AIG and the “Negative Basis”
In answer to a question.
Negative basis means that the CDS yields less than the underlying bond/CDO tranche. For exemple if the bond yields Libor + x bp, the CDS quote is lower than x.
According to AIG Form 8-K:
…during AIG’s December 5 Investor Conference, representatives of AIGFP indicated that the estimate of the decline in fair value of AIGFP’s super senior credit default swap portfolio during November was then being determined on the basis of cash bond prices for securities in the underlying collateral pools, with valuation adjustments made not only for the cash flow diversion features referred to above but also for “negative basis”, to reflect the amount attributable to the difference (the “spread differential”) between spreads implied from cash CDO prices and credit spreads implied from the pricing of credit default swaps on the CDOs.
In my view, AIG has no business adjusting for the “negative basis”, reason being that they don’t know which “price” is right and they aren’t in a “negative basis” trade but plain short protection on super senior swaps. The real problem is how to value these “things.”
The price of a CDS is always uncertain. There is no single answer, CDS trade on “spreads” but the price depends on models and inputs. To quote a famous textbook exemple, suppose that you bought protection on Marconi at 250 bp sometimes in the late 90s, in 2001 the spread had widened to 4000 bp. You are rich or so you think. The unwind price depends entirely on recovery assumptions, at 30% recovery you would get 51% of par, at 99% you would get 1% of par,that is you would “lose” even though you got the scenario right. No prize for guessing your dealer counterparty quote. Hope that helps.
Related :
AIG’s Arbitrary Write-Down
AIG’s subprime hit
AIG’s revised subprime losses raise fears of tough audits
AIG puts two-month subprime loss at $5bn
Added 15:00 gmt
AIG says potential derivatives loss not material
Of course, mark to market losses aren’t the same than losses due to defaults and have no predictive value as to what final losses will be, if any.
February 12th, 2008 at 6:28 am
Negative Basis .. is where you buy the bonds and then buy protection… so using your example of LIBOR + X on the bond… the CDS quote (i.e. where you can buy protection) is LOWER than X (let’s call that Y) .. so that means you can own a bond, and own protection on that same bond .. with positive carry which is X - Y
February 12th, 2008 at 6:51 am
Jason:
Correct and corrected, thx.
February 12th, 2008 at 12:21 pm
If they’re short CDS and premium falls, shouldn’t they be marking greater profits assuming recovery rate doesn’t change? Everything else is clear to me
February 12th, 2008 at 12:47 pm
Well, the premium aren’t falling at the moment, they are going through the roof…that’s the problem.
February 12th, 2008 at 5:06 pm
How are recovery rates assumptions independent of spreads? I don’t understand how a trader long protection at 250bps doesn’t make money by closing it out by shorting at 4000bps.
February 12th, 2008 at 5:30 pm
Shorting doesn’t close the position. The exemple given is about a long protection buyer who wanted to unwind [close] the position rather than shorting which is a new position. A dealer could have quoted for an unwind anything between 1% and over 50% of par depending on his view of recovery. At 99% recovery, he would have paid 1% of par to unwind and the protection buyer would have lost since he paid 250 bp/year for 100 bp final pay-off. This didn’t happen, it’s just an exemple to show that the value of a CDS is dependent on models and inputs and that with a given spread there is no such thing as a unique solution for pricing a CDS.
In any case Marconi went bust and recovery was about 30%, so the buyer could have gone buy the bonds at recovery price of 30 and deliver to get par so he made money.
February 13th, 2008 at 5:48 am
ed:
I would add that coupons are paid in arrears so that if you were to short at 4000 bp, you could end up having to pay (1-recovery) without having received a single premium if default happens before the first coupon is due. This is not possible any more, if a company is under threat of imminent default, CDS will trade at a fixed coupon usually 500 bp plus an upfront fee.
February 14th, 2008 at 2:06 pm
jck: Sure, I understand premiums are rising but here are the two prices they’re looking at: X = “fair value” assuming no negative basis, and X-A = market price. They are marking to market and then stating that these writedowns wouldnt occur if the adjusted for negative basis (marked to X instead of market). If they’re short protection and X > X-A, how can they possibly show greater MTM by marking to X?
Obviously I’m misunderstanding something
February 14th, 2008 at 3:29 pm
Gary:
There is no market price, it’s all model derived. They were using implied spreads on the cash CDOs and implied spreads on the CDS of the CDOs to derive a negative basis which has the benefit of reducing x the super senior cds premium to something < x , hence the benefit of spread differential.
If you look at note 6, they start their modelling from the asset side of the CDO to get to spreads on the tranches. There is huge model risk here particularly on the correlation side.
May 13th, 2008 at 7:53 am
[...] For a spread, there is more than one price and for a price there is more than one spread. The price of a CDS is always uncertain. There is no single answer, CDS trade on “spreads” but the price depends on models and inputs. [...]