Page 150 - untitled
P. 150
THE MORTGAGE MACHINE
the national drop, staying down over this short but multiple-year period, is more
stressful than the statistics call for.” Even as housing prices rose to unprecedented lev-
els, Moody’s never adjusted the scenarios to put greater weight on the possibility of a
decline. According to Siegel, in , “Moody’s position was that there was not a . . .
national housing bubble.”
When the initial quantitative analysis was complete, the lead analyst on the deal
convened a rating committee of other analysts and managers to assess it and deter-
mine the overall ratings for the securities. Siegel told the FCIC that qualitative
analysis was also integral: “One common misperception is that Moody’s credit rat-
ings are derived solely from the application of a mathematical process or model. This
is not the case. . . . The credit rating process involves much more—most importantly,
the exercise of independent judgment by members of the rating committee. Ulti-
mately, ratings are subjective opinions that reflect the majority view of the commit-
tee’s members.” As Roger Stein, a Moody’s managing director, noted, “Overall, the
model has to contemplate events for which there is no data.”
After rating subprime deals with the model for years, in Moody’s intro-
duced a parallel model for rating subprime mortgage–backed securities. Like M
Prime, the subprime model ran the mortgages through , scenarios. Moody’s
officials told the FCIC they recognized that stress scenarios were not sufficiently se-
vere, so they applied additional weight to the most stressful scenario, which reduced
the portion of each deal rated triple-A. Stein, who helped develop the subprime
model, said the output was manually “calibrated” to be more conservative to ensure
predicted losses were consistent with analysts’ “expert views.” Stein also noted
Moody’s concern about a suitably negative stress scenario; for example, as one step,
analysts took the “single worst case” from the M Subprime model simulations and
multiplied it by a factor in order to add deterioration.
Moody’s did not, however, sufficiently account for the deteriorating quality of the
loans being securitized. Fons described this problem to the FCIC: “I sat on this high-
level Structured Credit committee, which you’d think would be dealing with such is-
sues [of declining mortgage-underwriting standards], and never once was it raised to
this group or put on our agenda that the decline in quality that was going into pools,
the impact possibly on ratings, other things. . . . We talked about everything but, you
know, the elephant sitting on the table.”
To rate CMLTI -NC, our sample deal, Moody’s first used its model to simu-
late losses in the mortgage pool. Those estimates, in turn, determined how big the jun-
ior tranches of the deal would have to be in order to protect the senior tranches from
losses. In analyzing the deal, the lead analyst noted it was similar to another Citigroup
deal of New Century loans that Moody’s had rated earlier and recommended the same
amount. Then the deal was tweaked to account for certain riskier types of loans, in-
cluding interest-only mortgages. For its efforts, Moody’s was paid an estimated
,. (S&P also rated this deal and received ,.)
As we will describe later, three tranches of this deal would be downgraded less
than a year after issuance—part of Moody’s mass downgrade on July , , when
housing prices had declined by only . In October , the M–M tranches
