Behind the credit crunch: Statistical models helped drive "criminal enterprise"

"Creating complex structured assets that nobody can understand is a bad idea and bad public policy. The only beneficiaries are the dealers who sell this crap. It is borderline fraud, definitely not 'fair dealing' in the traditional American sense. Likewise, the use of statistical models to value this illiquid, opaque crap is just another aspect of this ongoing criminal enterprise. The only thing worse that the acts/omission of the dealers and the rating agencies is the idiocy and lack of care by the buy-side investors who bought these 'assets' and the regulators who enabled this fiasco in the first place...Just because you can do something technologically doesn't mean you should do it."

-- Christopher Whalen, Senior Vice President & Managing Director, Institutional Risk Analytics

Whalen was speaking about the creation, rating and marketing of structured securities, derived from mortgage notes, which are at the center of the credit crisis in the wake of the housing market bust. The context was how modeling tools were used to create new and risky  structured securities and how debt rating agencies used statistical models that overvalued them. Many institutional investors used borrowed money to buy the securities. When the value of those assets came into question, ratings were dropped. Loans were called in, and to cover those loans, some investment firms were forced to sell the securities into a market with no buyers. Some of those institutions have gone under.

Whalen's comments ended up on the cutting room floor as I put together my recent column, Greed, analytics and the mortgage lending crisis. For another perspective on this saga, read Anatomy of a Ratings Downgrade in Business Week.