Tuesday, December 30, 2008

Wash Post AIG Story Missing Something

Reading Day 3 of the Washington Post series on AIG, I sense they are leaving out crucial information. For example, the story starts:

The model showed that these swaps could be a moneymaker for the decade-old firm and its parent, insurance giant AIG, with a 99.85 percent chance of never having to pay out.

Saying "99.85% chance of never having to pay out" is stupid in two ways. One, the spurious precision, reflecting an inappreciation of standard errors (why not say 99.8%, at the very least?). At anyplace I worked such spurious precision would precipitate instant sarcasm, so if this statement stood, there were two guilty parties: the one who said it, and the one who didn't instantly mock the statement. As this is a story being retold, I'm unsure about the credibility of this little scene, because one of the guilty parties seems to be retelling his idiocy. Secondly, there is no time dimension here. I'll assume that is over 10 years, that's a AA- bond, usually very safe but spreads for AA- bonds are not sufficient to cover one's own internal funds transfer pricing (unless they stupidly cost out their funding at their top corporate rate, which may be AAA). So, this never made sense, even without the spurious ratings and not predicting the blow-up in spreads.

There's a seeming conflation between derivatives and swaps with all sorts of CDOs. A journalist would never write a story about how black people should all be put in jail because they commit a disproportionate amount of crime, because it is obviously a ludicrous propositon, but these same journalists will stereotype 'derivatives' as if an interest rate swap is the same as the swap on the Mezzanine tranche for subprime CDOs.

And I love the little arrogance bit: "He told us that in no uncertain terms, that he was -- that all of his people up there were -- smarter than anybody we had at AIG". Every story about financial types blowing up needs some stories about how they were arrogant, hubris leads to disaster, etc. True enough. Indeed, most jerks think they know more than they do, that's one big reason they are jerks. However, in real time, rich and arrogant people will have an infinite supply of toadies looking to get a piece of their business, and any arrogant person running a large business will have a ton of glad handlers, and only time will punish them--they are arrogant because they are ignorant, so they won't be impressed by you pointing out the whole 'Icarus' parable, or that they aren't quite that smart. As we used to say in banking: it's all about spread, volume, and risk--you can't have all three. But, ignorance more frequently begets confidence than does knowledge, and so those who are incompetent will have inflated self-assessments.

In some sense, this crisis is hurting the right people, because those with no money down, who now can't meet there payments on their home, are fine: in the US these loans are no-recourse, so the borrower walks away with nothing but a ding on his credit report, which is probably good for these people. No loss in wealth, just a house they never owned they no longer live in. Those who lost the most money in this crisis, are the right people.

2 comments:

Anonymous said...

it's not the 2 decimal precision, it's the fact that they don't understand what confidence intervals are and what they measure.
but i wanna ask you something else cause i started getting worried recently. what are the chances the system can actually self-destruct? i adjusted it to 20%-50%. bernanke is either a madman or winging it to the tune of trillions and that doesn't bode well.

wcw said...

This is not an interesting question. The chances that an unregulated financial system run by us human beings 'self-destructs' on its own appears to be 1, since even regulated systems appear to do so with some regularity. Another uninteresting question is the chances of a regulated system in the presence of external supporting institutions (let's call these 'government') self-destructing. That appears to be 0 wherever such institutions continue to exist.

The interesting question is the chances of the expected panic being so large it takes down the surrounding, imperfect institutions meant to rescue and restore it (call this 'revolution' or 'anarchy'). That answer, I am afraid, is beyond my powers to model. However, I am pretty sure that in the current world environment over a reasonable time period, it is not 20-50%.