SecProbe.io

CORRESP
1
filename1.htm

CORRESP

May 23, 2008

Mr. Jim B. Rosenberg

Senior Assistant Chief Accountant

United States

Securities and Exchange Commission

Division of Corporation Finance

Mail Stop 6010

Washington, D. C. 20549

    Re:

    American International Group, Inc.

Form 10-K for the Year Ended December 31, 2007

File No. 1-8787

Dear Mr. Rosenberg:

We are in receipt of your letter dated April 29, 2008 and thank you for your comments concerning
American International Group, Inc.’s (AIG) filing on
Form 10-K for the Year Ended December 31, 2007 (2007
Form 10-K) . We are pleased to respond to your
questions and to have the opportunity to work with the Staff of the Commission to further enhance
the overall presentation and disclosure in our filings, and especially appreciate the time that Mr.
Wyman spent discussing the comments with us on May 5 and May 7.

As more fully described below, we have modified certain of our disclosures effective with our Form
10-Q for the quarter ended March 31, 2008 (First Quarter 2008
Form 10-Q), which was filed on May 8. Our responses and those disclosures are provided below.
If further clarification or information is required, please let us know at your earliest
convenience.

The adequacy and accuracy of the disclosures in AIG’s filings is the responsibility of AIG. AIG
acknowledges that Staff comments or changes in response to Staff
comments on the disclosures in AIG’s filings on Form 10-K and Forms 10-Q do not foreclose the Commission from taking any action
with respect to the filings. AIG also acknowledges that Staff comments may not be asserted as a
defense in any proceeding initiated by the Commission or any person under the Federal securities
laws of the United States.

We have repeated your questions below to facilitate your review.

Capital Resources and Liquidity

Contractual Obligations, page 95

    1.

    Please revise your table of contractual obligations to include estimated payment
obligations associated with the underlying guarantees arising from your credit default swap
activities.

    AIG Response:

The majority of AIG’s credit default swaps require AIG to provide credit protection on a
designated portfolio of loans or debt securities. AIG provides such credit protection on a
“second loss” basis, under which AIG’s payment obligations arise principally after credit

losses in the designated portfolio exceed a specified threshold amount or level of “first
losses.”

AIG
had made no payments under these contracts as of the date of its 2007
Form 10-K and First Quarter 2008 Form 10-Q. AIG did not reflect estimated payment obligations associated with
its credit default swap activities in the table of contractual obligations at December 31, 2007
because (a) no defaults had occurred on these contracts to trigger an obligation to make
payments and (b) due to the high degree of uncertainty regarding the amount and long-term timing
of any potential future cash flows under these contracts, AIG is currently unable to make
reasonable estimates for any potential cash settlements.

In
the First Quarter 2008 Form 10-Q, AIG included the following statement in a footnote to the
table: (page 86)

The majority of AIG’s credit default swaps require AIG to provide credit
protection on a designated portfolio of loans or debt securities. AIG provides
such credit protection on a “second loss” basis, under which AIG’s payment
obligations arise only after credit losses in the designated portfolio exceed a
specified threshold amount or level of “first losses.” Through May 7, 2008 AIG has
made no payments under these contracts and because of the high degree of
uncertainty regarding the amount and long-term timing of any potential future cash
flows under these contracts, AIG is unable to make reasonable estimates for any
cash settlements at this time.

These disclosures will be updated or revised as the facts and circumstances change.

Credit Derivatives, page 121

    2.

    In 2007, you recognized an $11.5 billion unrealized market valuation loss related
primarily to the multi-sector CDO component of your super senior credit default swap
portfolio. You state that a modified version of the BET model was used to value the super
senior credit default swaps, and that this model utilizes diversity scores, weighted
average lives, recovery rates and discount rates, and that the most significant assumption
is the pricing of the securities within the CDO collateral pools. You provide a sensitivity
analysis with respect to the pricing of the securities within the CDO collateral pools, but
not with respect to the other significant assumptions. Please expand your disclosure to
provide sensitivity analysis that includes quantification of the impact on fair value of
reasonably likely changes for each of the other key identified assumptions. Also, please
revise to provide a discussion of the strengths and weaknesses of your modified BET model.

