Black swans and dark matter
There’s been a lot of talk about the idea that the GFC (the in-group shorthand for ‘global financial crisis’) is an example of a ‘black swan’, that is, an event that would be treated as impossible on the basis of induction from past experience, and hence that could not be encompassed by formal models of the kind used by risk managers. All this talk has of course been great for Nicholas Taleb who has a book with this title. It’s good in a lot of ways, but I found it ultimately insufferable in the continuous repetition of the message that only Taleb was smart enough to see all this. ( To be fair, Taleb predicted a global financial crisis, and didn’t simply claim it in retrospect as an unpredictable Black Swan).
I spend a lot of my time working on how to think about unforeseen contingencies and I’m not at all convinced that the GFC should be described in this way. Of course, the models used by the risk managers in investment banks didn’t include this as a possibility; if they had, the implication would have been that all sorts of much-desired deals should not go ahead. But as I pointed out a while ago, very simple models based on well-established principles predicted that the bubble economy would end badly.
The crisis then, involved something more like dark matter, the ‘missing’ matter in the universe that must exist if it is to work as it does, but can’t at presented be detected. Given that risk can’t easily be made to disappear*, it was obvious that the risk associated with lending of all kinds (most obviously, mortages offered to people with no capacity to repay) was being borne by someone, and probably someone who was unaware of it.
The big problem for the Cassandras (and we were certainly both correct and disregarded) was that it was easy to see that the bubble could not continue and much harder to foresee how it would end – it’s one thing to say that dark matter must exist and another to work out what it is really like. Like Brad and Brad, I expected that the problems would emerge first in the form of a run on the US dollar, given that holders of US dollar assets were receiving very little compensation for the obvious risk of large capital losses. In fact, the US dollar actually rose in the early stages of the meltdown, though it has been falling more recently.
One ‘black swan’ explanation of the mortgage crisis was that the mortgage derivatives created by Wall Street couldn’t fail except in the event of a simultaneous downturn in all major housing markets in the US, something that had never been observed, and therefore could not be included in the models. But of course the reason such a thing had never occurred was that local housing markets had been separate from each other, with their own sets of banks, S&Ls and other financial institutions. The very banks that were doing the modelling were creating the conditions under which a national bubble and bust could take place. This was both foreseeable and foreseen.
*Unlike matter, risk can be created with ease. Reducing it is more difficult, though pooling and sharing of uncorrelated risks works in many cases.