example where there are four banks engaging in seven CDS trades. Here, a positive number means the dealer is a buyer of CDS protection, and negative is a seller.

In the above, Bank A buys $12 CDS from Bank C, hence there is $12 of gross notional from that one trade. Looking at all seven trades, the total gross notional is in the system is $132, which you get by just adding up and not double-counting.

Provided the trades are fungible with each other (and the CDS market went through a huge standardisation a couple of years ago thus improving fungibility), then for any given bank, the buys and sells can be added up to get that bank’s overall net position. And since this is a contained universe, adding up all the net positions of all four banks gives zero.

The net position for just the overall buyers (11 + 44) or just the overall sellers (38 + 17) is $55. That $55 net number is the one that should be compared to the $132 gross.

If a credit event occurs, triggering payouts on the CDS, then an auction is held to determine the final settlement amount for all outstanding contracts. Whether a credit event has occurred is decided by the Isda determinations committee, who also decide what the deliverables (bonds) are for the auction.

The auction works by dealers submitting requests for physical settlement (i.e. actual bonds change hands), but the vast majority of trades will cash settle using the price at which the physical settlement orders were met. Here’s what the auction results for Allied Irish looked like:

If you had bought $10m of Allied Irish Senior CDS, you would have received a payout of $2.988m ($10m * 1-0.70125).

If the four banks in the table at the top were involved in this auction (and let’s assume that they were all senior CDS holders), then the total amount that would have exchanged hands would have been $55 * 1-0.70125 = $16.4. Notice how this substantially smaller than that gross figure of $132.

However, there’s one over-simplification and one nice technical point going on here, when it comes to extending the example out to CDS referencing Greece.

Gross, net, and counterparty risk

Let’s imagine Bank D had an insider trader whose unauthorised losses bought the place down. The table above would look like this instead:

All of Bank D’s trades are gone in the sense that either:

(a) They were closed out at fair value when Bank D was wiped out. This probably would have relied on collateral already being held by those counterparties of D that were in-the-money. It’s highly likely that they did have collateral.

(b) Bank D’s counterparties recklessly didn’t have collateral, or not enough, and are now queued up as an unsecured creditors to try to get a pay out.

If the latter scenario happened, Bank B would find that it wasn’t as hedged as it thought it was. If B was taking losses on say, Greek sovereign bonds, and its hedge has shrunk, than that might hurt a bit. At an extreme, it could hurt so much, that B itself runs into trouble. The dreaded contagion. Of course, Bank D was a relatively small player. Let’s knock out C instead.

This looks ugly! Looks like almost everyone was leaning on Bank C. Counterparty concentration risk if ever there was. Pick your own analogy:

1. C = CCP, creating an epicentre for counterparty risk to reside and the ultimate too big to fail.

2. C = AIG, a very popular counterparty with very few collateral posting requirements.

As you can see though, a counterparty going down exactly when they should be paying out (and receiving payments) is not so nice, to put it mildly. Let’s hope that no one was silly enough to buy protection on Greece from a Greek bank, cause that’d be one hell of a wrong-way risk!!

Old restructuring and maturity buckets in auctions

Get your geek hat on.

CDS come with different “restructuring clauses”, and these are agreed between counterparties at initiation of the trade. The clause determines whether certain types of restructuring count as a credit event and also have an effect on how auctions go down.

Most contracts that reference North American credits trade without any restructuring clause. This is a function of the prevalence of just filing for bankruptcy protection, which is the classic credit event. Restructuring clauses just aren’t needed in the US when companies can just file away.

In Europe, however, there are many jurisdictions and hence as many different takes on insolvency. Restructurings do happen and additionally, the credit that Basel II gives CDS hedges is greater for contracts that contain such clauses.

In a restructuring credit event, rather than having all the contracts lumped together to one auction result, the results are by maturity bucket. Here’s what that looks like:

Remember at the top in the first table we said that we had to assume the CDS were “fungible” to net them off against each other like we did?

Well, they are fungible if the credit event is a non-restructuring one, e.g. bankruptcy or failure to pay. But if the credit event is a restructuring then you’re going to have different buckets according to the tenor of the contract. Let’s do the same table again, but assuming there are four different contract maturities:

Here, Bank A goes from just being a $11 buyer to net $12 buyer at Tenor 1, $31 buyer at Tenor 2, and $32 seller at Tenor 4. Hence the amount settling in an auction that treats these tenors differently is going to have bigger cash settlement amounts.

Greece and other sovereign CDS

Western European sovereign CDS generally use the clause “old restructuring” (Old R). Under Old R, there aren’t maturity limits on deliverables, hence there would never be an auction with multiple maturity buckets.

If a credit event could be proven tomorrow and none of the current counterparties failed, the maximum amount that could settle for CDS referencing Greece is $3.7bn. Even that amount is an overestimate because it assumes a final price of zero being determined in the auction. If the final price was say, 50, then the total amount changing hands, on a net basis when all is said and done, would be $1.75bn.

The politics of uncertainty

How is it that such a small amount is causing such a huge distraction in political circles? There is already the naked short CDS ban coming in, but at least that’s written down on paper such that market participants can read it and prepare for it. And fine, if part or all of the CDS market is to be completely banned or whatever, then go to it. Come up with a proposal, do a request for comments, and go enact it with a timeline attached. Go forth and debate. Bonus points for looking at things like evidence and weighing up different viewpoints.

In the meantime though, please stop interfering by way of introducing sheer uncertainty into the market. The market has plenty of it without over-engineered debt restructurings and bailouts that come too late.

And take a moment to think of the quants.. Do you think they were sitting there a couple of years ago with the sovereign CDS models going, “probability of default – check, liquidity risk – check, interest rate risk – check, recovery rate assumption – check, political risk that a credit event will be engineered around for no apparent reason even though Basel II gives us relief for having CDS positions – check.”?

No way, mate.

(At which point we can only point you to Macro Man’s elegy for developed-world sovereign CDS. Very moving.)

Related links:
Confusion over Greek CDS continues – IFR
Bankers fear political moves will kill off CDS – FT
Who wants to be a Greek bond holdout? – FT Alphaville