    AIG Response:

    In the First Quarter 2008 Form 10-Q, AIG included additional sensitivity analysis and added
further discussion on the BET model as requested by the Staff as follows (page 46):

    The valuation of the super senior credit derivatives has become increasingly challenging
given the limitation on the availability of market observable information due to the lack of
trading and price transparency in the structured finance market, particularly during and

2

    since the fourth quarter of 2007. These market conditions have increased the reliance on
management estimates and judgments in arriving at an estimate of fair value for financial
reporting purposes. Further, disparities in the valuation methodologies employed by market
participants and the varying judgments reached by such participants when assessing volatile
markets has increased the likelihood that the various parties to these instruments may
arrive at significantly different estimates as to their fair values.

    AIGFP’s valuation methodologies for the super senior credit default swap portfolio have
evolved in response to the deteriorating market conditions and the lack of sufficient market
observable information. AIG has sought to calibrate the model to market information and to
review the assumptions of the model on a regular basis.

AIGFP employs a modified version of the BET model to value its credit default swap portfolio
written on the super senior securities issued by CDOs, including the embedded 2a-7 Puts. The
BET model uses default probabilities derived from credit spreads implied from market prices
for the individual securities included in the underlying collateral
pools securing the CDOs.

AIGFP obtained prices on these securities from the CDO collateral managers. The BET model
also uses diversity scores, weighted average lives, recovery rates and discount rates. The
determination of some of these inputs requires the use of judgment and estimates,
particularly in the absence of market observable data. AIGFP also employs a Monte Carlo
simulation to assist in quantifying the effect on valuation of the CDO of the unique
features of the CDOs’ structure such as triggers that divert cash flows to the most senior
level of the capital structure.

    AIG selected the BET model for the following reasons:

    •

    it is known and utilized by other institutions;

    •

    it has been studied extensively, documented and enhanced over many years;

    •

    it is transparent and relatively simple to apply;

    •

    the parameters required to run the BET model are generally observable; and

    •

    it can easily be modified to use probabilities of default and expected losses
derived from the underlying collateral securities market prices instead of using
rating-based historical probabilities of default.

    AIG’s implementation of the BET model uses a Monte Carlo simulation of the cash flows of
each underlying CDO for various scenarios of defaults by the underlying collateral
securities. The Monte Carlo simulation allows the model to take into account the cash flow
waterfall and to capture the benefits due to cash flow diversion
within each CDO. The BET model has certain limitations. A well known limitation of the BET model is that it
can understate the expected losses for super senior tranches when default correlations are
high. The model uses correlations implied from diversity scores which do not capture the
tendency for correlations to increase as defaults increase. Recognizing this concern, AIG
tested the sensitivity of the valuations to the diversity scores. The results of the testing
demonstrated that the valuations are not very sensitive to the diversity scores because the
expected losses generated from the prices of the collateral pool securities are currently
high, breaching the attachment point in most transactions. Once the attachment point is
breached by a sufficient amount, the diversity scores, and their implied correlations, are
no longer a significant driver of the valuation of a super senior tranche.

    The credit default swaps written by AIGFP generally cover the failure of payment on the
super senior CDO security. AIGFP does not own the securities in the CDO collateral

3

    pool. The credit spreads implied from the market prices of the securities in the CDO
collateral pool incorporate the risk of default (credit risk), the market’s price for
liquidity risk and, in distressed markets, the risk aversion costs. Spreads on credit
derivatives tend to be narrower than the credit spreads implied from the market prices of
the securities in the CDO collateral pool because, unlike investing in a bond, there is no
need to fund the position (except when an actual credit event occurs). In times of
illiquidity, the difference between spreads on cash securities and derivative instruments
(the negative basis) may be even wider for high quality assets. AIGFP was unable to reliably
verify this negative basis with market observable inputs due to the accelerating severe
dislocation, illiquidity and lack of trading in the asset-backed securities market during
the fourth quarter of 2007 and the first quarter of 2008. The valuations produced by the BET
model therefore represent the valuations of the underlying super senior CDO cash securities
based on AIG’s assumptions about those securities, albeit with no recognition of any
potential favorable effect of the basis differential on that valuation. AIGFP also
considered the valuation of the super senior CDO securities provided by third parties,
including counterparties to these transactions, and made adjustments as necessary.

