there's no model in housing credit going really above a multiplication, and there's no historical data whatsoever to calibrate anything since there never was a liquid market for those assets now or in the relevant past (and nobody knows the underlying loan portfolio anyway). So when MSM says "mark to model" valuation it's really just "mark to what I want".
Actually, the models are much more complicated than a multiplication, but I remain to be convinced that they actually fit the observational data so much better than a multiplication, indeed.
The calibration data is indirect when the product is new and was not traded in the past: you can calibrate to historical default statistics for instance (picking the time window that suits you best, like the phony past decade in housing).
Also, remember that you are only required, by law, to calibrate the model "to the market": that is, you calibrate the empirical constants in your model to replicate the risk perception that is "displayed" in the market at time T (for instance, the credit spread are supposed to embody the market's expectations of future default, and if they were low, it means you really could sell any crap to the crowds). This does represents the fair market value at which you could sell the asset at time T.
What regulation does not demand, is that you check the depth of the market (the volumes traded daily vs. what's in the book of a typical market player, and your books in particular). If the volumes are low in comparison, it means the prices could not hold very long if you had to liquidate. Forex to US$ is probably the deepest. Treasuries are very deep (except from the p.o.v. of the Chinese). Equity is pretty deep too. "classic" equity or treasury derivatives are almost as liquid as equity.
The CDO, CDS markets were never very deep to start with, and now they are down to zero. The spreads have shot up. In most models, it means the value is close to zero. In that, the models are not so wrong, actually !
It's just that bankers become reluctant to face the results the moment is displeases them. So it's also a stress-testing problem: regulators have been lax in the model-testing standards for credit derivatives, and no thorough risk assessment was ever made on these portfolios (risk assessment is mandatory and now pretty comprehensive on equity derivatives). Pierre
For calibration there are three big problems on the current brand of housing derivatives:
First in history means zero relevant observations, there is no magic model here.
What is really out the window, is the probability models that were used to predict the risk of such a market turnover (the 25-sigma probability, which was actually a +30%/year probability as common sense could have told 2 years ago). These models were based on the coincidental fact that optimistic market spreads ex ante from about 2000, did in fact match the exceptionnally low default rates observed ex post in 2005. When actually these low default rates were only due to the refi credit bubble.
Most senior bankers involved knew, but they chose to pretend not, so they could cash in and pack in time. And guess what ? most won't be liable to prosecution: regulations don't require bankers to hedge systemic risk (it's written in the law, I have it on my desk: "Règlement N°95-02 du 21/7/1995 relatif à la surveillance prudentielle des risques de marché", thick booklet). The systemic risk is explicitly of the responsibility of the market authorities and central banks (talk about bail out).
Because these authorities are highly politicized, whatever the claims of the contrary, it seems governments have chosen ratings agencies as a better scapegoat... Pierre
The issue isn't that the models are wrong, it's that the process should never have happened in the first place.
No one should be trying to model crap like this. It's madness even trying to get a sensible market value out of 'assets' that by any sane standard are either worthless, or close to worthless.
But as you say, it has happened for political reasons. This is one big political scam, and should be acknowledged as such.
It's been as much about social engineering, giving a sputtering economy the illusion of bouyancy so that approval ratings remain high among the peasants, and the have-mores can maintain their culture of entitlement, as about the specifics of risk modelling.
As the ARMs reset and the US starts to see an astonishing new wave of homeless bankrupts, it needs to be remembered that this was as much a political feint as a financial pyramid.
