Prudence and counterparty risk

Robert Peston raises the interesting issue of whether European banks (and Goldman Sachs) were taking high risks in dealing with AIG. Goldman, Barclays, Society Generale and Deutsche Bank have each received between $8bn and $13bn since the US government's AIG bailout.

Peston takes this as some indication that these banks were not being managed prudently - their reliance on AIG being a risky one.

But the main complaint against AIG is that it was selling credit default swaps and other financial insurance products while not having enough capital to cover them, and underpricing the risk of default. If this is the case, then it might have been perfectly legitimate for these banks to buy the insurance, even factoring in the risk that AIG would default.

A large proportion of the payments received by Goldman (Peston doesn't break down the payments to the other banks) are CDS-related - indicating that their insurance policy paid off. Even if it hadn't, they would not have "lost" the $13 billion, but only the much smaller amount paid as premium on the CDS.

It's also quite possible that the banks gambled that AIG would not be allowed to fail by the US authorities. If so, it looks like they have been proved right.

Of course you can't prove that a bet was a good one by the fact that it was successful. But in this case the indications don't point the way Peston suggests. These banks have come out of the situation quite well and it may be a bit unfair to criticise them for it.

p.s. Analysis of the economics word cloud now available.


Tim Joslin said…
"A large proportion of the payments received by Goldman... are CDS-related - indicating that their insurance policy paid off. Even if it hadn't, they would not have 'lost' the $13 billion, but only the much smaller amount paid as premium on the CDS."

Yes, though they would have lost on the insured debt, so would be £13bn down - if AIG hadn't failed or they'd insured with someone else they wouldn't have lost this. But do we know whether Goldman had bought CDS to insure debt it held, or was it "naked" - i.e. a punt? Or how much of each? Maybe Goldman's held Lehman's CDS which would make things really interesting! Do we know?
Leigh Caldwell said…
Good points - "lost" only has meaning relative to some counterfactual, which could be the counterfactual of AIG not paying up, or of Goldman not insuring at all, or of the underlying debt not defaulting, or of Goldman having no position in the underlying debt. And as you mention, we don't know whether Goldman was insuring debt that it really held or just speculating.

"Speculating" is a loaded word in this context of course - a more neutral locution would be "investing" or "arbitraging". In simple terms, Goldman bought an asset (a CDS) at a known price (probably a few hundred million) with a low probability of paying off at $13 billion. They evaluated the probability of the payoff and decided it was worth buying.

Did Goldman take into account the counterparty risk - the chance AIG would fail to pay up? Did they adjust that risk by their estimate of the chance that the government would step in? I wouldn't be at all surprised if they had.
Tim Joslin said…
Thanks. It's inconceivable Goldman's didn't assess the counterparty risk, and, as you say, they made the right call!

Goldman's are unlikely to let us know their precise reasoning, but we can speculate that they employed a numerical model to assess the risk on the CDS deal. And almost certainly their model would include probabilities of AIG failing and of not being "bailed out" - giving a very low risk on the CDS purchase. If it was true insurance, the risk of BOTH debtor default and insurance default would have been very low indeed. The creation of these "long tail" risks on complex transactions to other institutions is an important reason why modern banks can't be allowed to fail. It's not the 19th century, it's now *essential* that regulation is effective and that the authorities have effective fallback plans - such as provision of unlimited liquidity and mechanisms for ensuring banks can recapitalise - to prevent knock-on failures when something goes wrong. This financial crisis is a colossal screw-up and regulators and governments are most to blame, yet not being blamed!
Alexander said…
Even if it hadn't, they would not have "lost" the $13 billion, but only the much smaller amount paid as premium on the CDS.

In fact, if AIG had failed, Goldman's would have lost up to 13 billion (assuming all debts, insured by AIG CDSs, defaulted across the board) plus the aggregated premiums of the credit default swaps.
I am convinced that Goldman's must have factored into account the probability of a government bailout in the event of AIG failure. If "risk" is today's buzz word, it seems that Goldman's took a big one by assuming government intervention.
Mark said…
One thing to think about is the P&L on the CDS trades: as Lehmans heads towards going bust, the credit spread on it widens and so the CDS's written on it become worth more e.g. say they go from $100m to $700m. For the buyer, each CDS is an asset and so would be counted as such in the accounts and marked with a P&L of $600m. However, if Lehmans goes bust and AIG defaults this asset is now worthless and so Goldman have to write down its value to zero. Hence they will have lost not only their premiums but also the $600m in asset value.

Popular posts from this blog

Discussion 2 of 3: No spooky action at a distance - a theory of reward

The economics zeitgeist, 5 June 2011

The Cognitive Microfoundations Project: a behavioural economics world tour