Tuesday, May 05, 2009

Roubini the Risk Manager

Nouriel Roubini called the recent crisis, and for this his status as a prognosticator has elevated. But his call has been perenial, and the driver anything and everything. In today's Wall Street Journal, Roubini outlines how he would address the crisis, and
He outlines regulations he has in mind, including the following:
Consider also recent bank risk-taking. The media has recently reported that Citigroup and Bank of America were buying up some of the AAA-tranches of nonprime mortgage-backed securities. Didn't the government provide insurance on portfolios of $300 billion and $118 billion on the very same stuff for Citi and BofA this past year? These securities are at the heart of the financial crisis and the core of the PPIP. If true, this is egregious behavior -- and it's incredible that there are no restrictions against it.

Typical risk manager, looking in the rear view mirror. An AAA mortgage now is not like a AAA mortgage in 2007. Mortgages have been tanking for 18 months. They experienced historic losses in that time. So, using the same logic that got us in this mess, assumes that these assets, regardless of current quality (mortgages are a heterogeneous lot), all mortgages are now ultra risky.

As the rating agencies are especially wary about mortgages, any specific AAA rated mortgage security is probably the least risky AAA security out there. The incentives for doing this are all skewed in favor of being too pessimistic. Historically, the best ABS investments have been in areas with above average defaults in the recent cycle. I don't think commercial real estate mortgage backed securities (CMBS) had any losses from 1992 to 2000, as the rating agencies overcompensated. Thus, if the rating agencies are specifying some mortgage backed securities as AAA, I bet they are pretty safe.

But, using the reasoning that those assets that 'got us into this mess' are now ALL bad, he suggests micromanaging the banks to exclude such assets going forward. It's all too typical, and exactly the kind of stupid extrapolation that led investors to believe that ALL mortgages were safe simply because they had historically low losses. Details matter. For mortgages, the fact that so many were becoming no-documentation, teaser rate, high LTV loans, was significant, but those are details. For current mortgages, ignoring these details on mortgages is equally an error, even if in the other direction. Pushing banks into other 'regulator approved' safe assets was also a cause of this crisis, and we shouldn't do that again. Regulators, or academics who never worked for a bank, are not forward looking.


Anonymous said...

eff roubini. let's stick to more interesting subjects.
how do you solve the single vote probability? you know, the 'your vote don't matter' thing. i thought it's similar to the birthday problem with 50% chance and combinations of half the voters. instead i found this: p=1/sqrt((N/2)*pi)? when did pi get in?

Anonymous said...

"when did pi get in?"

probably s.o. introduced a gaussian mean 0 var 1 assumption earlier in.