Page 299 - untitled

 FINANCIAL CRISIS INQUIRY COMMISSION REPORT

The decision to develop a valuation model was not unanimous. In mid-Septem-
ber, Cassano and Forster met with Habayeb and others to discuss marking the posi-
tions down and actually recording valuation losses in AIG’s financial statements.
Cassano still thought the valuation process unnecessary because the possibility of de-
faults was “remote.”  He sent Forster and others emails describing requests from
Habayeb as “more love notes . . . [asking us to go through] the same drill of drafting
answers.”  Nevertheless, by October, and in consultation with PwC, AIG started to
evaluate the pricing model for subprime instruments developed and used by
Moody’s. Cassano considered the Moody’s model only a “gut check” until it was fully
validated internally.  AIG coupled this model with generic CDO tranche data sold
by JP Morgan that were considered to be relatively representative of the market. Of
course, by this time—and for several preceding months—there was no active market
for many of these tranches. Everyone understood that this was not a perfect solution,
but AIG and its auditors thought it could serve as an interim step. The makeshift
model was up and running in the third quarter.

“Confident in our marks”
On November , when AIG reported its third-quarter earnings, it disclosed that it
was taking a  million charge “related to its super senior credit default swap port-
folio” and “a further unrealized market valuation loss through October  of ap-
proximately  million before tax [on that] portfolio.” On a conference call, CEO
Sullivan assured investors that the insurance company had “active and strong risk
management.” He said, “AIG continues to believe that it is highly unlikely that AIGFP
will be required to make payments with respect to these derivatives.” Cassano added
that AIG had “more than enough resources to meet any of the collateral calls that
might come in.”  While the company remained adamant that there would be no re-
alized economic losses from the credit default swaps, it used the newly adopted—and
adapted—Moody’s model to estimate the  million charge. In fact, PwC had ques-
tioned the relevance of the model: it hadn’t been validated in advance of the earnings
release, it didn’t take into account important structural information about the swap
contracts, and there were questions about the quality of the data.  AIG didn’t men-
tion those caveats on the call.
Two weeks later, on November , Goldman demanded an additional  billion in
cash. AIG protested, but paid . billion, bringing the total posted to  billion. 
Four days later, Cassano circulated a memo from Forster listing the pertinent marks
for the securities from Goldman Sachs, Merrill Lynch, Calyon, Bank of Montreal,
and SocGen.  The marks varied widely, from as little as  of the bonds’ original
value to virtually full value. Goldman’s estimated values were much lower than those
of other dealers. For example, Goldman valued one CDO, the Dunhill CDO, at 
of par, whereas Merrill valued it at  of par; the Orient Point CDO was valued at
 of par by Goldman but at  of par by Merrill. Forster suggested that the marks
validated AIG’s long-standing contention that “there is no one dealer with more
knowledge than the others or with a better deal flow of trades and all admit to