This is exactly what’s happening when you’re buying a synthetic CDO (if you’ve seen The Big Short, you know what they are, nothing to be worried about) and want to measure your counterparty credit risk.
For my new followers, counterparty credit risk (or CCR), is the risk you’re taking if you trade a derivative or a repo with someone who goes bust.
As in Credit Suisse and Archegos, for example.
Let’s go back in time a bit.
In 2014, the Basel committee decided that the standard approach to measure CCR was just crap.
So, they changed the rule and introduced a new methodology.
It took a while for the EU to adopt it but it was finally done in the CRR2 back in 2019.
One innovation was the use of the “supervisory delta”.
If you’re buying simple stuff (FX, shares, futures on bonds, etc.) then the delta is simple: if you’re long the risk, delta = 1, if you’re short the risk delta = -1.
But what about options?
Then, the way your risk will move over time is not as simple and you have to use a “delta” to reflect how your risk changes as the price of the underlying changes.
That’s the pretty standard delta formula.
Intuitively, the sign is correct, of course: if you buy a call or sell a put, you make money when the share price goes up, so delta is positive, & the opposite if you buy a put or sell a call.
But what about those CDO tranches mentioned above?
Very quickly, a CDO tranche is a bunch of credit risks that have been bundled, sold into pieces, with some investors taking the first 3% of credit losses, others taking the 3%-10%, etc.
So if you’re buying a tranche, your risk is in the same direction as buying a bond and you use the formula below.
E.g. If your tranche takes the 3%-10% losses your delta is 4.4.
But what does that “long protection” thing I underlined mean ?
Well, we’re talking CCR here, so it’s mostly about derivatives, which means our CDO is most probably a synthetic CDO, i.e. a CDO based not on actual bonds, but on Credit Default Swaps (CDS).
CDS are effectively traded insurance contracts against the risk of default of an issuer.
This is why, confusingly, when you “buy” a CDS, you buy an insurance protection, which mean you take a SHORT position on the credit.
BUYING CDS is very much like SELLING the bonds. Yes, this is confusing, and I know people who have worked 20 years in credit and are still very cautious about this confusion.
But, and this is where it gets confusing^2, the jargon on CDS indices (less so on CDS single name contracts) is often to say “long protection” to mean “long the underlying credit index” or “seller of protection”.
Yes, long protection here means that you’ve sold protection and this is why the delta is positive (since your risk goes in the same direction as the underlying bonds or as a cash CDO tranche).
Don’t worry, you’re not the only one totally confused.
Because that’s how the CRR transposed the Basel document.
It was good of the EC to make the language much clearer and give the sign of the delta for cases where the bank had sold protection and purchased protection...
But it got it wrong!
If “credit protection has been obtained” through the transaction, you are protected so you have, purchased an insurance, so you are SHORT the credit risk and delta should be -1!
Symmetrically, if you have provided protection, you receive a premium (the spread) and you lose money if the company defaults. You are LONG the credit risk and delta should be +1.
The EC got confused because of the sloppy wording of Basel and didn’t look at the economics – that the delta should go in the same direction as the delta of the underlying credit risk!
Amusingly, with the formula as written, you can buy from DB 8bn of risk exposure on the Crossover, and buy an extra 1bn of 0-6% equity piece on the same index … and the CRR tells you DB has no counterparty risk on you 😊
this sounds so terribly weird that I sincerely do hope I got it wrong - and honestly it's very possible given how confusing this is !
It feels like a big mystery about Credit Suisse remained largely unnoticed.
OK, they took a 4.4bn hit from Archegos… but how did they manage to lose *only* 900m in Q1 and wrongfoot analysts who had all estimated much lower capital ratios?
To understand the magnitude of the mystery: the Q1 consensus *before* the Greensill mess was around 1.4bn.
1.4bn-4.4bn=-3bn. Where the hell did the extra 2.1bn come from?! Not to mention a potential Greensill provision.
Let’s be super generous and assume no Greensill provision (hmmm) and a fabulous Q1 with PBT at 2bn.
We’re still missing 1.5bn. Not exactly small change. Where could this come from?
We have the same kind of institution in France. Its effectiveness is not great. Pretty much every application is approved unless reckless. But there was an extraordinary story once... let me tell it to you
This is the kind of institution that will routinely allow members of the ministry of finance to go work for a bank.
But there was this normal guy who was working as a tax inspector. And he had enough of it, he wanted a life change. So he quit and said he wanted to become...
A travelling salesman selling umbrellas ! And believe it or not, the Commission de deontologie (the french equivalent) refused ! Why, will you ask? Surely there is no lobbying or conflict of interest involved !
Why can a treasury guy who authorised big M&A deals can go to a bank