Page 149 - untitled

 FINANCIAL CRISIS INQUIRY COMMISSION REPORT

financial institutions, relied on the ratings. Still, some investors who did their home-
work were skeptical of these products despite their ratings. Arnold Cattani, chairman
of Mission Bank in Bakersfield, California, described deciding to sell the bank’s hold-
ings of mortgage-backed securities and CDOs:

At one meeting, when things started getting difficult, maybe in , I
asked the CFO what the mechanical steps were in . . . mortgage-backed
securities, if a borrower in Des Moines, Iowa, defaulted. I know what it
is if a borrower in Bakersfield defaults, and somebody has that mort-
gage. But as a package security, what happens? And he couldn’t answer
the question. And I told him to sell them, sell all of them, then, because
we didn’t understand it, and I don’t know that we had the capability to
understand the financial complexities; didn’t want any part of it. 

Notably, rating agencies were not liable for misstatements in securities registra-
tions because courts ruled that their ratings were opinions, protected by the First
Amendment. Moody’s standard disclaimer reads “The ratings . . . are, and must be
construed solely as, statements of opinion and not statements of fact or recommen-
dations to purchase, sell, or hold any securities.” Gary Witt, a former team managing
director at Moody’s, told the FCIC, “People expect too much from ratings . . . invest-
ment decisions should always be based on much more than just a rating.” 

“Everything but the elephant sitting on the table”
The ratings were intended to provide a means of comparing risks across asset classes
and time. In other words, the risk of a triple-A rated mortgage security was supposed
to be similar to the risk of a triple-A rated corporate bond.
Since the mid-s, Moody’s has rated tranches of mortgage-backed securities
using three models. The first, developed in , rated residential mortgage–backed
securities. In , Moody’s created a new model, M Prime, to rate prime, jumbo,
and Alt-A deals. Only in the fall of , when the housing market had already
peaked, did it develop its model for rating subprime deals, called M Subprime. 
The models incorporated firm- and security-specific factors, market factors, regu-
latory and legal factors, and macroeconomic trends. The M Prime model let
Moody’s automate more of the process. Although Moody’s did not sample or review
individual loans, the company used loan-level information from the issuer. Relying
on loan-to-value ratios, borrower credit scores, originator quality, and loan terms
and other information, the model simulated the performance of each loan in ,
scenarios, including variations in interest rates and state-level unemployment as well
as home price changes. On average, across the scenarios, home prices trended up-
ward at approximately  per year.  The model put little weight on the possibility
prices would fall sharply nationwide. Jay Siegel, a former Moody’s team managing di-
rector involved in developing the model, told the FCIC, “There may have been [state-
level] components of this real estate drop that the statistics would have covered, but