    The most significant assumption used in developing the estimate is the pricing of the
securities within the CDO collateral pools. These prices are used to derive default
probabilities that are used in the BET model. If the actual pricing of the securities within
the collateral pools differs from the pricing used in estimating the fair value of the super
senior credit default swap portfolio, there is potential for material variation in the fair
value estimate. A decrease by five points (for example, from 87 cents per dollar to 82 cents
per dollar) in the aggregate price of the securities would cause an additional unrealized
market valuation loss of approximately $3.9 billion, while an increase in the aggregate
price of the securities by five points (for example, from 90 cents per dollar to 95 cents
per dollar) would reduce the unrealized market valuation loss by approximately $3.7 billion.
The effect on the unrealized market valuation loss is not directly proportional to the
change in the aggregate price of the securities.

    The following table presents other key inputs used in the valuation of the credit derivative
portfolio written on the super senior securities issued by multi-sector CDOs, and the
potential increase (decrease) to the unrealized market valuation loss at March 31, 2008
calculated using the BET model for changes in these key inputs. The adjustments to the key
inputs incorporated in the sensitivity analysis below are based on management’s judgment of
reasonably possible ranges for these inputs:

    Increase

    (Decrease) To

    Unrealized Market

    (in millions)

    Valuation Loss

    Weighted average lives

    Effect of an
increase of 1 year

    $
    375

    Effect of a
decrease of 1 year

    (620
    )

    Recovery rates

    Effect of an
increase of 10%

    (103
    )

    Effect of a
decrease of 10%

    194

4

    Increase

    (Decrease) To

    Unrealized Market

    (in millions)

    Valuation Loss

    Diversity scores

    Effect of an increase of 5

    (40
    )

    Effect of a decrease of 5

    15

    Discount curve

    Effect of an increase of 100 basis points

    70

    These results are calculated by stressing a particular assumption independently of changes
in any other assumption. No assurance can be given that the actual levels of the key inputs
will not exceed, perhaps significantly, the ranges assumed by AIG for purposes of the above
analysis. No assumption should be made that results calculated from the use of other changes
in these key inputs can be interpolated or extrapolated from the results set forth above.

    In the case of credit default swaps written on investment grade corporate debt and
CLOs, AIGFP estimates the value of its obligations by reference to the relevant market
indices or third-party quotes on the underlying super senior tranches where available.

    The following table represents the relevant market credit indices and index CDS maturity
used in the valuation of the credit default swap portfolio written on investment-grade
corporate debt and the increase (decrease) to the unrealized market valuation loss at March
31, 2008 corresponding to changes in these market credit indices and maturity:

    Increase (Decrease) To

    (in millions)

    Unrealized Market Valuation Loss

    CDS maturity (in years)

    5

    7

    10

    CDX Index

    Effect of an increase of 10 basis points

    $
    26

    $
    51

    $
    5

    Effect of a decrease of 10 basis points

    (26
    )

    (51
    )

    (5
    )

    iTraxx Index

    Effect of an increase of 10 basis points

    11

    37

    13

    Effect of a decrease of 10 basis points

    (11
    )

    (37
    )

    (13
    )

    These results are calculated by stressing a particular assumption independently of changes
in any other assumption. No assurance can be given that the actual levels of the indices and
maturity will not exceed, perhaps significantly, the ranges assumed by AIG for purposes of
the above analysis. No assumption should be made that results calculated from the use of
other changes in these indices and maturity can be interpolated or extrapolated from the
results set forth above.

AIG intends to include similar disclosure in future filings.

    3.

    You disclose on page 124 that you believe any realized losses relating to the super
senior credit default swap obligations will not be material to your consolidated financial

5

    condition, but could be material to your results of operations for any individual reporting
period. Please revise your disclosures to clarify the context of this statement in relation
to the $11.5 billion loss recorded during the quarter ended December 31, 2007. Clarify
whether your disclosure indicates that incremental realized losses beyond the previously
recorded $11.5 billion of unrealized losses may be incurred, or should be interpreted to
mean that when realized losses are incurred, all or a portion of the previously recorded
unrealized losses will be reversed.

    AIG Response:

    AIG believes that the losses AIGFP will ultimately realize in meeting its obligations under
the super senior credit default swaps will be significantly less than the unrealized market
valuation losses recognized to date. Thus, AIG’s disclosure was not intended to indicate
that AIGFP may incu