AIG: New Math for Understanding What Went Wrong

It is becoming increasingly apparent from the scale of the revised AIG (AIG) rescue plan announced yesterday, and the fact that there has still been no real “pricing” of the really toxic stuff by the TARP, that the financial engineers that created this monumental disaster were either unintentionally reckless beyond belief or, if the obfuscation was intentional, this must surely be counted as the largest theft ever perpetrated and taxpayers should be demanding ongoing investigations leading to criminal prosecutions.

If we just focus on the more benign interpretation (i.e. that it was not malicious intention) one is forced to the conclusion that the mathematics that underlies all of the risk modeling that went into these structured products is absolutely flawed. Not just slightly wrong but fundamentally mis-conceived.  

The saddest part of the whole mess is the hubris and arrogance of the people that signed off on the assessments made by the “quants” of the likelihood of defaults and the probability distributions of financial accidents.  Extreme events in time series data are of a completely different complexion to the tails of a normal distribution.

If risk managers really insist on finding a bit of math that can account for the likelihood of critical events taking place – such as collapses in real estate prices and implosions of liquidity – a better place to look would be the predator-prey modeling which has been conducted mainly in relation to bio and eco-systems.  

Predators eventually run out of sufficient prey and their populations decline substantially until there is a chance for the supply of prey to be replenished. At such time the predators can get back to business. This modeling gives rise to oscillations in the ratio of predators to prey and right now we are in a bear phase for the predators.  One slight modification to the model needs to be made for the government rescue plans (etc.) which will perhaps keep those marginal survivors from the last predation on a life support system long enough so that they can limp off and become ensnared in some new financially engineered scam.

Disclosure: no postions

Comments

[1 of 4:]

The historical data used to model did not reflect the positive feedback loop that increased availability of mortgage funds to the least sophisticated borrowers/buyers would have on home prices.

2008 Nov 11 11:05 AM

[Pat Shuff:]

Bravo! Insightful and hilARious.

2008 Nov 11 03:57 PM

[Tom Armistead:]

I never thought of it that way. You could be right.

2008 Nov 11 06:55 PM